facilitating quality use of medicines (qum) between hospital and community...

Third Community Pharmacy Agreement Research & Development

Grants Program

Facilitating Quality use of Medicines (QUM) between

hospital and community RFT 2003-03

FINAL REPORT TO THE PHARMACY GUILD OF AUSTRALIA

June 2006

This research was funded by the Australian Government Department of Health and Ageing through the Third Community Pharmacy Agreement Research and

Development Program

Professor Gregory Peterson

Miss Anna Tompson

Dr Shane Jackson

Dr Omar Hasan

Mr Peter Gee

Miss Rose McShane

Mr Craig Cooper

Ms Kimbra Fitzmaurice

Mrs Bronwen Roberts

Mrs Donielle Luttrell

UMORE (Unit for Medication Outcomes Research and Education)

School of Pharmacy

University of Tasmania

Hobart TAS

Mr Jeff Hughes

School of Pharmacy

Curtin University of Technology

Perth WA

Professor Kenn Raymond

Department of Pharmacy

School of Health and Environment

La Trobe University

Bendigo VIC

Contact person for correspondence Anna Tompson Clinical Research Project Manager UMORE (Unit for Medication Outcomes Research and Education) School of Pharmacy

University of Tasmania Locked Bag 83 HOBART TAS 7001 Phone: 61-3-62261032 Fax: 61-3-62267627

Table of Headings, Figures and Tables

of 767

1 Tables of Headings, Figures and Tables 1.1 Table of Contents 1 Tables of Headings, Figures and Tables ...................................................................3 1.1 Table of Contents...........................................................................................................................3 1.2 Table of Figures ............................................................................................................................7 1.3 Table of Tables ..............................................................................................................................8 2 Med eSupport Abstract ..............................................................................................12 3 Med eSupport Executive Summary ..........................................................................16 3.1 Introduction .................................................................................................................................16 3.2 Methodology ................................................................................................................................18 3.2.1 Med eSupport Trial .....................................................................................................................18 3.2.2 The warfarin focused aspect of the Med eSupport trial; a specific 'high-risk' example. ...........24 3.2.3 ICT systems evaluation ...............................................................................................................26 3.3 Results..........................................................................................................................................27 3.3.1 Key Findings ...............................................................................................................................29 3.3.2 Economics and Financial Analysis of Med eSupport.................................................................33 3.4 Med eSupport Recommendations................................................................................................35 4 Project Team Members ..............................................................................................37 5 Acknowledgements....................................................................................................38 6 How to use this document.........................................................................................39 7 Acronyms and Definitions.........................................................................................40 8 Introduction.................................................................................................................43 8.1 Adverse drug events ....................................................................................................................43 8.2 The elderly at particular risk of adverse drug events.................................................................44 8.3 Patient behaviours towards their medications influence risk of adverse drug events...............45 8.4 Adverse drug events relating to hospitalisation .........................................................................46 8.4.1 On admission to hospital .............................................................................................................46 8.4.2 During the hospital stay ..............................................................................................................47 8.4.3 At the time of hospital discharge ................................................................................................48 8.4.4 Hospital discharge is a particularly high-risk period for the elderly..........................................50 8.4.5 Peri-discharge adverse drug events can lead to unplanned readmissions ..................................51 8.4.6 Challenges and obstacles relating to transitional care - The current state of affairs..................51 8.5 Some strategies to improve the situation ....................................................................................55 8.5.1 Medication reconciliation at time of discharge ..........................................................................55 8.5.2 Production of a clear and concise discharge summary...............................................................55 8.5.3 Improved communication with community health providers ....................................................56 8.5.4 Discharge medication counselling ..............................................................................................58 8.5.5 Post-discharge follow-up telephone calls ...................................................................................58 8.5.6 Access for patients to medication information via the internet..................................................59 8.6 Local example of need for a medication liaison service ............................................................60 8.7 Study aims and objectives ...........................................................................................................62


of 767

9 Med eSupport trial and participant surveys ............................................................64 9.1 Methodology ................................................................................................................................64 9.1.1 Methodology of the Med eSupport trial .....................................................................................64 9.1.2 Methodology of anonymous participant surveys........................................................................78 9.1.3 Methodology for the post-hoc reclassification process. .............................................................79 9.2 Results..........................................................................................................................................82 9.2.1 Results of the Med eSupport trial ...............................................................................................82 9.2.2 Results of anonymous participant surveys................................................................................135 9.3 Discussion..................................................................................................................................149 9.3.1 Discussion of the Med eSupport trial........................................................................................149 9.3.2 Discussion of anonymous participant surveys..........................................................................149 10 ICT systems evaluation............................................................................................151 10.1 Methodology ..............................................................................................................................151 10.1.1 Planning stage............................................................................................................................151 10.1.2 Implementation..........................................................................................................................152 10.1.3 Procedural impact......................................................................................................................156 10.1.4 Patient telephone interview at thirty days after discharge........................................................160 10.1.5 Comparison of results for the anonymous patient satisfaction survey.....................................161 10.1.6 Community Health Professional anonymous satisfaction survey ............................................161 10.1.7 Data extraction techniques ........................................................................................................161 10.2 Results........................................................................................................................................161 10.2.1 Analysis of website utilisation - including usability and usefulness........................................161 10.2.2 Results of the patient telephone interview at thirty days after discharge.................................164 10.2.3 Comparison of results for the anonymous patient satisfaction surveys ...................................166 10.2.4 The participating Community Health Professional anonymous satisfaction surveys..............168 10.3 Discussion and interpretation...................................................................................................171 10.3.1 Website utilisation.....................................................................................................................171 10.3.2 Subjective data ..........................................................................................................................172 10.3.3 Barriers to implementation........................................................................................................172 10.3.4 Project team insights .................................................................................................................172 10.3.5 Ongoing maintenance................................................................................................................173 10.4 Conclusions ...............................................................................................................................174 10.4.1 Interpretation .............................................................................................................................174 10.4.2 Software vendors.......................................................................................................................174 10.4.3 Drug linking table......................................................................................................................175 11 Economic and financial analysis of the Med eSupport trial................................176 11.1 Introduction ...............................................................................................................................176 11.2 Methodology ..............................................................................................................................176 11.2.1 Quality of Life ...........................................................................................................................177 11.2.2 Readmission costs and rates of readmission for patients within thirty days of discharge.......177 11.2.3 Costs of activities, determined through time trials ...................................................................178 11.2.4 The clinical panel ......................................................................................................................179 11.2.5 Panel methodology and economic analysis ..............................................................................180


of 767

11.3 Results........................................................................................................................................189 11.3.1 Quality of Life ...........................................................................................................................189 11.3.2 Readmission costs/rates within thirty days of discharge..........................................................191 11.3.3 Additional health care costs for trial patients ...........................................................................192 11.3.4 Time trials..................................................................................................................................193 11.3.5 Clinical panel and economic analysis .......................................................................................194 11.4 Discussion..................................................................................................................................213 11.4.1 Time trials..................................................................................................................................213 11.4.2 Clinical Panel ............................................................................................................................213 11.4.3 The economic value of interventions........................................................................................214 12 The warfarin focused aspect of the Med eSupport trial; a specific 'high-

risk' example. ............................................................................................................216 12.1 Background................................................................................................................................216 12.2 Methodology ..............................................................................................................................219 12.2.1 Recruitment procedure ..............................................................................................................219 12.2.2 Post-discharge PDINR monitoring procedures ........................................................................219 12.2.3 Usual care group procedures.....................................................................................................223 12.2.4 Data collection...........................................................................................................................223 12.2.5 Data analysis..............................................................................................................................223 12.2.6 Exclusions..................................................................................................................................224 12.2.7 Cost-effectiveness analysis .......................................................................................................225 12.3 Results........................................................................................................................................228 12.3.1 Recruitment ...............................................................................................................................228 12.3.2 Cost-effectiveness .....................................................................................................................247 12.4 Discussion..................................................................................................................................248 12.5 Recommendations......................................................................................................................255 13 Overall Discussion ...................................................................................................256 13.1 Key findings ...............................................................................................................................256 13.2 Project complications................................................................................................................259 13.2.1 ICT issues ..................................................................................................................................260 13.2.2 Patient Exclusions .....................................................................................................................261 13.2.3 Withdrawals...............................................................................................................................263 13.2.4 Removal of data collected outside of trial protocol specifications. .........................................265 13.2.5 Grouping study patients by services received ..........................................................................266 13.3 Medication chart discrepancies in the hospital........................................................................267 13.3.1 RMO drug chart discrepancies in general.................................................................................267 13.3.2 Early resolution .........................................................................................................................268 13.3.3 Any resolution ...........................................................................................................................269 13.3.4 Admission discrepancy resolution impact on length of stay ....................................................269 13.3.5 New discharge discrepancies ....................................................................................................270 13.3.6 The value of the community pharmacist dispensing history ....................................................271 13.3.7 Economic analysis.....................................................................................................................272 13.4 Post-discharge medication review............................................................................................274


of 767

13.4.1 Uptake of the models for promoting conventional HMRs post-discharge ..............................274 13.4.2 Timeliness..................................................................................................................................275 13.4.3 Economic value .........................................................................................................................278 13.4.4 Uptake of PDMR/HMR recommendations...............................................................................279 13.5 Patient specific variables ..........................................................................................................283 13.5.1 Medication knowledge ..............................................................................................................283 13.5.2 Self-reported medication compliance .......................................................................................284 13.5.3 Self-reported drug-related problems .........................................................................................285 13.5.4 Quality of life ............................................................................................................................286 13.6 Drug Related Problems.............................................................................................................288 13.6.1 Total actual or potential DRPs ..................................................................................................288 13.6.2 Breakdown of the identified actual or potential DRPs .............................................................289 13.7 Annonymous participant satisfaction surveys ..........................................................................292 13.8 Website utilisation .....................................................................................................................293 13.9 Additional health care costs......................................................................................................293 13.10 Readmission rates of patients within 30 days of discharge......................................................294 13.11 Future direction.........................................................................................................................296 13.12 Project limitations .....................................................................................................................298 14 Conclusions and recommendations ......................................................................300 15 References.................................................................................................................303 16 Appendices................................................................................................................318 Trial process documents ................................................................................................................................................318 Appendix I Welcome page to patients .........................................................................................................318 Appendix II Where to find a computer (Tasmanian patients).......................................................................319 Appendix III Intervention patient introduction page (providers received a similar sheet)............................321 Appendix IV Information sheet sample ..........................................................................................................322 Appendix V Consent Form sample................................................................................................................324 Appendix VI RMO discrepancies letter ..........................................................................................................326 Appendix VII Patient data collection sheet ......................................................................................................327 Appendix VIII Cover letter faxed to Community Pharmacists at discharge for patients in the

Intervention – Streamlined HMR Recommendation Group.....................................................365 Appendix IX Cover letter faxed to the nominated GP at discharge for patients in the Intervention –

Streamlined HMR Recommendation Group.............................................................................366 Appendix X Discharge Summary to be sent to nominated GP and Community Pharmacist for

patients in the Intervention - Streamlined HMR Recommendation Group..............................367 Appendix XI Cover letter to faxed to the nominated Community Pharmacist at discharge for patients

in the Intervention – PDMR Group ..........................................................................................369 Appendix XII Cover letter to faxed to the nominated GP at discharge for patients in the Intervention –

PDMR Group ............................................................................................................................370 Appendix XIII Discharge Summary to be faxed to the nominated GP and Community Pharmacist at

discharge for patients in the Intervention – PDMR Group.......................................................371 Appendix XIV Med eSupport Trial protocol.....................................................................................................372 Appendix XV Examples of potential and real problems identified by the trial officers of Med

eSupport.....................................................................................................................................551 Anonymous Surveys .......................................................................................................................................................558


of 767

Appendix XVI Control Patient Satisfaction Survey ..........................................................................................558 Appendix XVII Intervention – Streamlined HMR Recommendation Patient Satisfaction Survey ...................561 Appendix XVIII Intervention PDMR Model Patient Satisfaction Survey ..........................................................566 Appendix XIX GP Satisfaction Survey for when they had a patient in the Intervention HMR

streamlined recommendation group..........................................................................................571 Appendix XX GP Satisfaction Survey for when they had a patient in the intervention PDMR model

group..........................................................................................................................................575 Appendix XXI Community Pharmacist Satisfaction Survey for when they had a patient in the

intervention – Streamlined HMR recommendation Group ......................................................579 Appendix XXII Community Pharmacist Satisfaction Survey for when they had a patient in the

Intervention – PDMR Model Group .........................................................................................583 ICT Appendices..............................................................................................................................................................587 Appendix XXIII Description of electronic communications pathway for medication information

developed in Med eSupport ......................................................................................................587 Appendix XXIV SQL Code ..................................................................................................................................619 Appendix XXV Entity relationship diagram (supplied by Phoenix) ..................................................................622 Appendix XXVI Med eSupport dataset (supplied by Phoenix) ...........................................................................623 Appendix XXVII QUM repository lookup tables (supplied by Phoenix).............................................................647 Economic Analysis.........................................................................................................................................................662 Appendix XXVIII Panel letter .................................................................................................................................662 Appendix XXIX Template for case studies ..........................................................................................................666 Appendix XXX Panel Discrepancies...................................................................................................................668 Appendix XXXI HMR Recommendation ............................................................................................................709 Appendix XXXII Tasks to be timed.......................................................................................................................750 Appendix XXXIII Template for Quality of Life Survey ........................................................................................751 Warfarin trial .................................................................................................................................................................755 Appendix XXXIV Example reviews for warfarin trial ...........................................................................................755 Appendix XXXV Report on the quality of Medication reviews conducted in the Med eSupport Project ...........757

1.2 Table of Figures Figure 1 Summary of primary Med eSupport trial process.......................................................................................22 Figure 2 Flowchart showing methodology of warfarin focused aspect of Med eSupport ........................................25 Figure 3 Med eSupport trial patient enrolment flowchart ........................................................................................28 Figure 4 Example Pharmcare® counselling sheet .....................................................................................................64 Figure 5 Flow chart indicating recruitment and randomisation procedure.............................................................69 Figure 6 Example Med eSupport HMR trial sticker..................................................................................................70 Figure 7 Flowchart of enrolment and admission process .........................................................................................74 Figure 8 Flowchart of inpatient interview process....................................................................................................75 Figure 9 Flowchart of discharge process ..................................................................................................................76 Figure 10 Flowchart of follow-up process...................................................................................................................77 Figure 11 Enrolment flowchart for RHH.....................................................................................................................88 Figure 12 Enrolment flowchart for LGH.....................................................................................................................89 Figure 13 Enrolment flowchart for SCGH...................................................................................................................90


of 767

Figure 14 Enrolment flowchart for HPH.....................................................................................................................91 Figure 15 Enrolment flowchart for BCGH ..................................................................................................................92 Figure 16 Enrolment flowchart for all sites.................................................................................................................93 Figure 17 Proposed network model ...........................................................................................................................152 Figure 18 Med eSupport Network system architecture .............................................................................................156 Figure 19 Automatic upload screenshot from the Rex® dispensing system for a test patient ...................................158 Figure 20 Automatic upload screenshot from the Winifred® dispensing system for a test patient .........................158 Figure 21 Number of users vs number of logins grouped by user type.....................................................................163 Figure 22 Comparison of AQoL scores at admission and 30 days post-discharge..................................................189 Figure 23 Comparison of AQoL scores (only patients with thirty day data)............................................................191 Figure 24 Pharmacist's checklist for warfarin counselling.......................................................................................220 Figure 25 One page guide to warfarin treatment for patients. .................................................................................221 Figure 26 Protocol for Post-discharge INR Monitoring Group ...............................................................................222 Figure 27 Patient recruitment flowchart ...................................................................................................................228 Figure 28 Proposed future model for the ICT delivery of Med eSupport .................................................................297

1.3 Table of Tables Table 1 Deficits in the delivery of care, as described by Forster et al46 .................................................................49 Table 2 Patient and provider surveys.......................................................................................................................79 Table 3 Details of patient exclusions from study prior to randomisation ...............................................................83 Table 4 Reasons patients were fully withdrawn from the trial ................................................................................85 Table 5 Reasons patients were withdrawn at the point of discharge ......................................................................86 Table 6 Reasons patients were withdrawn at thirty days post-discharge ...............................................................86 Table 7 Breakdown of patient enrolment details at each site ..................................................................................87 Table 8 Baseline characteristics of patients within the control and intervention groups, and the four sub

groups, including all patients enrolled in the trial .....................................................................................95 Table 9 Some key baseline variables of patients within the control and intervention groups, and the four

sub groups, including all patients enrolled in the trial ..............................................................................96 Table 10 Baseline characteristics of patients within the three group split, including all patients enrolled in

the trial.........................................................................................................................................................97 Table 11 Some key baseline variables of patients within the three group split, including all patients

enrolled in the trial ......................................................................................................................................98 Table 12 Baseline characteristics of patients within the control & intervention groups, & the four sub

groups, including patients enrolled in Tasmania & Victoria only .............................................................99 Table 13 Some key baseline variables of patients within the control & intervention groups, & the four sub

groups including patients enrolled in Tasmania & Victoria only ............................................................101 Table 14 Baseline characteristics of patients within the three group split, including patients enrolled in

Tasmania and Victoria only ......................................................................................................................102 Table 15 Some key baseline variables of patients within the three group split, including patients enrolled

in Tasmania and Victoria only ..................................................................................................................103 Table 16 Patients with medication discrepancies on admission .............................................................................105 Table 17 Number of medication discrepancies on admission per patient (compared with reconciled list)...........105 Table 18 Comparison of the resolution of discrepancies identified between the intervention and control

groups over the hospital stay period, measured by the median number of discrepancies per patient ........................................................................................................................................................107


of 767

Table 19 Comparison of the resolution of discrepancies identified between the intervention and control groups over the hospital stay period, measured by the percentage of patients with one or more discrepancies .............................................................................................................................................108

Table 20 Identification and resolution of new discrepancies at discharge .............................................................109 Table 21 Identification and resolution of new discrepancies at discharge, measured by the percentage of

patients with one or more discrepancies...................................................................................................109 Table 22 Differences in length of stay between those who did or did not have all discrepancies on their

admission drug chart resolved in the first 48 hours of admission for all patients...................................110 Table 23 Differences in length of stay between those who did or did not have all discrepancies on their

admission drug chart resolved in the first 48 hours of admission for those who had admission drug chart discrepancies ...........................................................................................................................111

Table 24 Admission drug chart discrepancies not acted on during the hospital stay versus length of stay for all patients............................................................................................................................................111

Table 25 Admission drug chart discrepancies not acted on during the hospital stay versus length of stay for those patients who had admission drug chart discrepancies..............................................................112

Table 26 Differences in length of stay between those who did or did not have discrepancies on their admission drug chart.................................................................................................................................112

Table 27 Community pharmacy originated discrepancies on admission, measured as percentage of patients with at least one discrepancy of this origin ................................................................................113

Table 28 Community pharmacy originated discrepancies on admission, measured as a median of discrepancies of this origin per patient.....................................................................................................113

Table 29 Comparisons of patient knowledge of their medications, as inpatients and at thirty days post-discharge....................................................................................................................................................114

Table 30 Comparison of self-reported patient compliance as inpatients and at thirty days post-discharge .........115 Table 31 Drug-related problems identified with time, measured by the median number of issues per

patient ........................................................................................................................................................117 Table 32 Drug-related problems identified with time, measured by percentage of patients with at least one

or more issue .............................................................................................................................................117 Table 33 Drug-related issues classified as significant or moderate per Cognicare® identified with time,

measured by median number of issues per patient ...................................................................................118 Table 34 Drug-related problems classified as significant or moderate per Cognicare® identified with time .......119 Table 35 Drug interactions identified using DIF® (level 1 & 2) identified with time............................................120 Table 36 Drug interactions identified using DIF® (level one & 2) identified within the three group split at

the three critical assessment points...........................................................................................................121 Table 37 Drug-related problems identified by the patient, measured by median number of issues per

patient ........................................................................................................................................................122 Table 38 At least one drug-related problem identified by the patient .....................................................................123 Table 39 Category ‘D’ DRPs identified per patient, measured by median number of problems per patient .......124 Table 40 Percentage of patients who had at least one category ‘D’ DRP(s) identified. ........................................125 Table 41 Category ‘C’ DRPs identified per patient, measured by median number of problems per patient.........126 Table 42 Percentage of patients who had at least one category ‘C’ DRP(s) identified. ........................................127 Table 43 Category ‘U’ DRPs identified per patient, measured by median number of problems per patient ........128 Table 44 Percentage of patients who had at least one category ‘U’ DRP(s) identified .........................................129 Table 45 Category ‘T’ DRPs identified per patient, measured by median number of problems per patient .........130 Table 46 Percentage on patients with at least one category ‘T’ DRP(s) identified................................................131 Table 47 Uptake of each model of medication reviews............................................................................................132 Table 48 Time after discharge to conduct home visit ..............................................................................................132 Table 49 Key findings of the Med eSupport trial .....................................................................................................134 Table 50 Number of patient surveys returned from each trial site ..........................................................................137


of 767

Table 51 Provider responses by trial site.................................................................................................................137 Table 52 Patient survey responses ...........................................................................................................................138 Table 53 General Practitioner survey responses indicating median and range.....................................................141 Table 54 Community Pharmacist survey responses indicating median and range.................................................145 Table 55 Timeline of work done with regards to ICT ..............................................................................................152 Table 56 Number of logins per user group ..............................................................................................................161 Table 57 Number of users per user group................................................................................................................162 Table 58 Monthly breakdown of website use by number of logins from 1/12/2004 to 5/10/2005 for each

user type.....................................................................................................................................................162 Table 59 Monthly breakdown of website use by number of users from 1/12/2004 to 5/10/2005............................163 Table 60 Dispensing script item automatic uploads by dispensing system from 1/12/2004 to 5/10/2005 .............164 Table 61 Reasons given by patients at phone interview for not using the website .................................................165 Table 62 Patient’s response on website ease of use ................................................................................................167 Table 63 Patient’s response of website services used and their usefulness ............................................................167 Table 64 Reasons why people did not use the web site............................................................................................168 Table 65 Responses for which form respondent would like to receive discharge medication information in

future..........................................................................................................................................................168 Table 66 Reasons providers did not use the web site...............................................................................................169 Table 67 Aspects of Med eSupport website that respondents found useful .............................................................170 Table 68 Did you find there were any errors or inconsistencies in the patient’s initial medication

information provided on the website?.......................................................................................................170 Table 69 Marking tool provided to guide the clinical panel ...................................................................................181 Table 70 Examples of clinical consequences and their severity descriptions, derived from the PROMISe

project ........................................................................................................................................................183 Table 71 Examples of values assigned to consequences, derived from the PROMISe project116...........................186 Table 72 Comparison of patient Quality of Life survey (Tas and Bendigo only) as inpatients and at 30

days post-discharge for each service level received.................................................................................189 Table 73 Comparison of patient Quality of Life survey (Tas and Bendigo patients with both survey results)

as inpatients and at thirty days post-discharge ........................................................................................190 Table 74 Proportion of patients from each group who visited the indicated practitioner within thirty days

post-discharge ...........................................................................................................................................192 Table 75 Medical consultations for each level of service........................................................................................193 Table 76 The value of discrepancy reviews - 19 reviewed cases.............................................................................196 Table 77 The effect of HMRs – 20 reviewed cases...................................................................................................197 Table 78 Scenario 1 – full implementation of the recommendations ......................................................................200 Table 79 Scenario 2 Adjusted for uptake of recommendations by prescriber.........................................................201 Table 80 Costs of program: payments .....................................................................................................................203 Table 81 Costs of program: program fixed costs at sites and nationally................................................................204 Table 82 Costs of program: activity per site............................................................................................................206 Table 83 Costs of program: total costs per annum of a national program.............................................................208 Table 84 Outcomes per site, two scenarios..............................................................................................................209 Table 85 Economic value of a national program.....................................................................................................212 Table 86 Costs associated with major bleeding sub-types ......................................................................................227 Table 87 Baseline characteristics of trial participants............................................................................................229 Table 88 Reason for initiation of warfarin...............................................................................................................230


of 767

Table 89 Anticoagulant control at discharge...........................................................................................................230 Table 90 Anticoagulant control at day eight after discharge ..................................................................................231 Table 91 Anticoagulant control at day eight by discharge INR range and follow-up type ....................................231 Table 92 Examples of anticoagulant-related problems noted at day eight in the UC group .................................232 Table 93 Anticoagulant control by day of follow-up for the Post-discharge INR monitoring group.....................233 Table 94 Median time taken for home visits in minutes (range)..............................................................................234 Table 95 Adverse outcomes occurring up to ninety days after discharge ...............................................................235 Table 96 Classification of identified drug-related problems in the PDINR group .................................................235 Table 97 GP responses to questionnaire..................................................................................................................237 Table 98 Comments about the monitoring from general practitioners ...................................................................238 Table 99 Responses to patient satisfaction survey from patients in the PDINR group...........................................240 Table 100 Unsolicited comments from patients .........................................................................................................242 Table 101 Patient knowledge ninety days after discharge ........................................................................................243 Table 102 Statistics comparing PDINR and UC groups ...........................................................................................246 Table 103 Cost savings associated with home monitoring compared with usual care.............................................247

Med eSupport Abstract

of 767

2 Med eSupport Abstract Background:

A major contributor to inadvertent polypharmacy and drug-related problems,

particularly in the elderly, appears to be hospitalisation and the consequent changes in

medication during the transitions between the community and hospital settings. It has

been noted previously that the management of prescribed medications among

chronically ill patients recently discharged from acute hospital care is often sub-optimal.

It has also been noted that assessment of medication management in the home following

discharge provides an invaluable opportunity to detect and address problems likely to

result in poorer health outcomes.

While being a very valuable drug, warfarin is also a recognised high-risk drug for

adverse events (bleeds). Adverse events from warfarin use in Australia were estimated

to cost over $100 million per annum in direct hospital costs alone, in 1992. The major

complication of anticoagulant therapy is bleeding and a number of studies have reported

that the risk of bleeding associated with warfarin is highest early in the course of

therapy; in fact the risk for major bleeding during the first month of therapy is

approximately 10 times the risk after the first year of therapy.

Methods:

Med eSupport, an innovative medication support program, tackled this major issue of

sub optimal medication management at the community-hospital interface. The program

utilised information and communications technology solutions and included (i)

provision of a secure bi-directional electronic pathway for medication profiles between

community and hospital pharmacies to facilitate medication reconciliation, (ii)

supplying a comprehensive medication information sheet to the patient/carer, general

practitioner and community pharmacist at the time of discharge from hospital, (iii)

uploading of the discharge medication information to a secure website for viewing and

printing by the patient/carer, general practitioner or community pharmacist, (iv)

providing a model whereby suitable patients were automatically referred for a home

medicines review after discharge from hospital, and (v) providing home follow-up

education, medication review and monitoring of the International Normalised Ratio for

patients initiated on warfarin during hospitalisation. The program was assessed in a


of 767

randomised, controlled evaluation involving 487 medical patients and five hospitals (2

in Western Australia, 2 in Tasmania and 1 in Victoria). The range of outcome measures

assessed included medication history discrepancies on admission and discharge,

compliance, medication knowledge, drug-related problems, unplanned readmissions to

hospital, quality of life, and patient and health care professional satisfaction. The

program of home follow-up for patients initiated on warfarin during hospitalisation was

separately evaluated in a randomised, controlled study involving 161 patients. An

economic analysis was applied to the studies’ data to determine whether the overall

program and its components were cost-effective, and the implications for a national

rollout.

Results:

There was clear evidence of problems relating to sub-optimal use of prescribed

medications, particularly with regard to discrepancies in medication histories at the

community-hospital interface. For instance, 66% of initial hospital drug charts had at

least one error. A significantly greater number of discrepancies per patient were

resolved within the first 48 hours of hospital admission for the intervention group than

for the control. There appeared to be a weak relationship between increased length of

stay and the number of discrepancies not resolved at 48 hours. Significantly more

discrepancies were also resolved prior to discharge for intervention patients than for

control patients. The intervention group displayed a significant improvement in their

compliance and drug knowledge over the 30-day post-discharge period, along with a

significant decrease in the total number of major and moderate drug-related problems

per patient. Although the numbers were small, 3% of control group patients vs. 0.6% of

intervention group patients were readmitted to hospital within 5 days of initial discharge

(“rebound readmissions”). Also, 44% of the patients who had such a rebound admission

had left hospital with apparent medication discrepancies at discharge, compared with an

overall figure of 24% of all the study patients. Patients who had a medication review

were more likely to feel confident about their medications after discharge. They also

displayed improved compliance and drug knowledge. The Med eSupport program was

welcomed by the patients and their general practitioners and community pharmacists.

Our economic analysis, based on conservative assumptions, indicated that Med

eSupport could save the Australian health sector about $60M annually on a national


of 767

level (50 sites initially). With the current rate of pharmacists’ recommendation uptake

being only partial, the sector can save $25M (additional) and $34M (total) annually at a

national level. Med eSupport also clearly established the benefits of patients initiated

on warfarin in hospital subsequently receiving home visits by a pharmacist after

discharge from hospital. Control of anticoagulation was significantly improved, and

there was a significantly lower incidence of total, major and minor bleeding

complications within 90 days in the intervention patients. The program was highly cost-

effective, and could save over $10M in reduced bleeding costs per year if implemented

across the country. The program was well received by patients and doctors.

Conclusion:

Based on the conduct and results of Med eSupport, the Project Team makes the

following recommendations.

1. A strategy for the national roll-out of a medication information sharing process

between hospitals and community pharmacies should be developed and

consequently implemented. Ideally, this would incorporate an automated ICT

system to transfer medication information efficiently. With some modifications,

the approach utilised in Med eSupport and successfully trialled with the

principal Australian pharmacy software vendor, could be expanded. Transfer of

information to GPs and community pharmacists regarding initiation of warfarin

in hospitals is one priority.

2. A strategy for the national implementation of automatic post-discharge home

medication reviews in high-risk patients, identified during hospitalisation,

should be developed and implemented. There was very strong support for this

amongst patients and other stakeholders exposed to the Med eSupport program.

This would include patients commenced on warfarin in hospitals as a priority

group.

3. In the event of national implementation of automatic post-discharge medication

reviews, existing MMR Facilitators should be trained to act as liaison officers,

working to co-ordinate accredited pharmacists for the post-discharge medication

reviews.


of 767

4. There should be further examination of factors influencing the uptake of

recommendations from home medication reviews. One strategy could be

development and implementation of educative and monitoring procedures to

continually improve the quality and presentation of home medication reviews by

accredited pharmacists.

5. When considering the implementation of new services, (such as transferring of

community pharmacy dispensing histories to hospitals, creation of a community

liaison role, or PDMR), whether within a trial framework, or on a larger national

scale, all sites should be considered individually to ensure the roll out is

successful, and ongoing quality assurance measures must be put in place to

ensure the ongoing integrity of the new service.

6. Training and accreditation programs should be developed for accredited

pharmacists to undertake, for the purposes of developing a system for

pharmacists to monitor the INR of patients after discharge from hospital.

7. All patients who are initiated on warfarin in the hospital setting should receive a

PDINR after discharge, as outlined in this study. This service should be funded

similarly to the existing HMR program, although funding would need to be

significantly increased. The PDINR program should comprise point of care INR

monitoring, patient-focussed anticoagulant education and medication review.

Med eSupport Executive Summary

of 767

3 Med eSupport Executive Summary This 21 page section is a summary of the complete Med eSupport report. Throughout

this document, references are made to the sections within the complete report that are

relevant to the summarised information. These are indicated at the top of each section as

follows: Refer Section X

3.1 Introduction Refer Section 8

A major contributor to inadvertent polypharmacy and drug-related problems in the

elderly appears to be hospitalisation and the consequent changes in medication during

the transitions between the community and hospital settings. It has been noted

previously that the management of prescribed medications among chronically ill

patients recently discharged from acute hospital care is often sub-optimal. It has also

been noted that assessment of medication management in the home following discharge

provides an invaluable opportunity to detect and address problems likely to result in

poorer health outcomes. More specifically, while being a very valuable drug, warfarin

is also recognised as a major contributor to adverse drug events, particularly soon after

initiation.

Med eSupport tackled this major issue of poor medication management at the

community-hospital interface. Confusion about medications, poor compliance and

adverse outcomes in recently hospitalised patients were a major focus of this project.

The Project Team implemented and evaluated an innovative medication support

program (Med eSupport) for high-risk patients. The program utilised information and

communications technology solutions and included:

(i) provision of a secure bidirectional electronic communication pathway for

medication profiles between community and hospital pharmacies to facilitate

medication reconciliation,

(ii) supplying a comprehensive medication information sheet to the patient/carer,

general practitioner and community pharmacist at the time of discharge

from hospital,


of 767

(iii) uploading of the discharge medication information to a secure website for

viewing and printing by the patient/carer, general practitioner or community

pharmacist,

(iv) providing a model whereby suitable patients were automatically referred for

a home medicines review after discharge from hospital, and

(v) providing home follow-up education, medication review and monitoring of

the International Normalised Ratio for patients initiated on warfarin during

hospitalisation.

Specifically, the desired outcomes of the Med eSupport project were expected to

include the following:

• Better correlation between the list of prescribed medications during

hospitalisation with the actual medications being taken by the patient

immediately prior to admission,

• Improvements in quality use of medicines (QUM) by patients,

• Improved provision of medication-related information and follow-up of

patients after discharge from hospital,

• Improved communication of discharge medication-related information

between the hospital and community-based health professionals, ensuring

continuity of treatment to promote patient care at the time of discharge from

hospital,

• Improved patient compliance/adherence,

• Improved patient knowledge of their medications,

• Less medication-related adverse events following hospital discharge,

• Demonstration of the benefits of home follow-up visits on patient outcomes,

targeting newly initiated warfarin patients as a specific example, and

• Favourable acceptance of the program by patients, community pharmacists

(CPs) and general practitioners (GPs).


of 767

3.2 Methodology

3.2.1 Med eSupport Trial Refer Section 9.1

Med eSupport consisted of a primary randomised controlled trial of a range of services

as a package, with a smaller study specifically focused on patients initiated on warfarin.

Med eSupport primary trial

The primary Med eSupport trial was implemented in five hospitals in Tasmania,

Western Australia and Victoria over a period of 9 months from December 2004 to

August 2005.

Patients were eligible for enrolment if:

• they were 50 years or older,

• had at least two chronic conditions, (one of which was cardiovascular

disease, diabetes mellitus or chronic obstructive airways disease),

• were taking at least three chronic medications,

• could nominate a regular GP and CP,

• did not live in a aged or residential care facility, and

• were able to provide informed consent.

The primary study consisted of a control and intervention group. These two groups were

further divided into two sub groups, dependant on the form of home medicines review

(HMR) recommendation utilised. Patients were randomly allocated to one of the four

sub groups using block allocation concealment.

These groups were:

• Control

No HMR recommendation

HMR recommendation

• Intervention

streamlined HMR recommendation

post-discharge medication review (PDMR) model


of 767

For each patient enrolled, their progress was followed from the point of admission to 30

days post discharge. The following is a summary of the processes at each point in time,

and a flow chart is provided on page 22 (see Figure 1).

At admission

For all patients:

• Patients interviewed to obtain basic demographic data, medication history,

baseline knowledge, compliance and quality of life (QoL) score and self-

reported drug-related problems (DRPs).

• CP telephoned and a 6 month dispensing history was obtained.

• GP was contacted for a medication history where required.

• Current medical notes checked for information regarding admission

medications.

• All obtained information compared and collated to form a reconciled list of

the most likely medications the patient was taking when they entered

hospital.

For intervention patients only:

• The reconciled list, along with highlighted discrepancies between it and the

initial drug chart were discussed with the resident medical officer (RMO)

within 24 hours of admission.

Progress of resolution of identified discrepancies was followed for all patients

throughout their hospital stay, actively for intervention patients and silently for control

patients.

Prior to discharge

For all patients:

• Discharge prescriptions were checked against current drug charts and

medical notes and any new discrepancies found were highlighted.


of 767

For Intervention patients:

• Newly identified discrepancies were discussed with the RMO prior to

discharge.

• Discharge medication counselling and a counselling sheet were provided.

All control patients received usual care in accordance with regular hospital practices.

At discharge

Control – No HMR recommendation

• Patients received usual care

Control – HMR recommendation

• Patients had a sticker, suggesting an HMR would be beneficial, placed on

their discharge summary

Intervention streamlined HMR recommendation

• All medications and relevant medication information uploaded to the Med

eSupport trial website in the form of a discharge medication summary,

counselling sheet and weekly checklist.

• Summary was sent, via fax, to the patient’s nominated GP(s) and CP(s)

within 24 hours of discharge, along with access details for the website and

information regarding the project and their patient’s group allocation.

• Summary was in a HMR referral format, so all the GP had to do to refer the

patients for an HMR was check the information, sign it and send it the CP.

• Nominated GP(s) and CP(s) were telephoned at this point and informed what

they were going to receive and given the opportunity to ask questions and

clarify any information.

Intervention – PDMR model

• All medications and relevant medication information uploaded to the Med

eSupport trial website in the form of a discharge medication summary,

counselling sheet and weekly checklist.


of 767

• Summary was sent, via fax, to their nominated GP(s) and CP(s) within 24

hours of discharge, along with access details for the website and information

regarding the project and their patient’s group allocation.

• Patients were given an automatic post-discharge medication review, funded

by the trial, within 5-7 days post-discharge. The GP(s) and CP(s) were

telephoned and informed what they were going to receive, how a PDMR

worked, and given the opportunity to ask questions and clarify any

information.

• Nominated CP(s) were encouraged to perform the PDMR themselves, but

accredited pharmacists were available through the project team, if this was

not possible.

• Trial Officers continued to keep in close contact with the nominated CP(s)

and GP(s) to ensure the PDMR was performed in a timely manner.

30 days post-discharge

All patients:

• Telephoned and asked about their current medications and changes since

discharge.

• Follow-up knowledge, compliance and QoL scores and self-reported DRPs

collected.

• Asked if they had received a home visit from a pharmacist and when.

Intervention patients only:

• Asked about their use of the website.

After the 30 days post-discharge phone call

• CP(s) (and GP(s) where required) contacted to confirm current medication

list and PDMR/HMR activity and reports collected where applicable.

• Satisfaction surveys sent to all patients.

• Satisfaction surveys sent to the CP(s) and GP(s) of intervention patients.


of 767

Figure 1 Summary of primary Med eSupport trial process

Patient admitted to hospital

Identified as High-Risk Patient

Yes

No Excluded

Patient or carer provides informed consent

Yes

No

RandomisedControl Group Intervention Group

TO collects 6 months dispensing history from the patient’s nominated CP(s)

Compilation of a reconciled medication list using CP dispensing data, meds on admission list (GP

account if necessary) & patient interview


Patient reaches point of discharge

Otherwise normal discharge processes as per usual hospital care.

Discharge medication counselling sheet supplied, with verbal counselling.

Discharge medication information uploaded to the web & sent via fax to nominated CP(s) & GP(s)

within 24 hrs of discharge.

HMR not recommendedat the point of discharge

HMR recommended via sticker on discharge

summary

HMR recommendedusing a streamlined

referral model

PDMR facilitated by Project Team

All patients telephoned by TO at 30 days post discharge to collect follow-up data

Postage of satisfaction surveys to participating health care professionals and patients/carers

Final analysis by research team

30 days post discharge

Excluded



Discrepancies between reconciled list and initial drug chart identified and kept ‘silent’ as a measure

of discrepancy resolution through usual care.

Discrepancies between reconciled list and initial drug chart identified and discussed with RMO


Discrepancies reported and rectified where necessary.

RMO’s discharge script compared with recent drug charts to detect any discrepancies.

Discrepancies documented but not reported.


Identified as High-Risk Patient

Yes

No Excluded

Patient or carer provides informed consent

Yes

No

RandomisedControl Group Intervention Group





Patient reaches point of discharge

Otherwise normal discharge processes as per usual hospital care.

Discharge medication counselling sheet supplied, with verbal counselling.

Discharge medication information uploaded to the web & sent via fax to nominated CP(s) & GP(s)

within 24 hrs of discharge.

HMR not recommendedat the point of discharge

HMR recommended via sticker on discharge

summary

HMR recommendedusing a streamlined

referral model

PDMR facilitated by Project Team

All patients telephoned by TO at 30 days post discharge to collect follow-up data

Postage of satisfaction surveys to participating health care professionals and patients/carers



Excluded



Discrepancies between reconciled list and initial drug chart identified and kept ‘silent’ as a measure

of discrepancy resolution through usual care.

Discrepancies between reconciled list and initial drug chart identified and discussed with RMO


Discrepancies reported and rectified where necessary.

RMO’s discharge script compared with recent drug charts to detect any discrepancies.

Discrepancies documented but not reported.


of 767

Post-Hoc Reclassification process

Refer Section 9.1.3

At the end of the trial it was found that, due to the nature of the trial design, which

group a patient was allocated to and what services they received were not always

uniform. The primary difference was whether they received a PDMR or HMR. This

was variable due to the differing extents of uptake between the different models of

recomendation.

On reflection, it was felt that these differences within the original groups could not be

ignored, as it was the impact of the services received that was most important in terms

of analysis of different aspects of the program.

Initial group allocation was still vital to measure uptake of the different methods of

PDMR/HMR promotion and to assess for most aspects of patient satisfaction. However,

for a more realistic analysis of the services offered by Med eSupport, the patients were

reallocated to a new group, dependant upon the services received.

The new grouping system was comprised of three groups, as follows.

1. Minimal Intervention

• Control Patients who did or did not receive discharge medication counselling

2. Partial Intervention

• Discharge medication counseling received and discrepancies reported to the

RMO

3. Full Intervention

• Discharge medication counseling received, discrepancies reported to the

RMO and PDMR/HMR performed

This new grouping, termed the ‘services received’ grouping, has been used to analyse

those parameters that involve comparisons from baseline to 30 days and allow accurate

comparison of the impact the services provided had on the outcomes. It is to be noted

that all data was initially analysed using all three group allocations, control vs

intervention, the four sub groups and the post-hoc ‘services received’ grouping.


of 767

3.2.2 The warfarin focused aspect of the Med eSupport trial; a specific 'high-risk' example.

Refer Section 12

For the warfarin focused section of the trial, a different trial process was undertaken.

Patients initiated on warfarin at the Royal Hobart Hospital (RHH) were prospectively

identified and randomised to a control (usual care; UC) or intervention (Post-discharge

INR monitoring; PDINR) group. PDINR group patients received a home-visit by the

project pharmacist on alternate days on 4 occasions, with an initial visit two days after

discharge from hospital. The pharmacist, using Point-of-Care (POC) testing obtained

international normalised ratio (INR) results and educated the patients regarding

anticoagulant therapy. A review of the patients’ medication was also undertaken during

the visits (effectively a Home Medicines Review on the first visit, followed by short

reviews on subsequent visits). The UC group was solely managed by the general

practitioner (GP) and only received a visit from the project pharmacist eight days after

discharge to determine anticoagulant control. A number of outcome variables were

assessed, including the achievement of a therapeutic INR value on day eight after

discharge. Bleeding, thromboembolic outcomes and warfarin knowledge were assessed

90 days after discharge. Patients and general practitioners were anonymously surveyed

to assess satisfaction with the program.

See Figure 2 for a flowchart representation of this trial process.


of 767

Figure 2 Flowchart showing methodology of warfarin focused aspect of Med eSupport

Patient initiated on warfarin in hospital

Randomised

Usual Care Group PDINR Group

Home visit

Day 6

Home visit

Day 2

Home visit

Day 4

Home visit

Day 8

•INR measured and reported to GP

•Warfarin dose changed in consultation with GP as necessary

•Patients provided with warfarin education

•Medication Review conducted

Discharged

Day 0

Home visit

Day 8

Bleeding and thromboembolic outcomes and warfarin knowledge assessed

Day 90

GPs and patients anonymously surveyed to assess satisfaction

INR reported to GP if deemed serious or life-

threatening

Patient initiated on warfarin in hospital

Randomised

Usual Care Group PDINR Group

Home visit

Day 6

Home visit

Day 2

Home visit

Day 4

Home visit

Day 8

•INR measured and reported to GP

•Warfarin dose changed in consultation with GP as necessary

•Patients provided with warfarin education

•Medication Review conducted

Discharged

Day 0

Home visit

Day 8

Bleeding and thromboembolic outcomes and warfarin knowledge assessed

Day 90

GPs and patients anonymously surveyed to assess satisfaction

INR reported to GP if deemed serious or life-

threatening


of 767

3.2.3 ICT systems evaluation Refer Section 10.1

Med eSupport sought to improve sub optimal medication management at the

community-hospital interface. One focus of the study was to investigate the use of ICT

in securely transferring patient’s information between key stakeholders.

The ICT model implemented provided an innovative bi-directional transfer of

prescription information between community pharmacies and hospitals (Figure 18). It

also allowed patients and their GP’s access to discharge information, including

discharge diagnosis, relevant medical conditions, and a current list of medications on

discharge from hospital. The model also allowed for the secure, automatic uploading of

patient’s prescription details from the community pharmacy to the server based at the

University of Tasmania.

Two community pharmacy dispensing software vendors were contracted to enable their

software to communicate patient's medication information to the central database. For

incompatible dispensing systems, medication lists were faxed from the pharmacy and

manually entered into the database by trial officers using the website. All transfers of

patient information occurred using recommended security technologies, including 128-

bit encrypted HL7 messages with PKI keys.

Retrieving information from the central database was achieved using the secure

interactive website. Different access privileges were given to patients, health care

providers and trial officers. Patients were encouraged to use the website to obtain a

current discharge summary. Health care providers had the ability to print medication

lists and update medication lists for patients in their care.

Transactions to the website were logged to the database. These data were used to

determine the uptake and usage by health care providers and patients, along with the

number of automated uploads that occurred. Intervention patients and their GPs and CPs

were surveyed to establish uptake and barriers of the ICT solution.

As can be expected with a trial of this nature, there were many technical obstacles to

overcome. Examples include: full logging requested was not available until the trial

had ended and one dispensing software vendor delivered automatic uploading towards

the end of the data collection period.


of 767

3.3 Results Refer Section 9.2

Four hundred and eighty seven patients remained enrolled across all sites and available

for initial data analysis; four hundred and twenty seven for analysis at the point of

discharge, and 378 for the 30-day phone call. Significant losses occurred due to

concerns with data integrity and alternative interpretations of the trial protocol at two of

the trial sites.

Further discussion surrounding withdrawals of patients from the primary Med eSupport

trial can be found in section 9.2.1.1 of the full report. Figure 3shows the flow of

patients through the study. Similar flowcharts for each individual site can be found

starting from page 88 of the full report.


of 767

Figure 3 Med eSupport trial patient enrolment flowchart

4176 Patients Screened

3361 Excluded

Intervention –PDMR model:

N =102

Intervention: N = 256

Control: N = 303

253 Refused

562 Consented

N = 815

Intervention –Streamlined HMR

model:N =101

Control –HMR

recommendation:N =143

Control –No HMR


26 withdrawn20 withdrawn10 withdrawn4 withdrawn

Primary Discharge from Hospital

9 withdrawn18 withdrawn 8 withdrawn14 withdrawn

Intervention –PDMR model

N = 67


modelN = 73


N = 115

Control –No HMR

recommendationN = 123

14 withdrawn 5 withdrawn 30 withdrawn 23 withdrawn

3 withdrawn


N =125


model:N =131

Control –HMR


Control –No HMR


Initial Intervention completed


3361 Excluded


N =102


Control: N = 303

253 Refused

562 Consented

N = 815


model:N =101

Control –HMR


Control –No HMR






N = 67


modelN = 73


N = 115

Control –No HMR



3 withdrawn


N =125


model:N =131

Control –HMR


Control –No HMR




of 767

3.3.1 Key Findings Refer Section 9.2.1

3.3.1.1 Key findings of the Med eSupport trial Section Finding

9.2.1.5.2.1 A significantly greater number of discrepancies per patient were resolved within the first 48 hours of admission for the intervention group than for the control.

9.2.1.5.3.1 Significantly more discrepancies were resolved prior to discharge for intervention patients than for control patients.

9.2.1.5.4 LOS increased with number of discrepancies not resolved at 48 hours.

9.2.1.6 Generally, all patients were found to improve their knowledge over time. However, at 30 days after discharge patients who received the full intervention had significantly higher drug knowledge than minimal and partial intervention patients.

9.2.1.7 The full intervention group displayed a significant improvement in their compliance over the 30 day post-discharge period.

9.2.1.8.2.1 During the peri-discharge period (discharge to 30 days post-discharge), the full intervention group and the partial intervention group experienced a significant decrease in the total number of significant and moderate DRPs per patient.

9.2.1.8.3.1 Over the full study period (admission to 30 days post-discharge) the full intervention patients did not have an increase in the total number of drug interactions identified. In comparison, over the same period the minimal intervention and partial intervention patients did have a significant increase in drug interactions identified by Drug Interaction Facts software.

9.2.1.8.4.1 Patient identified drug-related problems reported by the full intervention group were significantly fewer than the other groups over the period from admission to 30 days post-discharge.

9.2.1.8.5.1.1 Generally, all patients experienced an increase in drug selection DRPs over the period from admission to 30 days post-discharge. However, in the full intervention group of patients this increase was not significant, where it was in the other groups.

9.2.1.8.5.2.1 Overall, from admission to 30 days post-discharge, the full intervention group displayed a significant decrease in total number of recorded compliance DRPs.


of 767

9.2.1.8.5.3.1 Generally, all patients experienced an increase in untreated indications DRPs over the period from admission to 30 days post-discharge. However, in the full intervention group of patients this increase was not significant, where it was in the other groups.

9.2.1.9.1 The current method of HMR referral or a passive indicator on the discharge summary does not induce a medication review after discharge.

9.2.1.9.1 The automatic post-discharge medication review is currently the best referral process to ensure a review occurs in a timely manner after discharge.

11.3.1 Quality of Life improved significantly for all patients from admission to 30 days post-discharge.

11.3.3 The average number of medical consultations and their associated costs were slightly higher in the full intervention group. This was not surprising as the conduct of the medication review would generally have necessitated at least one GP visit.

11.3.2 There was no statistically significant difference in readmissions across the groups. Most (82%) of the readmissions were seemingly unplanned.

11.3.2.1 It was interesting to note that of the 9 patients who reported they were readmitted within 5 days of initial discharge (“rebound readmission”), 8 were control patients and only 1 was an intervention patient. Unfortunately, due to the small numbers, this was not found to be statistically significant, but a trend was seen.

3.3.1.2 Patient survey responses Refer Section 9.2

Section Finding

9.2.2.2.1 Patients who had a PDMR were more likely to want a home visit by a pharmacist to be available in the future. It is likely that exposure to a service increases future uptake.

9.2.2.2.1 When asked how much money they felt they would pay for a home visit by an accredited pharmacist most replied they would pay less than $20. A common reason for not wanting to pay was the perception they could get the same information from their GP or Community Pharmacist.

9.2.2.2.1 Patients in the PDMR group were more likely to feel confident about their medications after discharge.


of 767

3.3.1.3 General Practitioner survey responses Refer Section 9.2

3.3.1.4 Community Pharmacists survey responses Refer Section 9.2

Section Finding

9.2.2.2.2 Most GP respondents thought that the medication summary was provided within an adequate timeframe and they strongly agreed that receiving discharge medication information in the future would be valuable. There were no statistical differences across groups for the timeframe that General Practitioners received information.

9.2.2.2.2 General Practitioners who had patients in the PDMR trial were more likely to think that Med eSupport gave them a clearer picture of their patient’s medication on discharge.

9.2.2.2.2 The General Practitioners who had patients in the PDMR trial arm were more likely to use the website. However, there was a trend for the group of General Practitioners who had patients in the PDMR trial arm to state the website was more difficult to use.

9.2.2.2.2 There was a very positive response when General Practitioners were asked if the PDMR/HMR assisted them in the medication management of their patient.

9.2.2.2.2 When General Practitioners were asked if they would like to see an automatic PDMR for their patients in the future 74% responded that they wanted the service. There were no significant differences between the study groups.

9.2.2.2.2 There was a positive response from GPs when asked if they thought that the study benefited them in optimising the patient’s medication management through improved communication of medication related information.

Section Finding

9.2.2.2.3 Community Pharmacists whose patients received a PDMR were more likely to use the website.

9.2.2.2.3 Having a PDMR did not influence the Community Pharmacist’s decision on whether the service ought to be automatic.

9.2.2.2.3 Comparing across provider groups, Community Pharmacists found that receiving a discharge summary gave them a clearer understanding of the patient’s medication history than General Practitioners.


of 767

3.3.1.5 Results of note for the Warfarin focused section of Med eSupport

Refer Section 12.3

3.3.1.6 ICT systems-related results Refer Section 10.2

Section Finding

12.3.1 The two study groups were found to be well matched in baseline demographics. A total of 161 patients were enrolled in the study (75 PDINR and 86 UC).

12.3.1 Thirty-nine and 48 percent of the PDINR and UC groups respectively had therapeutic INRs at discharge.

12.3.1 The PDINR group had 67% of patients with a therapeutic INR at day 8, compared with 38% of UC patients

12.3.1 Also, 27% of the UC patients had a high INR, compared with only 8% of the PDINR patients

12.3.1 There was a significantly lower incidence of all bleeding complications within 90 days in the PDINR group. Total bleeding was 14% in the PDINR group, compared with 34% in the UC group (P<0.004).

12.3.1 The incidence of major bleeding was 1% in the PDINR group compared with 11% in the UC group (P=0.02). The incidence of minor bleeding was also reduced in the PDINR group compared to the UC group, 12% to 30% respectively (P=0.01).

Section Finding

10.2.1 The trial protocol dictated that only intervention patients and their health-care professionals received access to the website. Twenty-eight patients (12%), 36 GPs (19%) and 63 Pharmacists (45%) used the website.

10.2.2 Of the 159 intervention patients telephoned, 12 reported they had used the website and eight reported they did so alone. Three patients reported they printed a counselling sheet from the website and one patient reported they printed a weekly checklist from the website

10.2.2 Of the people who did not use the web site when telephoned, a great majority reported they either did not have a computer (68%), or had no interest in using a computer (15%).

10.2.3 Of the anonymous questionnaire respondents, eleven intervention patients (10%) reported that they used the web site. Of these eleven respondents, nine thought that the website was ‘OK’ or ‘Easy’ to use, whereas two thought it was difficult to use.


of 767

3.3.2 Economics and Financial Analysis of Med eSupport Refer Section 11.3

An economic analysis of the Med eSupport project was performed to assist in the

evaluation of the future sustainability of the program. The analysis was applied to

isolate the outcomes of the two key components of the intervention program (the

services that would be critical in a national roll-out): the medication list reconciliation

process undertaken at the time of hospital admission and discharge, and performance of

post-discharge medication reviews. The analysis included the determination of direct

health service costs utilising time trials to estimate staff time required to perform the

critical activities and an evaluation by an independent 5-member clinical panel of a

random selection of medication reconciliation cases and post-discharge medication

reviews from the trial. The economic analysis was intentionally conservative in that

only one drug-related problem was considered per medication reconciliation case and

post-discharge medication review. The clinical panel members were asked to consider

the probability of a consequence occurring (with the Med eSupport program’s

intervention and without the intervention) and also the “attributability” of the

intervention to the program. Savings to the health care system were based on hospital

admission, general practitioner and medical specialist consultation, and investigation

and pathology costs avoided.

10.2.3 Most people thought that the current medication summary was the most useful feature of the web site.

10.2.4 Of the providers who did not use the web site, 46% of community pharmacists and 35% of GPs responded they did not need to look at the website,

10.2.4 A majority of providers would prefer to receive patient discharge information by fax.

10.2.4 Of the respondents who did use the website, access to dispensed medication history, a current medication list, discharge medication information and the ability to maintain an up to date medication list were the most useful aspects.

10.2.4 In general, health care providers agreed that the information presented on the website was accurate.

10.2.3 People using the website tended to be younger, with approximately 30% of users aged between 50-59 age group used the website in comparison to 10% of users in the 70-79 year age group.


of 767

All estimates were subsequently adjusted (diluted) to take into account that not all

medication reconciliation cases and post-discharge medication reviews identify drug-

related problems and have associated clinical recommendations, and the likely uptake

rate of the recommendations from medication reconciliation cases and post-discharge

medication reviews. Discrepancy review recommendations, which occurred in the

hospital, were implemented 78% of the time. However, post-discharge medication

review recommendations were implemented only 29% of the time. The costs of

implementation at site and national levels were identified. The number of sites at which

the program could be implemented in an initial roll-out was estimated to be 50, with

each site essentially representing a combination of a Division of General Practice and

one or two mid-size (approximately 400-bed) general public hospitals.

The analysis determined that an average medication reconciliation review would

prevent 41 days of health loss and save $205 in financial savings to the health sector.

Similarly, just one randomly selected post-discharge medication review

recommendation would prevent 46 days of health loss and $206 in financial savings to

the health sector. When adjusted, particularly for the likely uptake rate of clinical

recommendations, it was estimated that, on average, a medication discrepancy review

produces $103 financial savings to the health sector and a post-discharge medication

review produces $51 financial savings to the health sector.

Our analyses indicated a saving to the health care system of approximately $0.5M per

annum for each roll-out site for the extra post-discharge medication reviews and

medication discrepancy checks performed, taking into account the partial uptake of

recommendations. Overall, the economic effect of the Med eSupport program is

relatively modest (i.e. around $111 savings per patient) and it is not surprising that this

effect was not observed in clinical outcomes when dispersed across the relatively small

number of patients in our trial. However, when applied to a hospital over one year, these

savings are substantial and important. Although based on conservative assumptions,

Med eSupport can save the health sector between $54M (additional) and $69M (total) in

financial savings annually on a national level (50 sites initially), assuming full

implementation of the key post-discharge medication review recommendations. With

the current rate of pharmacists’ recommendation uptake being only partial, the sector

can save $25M (additional) and $34M (total) annually at a national level. Even


of 767

assuming partial compliance and excluding the expected financial savings, the program

represents value for money at an additional $10.84 per day of health loss prevented.

3.4 Med eSupport Recommendations Refer Section 14









information to GPs and community pharmacists regarding initiation of warfarin

in hospitals is one priority.






group.




reviews.







of 767













significantly increased. The PDINR program should comprise point of care INR


Project Team Members

of 767

4 Project Team Members

Professor Gregory Peterson (Chief Investigator)

Anna Tompson

Dr Shane Jackson

Dr Omar Hasan

Peter Gee

Rose McShane

Craig Cooper

Kimbra Fitzmaurice

Bronwen Roberts

Donielle Luttrell

Rhonda Niemann

Barbara Van Der Werf

Jessie Loh

Jean Chow

Rebecca Dunn

Acknowledgements

of 767

5 Acknowledgements The Project Team would like to thank the following individuals and organisations for

valuable assistance with various aspects of the project:

• The funding body (Australian Government Department of Health and Ageing,

through the Third Community Pharmacy Agreement Research and Development

Program) and the Pharmacy Guild of Australia;

• Dr Simone Jones, Ms Erica Vowles and the members of the Expert Advisory

Group;

• Mr James Reeve, Dr Paul Turner, Ms Brita Pekarsky, Mr Peter Tenni and Mr

John Elkerton;

• Kate Cassidy, David Peachy, Cassie Smith and Jessica Howard;

• The community pharmacists, hospital pharmacists and general practitioners who

willingly participated, and gave us their time to test our ideas and provide us

with valuable feedback;

• The participating hospitals and their staff;

• Those patients who willingly volunteered to be involved in this trial and gave us

not only their time, but accepted without question, our intrusion into their lives;

• The Project Steering Committee: Louise Sullivan, Sue Leitch, and Lyle Borlase.

• The Project Advisory Committee: Louise Sullivan, Sue Leitch, Lyle Borlase, Di

Aldous, William Flassman, Chris Showell, and Sarah Male; and

• Cognicare® Solutions, ICS Multimedia, MIMS, Phoenix Computer Systems and

PCA Nu Systems.

How to use this document

of 767

6 How to use this document The Med eSupport trial and its evaluation is a substantial piece of work covering many

disciplines. As a result, this report is a large document, totalling 767 pages. Evaluation

of the Med eSupport project covers: the Med eSupport trial, health care and provider

surveys, ICT systems, economic analyses, and post-discharge home INR monitoring.

To improve readability of this report each area will be presented separately, with its

own methods, results and discussion. Recommendations for the future of the project

will be presented at the end of the report.

Appendices referred to in the text are in a separate volume when viewed in hard copy

format. Program source code has also been provided on an accompanying CD along

with electronic copies of this document.

All headings are numbered, and there are tables of headings, figures and tables at the

beginning of the document.

The essential layout of this report is as follows:

• Abstract,

• Executive summary,

• Project team members,

• Acknowledgments,

• How to use this document,

• Acronyms used and definitions,

• Introduction,

• Med eSupport trial and participant survey evaluation,

• ICT systems evaluation,

• Economic and financial analysis of Med eSupport,

• The warfarin focused aspect of the Med eSupport trial; a specific 'high-risk' example,

• General discussion,

• Conclusions and recommendations,

• References, and

• Appendices (separate volume when viewed in hard copy format).

Acronyms and Definitions

of 767

7 Acronyms and Definitions ADE = Adverse Drug Events

ADR = Adverse Drug Reactions

ADSL = Asymmetric Digital Subscriber Line - a technology for transmitting

digital information on existing phone lines to homes and businesses.

Unlike regular dial-up Internet service, ADSL provides an "always on"

connection. It is ‘asymmetric’ which means the upload speed is different

to the download speed (e.g. 256/64).

AF = Atrial Fibrillation

AMI = Acute Myocardial Infarction

APAC = Australian Pharmaceutical Advisory Council

AQoL = An Australian-developed quality of life (QoL) instrument; “Assessment

of Quality of Life”. It consists of 15-questions, organised into 5 scales

measuring Illness, Independence, Social Relationships, Physical Senses

and Psychological Wellbeing. Scores from these five dimensions can be

used to provide a QoL profile for respondents.1

BHCGH = Bendigo Health Care Group Hospital – a participating hospital in

Victoria.

CDHA = Commonwealth Department of Health and Ageing

CMMS = Cognicare® Medication Management System

CP = Community Pharmacist/Pharmacy

DIF = Drug Interactions Facts® - a computer program used in this project as a

standardised measure of drug interactions

DRP = Drug related problem

Firewall = A set of related programs, located at a network gateway server that

protects the resources of a private network from users from other

networks. (The term also implies the security policy that is used with the

programs.) A business with an intranet that allows its workers access to

the wider Internet installs a firewall to prevent outsiders from accessing

its own private data resources and for controlling what outside resources

its own users have access to.

GP = General Practitioner


of 767

HL7 = Health Level 7 - a standard communications language for healthcare and

is the interface standard for communication between various systems

employed in the medical community

HMR = Home Medication Review

HP = Hospital Pharmacist

HPH = Hollywood Private Hospital – a participating hospital in Western

Australia

ICS® = ICS multimedia - a software vendor based in Hobart2

ICT = Information and Communications Technology

INR = International Normalised Ratio – blood test used to monitor the effects of

warfarin therapy; ideal range while on warfarin is 2.0 - 3.0. Some

patients may require higher or lower ranges depending on other medical

conditions.

LGH = Launceston General Hospital – a participating hospital in Tasmania

LOS = Length of stay

MDC = Main Diagnostic Categories

NSAID = Non steroidal Anti Inflammatory Drug

ODBC = Open Data Base Connectivity – a method that allows different computer

programs to send data to each other.

OTC = Over the Counter - medication that can be purchased at a pharmacy

without a prescription

POD = Patients Own Drugs

PBS = Pharmaceutical Benefits Scheme

PCANu = PCA NU systems - a software vendor based in Victoria. They market a

suite of programs for pharmacists, including the Winifred®dispensing

software3

PDINR = Post-discharge INR monitoring

PDMR = Post Discharge Medication Review

PGA = Pharmacy Guild of Australia

Phoenix = Trading as Phoenix Corporation - a software vendor based in Tasmania.

They market a suite of programs including the Rex® dispensing

software.4


of 767

PKI = Public Key Infrastructure - a security management system including

hardware, software, people, processes and policies, dedicated to the

management of Digital Certificates for the purposes of secure exchange

of electronic messages5

POC = Point of Care

QoL = Quality of Life

RHH = Royal Hobart Hospital – a participating hospital in Tasmania

RMO = Registered Medical Officer

SCGH = Sir Charles Gairdner Hospital – a participating hospital in Western

Australia

SQL = Structured Query Language - the most popular computer language used

to create, modify and retrieve data from relational database management

systems. The language has evolved beyond its original purpose to

support object-relational database management systems. It is an

ANSI/ISO standard6

SSL = Secure Sockets Layer - a commonly-used protocol for managing the

security of a message transmission over the Internet.

TCP/IP = Transmission Control Protocol/Internet Protocol - a protocol for

communication between computers, used as a standard for transmitting

data over networks and as the basis for standard Internet protocols.

TIA = Transient Ischaemic Attack

TO = Trial Officer

UC = Usual Care

USA = United States of America

VTE = Venous Thromboembolism

WHO = World Health Organisation

Introduction

of 767

8 Introduction 8.1 Adverse drug events Adverse events associated with the use of pharmaceutical drugs in society continue to

be a significant public health burden.7, 8 In a recent survey of South Australian

households, it was found that of the 3,015 responding households, 46.8% were using

some form of prescribed medication, and 5.7% were using more than six prescribed

medications.9 Rates of use such as these show the great potential for harm in the

community.9

Adverse Drug Events (ADEs) can be both non-preventable and preventable and are

made up of Adverse Drug Reactions (ADRs) and medication errors.10 An ADR has

been defined by the World Health Organisation (WHO) as “Any response to a drug

which is noxious, unintended and occurs at doses used for prophylaxis, diagnosis or

therapy” whereas, as defined by the United States of America (USA) National Co-

ordination Council for Medication Error Reporting and Prevention

“A medication error is any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of health professional, patient or consumer”11

Of particular concern is the prevalence of ADEs that lead to hospitalisation. In 1999, a

comprehensive literature review of international studies relating to the incidence of

preventable drug-related admissions, performed by Winterstein et al,12 found fifteen

studies reporting a median prevalence of preventable drug-related admissions of 4.3%.

These statistics mirror other studies performed around the world. A prospective analysis

of the hospital admissions of 18,820 patients in Liverpool, England in 2004 found that

1,225 admissions were related to an ADR, giving a prevalence of 6.5%, with the ADR

directly leading to the admission in 80% of cases. Another study in Nottingham

produced similar results, finding 4.3% of the study admissions were due to potentially

preventable ADEs.13 Bates et al10 stated in 2003 that the overall rate of ADEs in the

USA is estimated to be 6.5 per 100 hospital admissions, with 28% of these being

preventable. A study in France by Peyriere et al14 of 156 patients admitted to hospital

found that 38 ADEs occurred in 32 of the patients and in 15 cases, the ADE was the

reason for the hospital admission.

Introduction

of 767

Here in Australia, Runciman et al15 reviewed all available information regarding the

incidence of ADEs and medication errors in Australia in 2003. It was found that 2-4%

of all hospital admissions were medication-related and of these, up to three quarters

were potentially preventable. Runciman et al15 also found that routine death certificate

and hospital discharge record coded data capture less than half as many ADEs as does

manual medical record reviews. Of coded adverse events that contributed to death, 27%

involved an ADE, as did 20% of adverse events identified at discharge and 43% at

general practice encounters.

Similarly, the Second National Report on Patient Safety8 by the Australian Council for

Safety and Quality in Health Care (‘Improving Medication Safety’) concluded the

following:

“Data show that between 2 to 3% of hospital admissions are related to problems with medicines (approximately 140,000 per year). This is more than the combined number of admissions for asthma and heart failure. About one-half of these hospital admissions are considered to be avoidable. Of the 100 million encounters recorded in general practice each year, it is estimated that around 400,000 of these are related to adverse medication events.”8

These statistics are alarming to say the least, particularly considering it is well

recognised that ADEs are vastly under-reported.15, 16

8.2 The elderly at particular risk of adverse drug events

The elderly appear to be at particular risk of ADEs.17-23 Across the globe, the size of the

elderly population has been increasing steadily for several years.20, 24, 25 Merle et al20

identified some of the reasons why the elderly population is at greater risk of

medication misadventure. Not only does this age group commonly have several

concomitant diseases that require treatment with multiple medications, but some

pathophysiological consequences of aging soon begin to have an influence.20 Drug

absorption, distribution and elimination all can decline with age and furthermore, while

hepatic metabolic function is fairly normal, renal function is usually markedly

depressed in very old individuals.20

These concerns reflect the findings of increased numbers of hospital admissions due to

ADEs in the elderly population around the world. A Medline search over 2002-3,

performed by Hanlon et al17 found 7 articles relating to incidence of drug-related

Introduction

of 767

problems in the elderly. They identified a number of studies in this population which

reported incidences ranging between 5% and 20% of at least one ADE within the

previous year. A study performed in Boston by Gurwitz et al18 examining the incidence

and severity of ADEs among older persons in the ambulatory care setting found an

overall rate of ADEs of 50.1 per 1,000 person years with a rate of 13.8 preventable

ADEs per 1,000 person years. Of the ADEs identified in this study, 38% were

identified as serious, life-threatening or fatal. In Canada, a study performed in 1998 by

Hohl et al22 found that over the study period, ADEs accounted for 10.6% of all

emergency department visits in an elderly cohort to the study hospital.

In the Australian study by Runciman et al15 up to 30% of hospital admissions for

patients older than 75 years of age were found to be medication related compared to the

previously documented statistic of 2-4% for the general population. Here in Tasmania, a

study performed at the Royal Hobart Hospital (RHH) over an 8 week period in 1998

found that 73 of 240 (30.4%) admissions of elderly persons may have been the result of

an ADE and of these, 53.4% were considered definitely preventable.23

8.3 Patient behaviours towards their medications influence risk of adverse drug events

Low compliance to prescribed medical interventions is an ever present and complex

problem, especially for patients with a chronic illness.26, 27 It has been suggested that

patient compliance with medication prescriptions after hospital discharge should be a

major concern for all hospital staff, who need to play a role in determining whether

discharge medications are used as ordered, whether complications which could lead to

readmission are likely to result from poor compliance and whether adequate measures

have been instituted to maximise compliance.28

In a recently published comprehensive literature review by Vik et al29 examining the

measurement, correlates and health outcomes of medication compliance among the

elderly population, they found that the available evidence suggests that polypharmacy

and poor patient-health care provider relationships may be major determinants of non-

compliance, however there have been few other investigations into other possible

determinants of non-compliance. In recent years the concordance model has begun to

be favoured as an alternative way to view compliance. According to Vermeire et al,26

the concordance model points at the importance of the patient’s agreement and harmony

Introduction

of 767

in the patient-health care provider relationship. The concordance model sets the patient

in the role of decision maker and focuses highly on the importance of the patient-health

care provider relationship model.26 In order to fulfil this role of decision maker, a

patient must be provided with an adequate understanding of the decisions they have to

make. It has been recognised that patient knowledge is a necessity for empowering

patients to perform self-care, especially when dealing with new, ill-defined and

unknown situations.30

In his investigations into improving patient compliance, Rosenow III27 suggested a

number of initiatives that may lead to improved compliance and hence, health

outcomes. Among other things, these suggestions included:

• Educating patients and the public at large about the importance of

compliance should begin in primary school. It is imperative that patients

begin to assume more responsibility for their own health.

• With a new diagnosis, the patient and family could be given information,

written at their level, about the disease or condition.

• A discharge summary for the hospitalised patient written directly for him or

her in a language the patient can understand would be of great value.

8.4 Adverse drug events relating to hospitalisation Unfortunately, once patients have entered the hospital system, they cannot consider

themselves safe from ADEs. Conversely, the process of being admitted to hospital and

returning to the community setting can be one of the most high-risk periods for

incidence of ADEs31-33. These errors can occur at any point during the hospital

admission stay.

8.4.1 On admission to hospital It has been suggested that over a quarter of hospital prescribing errors are attributable to

incomplete medication histories being obtained at the time of admission.32

Tam et al34 reviewed 22 published studies, involving a total of 3755 patients, that

compared prescription medication histories obtained by physicians at the time of

hospital admission with comprehensive medication histories. Errors in prescription

medication histories occurred in up to 67% of cases: 10%–61% had at least one

omission error (deletion of a drug used before admission), and 13%–22% had at least

Introduction

of 767

one commission error (addition of a drug not used before admission). It was concluded

that medication history errors at the time of hospital admission are common and

potentially clinically important. Improved physician training, accessible community

pharmacy databases and closer teamwork between patients, physicians and pharmacists

could reduce the frequency of these errors.

The recently published Canadian study by Cornish et al35 found that medication errors

at the time of hospital admission are common and some have the potential to cause

harm. In their study cohort, 53.6% of patients had at least one unintended medication

discrepancy identified on their admission drug chart. Of these, omission of a regularly

used medication was the most common error (46.4%). However, their findings state

that most of the discrepancies were judged to have no potential to cause serious harm.

Here in Australia, Stowasser et al36 found that a mean of 0.5 +/- 1 medication was

omitted from the admission drug charts of their study cohort.

The other growing concern, besides errors relating to conventionally prescribed

medications on initial hospital drug charts, is the omission of complementary and

alternative medicines. It has been recognised that despite increasing use of

complementary and alternative medicines among those admitted to hospitals, accurate

documentation of usage by hospital practitioners is still low.37, 38

8.4.2 During the hospital stay Similarly, concerns have been raised regarding prescribing errors for hospital inpatients.

The well-publicised 1999 report by the USA Institute of Medicine, To Err is Human39

estimated that between 44,000 to 98,000 people die in American hospitals each year as

the result of medical errors. About 7,000 of these deaths were estimated to be due to

medication error alone. This figure equated to approximately 16% more deaths than the

number attributable to work-related injuries at the time. In 2002, Dean et al40

investigated the incidence of prescribing errors in one UK hospital. Over a 4 week

period, pharmacists prospectively recorded details of all prescribing errors identified in

non-obstetric inpatients. In that time, 36,200 medication orders were written and a

prescribing error was identified in 1.5% of these, a potentially serious error was

identified in 0.4% and the majority of the errors (54%) were associated with choice of

dose.

Introduction

of 767

8.4.3 At the time of hospital discharge The period surrounding a patient discharge from hospital and subsequent return to

community-based care appears to represent the most complicated time for increased risk

of errors. The discharge medication list can contain medication errors, as described in

1997 by Stowasser et al36 They found an average of 1.38 +/- 2.04 medications current

on the medication chart were omitted in error from the discharge prescription.

Similarly, a review by Coombes et al41 in 2001 of discharge prescriptions for 68

patients at one Australian hospital found that 15% of the regular medications intended

to be continued were omitted at discharge.

However, the transfer of patients from the hospital to community setting, if done poorly,

can result in a further breakdown of patient medication management often resulting in

ADEs.42-49 As defined by Coleman50 in 2003, transitional care is “a set of actions

designed to ensure the coordination and continuity of health care as patients transfer

between different locations or different levels of care in the same location”. Medication

errors in particular are widely recognised as a significant hazard.28

This concern has been studied extensively around the world in recent years, all with

similarly alarming results. A nationwide prospective study was performed in France by

Letrilliart et al48 to describe and estimate the incidence and preventability of post-

discharge ADRs. GPs were asked to report all cases of an adverse reaction to a drug

instituted in hospital over a 30 day period post-discharge. The results yielded a minimal

incidence of post-discharge ADR of 0.4 per 100 admissions. Of these ADRs, 60% were

considered serious and 59% of cases considered preventable. Forster et al46 performed a

prospective cohort study in a large Canadian hospital in 2004 of consecutive medical

patients discharged home from a general medical service who were reviewed

approximately 24 days post-discharge. Of the 400 reviewed, 45 (11%) had experienced

an ADE. Of these, 27% were preventable and 33% were ameliorable. Of the ADEs

experienced, 32 were significant, 6 serious and 7 life-threatening. Forster et al went on

to identify the following deficiencies (Table 1):

Introduction

of 767

Table 1 Deficits in the delivery of care, as described by Forster et al46

In the USA, a study performed by Moore et al42 investigating the incidence of

continuity errors during the peri-discharge period found 42% of patients experienced at

least one medication continuity error at some time during the first two months post

initial hospital discharge. Elsewhere in the USA, a comprehensive literature review of

studies examining medication reconciliation at the time of transfer across health settings

stated that considerable evidence indicates the potential for preventable ADEs at

transitional points of care and that an estimated 46% to 56% of all medication errors

occur at such points.45 Stuffken et al51 investigated the nature and frequency of changes

in drug treatment upon discharge from a Netherlands hospital. They found that of all

prescriptions dispensed over a one month period by their outpatient pharmacy, 40% had

some form of discontinuity from the most recent medication discharge list.

Again, Australian hospitals appear to be in a similar situation. The Second National

Report on Patient Safety8 by the Australian Council for Safety and Quality in Health

Care (‘Improving Medication Safety’) concluded the following:

“Patients can leave hospital with a particular medicine, but may experience a breakdown in communication between their specialist and

* Reviewers were asked whether they felt that system problems contributed to the occurrence or severity of the adverse event. If so, they were asked to identify deficits in the delivery of care as described above.

Deficits in the Delivery of Care* Inadequate patient education regarding the medical condition or its treatment

Poor communication between patient and physician

Poor communication between hospital and community physicians

Inadequate monitoring of the patient’s illness after discharge

Inadequate monitoring of the patient’s treatment after discharge

No emergency contact number given to the patient to call about problems

Patient problems getting prescribed medications immediately

Inadequate home services

Delayed follow-up care

Premature hospital discharge

Other

Introduction

of 767

regular GP leading to inappropriate medicines being used. Patients and their carers may be confused by complex instructions, particularly when taking multiple drugs.”8

For example, in developing a new form to enhance continuity of medication use post-

discharge, Parke et al52 found that of the 404 charts reviewed at the time of discharge,

125 (30%) were considered to fit into the ‘at risk’ category for an ADE.

8.4.4 Hospital discharge is a particularly high-risk period for the elderly

A major contributor to inadvertent polypharmacy and resultant ADEs in the elderly

would appear to be hospitalisation and the consequent changes in medication (drugs or

brands of drugs) at the transition from out-patient to in-patient care and back53, 54 In the

study by Parke et al,52 those categorised as ‘at risk’ of medication misadventure at time

of discharge were most commonly aged 65 years or older (34%), receiving more than

five medications (27%) and/or undergoing significant changes to their medications

(25%).

Midlöv et al44 monitored the transfer between the community and hospital setting for a

group of elderly patients taking an average of 10 medications. Out of 758 medication

transfers, 142 medication errors were identified. In this study, the most common error

at hospital discharge was drugs being erroneously added. It would appear that ADEs

continue to be a problem post-discharge. In a study of elderly patients receiving home

health services, Gray et al53 determined that self-reported adverse drug events were

common during the month following hospital discharge.

In a recently published study by Coleman et al49 investigating post-discharge

medication discrepancies in patients aged 65 years or older, a comprehensive

medication assessment in the patient’s home between 24 and 72 hours post-discharge

was performed by a nurse practitioner. It was found that 14.1% of patients experienced

one or more medication discrepancies. 50.8% of identified contributing factors for

discrepancies were categorised as patient-associated and 49.2% were categorised as

system-associated. A total of 14.3% of the patients who experienced medication

discrepancies in the study period were readmitted within the first 30 days post-discharge

compared with 6.1% of patients who did not experience a discrepancy.49

Introduction

of 767

8.4.5 Peri-discharge adverse drug events can lead to unplanned readmissions

Unplanned readmissions to hospital have a substantial impact on health care costs55, 56

and it has been found that approximately 15% of older patients have an unplanned

readmission.55 Locally in 1999, a follow-up study was performed of over 500 medical

patients discharged from the RHH aged over 60 years and taking two or more regular

medications.57 The unplanned admission rate within 6 months of initial discharge was

33.8% and 8.4% of these readmissions were noted in the medical record as being drug-

related.57

Between 9% to 48% of all readmissions have been judged to be preventable because

they were associated with indicators of substandard care during the original admission,

such as poor resolution of the main problem, unstable therapy at discharge, and

inadequate post-discharge care.56 Many studies have attempted to identify precise

factors that may lead to patient readmission. Problems with drug therapy have

previously been identified as a major cause of readmission to hospital55 and it has been

found that the majority of the drug-related problems identified in readmission studies

are potentially preventable and the types of problems found have indicated that

interventions should be focused on both hospital staff and patients.58

In a review of randomised prospective trials, Benbassat et al56 found that 12 to 75% of

all readmissions can be prevented by patient education, pre-discharge assessment and

domiciliary aftercare.56 Poor compliance with therapy has also frequently been

identified as one cause of hospital readmission, resulting in 17 to 48% of all drug-

related readmissions.28

8.4.6 Challenges and obstacles relating to transitional care - The current state of affairs

In the 2003 paper by Coleman exploring challenges and opportunities for improving

transitional care,59 he categorises the barriers to effective care transitions into 3 levels –

the delivery system, the clinician and the patient. He goes on to highlight that when

patients are discharged from hospital, they may be uncertain about whether they should

resume their previous medication regimen or only take the medications listed on their

discharge instructions.59 From a peer to peer point of view, other potential problems can

arise when the sending clinician fails to communicate critical elements of the care plan

to the receiving clinician or when patients are not adequately prepared for care in the

Introduction

of 767

next setting (including being informed about their care plan, what to expect in the next

setting, playing an active role in determining their care plan and expressing their care

preferences.).59

Here in Australia, the importance of communication on all levels has also been

identified as essential in ensuring continuity of care at time of hospital discharge. In

1998, The Australian Pharmaceutical Advisory Council (APAC) produced the National

guidelines to achieve the continuum of quality use of medicines between hospital and

community.60 The Guidelines consisted of 7 principles which were as follows:

1. It is the responsibility of the admitting institution to ensure the development and coordination of a medication discharge plan for each patient. The person responsible for coordinating the development, implementation, and monitoring of the medication discharge plan, including medication supply and medication information, should be identified as soon as practicable after admission.

2. Hospital staff should obtain an accurate medication history, including prescription and over-the-counter medicines and other therapies such as herbal products, at the time of admission.

3. Hospital staff should evaluate the current medication at the time of admission, in consultation with the patient’s general practitioner, with a view to:

a. Identifying the appropriateness and effectiveness of current medication, and rationalising current medications if appropriate;

b. Paying particular attention to any problems associated with current drug therapy including any possible relationship with the current medical condition; and

c. Documenting allergies and any previous adverse drug reactions.

4. During the hospital stay, treatment plans relating to the probable medication management during the stay and where applicable at discharge should be developed in consultation with the patient and/or carer. Hospital staff should negotiate with the patient issues relating to treatment and the development of a discharge plan, and these discussions should be documented in the patient’s notes. This plan should form part of the overall care plan or critical pathway.

a. The use of interpreters may be required to ensure good communication with people from non-English speaking backgrounds.

b. To enable the discharge process to be successful, there needs to be effective communication and coordination between all relevant parties in the hospital environment.

c. Where appropriate, community health providers, especially the patient’s general practitioner, should be consulted.

d. Carer should also be consulted where appropriate.

5. Prior to discharge, pre-discharge medication review and dispensing of adequate medication should take place in a planned and timely fashion. Adequate medication means sufficient medication to carry the patient through to the next

Introduction

of 767

arranged review (by their general practitioner, outpatient clinic or some other arrangement), or to complete the course of treatment. If patients are discharged with inadequate supplies of medications, this can compromise quality of care for the patients. Supply of the medication from the hospital facility must be adequate to ensure continuity of medication is not interrupted by the inability to obtain further ongoing supplies if required, within a reasonable timeframe.

6. At the time of discharge, each patient should be provided with a discharge folio containing relevant information such as Consumer Medicine Information, a medication record, patient/carer plan, and information on the availability and future supply of medication.

7. No patient should be discharged from hospital until the details of the admission, medication changes (including additions/deletions) and arrangements for follow up have been communicated to the healthcare provider(s) nominated by the patient as being responsible for his or her ongoing care.

Unfortunately, the implementation of the APAC Guidelines has not been overly

successful. In 2001, Mant et al61 conducted a survey of GPs to evaluate compliance

with the APAC Guidelines, particularly focusing on how hospital information was

communicated to GPs. Interestingly, Mant et al found that compliance with the

Guidelines was not good at that time. 106 GPs answered questionnaires about the type

of information they had received from the hospital for 203 of their patients. In only 22%

of cases did the hospital directly notify the GP of the patient’s admission. In 27% of

cases the patient notified the GP, while in the remaining 52% of cases there was no

notification given to the GP. A change to the patient’s medicines was made in hospital

in 87% of the cases, with the patient’s medicine at discharge differing from what the GP

understood the patient to be taking before they went to hospital in 72% of cases.

Consultation with the GP about the patient’s medication during the hospital stay

occurred for only 11% of all patients. The mean time taken for GPs to receive the

discharge summary from the hospital was 3 days, with a maximum of 21 days.

A survey of Australian hospital pharmacy departments performed by O’Leary et al62 in

2003 found similarly that the Guidelines had only been partially implemented in

Australian Hospitals. Although responses varied, only 3 of the 101 hospitals surveyed

indicated that their hospital had fully implemented all of the 7 principles of the

Guidelines.62 In July 2005, after an extensive review period, considering findings such

as O’Leary et al’s, the APAC Guidelines were revised and re-released as a set of 10

Guiding Principles to Achieve Continuity in Medication Management63. They are titled

as follows:

1. Leadership for medication management

Introduction

of 767

2. Responsibility for medication management

3. Accountability for medication management

4. Accurate medication history

5. Assessment of current medication management

6. Medication Action Plan

7. Supply of medicines information to consumers

8. Ongoing access to medicines

9. Communicating medicines information

10. Evaluation of medication management

In a study in 2001, Wilson et al64 found that GPs in the Macarthur Health Sector of New

South Wales only received summaries from the hospital for 27.1% of 569 discharged

patients. Of more concern was that 36.4% of discharge summaries contained

information that did not reflect the information recorded in the hospital notes. These

inaccuracies included medication (17.5%), clinical (17.3%), follow-up (14.4%) and

clerical (2.5%) inconsistencies. Medication errors included incorrect medications

recorded, medications omitted from the summary, and omission of dose or frequency.

The same study showed that the recording rate of medications on discharge was 79.3%.

This suggests that 21% of the summaries contained no indication of whether there were

any variations in the existing medications or indeed any medications at all. Wilson et

al64 suggested that perhaps junior medical officers may not be the ideal authors for

writing the summaries due to their high turnover and that nursing and allied health

professionals could contribute to the partial or total production of the summary, which

may improve the quality and accuracy of the information.

The Second National Report on Patient Safety8 by the Australian Council for Safety

and Quality in Health Care (‘Improving Medication Safety’), published in 2002,

concluded the following:

“It is essential for patient care that information about a patient’s medicines is communicated to the hospital when the patient is admitted and back to their community health professionals when they are discharged. Poor communication at the time of discharge from hospital or errors in prescribing or transcribing at discharge can contribute to medication incidents. ……….. Accurate, well-timed transfer of information between hospital and community settings is important for ensuring appropriate medication use. This includes information about a patient’s current medicines, any changes that were made in hospital and why, as well as allergies and relevant medical history. This information

Introduction

of 767

needs to be transferred between hospital and community pharmacies, as well as between hospitals and general medical practices.”8

However, it is clear this situation is not unique to Australia. In the UK, an audit of

hospital discharge and outpatient information from a primary care perspective found a

similarly concerning situation.65 Information required for appropriate prescribing post-

discharge was missing from 11% of discharge advice notes and 15.2% of discharge

letters. Discharge advice notes and discharge letters were both available when needed

in only 8.1% of cases. Even more alarmingly, comparison between discharge advice

notes and discharge letters showed discrepancies in one-third of cases. Important

information was missing from 25% of out-patient letters and 96.1% of discharge advice

notes arrived within 8 days of discharge. The results of a national survey in the UK

aimed at identifying services that hospital pharmacies were providing in 1999 to

facilitate seamless care at discharge66 were also not encouraging. They found that

hospitals used a wide variety of methods to communicate information about medicine

regimens to GPs and patients, and few hospitals involved community pharmacies

routinely in the discharge process.

8.5 Some strategies to improve the situation

8.5.1 Medication reconciliation at time of discharge In a randomised controlled trial of a Medication Liaison Service (MLS) by Stowasser et

al,67, 68 one of the services provided was a full review of the medication profile during

hospital and immediately prior to discharge. A discharge medication summary was

produced by the MLS pharmacist, in conjunction with the Resident Medical Officer

(RMO) to ensure accuracy and appropriateness of therapy. During the hospital

admission it was found that a significantly greater proportion of MLS subjects

experienced a clinical pharmacist intervention (MLS 68% versus control 44%) and

more subjects had at least one change to their therapy (MLS 97% versus control 90%).

In the 30 days following discharge there was a tendency for fewer readmissions per

subject in the MLS group.

8.5.2 Production of a clear and concise discharge summary

Van Walraven et al69 performed a study aiming to determine what physicians perceived

to be necessary for high-quality discharge summaries. Quality was perceived to

Introduction

of 767

decrease significantly when summary length exceeded 2 pages and when the delay from

patient discharge to summary delivery increased. Conversely, summary content that

was considered to increase quality the most included, among other factors, a list of

discharge medications.

Pharmacy initiated discharge plans have also been successfully trialled. In a UK study

by Norris et al70, pharmacists trialled a process of electronically prescribing discharge

medication and providing information to GPs. They found that the introduction of a

pharmacy discharge letter had enabled the pharmacist to transfer discharge medication

information electronically and provide reasons for all medication changes during an

inpatient stay. GPs were surveyed regarding satisfaction with the service and 98% of

respondents agreed or strongly agreed with the benefits of the letter.70 The investigators

went on to suggest that the letter may be useful in informing community pharmacists of

discharge medication and as a patient medication card.70

Here in Australia, Parke et al52 developed and trialled a form to enhance the continuum

of medication from admission through to and beyond discharge. When trialled by the

research pharmacist, a marked increase in the quality of medication management was

found.52

8.5.3 Improved communication with community health providers

It is traditional for hospitals to communicate only with GPs at the time of patient

discharge and this is continuing to be the case. A national survey in the UK in 1999,

aimed at identifying services that hospital pharmacies were providing to facilitate

seamless care upon patient discharge, found that few hospitals involved community

pharmacies routinely in the discharge process.66

However, in their trial of a medication liaison service, Stowasser et al67, 68 demonstrated

that when a hospital pharmacist communicated via facsimile within 24 hours of

discharge to both a patient’s GP and community pharmacist, with a detailed summary of

medication issues, there was a tendency for a reduction in readmissions within 30 days

of discharge and there was a significant decrease in community healthcare professional

visits (e.g. GP, domiciliary nurse, community pharmacist, medical specialist). The

intervention’s discharge communication contained information related to the following:

Introduction

of 767

• list of medications on admission and discharge;

• corresponding therapy changes;

• intended duration of therapy, allergies and adverse drug reactions;

• new therapeutic devices; and

• medication-related problems and action required by the general practitioner.

The Australian study by Wilson et al64 also suggested that the preferred method for

delivering the discharge summary was via facsimile, combined with giving a copy to

the patient, to ensure that if the patient visited a general practitioner other than that

nominated at admission, then appropriate information could be passed on. It was also

stated that in 2001, email is still not a preferred way of improving transmission of

information, as many general practitioners still have not embraced this method of

communication.

A study in 112 community pharmacies across Europe was performed to review

community pharmacist interventions to detect, prevent and solve drug-related problems

(DRPs) after discharge.71 It was found that DRPs were identified in 63.7% of patients

investigated. Of these, the most common were a lack of knowledge or understanding

regarding the medication they were taking (29.5%) and side effects (23.3%). Practical

problems were identified by 12.4% of patients and in 24% of cases, pharmacists

identified potential ADEs.71 This study is a good illustration of the importance of the

role of the community pharmacist at the time of discharge from hospital.

In their survey of patients regarding their satisfaction with medication counselling in

hospital, Robinson et al72 found that 56% of patients expected to speak to their

community pharmacist after discharge regarding their medications. If this is the case

generally, access to the discharge medication information shortly after discharge would

facilitate this important role of the community pharmacist.

A study by Paquette-Lamontagne et al73 in Canada found the provision of a single form

including admission medications, in-hospital changes and discharge medications

assisted in increasing the conformity rates of community pharmacy patient profiles after

hospitalisation. They found that the form was well accepted by both pharmacists and

physicians, and stated that it may lead to a major decrease in drug-related problems.73

Introduction

of 767

8.5.4 Discharge medication counselling Providing patients with adequate education regarding their medications is likely to

improve their understanding and knowledge of and compliance with their

medications.24, 74 Kerzman et al30 examined factors that influence patient knowledge of

their medications at 1-2 weeks post-discharge. It was found that the only factor that

significantly influenced patient knowledge was whether or not they received medication

counselling during hospitalisation.30

Similarly, Al-Rashed et al75 conducted a randomised controlled trial in the UK in 2002

involving elderly patients who were taking four or more medications and were being

discharged to their own homes. A program was implemented for the intervention

patients involving inpatient discharge medication counselling, provision of a medication

information discharge summary and a medicine reminder card. Patients were followed

up at home 2-3 weeks post-discharge and then again at 3 months post-discharge. It was

concluded that the program contributed to better drug knowledge and compliance

together with reduced unplanned visits to the doctor and readmissions over a 3 month

period post-discharge.75 Improvements in compliance with and knowledge of

medications were seen most prominently in the initial interview 2-3 weeks post-

discharge.75 Louis-Simonet et al76 did a similar trial in Switzerland and found similar

improvements in patient knowledge of their medications, as measured via a telephone

call one week post-discharge, for those who received a patient-centred discharge

interview by residents, using a standardised treatment card.76

Not only does discharge medication counselling have the capacity to improve patient

medication management, it has also been found to improve patient satisfaction as found

in a study by Robinson et al72 Inpatients’ satisfaction with clinical pharmacist

counselling was measured via interview. Patients interviewed were found to be either

satisfied or very satisfied with counselling.72

8.5.5 Post-discharge follow-up telephone calls In the study performed by Fallis et al,77 home visit and telephone call follow up after a

specific surgical procedure were compared. It was found that patients who received a

follow-up telephone call from a registered nurse to discuss patient concerns and identify

any post-surgical management issues were significantly more satisfied than those who

Introduction

of 767

received the home visit. Home visits are also not always a practical option, particularly

in regional or remote areas.78

As stated by Nelson, telephoning patients within a few days or weeks of discharge can

be an effective way to follow up a patient post-discharge.78 A number of previous

studies have been performed proving the usefulness of a post-discharge telephone call to

a patient, but these studies have all involved nursing staff.77, 79 There have apparently

been no studies involving pharmacists performing a follow-up phone call,80 except for

the randomised controlled trial performed by Dudas et al80 This study analysed whether

pharmacists can improve patient satisfaction and outcomes by providing telephone

follow-up after hospital discharge. The intervention consisted of a follow-up phone call

by a pharmacist two days after discharge. During the phone call, pharmacists asked

patients about their medications, including whether they had obtained and understood

how to take them. The follow-up by phone call was associated with increased patient

satisfaction, resolution of medication-related problems and significantly fewer return

visits to the emergency department.80

8.5.6 Access for patients to medication information via the internet

The use of information technology has been investigated for its potential role in

improving patient safety for some time now, however, despite great leaps in the

technology available, it is a widely held belief that it could do much more.81-87 In a

2001 paper published by Prady et al,88 it was stated that:

“Interest in the use of other Internet-based tools, such as the World Wide Web, to enhance clinical communication is increasing. In such systems, static information can be made centrally available to patients and interactive tools such as messaging systems, schedules, and individualised care regimens can be integrated within the site.”88

Since then, a number of groups have attempted to do just that. In the USA, the

institution of David Bates, (Brigham and Women’s Hospital in Boston; world leaders in

the prevention of medication errors,84, 85) has introduced a secure Internet site (Patient

Gateway89 for patients and health care professionals. It is aimed at improving the flow

of communication and facilitating seamless care upon hospital admission and discharge.

Patients are able to access comprehensive health information, details of their own

prescribed medications and email questions to health care providers.

Introduction

of 767

In 2002, here in Australia, Calabretto et al86, 87 implemented an Internet Medicine

Cabinet named Winston. They reported that the information needs of patients and

pharmacists must be supported as the trend toward over the counter (OTC) based

therapies grows more popular. It was their belief that providing patients with secure on-

line access to a personal electronic medication record that could be updated and

accessed by patients, community pharmacists and hospital pharmacists for both

prescribed and OTC medications could improve communication. Patients could also

raise issues via the medication record screen and maintain lists of prescribers, allergies,

diagnoses and hospital visits.

Here in Tasmania, a similar internet communication tool called MediConnect was

trialled by the Commonwealth Department of Health and Ageing (CDHA),

(http://www.mediconnect.gov.au), in 2002-2003. The objective of the program was to

draw together information about patient’s medications and hence, ensure that health

professionals across the settings are better able to help prevent ADEs. The trial was

considered successful by the CDHA and is now being incorporated into HealthConnect,

which, as per its current website, (http://www.healthconnect.gov.au/), is “a trial network

of electronic health records that aims to improve the flow of information across the

Australian health sector. It involves the electronic collection, storage and exchange of

consumer health information via a secure network and within strict privacy safeguards.”

The first trial of HealthConnect in Tasmania ended in November 2004, and the next trial

is currently in the planning stage.

8.6 Local example of need for a medication liaison service

Currently around 40% of Tasmanians aged 65 or over are older than 75 years. By 2011

it is estimated that Tasmania will have more people over 65 years per head of

population than anywhere else in Australia.

Clinical Pharmacists at the RHH aim to provide services based on the APAC Guiding

Principles. At the time of this study’s implementation, clinical pharmacists routinely

interviewed patients regarding their medications on admission, reviewed medication

profiles for their patients while they were inpatients, and for those they considered at

risk, they reviewed their discharge medication lists and provided discharge medication

counselling and a counselling sheet where possible. At the time, approximately 60% of

Introduction

of 767

patients considered at ‘high-risk’ of potential adverse drug events were provided with a

discharge medication counselling sheet and counselling. It was not common practice to

transfer this information to GPs or community pharmacists unless specific requests were

made.

The usefulness of providing admission histories in order to ascertain accurate

medication histories has been shown locally. A study of 200 elderly patients undergoing

surgery at the RHH revealed inconsistencies for 57% of patients between the medication

history (drug therapy being taken prior to presenting to hospital) obtained by the

admitting doctor and that obtained after a thorough review by a clinical pharmacist.90

That study did not examine medication management issues at discharge from hospital

and thereafter, which the current study has focused on.

A recently concluded study at the RHH examined medication management amongst a

cohort of older, chronically ill patients discharged from hospital.91 Interventions

included a home visit at 5 days after discharge from hospital by a pharmacist, who

educated patients on their drug therapy, promoted compliance with therapy, assessed

patients for drug-related problems and intervened when appropriate and communicated

relevant findings to their community-based health professionals. All patients were

visited at 90 days after initial discharge from hospital to assess outcomes. Key findings

were:–

• The quality of medication usage was relatively poor, with a median of 3 drug-related problems identified per patient in the intervention group at 5 days post-discharge;

• Almost 39% of patients in the intervention group in this study had either taken more or omitted some of their medication within 5 days of their hospital discharge;

• The unplanned readmission rate at 90 days after initial discharge from hospital was 45% in the control group and 28% in the intervention group;

• Over 20% of patients had experienced difficulty obtaining a continued supply of discharge medication when visited at 5 days post-discharge; and

• Conversely, over 10% of patients were found to be taking discontinued medications at 5 days post-discharge.

The study demonstrated that local medication use post-discharge among the elderly

population is poor and that a pharmacist-conducted follow-up of hospitalised high-risk

patients at home is valuable in identifying and addressing DRPs in high-risk medical

patients, and significantly reducing the risk of unplanned readmission to hospital within

Introduction

of 767

90 days. Although this study demonstrated some excellent improvements in medication

outcomes, continuation of such a service is not feasible in the current setting without

additional funding. A more practical and cost-effective form of follow-up needs to be

explored.

During the study be Naunton et al91 it was common for local GPs to mention to the

study pharmacist that there was poor communication from the hospital. Without

prompting or questioning, approximately 20% of the patients’ general practitioners

made direct comments stating they were unhappy with the current discharge

arrangements and appreciated the pharmacist contacting them to inform them of

changes in their patient’s drug regimen.

8.7 Study aims and objectives The research funded under this Tender was to develop a comprehensive package

approach to the problems associated with medication management during the transition

between the hospital and community health care sectors, including ICT solutions. In

particular, the study was conducted to address the major issue of poor medication

management, including non-compliance, and adverse outcomes in recently hospitalised

‘high-risk’ patients. Further, improvement in the transfer of medication information

between health care providers in the hospital and community settings at the time of

patient admission to, and discharge from, hospital was targeted.

The multifaceted program was to be implemented and trialled at sites in Western

Australia, Victoria and Tasmania, with a comprehensive analysis of the results of the

trial.

Introduction

of 767

The specific objectives were to examine the following:

• The role of the community pharmacist in improving QUM outcomes through involvement in pre-admission activities,

• The development of an effective model to ensure that patients that would benefit from a HMR post-discharge are identified (hospital) and provided the service (community) within an appropriate time,

• The model should ensure timely and effective communication between hospital discharge staff, the patient’s GP and community pharmacist,

• Fast-tracking HMR and appropriate monitoring by community pharmacists for patients discharged on warfarin,

• The potential role of MMR facilitators in an extended MMR service model to improve continuity of care, and

• Improving communication between hospital- and community-based doctors and pharmacists, other care providers, patient and carers.

Specifically, the desired outcomes of the Med eSupport project were expected to

include the following:

• Better correspondence between the list of prescribed medications during hospitalisation with the actual medications being taken by the patient immediately prior to admission,

• Improvements in QUM by patients,

• Improved provision of medication-related information and follow-up of patients after discharge from hospital,

• Improved communication of discharge medication-related information between the hospital and community-based health professionals, ensuring continuity of treatment to promote patient care at the time of discharge from hospital,

• Improved patient compliance,

• Improved patient knowledge of their medications,

• Less medication-related adverse events following hospital discharge, and

• Favourable acceptance of the program by patients, community pharmacists and GPs.

• Demonstrate the benefits of home follow-up visits on patient outcomes, targeting newly initiated warfarin patients as a specific example.

Med eSupport trial and participant surveys

of 767

9 Med eSupport trial and participant surveys 9.1 Methodology

9.1.1 Methodology of the Med eSupport trial

9.1.1.1 Routine supply of a comprehensive medication information sheet to the patient/carer prior to discharge from hospital

Some hospitals, including those participating in the trial, provide patients with

computer-generated customised profiles of their discharge medications, such as that

shown below (Figure 4). Various in-house programs are used in different hospitals,

including Med Profs® at BHCGH and the LGH and Pharmcare® at the RHH, Stocca® at

HPH and the Clinical Governance System® at SCGH.

Figure 4 Example Pharmcare® counselling sheet

The Project Team believes that these sheets should be routinely provided to all patients

on discharge from Australian hospitals. For those trial sites that did not have an in-

house counselling sheet generating program, the counselling sheet generation facility

within the Cognicare® Medication Management System (CMMS) was utilised for the


of 767

duration of the trial. Utilising a program that was based on standardised information

was deemed a high priority and CMMS is based on an Australianised version of the

First DataBank database (http://www.firstdatabank.com), a standardised drug

information database originating from the USA and an international leader. First

DataBank creates and maintains some of the largest and most comprehensive drug and

healthcare knowledge databases in the world.

The production of counselling sheets, discharge medication summaries, and weekly

checklists on the Med eSupport website was a vital aspect of the website’s functionality

for patients and providers. Initially, the Pharmcare® product was to be used to produce

the sheets, and transmit the data to the repository. It was anticipated that Pharmcare®

would require minimal modification; however, this did not prove to be the case. The

product proved to be unsuitable for a number of reasons; therefore, the project team

invested time into seeking an alternate solution. A decision was taken to seek approval

from Cognicare® to use the First DataBank database to supply clinical drug information

for use in the project. It was decided that the project would develop its own web-based

counselling sheet and weekly checklists, integrated into the Med eSupport web

interface. This required significantly more work than what would have originally been

required through the incorporation of an existing product.

Extra functionality has been included that allows pharmacists and other health

professionals to produce counselling sheets and weekly checklists directly from the Med

eSupport website. The counselling sheets include advice notes for patients. To achieve

this, each medication had to have its own advice note generated by the website. The

information came from Cognicare, where available and if Cognicare® did not have

advice notes, information from the APF 19th Edition92 was used instead. This ensured

that the information provided was standard for all patients receiving the datasheet. The

indication for the drug was entered by the project pharmacist or health provider.

MIMS Australia provided permission for their electronic drug database to be utilised in

developing an interface simplifying the entry of drug names into the repository and in

the provision of CMI leaflets.

At the time of discharge, every intervention patient was provided with a counselling

sheet and verbal counselling by a pharmacist. Bearing in mind the situation described

above, it was decided that four possible processes could be implemented to achieve this:


of 767

1. Counselling by a hospital pharmacist and provision of a medication information sheet using the hospital’s own system (providing it had all the required elements to ensure trial patients were receiving uniform information),

2. Counselling by a trial officer and provision of a medication information sheet using the hospital’s own system,

3. Counselling by a hospital pharmacist and provision of a medication information sheet using CMMS or the Med eSupport website, and

4. Counselling by a trial officer and provision of a medication information sheet using CMMS or the Med eSupport website.

9.1.1.2 Automated faxing of the medication information sheet to the patient’s general practitioner and community pharmacist at discharge

The objective of this facet of the project was to ensure that information on patients’

discharge medication was delivered in a safe, efficient, timely and user-friendly manner,

in order to promote ‘seamless care’ across the interface between primary and secondary

care. For each intervention patient, a copy of the discharge medication summary,

produced on the Med eSupport website at time of discharge, was faxed, along with

explanatory information as needed, to their regular GP and CP within 24 hours of the

patient’s discharge from hospital.

The nominated community health providers were also advised that their patient had

nominated them to be given access to the Med eSupport website and they were provided

with brief instructions and a username and password to access the website. This access

allowed them to view the patient’s community pharmacy dispensing history, discharge

medication information and to make further changes as they arose, as previously

described.

Each nominated provider was telephoned at the time of discharge to re-introduce them

to the project and verbally discuss the information prior to the fax being sent. The aim

was to prevent misunderstanding or misinterpretation of the new service and to further

bolster communication. This aspect was considered a function of the trial and would not

necessarily be included in an implementation strategy of the project.

For those patients in the intervention – streamlined HMR recommendation group (see

below), the discharge medication summary produced by the Med eSupport website was

designed to double directly as a pre-filled HMR referral form for the GP - so if they

desired, they only had to sign it and send it on to the community pharmacy to initiate

the process of the conventional HMR. Examples of these letters and discharge


of 767

summaries can be found in Appendix X, along with examples of the information sent to

the GP in Appendix IX and CP in Appendix VIII on admission from the trial

office.Appendix I

9.1.1.3 A model whereby suitable patients are referred for a HMR after discharge from hospital

A model was developed that involved and reimbursed both community and hospital

pharmacists in the provision of timely medication reviews for patients discharged from

hospital. Our previous study demonstrated the usefulness of a pharmacist in assisting

with the management of medications after discharge from hospital, resulting in a

reduction in drug-related problems and unplanned readmissions.93 We concluded that:

“A pharmacist following-up high-risk medical patients at home resulted in a reduction in drug-related problems and unplanned readmissions. The outcomes suggest that this sort of program merits a more widespread application in Australian hospitals. More assistance needs to be given to patients, particularly those at ‘high-risk’ with multiple medications and several chronic medical conditions, when they are discharged from hospital. An in-home pharmacy assessment reveals many problems with drug administration not otherwise detected easily. These assessments can lead to potentially useful interventions that can improve medication regimens and patient compliance. These programs may be particularly cost-effective if applied selectively to patients with a history of frequent unplanned hospital admission. Ideally, the present Home Medication Reviews (HMR) scheme would enable and fund hospital pharmacists to perform, or assist in, HMRs for ‘high risk’ patients soon after discharge from hospital. Community-based pharmacists would find it difficult to perform these types of review in isolation and without having access to comprehensive information on the patient’s hospitalisation.”93

Elsewhere, we have stated the following based on our research.60

“In the previously reported study, a median of 3 medication problems was identified in the intervention patients who were visited at home by a pharmacist at 5 days post-discharge. Drug-related issues included continuation of a ceased medication or not taking a newly commenced medication, drug-interactions, difficulty obtaining a continued supply of medication, and duplication of drug intake. Communication flow between hospital and community sectors was generally poor. Many of the problems could not be addressed without access to detailed information on the patient’s hospitalisation. Information that may not be readily available to a community pharmacist performing a home-medication review may include laboratory results or reasons why medications were ceased or why a particular medication was started in preference to another. Hospital pharmacists can comment


of 767

on medication changes because of their access to medical records and their involvement in the patient’s care during the hospitalisation. In addition, hospital pharmacists are ideally placed to comment on the need and frequency for therapeutic drug monitoring. To ensure quality use of medicines, hospital-based pharmacists should be routinely used or consulted in the provision of home medication reviews following the hospitalisation of “high-risk” patients. The current model for the provision of these reviews should be expanded to involve and remunerate hospital pharmacists.”

Our experience indicates that both community and hospital pharmacists need to have

input into the post-discharge follow-up process and that is the model that was developed

for the trial. “Recent discharge from a facility/hospital (in the last 4 weeks)” is

currently a recognised risk factor known to predispose people to medication-related

problems and is one of the eligibility criteria of patients for HMR, but it is not well

promoted and utilised.

A recent report of the Medication Alert Project, funded by the Department of Human

Services of Victoria94 , has endeavoured to promote Home Medicines Review (HMR)

after discharge from hospital for high-risk patients. In this report, a section titled “What

have been the factors that have hindered the project” had direct implications for the

success of the Med eSupport project. It reports that “The reluctance of general

practitioners to instigate HMRs - this remains a frustration despite close working with

HMR facilitators.”

The disappointing aspect of the Medication Alert Project was that only 10% of patients

referred for an HMR received a referral from the general practitioner to the patient’s

community pharmacy. If this were the case in our intervention arm, despite promotion

of HMRs, we would likely have in the order of only 40 reviews conducted in our

intervention group. This is simply not enough patients to evaluate, given the likely rate

of hospital readmissions in the intervention (10%) compared to the control arm (20%).

With this in mind, the project team believed it was essential that we studied the

effectiveness of the process of medication review after discharge, and we would run into

difficulty if we relied on the current procedures, even with their active promotion, for

HMRs to evaluate the success of our project.

To overcome this potential problem, a system of four randomised groups was instigated.

This incorporated an additional arm in the intervention group in which all enrolled


of 767

patients received an automatic post-discharge medication review (PDMR) as shown in

Figure 5.

Figure 5 Flow chart indicating recruitment and randomisation procedure

The PDMR was performed, wherever possible, within seven days post-discharge to

ensure timely discovery and resolution of any post-discharge issues. At the time of

discharge for a patient in the PDMR group, the nominated Community Pharmacist was

contacted to ascertain if they had an accredited pharmacist available to perform the

PDMR. If possible, this was organised to ensure continuity of care and that the process

reflected, wherever possible, the real life process. If the Community Pharmacist did not

have an accredited pharmacist they could call on, one was organised by the project

team. Accredited Pharmacists, CPs and GPs were all remunerated the same amounts by

the Med eSupport project as they would receive under the current HMR model. The

hospital pharmacy departments were reimbursed $20 for the provision of clinical

Control Group Intervention Group

Control Group -HMR referral

Control Group-No HMR referral

Intervention Group – PDMR

Intervention Group- Streamlined HMR referral


Identified as suitable by HP or TO

Yes

No Excluded

Randomised

Trial explained and informed consent obtained

Yes

No

Collect RMO’s initial Drug Chart

Patient interviewed to obtain their account of the medications they were taking at time of admission.

Excluded


Control Group -HMR referral

Control Group-No HMR referral

Intervention Group – PDMR

Intervention Group- Streamlined HMR referral



Yes

No Excluded

Randomised


Yes

No



Excluded


of 767

information related to the hospital admission that assisted with completion of the

review. Also, Community Pharmacists were remunerated $5 for the provision of each

patient dispensing history, either via fax or direct upload.

This model fulfils the tender guidelines by assessing two methods of promotion of

HMRs. Firstly, the control group was split into two - with one half, randomly selected,

receiving no interventions and being therefore, purely control. The other half had a

sticker (Figure 6) placed on their hospital discharge summary as a process of alerting

the GP to the potential need for an HMR. This information could also be passed to the

CP if the hospital staff sent the discharge summary on to them, however, in this model,

there was no active pursuit of the recommendation, making it a purely passive alerting

system.

Figure 6 Example Med eSupport HMR trial sticker

There was also a more active streamlined process of HMR alerting and referral in the

second arm of the intervention group. This included telephone contact with the

community providers at discharge and the faxing of a discharge medication summary to

the GPs (and CPs), that could double as a pre-filled HMR referral form (see Appendix

X). If the GP wanted to initiate an HMR, they could do so by simply checking the

information provided by the hospital, signing the form and sending it on to the patient’s

preferred Community Pharmacy. This HMR referral form could also be downloaded

from the Med eSupport website by the GP who was also given permission by the patient

to access their information.

It was envisaged that the small number of patients who would receive an HMR in these

two groups would allow evaluation of the process of referral, but it was unlikely that the

number of reviews successfully completed would allow formal statistical analysis

across the groups.

9.1.1.4 Implementation and evaluation The Med eSupport trial was a randomised controlled trial based at five hospitals

including the Royal Hobart Hospital and Launceston General Hospital in Tasmania, Sir


of 767

Charles Gairdner Hospital and Hollywood Private Hospital in Perth, Western Australia

and the Bendigo Hospital Campus, Bendigo Health Care Group in Victoria. Ethical

approval was obtained at each site.

Patients were eligible for entry to the trial if they were admitted to medical units at each

of the hospitals, were aged 50 years or older, had at least two chronic medical

conditions requiring drug therapy (including at least one of: cardiovascular disease,

chronic obstructive airways disease or diabetes mellitus) and were prescribed three or

more regular medications (‘high-risk’ patients). These criteria are very similar to those

for guiding eligibility of patients for HMR. Patients who live in a aged or residential

care facility or who are unwilling or unable to provide informed consent were excluded

(numbers were recorded).

The evaluation of the program included a randomised, controlled trial design. At each

site, consecutive patients fulfilling the high-risk criteria and who provided informed

consent were randomly allocated (using pre-determined, blocked allocation

concealment) to one of the four groups previously described. The project design was

very similar to the Team’s recent randomised, controlled evaluation of a pharmacist

follow-up of ‘high-risk’ patients at home after discharge from the Royal Hobart

Hospital.93 The design was also very similar to a recent major Australian study of the

effectiveness of case management and post-acute services in older people after hospital

discharge.95

Patients in the intervention group entered the multi-faceted program to improve

medication management between the community and hospital sectors. Patients in the

control group received routine care according to each hospital’s ‘normal’ admission and

discharge procedures. All patients received their routine medical and nursing care as

hospital inpatients.

The multi-faceted program for the intervention group complied with the Australian

Pharmaceutical Advisory Council’s guidelines on the continuum of quality use of

medicines between hospital and the community,96 and the principles of quality use of

medicines.97 It possessed the elements outlined previously and listed below.

1. An electronic communication pathway for medication information between

community and hospital pharmacies.


of 767

2. Verbal counselling and routine supply of a comprehensive

medication information sheet to the patient/carer prior to discharge from

hospital.

3. Automated faxing of a project generated discharge medication summary to the

patient’s general practitioner at discharge.

4. A model whereby suitable patients were referred for a HMR after discharge

from hospital.

The follow-up of patients initiated on warfarin was evaluated separately, as this is likely

to be sustained as a professional pharmacy service in its own right.

The intervention and control groups were statistically compared with regard to

demographics, quality of life and clinical variables such as age, gender, inclusion

criteria satisfied, length of hospital stay for index admission, number of chronic medical

conditions and number of regular medications, to ensure the groups were similar at

baseline.

Outcome measures collected for the four study groups included the following:

1. Whether a HMR or PDMR was performed.

2. Self-reported compliance to prescribed medication, as assessed using a validated

questionnaire (Morisky Compliance Questionnaire).99-101 The term

‘compliance’ is used throughout this report. For the purposes of this report, this

is considered synonymous with the term ‘adherence,’ which is also commonly

found in the literature.

3. Knowledge of prescribed medications, using the questionnaire designed by Al

Rashed et al98 Compliance and knowledge were measured at both baseline and

at 30 days post-discharge from hospital (by telephone follow-up).

4. Self-reported adverse drug events, as assessed by questionnaire53 at both

baseline and at 30 days post-discharge from hospital (by telephone follow-up).

5. Total unplanned readmissions (plus cause and length of stay) within 30 days of

initial discharge from hospital.

6. Unplanned readmissions due to preventable adverse drug events within 30 days

of initial discharge from hospital.


of 767

7. Out-of-hospital deaths within 30 days of initial discharge from hospital.

8. Quality of Life, as assessed with the Assessment of Quality of Life (AQoL)

questionnaire,102 at baseline and at 30 days after initial discharge (by telephone

follow-up). This Australian questionnaire has five dimensions (Illness,

Independent Living, Social Relationships, Physical Senses and Psychological

Well-being).

The measurement tools were very similar to those employed by the Project Team in a

number of other studies directed at improving the quality use of medicines in a variety

of settings. These outcome measures were statistically compared across the groups. In

addition to the clinical outcomes, a number of process outcome measures were collected

(e.g. number/proportion of patients in the intervention group for whom the electronic

communication pathway for medication support was utilised).

Satisfaction with the Med eSupport services were also evaluated by anonymously

surveying the patients exposed to them to assess their opinion of their usefulness.

Control group patients were similarly surveyed to assess their satisfaction with the

routine discharge process. Those community health providers whose patients were

exposed to at least one of the services were also surveyed in a similar manner to assess

their views. The questionnaires used a combination of questions with either a multiple

choice of answers or a standard visual analogue scale to obtain answers and were posted

to each participant after the 30-day follow-up phone call was complete, along with a

reply-paid envelope.

The following flowcharts provide a more detailed picture of the trial design.


of 767

Figure 7 Flowchart of enrolment and admission process

TO = Trial Officer; HP = Hospital Pharmacist; RMO = Prescribing Hospital Doctor; CP = Community Pharmacist; GP = General Practitioner



Yes

No Excluded

Randomised

TO collects GPs medication history if no evidence of previous collection (eg Medical Director Printout).

TO generates a standard trial form containing the four lists, an outline of discrepancies found & suggested

solutions

TO’s form presented to RMO and discussed. Uptake of discrepancies/issues found & reasons for non-

uptake recorded

TO calls the patient’s nominated CP(s) within 24 hrs of arrival on the ward & confirms patient is theirs & they are happy to transfer the dispensing information

TO faxes consent form and a cover sheet requesting 6 months dispensing history and any other relevant

information, for example: dose administration aids, known OTCs


Yes

No









solutions

TO’s form filed as a reference to passively observe & record progress of discrepancies/issues found at a

later date

Excluded




Yes

No Excluded

Randomised



solutions

TO’s form presented to RMO and discussed. Uptake of discrepancies/issues found & reasons for non-

uptake recorded





Yes

No









solutions

TO’s form filed as a reference to passively observe & record progress of discrepancies/issues found at a

later date

Excluded


of 767

Figure 8 Flowchart of inpatient interview process

The following items are to be covered (in any order that suits the situation)

Confirmation of demographic details

Explain follow-up phone call at 30 days post-discharge and document preferred time for call

if requested

Inform patient they have been randomised to the intervention group

Provide a more detailed explanation of the trial and services they may/will receive:

Discharge counselling sheet and counselling

Indicate on Information Sheet trial staff to contact with any concerns

Transfer of discharge information to GP and Community Pharmacy

Website FunctionalityConduction of a

PDMR/HMR

Conduct Quality of Life survey

NB. Where not completed in first interview, conduct knowledge and compliance surveys and collect self-

reported DRPs

Explain the HMR/PDMR process &

potential benefits

Present patient with their Project Package

NB. Where applicable, use the document

provided in the Project Package, to explain the website functionality.

Outline difference to current practice and why this information

sharing is/may bebeneficial to them

Explain process, what they will/may receive

and the potential benefits to them


Inform patient they have been randomised to the control group

If requested, provide a brief explanation of the trial and services they may receive:

Emphasise positives of current care, the possibility the intervention is not beneficial and the fact they’ll receive a follow-up call

The following items are to be covered (in any order that suits the situation)

Confirmation of demographic details

Explain follow-up phone call at 30 days post-discharge and document preferred time for call

if requested

Inform patient they have been randomised to the intervention group

Provide a more detailed explanation of the trial and services they may/will receive:

Discharge counselling sheet and counselling

Indicate on Information Sheet trial staff to contact with any concerns

Transfer of discharge information to GP and Community Pharmacy

Website FunctionalityConduction of a

PDMR/HMR

Conduct Quality of Life survey

NB. Where not completed in first interview, conduct knowledge and compliance surveys and collect self-

reported DRPs

Explain the HMR/PDMR process &

potential benefits

Present patient with their Project Package

NB. Where applicable, use the document

provided in the Project Package, to explain the website functionality.

Outline difference to current practice and why this information

sharing is/may bebeneficial to them

Explain process, what they will/may receive

and the potential benefits to them


Inform patient they have been randomised to the control group

If requested, provide a brief explanation of the trial and services they may receive:

Emphasise positives of current care, the possibility the intervention is not beneficial and the fact they’ll receive a follow-up call


of 767

Figure 9 Flowchart of discharge process

Point of discharge

Intervention Group- HMR Referral

Control Group- No HMR

Discharge medication counselling sheet

supplied, with counselling. Produced either by a TO or HP,

using either the hospital’s program or

CMMS.

Normal discharge processes

(Patient may or may not receive a discharge

medication counselling sheet and/or verbal

counselling)

HMR recommended via a sticker on the

Discharge Summary

Discharge medication information, with

notification of HMR referral sent, via automated fax, to

patient’s nominated CP(s) within 24 hrs of

discharge.

Intervention Group- PDMR

Control Group- HMR Referral

Otherwise normal discharge processes



counselling)




CMMS.

Discharge medication information required to perform PDMR sent, via automated fax, to

nominated CP within 24 hrs of discharge.

Discharge medication information, in form of HMR referral sent, via

automated fax, to patient’s nominated

GP(s) within 24 hrs of discharge.

Nominated CP(s) and GP(s) telephoned and explained process of

streamlined HMR model and consent to transfer information

confirmed & registration to website

offered

Discharge medication information uploaded to

repository by TO

NB Where necessary, same information transferred to

Accredited Pharmacist


PDMR and consent to transfer information


offered


repository by TO

Discharge medication information, sent, via




Point of discharge






CMMS.

Normal discharge processes



counselling)

HMR recommended via a sticker on the

Discharge Summary

Discharge medication information, with

notification of HMR referral sent, via automated fax, to

patient’s nominated CP(s) within 24 hrs of

discharge.



Otherwise normal discharge processes



counselling)




CMMS.

Discharge medication information required to perform PDMR sent, via automated fax, to

nominated CP within 24 hrs of discharge.

Discharge medication information, in form of HMR referral sent, via




streamlined HMR model and consent to transfer information


offered


repository by TO

NB Where necessary, same information transferred to

Accredited Pharmacist


PDMR and consent to transfer information


offered


repository by TO

Discharge medication information, sent, via





of 767

Figure 10 Flowchart of follow-up process

Telephone Patient/Carer at 30 days post discharge to collect follow-up data & surveys, self-reported DRPs

and whether an HMR/PDMR was performed

Collect all local & “global”data required for trial analysis

Post & follow-up return of satisfaction surveys to participating community health care professionals



Within first 5-7 days post discharge

Where required, liaise with CP &/or AP to

facilitate timely performance of

PDMR

Where requested, liaise with CP &/or

AP to facilitate timely performance of HMR









No follow-up required at this point


Where necessary, follow-up GPs/CPs and collect data for HMRs performed in

this time frame


this time frame

Follow-up all CPs/APs and collect data from PDMRs

performed


this time frame

Post & follow-up return of satisfaction surveys to patients/carers

TO = Trial Officer; HP = Hospital Pharmacist; RMO = Prescribing Hospital Doctor; CP = Community Pharmacist; GP = General Practitioner; AP = Accredited Pharmacist

Telephone Patient/Carer at 30 days post discharge to collect follow-up data & surveys, self-reported DRPs

and whether an HMR/PDMR was performed

Collect all local & “global”data required for trial analysis

Post & follow-up return of satisfaction surveys to participating community health care professionals



Within first 5-7 days post discharge

Where required, liaise with CP &/or AP to

facilitate timely performance of

PDMR

Where requested, liaise with CP &/or

AP to facilitate timely performance of HMR












this time frame


this time frame

Follow-up all CPs/APs and collect data from PDMRs

performed


this time frame

Post & follow-up return of satisfaction surveys to patients/carers

TO = Trial Officer; HP = Hospital Pharmacist; RMO = Prescribing Hospital Doctor; CP = Community Pharmacist; GP = General Practitioner; AP = Accredited Pharmacist


of 767

Expected or anticipated outcomes were:

• Improved provision of information and follow-up of patients.

• Improved communication between the hospital and community-based health

professionals, ensuring continuity of treatment to promote patient care.

• Better correspondence between the list of prescribed medications during

hospitalisation with the actual medications being taken by the patient

immediately prior to admission.

• Improved patient compliance.

• Improved patient knowledge about medications.

• Improved quality of life.

• Less medication-related adverse events following hospital discharge.

• Reduced number of unplanned readmissions within 30 days of initial

discharge from hospital.

• Reduced number of out-of-hospital deaths within 30 days of initial discharge

from hospital.

• Favourable acceptance of the program by patients, general practitioners and

pharmacists.

• Health care cost savings, specifically hospital costs by reducing re-admission

rates due to adverse drug reactions and non-compliance.

9.1.2 Methodology of anonymous participant surveys As previously described, patients and primary health providers involved in the trial were

sent surveys to assess their opinions of the project. Seven different surveys were sent.

Patients enrolled into both arms of the control group were sent a common survey

whereas patients allocated to the Streamlined HMR recommendation group and PDMR

model group were sent different surveys. General practitioners and community

pharmacists were also sent one of two surveys, depending on the group allocation of

their paitnets - Streamlined HMR recommendation or PDMR model. A cover letter was

attached to the survey reintroducing the project and reminding them of the patient

involved and their group allocation for each individual patient.


of 767

A list of surveys sent with page number is presented below.

Table 2 Patient and provider surveys

Name of survey Page Number

Appendix XVI Control Patient Satisfaction Survey 558

Appendix XVII Intervention – Streamlined HMR Recommendation Patient Satisfaction Survey 561

Appendix XVIII Intervention PDMR Model Patient Satisfaction Survey 566

Appendix XIX GP Satisfaction Survey for Intervention - Streamlined HMR Recommendation patients 571

Appendix XX GP Satisfaction Survey for PDMR Model patients 575

Appendix XXI Community Pharmacist Satisfaction Survey for Intervention - Streamlined HMR Recommendation patients 579

Appendix XXII Community Pharmacist Satisfaction Survey for PDMR Model patients 583

Many of the questions asked were similar across the groups, allowing comparisons and

grouping of information for analysis.

Although not previously validated, these surveys were modified versions of satisfaction

surveys used in similar post-discharge follow-up studies performed by the project

team.92, 100, 146

Data from the returned questionnaires were entered into Statview® for collation and

analysis. Non-parametric statistical tests were used to determine significance between

groups for key questions.

9.1.3 Methodology for the post-hoc reclassification process.

Patients were initially randomly allocated (using pre-determined, block allocation

concealment) to one of the four sub-groups at time of enrolment (Intervention – PDMR

model, Intervention – Streamlined HMR recommendation, Control – HMR

recommendation or Control – No HMR recommendation). As the project progressed,

patients did or did not receive one of, or a combination of, the three major interventions

- namely reporting of drug chart discrepancies in hospital, discharge medication

counselling and a home medicines review (PDMR/HMR) post-discharge.


of 767

However, due to the nature of the trial design, which group a patient was allocated to

and what services they received was not always uniform. Most commonly, the patients

in the streamlined-intervention group could be distinctly split into those who did receive

an HMR and those who did not. Also, through usual hospital care, it was found that

many control patients actually received discharge medication counselling.

On reflection, it was felt that these differences could not be ignored, as it was the impact

of the services received that was most important in terms of analysis of different aspects

of the program.

Initial group allocation was still vital to measure uptake of the different methods of

PDMR/HMR promotion and to assess for most aspects of patient satisfaction. However,

for a more realistic analysis of the services offered by Med eSupport, the patients were

reallocated to a new group, dependant upon the services received.

The new grouping system was comprised of three groups, as follows.


• Patients who either did not receive any of the intervention services or only

received discharge medication counselling. This group comprised of control

patients who received none of the interventions and control patients who

received discharge counselling as part of usual care.

2. Partial Intervention (Discrepancies reported to the RMO)

• The partial intervention group consisted of patients who received discharge

medication counselling and had their admission and discharge drug chart

discrepancies reported to the RMO. This group was made up of intervention

streamlined patients who did not receive a HMR post-discharge.

3. Full Intervention (Discrepancies reported to the RMO and PDMR/HMR

performed)

• The full intervention group consisted of patients who received all three

intervention elements and consisted of intervention PDMR patients and

intervention streamlined HMR recommendation patients who did receive a

PDMR/HMR.

It is to be noted that two control patients were not included in this grouping process as

they received a HMR through usual care, but neither of the other interventions. This


of 767

meant that a fourth group would have to have been formed for these two patients, which

was considered not viable.

This new grouping, termed the ‘services received’ grouping, has been used to analyse


comparison of the impact the services provided had on the outcomes. More detailed

explanation of group allocation in relation to analysis is provided in the results section.


of 767

9.2 Results

9.2.1 Results of the Med eSupport trial

9.2.1.1 Patient recruitment process It was initially planned that the data collection period for all five sites would run for six

months with the aim to enrol 150 patients per site to achieve a total of 750 project

participants.

Due to delays in the implementation of the ICT section of the project, and recruitment

of trial staff in all sites other than Hobart, the patient recruitment period varied between

sites. In summary, patient recruitment ran for nine months at the RHH, with at least

three team members. This number increased to five at one point. Sir Charles Gairdner

and Hollywood Private Hospital ran for the same period, but for some time, with only

one team member to cover both hospitals. One team member was recruited to work at

each of the LGH and Bendigo sites. Recruiting ran for three months at the LGH and

two months at Bendigo.

Each individual patient was followed from their time of enrolment, within 24 hours of

admission to 30-days post-discharge across all the sites, as per trial protocol.

9.2.1.1.1 Pre-enrolment exclusions The method used to identify eligible patients varied slightly at each site, depending on

admission processes and documentation available. Generally, admission lists were first

scanned for potential candidates - indicators included age and admission diagnosis.

From there, these identified patients were investigated further for each of the selection

criteria until they were either excluded or asked to enrol. In total, 4176 patients were

screened for eligibility. Of these, 3361 were excluded for a variety of reasons. The

reasons for exclusion at each site are presented in Table 3.

.


of 767

Table 3 Details of patient exclusions from study prior to randomisation

Hospital

Reason BHCGH SCGH HPH LGH RHH Total

Admitted for < 24 hours prior to discharge 12 21 11 31 183 258

Admitted for > 24 hours prior to identification 47 0 1 194 423 665

Enrolled previously 1 14 2 0 49 66

Either not enough chronic meds or conditions on admission

54 3 14 136 247 454

Non-English speaking 0 21 0 1 9 31

Not planned for discharge to a local private residence 80 4 6 13 52 155

Patient not on ward 20 0 2 11 137 170

Planned for discharge to a nursing home 15 99 78 27 96 315

Too unwell to provide consent 23 111 28 43 376 581

Too young to be enrolled 4 1 2 135 359 501

Transferred to/from another hospital 15 0 0 43 63 121

Miscellaneous 5 4 0 4 31 44

Total of column 276 278 144 638 2025 3361

9.2.1.1.2 Post-enrolment exclusions Eight hundred and fifteen patients were identified as eligible to enrol in the trial. Of

these, 253 declined to participate. Reasons patients gave for not wishing to take part in

the study were varied, but commonly included a belief that the project package would

not benefit them, that they did not have time to participate, that they were too unwell to

participate or they had concerns with the privacy of their information.

Once enrolled, a total of 183 patients were removed from the trial, either in part or in

full, before the 30 day end point was reached. Seventy five were removed completely

from the trial, 49 before they reached the point of discharge from hospital and 59 before

they reached the 30 day phone call. For the majority of patients, reasons for removal

were beyond the control of the project team and included: being transferred to another


of 767

health care facility, passing away during the trial period and being discharged over the

weekend when no hospital or trial staff were available to complete the discharge

process.

Above and beyond expected withdrawals (as described above), 40 patients, were lost

from the SCGH and HPH sites in Perth due, in part, to a difference in hospital

procedure that made implementation of the discharge trial protocol difficult. At these

hospitals, clinical pharmacists were not fundamentally involved in the discharge

process, causing difficulties in implementing the discharge medication counselling with

verbal counselling and the reporting of the discharge prescription discrepancies process

at the time of discharge. In many cases this would have required a complete change in

hospital procedure which was considered not achievable by the site trial officer. As a

result, 15 intervention patients had to be removed from the trial as they either did not

receive full discharge counselling or did not have their new discharge prescription

discrepancies reported to the RMO prior to discharge. For 25 intervention patients, it

was found that discrepancy identification had not been completed, resulting in some or

all of the initial drug chart discrepancies not being reported to the RMO. As this meant

that the initial intervention was not completed, these patients were fully withdrawn from

the trial.

At the end of the recruitment phase, 182 patients (109 control, 71 intervention, 2

withdrawn prior to randomisation) were enrolled at the two Perth sites. Of the 182

patients recruited, some or all of the data had to be removed for 88 patients, due to an

apparent lack of data integrity, leaving 94 patients available for analysis. Of these 94

patients, only 16 were from the intervention group. Not only were more control

patients recruited, but the integrity of the reported data for control patients was more

robust as it was essentially a measurement of usual care practices, not reliant upon the

intervention of the trial officer.

Therefore, in summary, 487 patients remained enrolled across all sites and available for

initial data analysis, 427 for analysis at the point of discharge, and 378 for the 30-day

phone call.

To provide optimal data, the decision was made to use the maximum number of patients

available at the point in question for each individual statistical measure. The other and

more simplistic option of only including the 378 patients whose data was complete


of 767

reduced the numbers considerably and was not considered best utilisation of available

data.

Table 4 shows reasons for full withdrawal from the trial and indicates how many

patients from each group.

Table 4 Reasons patients were fully withdrawn from the trial

Intervention Control Total

PDMR model n (%)

Streamlined HMR n (%)

HMR Rec.

n (%)

No HMR Rec.

n (%)

n (%) Transferred to another facility 2 (9%) 0 (0%) 0 (0%) 1 (7%) 3 (4%)

Previously enrolled in this, or another study 1 (4%) 2 (7%) 1 (20%) 0 (0%) 4 (5%)

Unable to confirm medications on admission

1 (4%) 2 (7%) 0 (0%) 0 (0%) 3 (4%)

In another medical facility prior to recruitment

2 (9%) 0 (0%) 0 (0%) 2 (14%) 4 (5%)

Discharged prior to completion of enrolment

3 (13%) 2 (7%) 0 (0%) 4 (29%) 9 (12%)

Voluntary withdrawal 0 (0%) 3 (10%) 0 (0%) 1 (7%) 4(5%)

Incomplete intervention 13 (57%) 19 (62%) 0 (0%) 0 (0%) 32 (43%)

Miscellaneous 1 (4%) 2 (7%) 4 (80%) 6 (43%) 13 (18%)

Withdrawn before randomised N/A N/A N/A N/A 3 (4%)

TOTAL 23 (31%) 30 (40%) 5 (7%) 14 (18%) 75 (100%)

Table 5 and Table 6 indicate reasons for loss at the point of discharge and 30 days post-

discharge respectively.


of 767

Table 5 Reasons patients were withdrawn at the point of discharge


PDMR model n (%)


HMR Rec.

n (%)

No HMR Rec.

n (%)

n (%) Discharge intervention incomplete 20 (76%) 18 (90%) 0 (0%) 1 (25%) 39 (65%)

Discharged to another care facility 3 (12%) 2 (10%) 2 (20%) 3 (75%) 10 (17%)

Deteriorated in health 1 (4%) 0 (0%) 1 (10%) 0 (0%) 2 (3%)

Deceased in hospital 1 (4%) 0 (0%) 6 (60%) 0 (0%) 7 (12%)

Still in hospital at trial end date 1 (4%) 0 (0%) 1 (10%) 0 (0%) 2 (3%)

TOTAL 26 (43%) 20 (33%) 10 (17%) 4 (7%) 60 (100%)

Table 6 Reasons patients were withdrawn at thirty days post-discharge


PDMR model n (%)


HMR Rec.

n (%)

No HMR Rec.

n (%)

n (%) Patient too unwell to complete interview 0 (0%) 1 (13%) 1 (6%) 1 (7%) 3 (6%)

In other care facility at time of call or most of 30 day period

0 (0%) 3 (37%) 3 (17%) 3 (21%) 9 (18%)

Unable to be contacted 4 (45%) 2 (25%) 4 (22%) 7 (50%) 17 (35%)

PDMR/HMR not complete post-discharge

1 (11%) 0 (0%) 2 (11%) 0 (0%) 3 (6%)

Readmitted within 5 days of discharge 1 (11%) 0 (0%) 6 (33%) 2 (15%) 9 (18%)

Deceased during 30 day period 2 (22%) 2 (25%) 2 (11%) 1 (7%) 7 (15%)

Data insufficient for use in analysis 1 (11%) 0 (0%) 0 (0%) 0 (0%) 1 (2%)

TOTAL 9 (18%) 8 (16%) 18 (37%) 14 (29%) 49 (100%)


of 767

Flowcharts of patient enrolment and withdrawal are provided on the next 5 pages, see

Table 7 for details.

Table 7 Breakdown of patient enrolment details at each site

Site Figure Number Page Number RHH Figure 11 88

LGH Figure 12 89

SCGH Figure 13 90

HPH Figure 14 91

BCGH Figure 15 92

Total Figure 16 93


of 767

Figure 11 Enrolment flowchart for RHH


2025 Excluded


N = 72


Control: N = 166

47 Refused

329 Consented

N = 376


model:N = 76

Control –HMR

recommendation:N = 82

Control –No HMR






N = 52


modelN = 63


N = 70

Control –No HMR



1 withdrawn


N = 80


model:N = 82

Control –HMR


Control –No HMR




2025 Excluded


N = 72


Control: N = 166

47 Refused

329 Consented

N = 376


model:N = 76

Control –HMR


Control –No HMR






N = 52


modelN = 63


N = 70

Control –No HMR



1 withdrawn


N = 80


model:N = 82

Control –HMR


Control –No HMR




of 767

Figure 12 Enrolment flowchart for LGH


638 Excluded


N = 10


Control: N = 22

24 Refused

43 Consented

N = 67


model:N = 6

Control –HMR


Control –No HMR






N = 4


modelN = 3


N = 5

Control –No HMR

recommendationN = 9


0 withdrawn


N = 13


model:N = 8

Control –HMR


Control –No HMR




638 Excluded


N = 10


Control: N = 22

24 Refused

43 Consented

N = 67


model:N = 6

Control –HMR


Control –No HMR






N = 4


modelN = 3


N = 5

Control –No HMR

recommendationN = 9


0 withdrawn


N = 13


model:N = 8

Control –HMR


Control –No HMR




of 767

Figure 13 Enrolment flowchart for SCGH


278 Excluded


N =17


Control: N = 65

164 Refused

123 Consented

N = 287


model:N = 17

Control –HMR


Control –No HMR






N = 8


modelN = 6


N = 30

Control –No HMR



0 withdrawn


N = 27


model:N = 31

Control –HMR


Control –No HMR




278 Excluded


N =17


Control: N = 65

164 Refused

123 Consented

N = 287


model:N = 17

Control –HMR


Control –No HMR






N = 8


modelN = 6


N = 30

Control –No HMR



0 withdrawn


N = 27


model:N = 31

Control –HMR


Control –No HMR




of 767

Figure 14 Enrolment flowchart for HPH


144 Excluded


N = 1


Control: N = 44

16 Refused

59 Consented

N = 75


model:N = 2

Control –HMR


Control –No HMR






N = 1


modelN = 1


N = 9

Control –No HMR



2 withdrawn


N = 3


model:N = 10

Control –HMR


Control –No HMR




144 Excluded


N = 1


Control: N = 44

16 Refused

59 Consented

N = 75


model:N = 2

Control –HMR


Control –No HMR






N = 1


modelN = 1


N = 9

Control –No HMR



2 withdrawn


N = 3


model:N = 10

Control –HMR


Control –No HMR




of 767

Figure 15 Enrolment flowchart for BCGH


276 Excluded


N = 2

Intervention: N = 2

Control: N = 6

2 Refused

8 Consented

N = 10


model:N = 0

Control –HMR


Control –No HMR






N = 2


modelN = 0


N = 1

Control –No HMR

recommendationN = 4


0 withdrawn


N = 2


model:N = 0

Control –HMR


Control –No HMR




276 Excluded


N = 2

Intervention: N = 2

Control: N = 6

2 Refused

8 Consented

N = 10


model:N = 0

Control –HMR


Control –No HMR






N = 2


modelN = 0


N = 1

Control –No HMR

recommendationN = 4


0 withdrawn


N = 2


model:N = 0

Control –HMR


Control –No HMR




of 767

Figure 16 Enrolment flowchart for all sites


3361 Excluded


N =102


Control: N = 303

253 Refused

562 Consented

N = 815


model:N =101

Control –HMR


Control –No HMR






N = 67


modelN = 73


N = 115

Control –No HMR



3 withdrawn


N =125


model:N =131

Control –HMR


Control –No HMR




3361 Excluded


N =102


Control: N = 303

253 Refused

562 Consented

N = 815


model:N =101

Control –HMR


Control –No HMR






N = 67


modelN = 73


N = 115

Control –No HMR



3 withdrawn


N =125


model:N =131

Control –HMR


Control –No HMR




of 767

9.2.1.2 Explanation of baseline characteristics and group allocation

Patients were randomly allocated (using pre-determined, block allocation concealment)

to one of the four sub-groups at time of enrolment (Intervention – PDMR model,

Intervention – Streamlined HMR recommendation, Control – HMR recommendation or

Control – No HMR recommendation).

All parameters were measured and analysed using the original grouping. However, as

previously explained in section 8.1.3 of the methodology section, the ‘services received’

grouping system was found to provide a more valuable interpretation of the data.

Therefore, the ‘services received’ grouping analysis is presented in this report to analyse


comparison of the impact the services provided had on the outcomes.

In some cases, such as analysis of discrepancies, data was analysed using the simple 2

group split of control and intervention, as opposed to the four subgroup split or the

‘services received’ grouping. This was because patients either received discrepancy

reviews or they did not and HMR performance was irrelevant at this point. It was at

this later point, post-discharge, that the ‘services received’ grouping was used to analyse

the data.

Further explanation is provided throughout the results section.

9.2.1.3 Removal of data collected outside of trial protocol specifications

During data analysis and after discussion with the trial officers at the two sites in Perth,

it was found that the trial protocol (Appendix XIV) had not been rigorously adhered to

as required in a multi-centre study of this nature. It became apparent that the integrity

of the data relating to key outcome measures (knowledge, compliance and AQoL) and

the identification and reporting of DRPs was not robust or thorough. Of particular note,

nearly all compliance and knowledge scores at 30 days were 100%, which raised doubts

about the validity of the data. It is possible that the trial officer influenced the outcomes

of this section. This phenomenon has been recognised as a problem in short

questionnaires designed to be conducted orally. Boynton el al103 stated “Questionnaires

completed by researchers can be a legitimate approach if they are a planned part of the

study protocol, but researchers can subtly influence responses by inflections of the


of 767

voice, facial expressions, or gestures. For example, they may unconsciously hurry

through questions they find uncomfortable or perceive as unimportant. A bored or tired

researcher will convey a lack of enthusiasm, which might be interpreted as ‘it doesn’t

really matter which answer you choose.” 103 These observations, coupled with the other

findings relating to the integrity of the Perth data led the Project Team to believe this 30

day follow-up data was not useable.

In addition, analysis of baseline characteristics of patients enrolled at SCGH and HPH

had a number of statistical differences, questioning the random allocation of these

patients. Given the concerns over the integrity of these data (knowledge, compliance,

AQoL and DRPs) they were not included in the analyses.

To ensure it is clear which data was used at which point, data included is indicated at

each statistical parameter.

9.2.1.4 Baseline Characteristics of enrolled patients Baseline patient characteristics and medication-related parameters have been presented

according to the original two and four group randomisation in Table 8 and

Table 9 respectively for all patients enrolled in the Med eSupport trial.

Table 8 Baseline characteristics of patients within the control and intervention groups, and the four sub groups, including all patients enrolled in the trial

Intervention (n = 203)

Control (n = 284)

Baseline Patient Characteristics PDMR model

(n =102)

Stream- lined HMR

(n =101)

HMR Rec.

(n =143)

No HMR Rec.

(n =141)

46.8 52.5 Gender – Female (%) 57.5 47.5 42.0 51.8

70.7 (10.3) 73.8 (9.5) Age – years, mean (STD) 70.7(11) 70.7(10) 74.4(9.4) 73.3(9.5)

38.6 32.7 Living alone (%) 37.3 40 33.6 31.9

65.3 61.1 Receives some no external help at home (%) 65.7 65 60.8 61.4

28.6 24.3 Help with medicines (%) 32.4 24.8 28.7 19.9

84.2 83.8 Nil or occasional alcohol intake (%) 80.4 88.1 85.3 82.3

63.1 62 Smoker or ex-smoker (%) 65.7 60.4 62.2 61.7


of 767

6(2-15) 6(2-17) Number of concurrent chronic conditions, median (range) 6(2-12) 6(2-15) 5(2-13) 6(2-17)

0(0-11) 0(0-9) Number of unplanned admissions in the last 12 months, median (range) 0(0-11) 0(0-7) 0(0-6) 0(0-9)

Primary admission diagnosis (%)

44.8 42.6 Cardiovascular Disease (%) 46.1 43.6 44.8 40.4

3.9 1.1 Diabetes Mellitus (%) 4.9 3 2.1 0

11.3 13 Chronic Obstructive Airways Disease (%) 9.8 12.9 14.7 11.3

15.0 14.4 Infection (%) 13.7 17.8 11.9 17.0

26.9 28.9 Other (%) 25.5 22.8 26.6 31.2

7.4 9.2 % of the admissions that were planned 5.9 8.9 8.4 9.9

4(0-31) 4(0-52) Length of stay, median (range) 4(0-26) 4(1-31) 4(0-36) 4(0-52)

44.3 42.96 PODs in hospital (%) 45.1 43.6 39.2 46.8

12.4 11.6 PODs at interview (%) 10.8 14.1 11.2 12.1

28.1 26.4 Clinical Pharmacist review prior to enrolment (%) 21.6 34.7 26.6 26.2

12.8 7.7 Carer assisting at interview (%) 11.8 13.86 6.99 8.5

Table 9 Some key baseline variables of patients within the control and intervention

groups, and the four sub groups, including all patients enrolled in the trial


Control (n = 284)

Key Baseline Variables PDMR model

(n =102)

Stream- lined HMR

(n =101)

HMR Rec.

(n =143)

No HMR Rec.

(n =141)

7(3-19) 7.5(3-17) Regular medications on admission, median (range) 7(3-19) 7(3-17) 7(3-17) 8(3-16)

7(0-20) 7(0-17) Regular medications on discharge, median (range) 7.5(3-20) 7(0-17) 7(0-17) 7(2-16)

67.8 65.02 Total number of medication chart discrepancies found on admission, (%) 70.6 65.0 59.9 70.2

5 (0-23) 5 (-022) Total number of drug related problems on admission, median (range) 5 (0-17) 5 (0-23) 4 (0-21) 5 (0-22)


of 767

1 (0-8) 1 (0-8) Number of drug interactions on admission, median (range) 1 (0-8) 1 (0-7) 1 (0-7) 1 (0-8)

1 (0-7) 1 (0-9) Number of self-reported drug-related issues on admission, median (range) 1 (0-5) 1 (0-7) 1 (0-6) 1 (0-9)

67 (0-100) 67 (0-100) Baseline knowledge questionnaire score (%) 58(0-100) 67(0-100) 67(0-100) 67(0-100)

56.4 36 Baseline compliance questionnaire score (% fully compliant) 50 63 67.8 58.2

30.9 (5.9) 30.3 (5.7) Baseline Quality of Life survey score, mean (STD) 30.9(6.2) 30.8(5.7) 30.2 (5.9) 30.5(5.5)

Table 10 and Table 11 demonstrate these same parameters according to the three

‘services received’ groups.

Table 10 Baseline characteristics of patients within the three group split, including all patients enrolled in the trial

Baseline Patient Characteristics Full (n = 83)

Partial (n = 115)

Minimal (n = 287)

Gender – Female (%) 57.8 48.2 47.4

Age – years, median (range) 69.4(10) 71.7(10.6) 73.8(9.4)

Living alone (%) 36.1 39.5 33.1

Receives some no external help at home (%) 71.1 60.5 61.2

Help with medicines (%) 27.7 29.6 24.04

Nil or occasional alcohol intake (%) 84.3 84.3 84

Smoker or ex-smoker (%) 63.9 62.6 61.7

Number of concurrent chronic conditions, median (range) 5(2-12) 6(2-15) 6(2-17)

Number of unplanned admissions in the last 12 months, median (range) 0(0-11) 0(0-8) 0(0-9)


Cardiovascular Disease (%) 45.8 45.2 42.2

Diabetes Mellitus (%) 7.2 1.7 1.0


of 767

Chronic Obstructive Airways Disease (%) 12.0 10.4 13.2

Infection (%) 12.0 18.3 14.3

Other (%) 22.9 24.3 29.3

% of the admissions that were planned 3.6 10.4 9.1

Length of stay, median (range) 4(0-26) 4(1-31) 4(0-52)

PODs in hospital (%) 42.2 47.8 41.8

PODs at interview (%) 10.98 14.0 11.2

Clinical Pharmacist review prior to enrolment(%) 16.9 34.8 27.2

Carer assisting at interview (%) 14.5 12.2 7.7

Table 11 Some key baseline variables of patients within the three group split, including all patients enrolled in the trial

Key Baseline Variables Full (n = 83)

Partial (n = 115)

Minimal (n = 287)

Regular medications on admission, median (range) 7(3-19) 7(3-18) 7(3-17)

Regular medications on discharge, median (range) 7(3-20) 7(0-17) 7(0-17)

Total number of initial medication chart discrepancies found on admission, (%) 68.7 66.7 65.0

Total number of drug related problems on admission, median (range) 5 (0-20) 5 (0-23) 5 (0-22)

Number of drug interactions on admission, median (range) 1 (0-8) 1 (0-7) 1 (0-8)

Number of self-reported drug-related issues on admission, median (range) 1 (0-5) 1 (0-7) 1 (0-9)

Baseline knowledge questionnaire score (%) 67(0-100) 67(0-100) 67(0-100)

Baseline compliance questionnaire score (% fully compliant) 51.8 57.9 63.8

Baseline Quality of Life survey score, median (range) 31.0 (6.7) 31.0 (5.3) 30.2 (5.7)


of 767

All baseline characteristics are not significantly different except for three parameters.

These included:

Age: When comparing intervention and control groups (ANOVA test: F = 12.0; df =

1/485; p = 0.0006), and when comparing the four groups (ANOVA test: F = 4.3; df =

3/483; p = 0.005),

Previous unplanned admissions: The number of unplanned admissions in the last 12

months was found to be statistically different between the intervention and control

groups (Mann-Whitney test U = 25550.5, z = -1.9, p = 0.05), and

Compliance Questionnaire score: A statistically significant difference was found

across the four groups for the baseline compliance questionnaire score (Chi-square =

8.5, df = 3, p = 0.04).

When the data from Perth was analysed alone, differences were found between these,

plus a number of other parameters. The data was then analysed excluding that collected

in Perth, and no statistically significant differences were found when compared

according to the original randomised two and four group split, and the post-hoc three

group split. Table 13 show baseline patient characteristics and medication-related

parameters for patients enrolled in Tasmania and Victoria only according to the original

randomisation. The identified differences provide further justification for doubting the

integrity of the Perth data, and excluding it from the analysis.

Table 12 and Table 13 show baseline patient characteristics and medication-related

parameters for patients enrolled in Tasmania and Victoria only according to the original

randomisation. The identified differences provide further justification for doubting the

integrity of the Perth data, and excluding it from the analysis.

Table 12 Baseline characteristics of patients within the control & intervention groups, & the four sub groups, including patients enrolled in Tasmania & Victoria only


Control (n = 186)

Baseline Patient Characteristics PDMR model

(n = 84)

Stream- lined HMR

(n = 82)

HMR Rec.

(n = 94)

No HMR Rec.

(n = 92)

51.8 47.3 Gender – Female (%) 55.95 47.6 43.6 51.1

70 (10.4) 71.3 (9.1) Age – years, median (range) 70.1(11) 69.6(9.8) 72.2(8.5) 70.4(9.6)


of 767

36.4 28.5 Living alone (%) 36.9 35.8 31.9 25.0

67.9 65.4 Receives some no external help at home (%) 67.9 67.9 62.8 68.1

31.9 26.3 Help with medicines (%)

36.9 26.8 28.7 23.9 84.9 87.1 Nil or occasional alcohol intake (%)

80.95 89 84 90.2 63.9 69.9 Smoker or ex-smoker (%)

69.0 58.5 69.1 70.7 6 (2-15) 6 (2-14) Number of concurrent chronic conditions, median

(range) 5(2-12) 6(2-15) 5(2-12) 6(2-14) 0 (0-11) 0 (0-6) Number of unplanned admissions in the last 12

months, median (range) 0(0-11) 0(0-7) 0(0-6) 0(0-6)


50.0 51.6 Cardiovascular Disease (%) 50.0 50.0 54.3 48.9

4.8 1.1 Diabetes Mellitus (%) 5.95 3.7 2.1 0

10.8 11.8 Chronic Obstructive Airways Disease (%) 10.7 10.98 13.8 9.8

13.9 16.1 Infection (%) 13.1 14.6 14.9 17.4

20.5 19.4 Other (%) 20.2 20.7 14.9 23.9

7.2 6.5 % of the admissions that were planned 6 8.5 6.4 6.5

4 (0-29) 4 (1-52) Length of stay, median (range) 5(0-26) 4(1-29) 4(1-20) 4(1-52)

49.4 54.3 PODs in hospital (%) 48.8 50 50 58.7

12.2 15.6 PODs at interview (%)

9.5 15 14.9 16.3 17.5 12.9

Clinical Pharmacist review prior to enrolment (%) 13.1 22 12.8 13.04

15.7 10.2 Carer assisting at interview (%)

14.3 17.1 8.5 11.96


of 767

Table 13 Some key baseline variables of patients within the control & intervention groups, & the four sub groups including patients enrolled in Tasmania & Victoria only


Control (n = 186)

Key Baseline Variables PDMR model

(n =84)

Stream- lined HMR

(n = 82)

HMR Rec.

(n = 94)

No HMR Rec.

(n =92)

7 (3-19) 7 (3-15) Regular medications on admission, median (range) 7(3-19) 7(3-17) 6(3-14) 7(3-15)

7 (0-20) 7 (0-16) Regular medications on discharge, median (range) 7(3-20) 7(0-17) 7(0-14) 7(2-16)

67.9 62.4 Total number of initial medication chart discrepancies found on admission, (%) 72.6 62.96 59.6 65.2

5 (0-23) 5 (0-22) Total number of drug related problems on admission, median (range) 5 (0-17) 5.5 (0-23) 4 (0-21) 5.5 (0-22)

1 (0-7) 1 (0-7) Number of drug interactions on admission, median (range) 1 (0-7) 1 (0-6) 1 (0-7) 1 (0-5)

1 (0-7) 1 (0-9) Number of self-reported drug-related issues on admission, median (range) 1 (0-5) 1 (0-7) 1 (0-6) 1 (0-9)

58 (0-100) 58 (0-100) Baseline knowledge questionnaire score (%) 58(0-100) 67(0-100) 58(0-100) 67(0-92)

49 50 Baseline compliance questionnaire score (% fully compliant) 44 54 54.3 45.7

31.4 (6.0) 31.6 (5.9) Baseline Quality of Life survey score, median (range) 31.5 (6.3) 31.6 (5.5) 31.5 (6.5) 31.8 (5.5)

Table 14 and Table 15 demonstrate these same parameters according to the three

‘services received’ groups for patients enrolled in Tasmania and Victoria only.


of 767

Table 14 Baseline characteristics of patients within the three group split, including patients enrolled in Tasmania and Victoria only

Baseline Patient Characteristics Full (n = 73)

Partial (n = 91)

Minimal (n = 186)

Gender – Female (%) 60.3 45.1 47.8

Age – years, median (range) 68.8 (10.1) 70.7 (10.7) 71.4 (9)

Living alone (%) 37 35.6 28.5

Receives some no external help at home (%) 69.9 65.5 65.4

Help with medicines (%) 68.5 68.1 73.7

Nil or occasional alcohol intake (%) 84.9 84.6 87.6

Smoker or ex-smoker (%) 64.4 63.7 69.4

Number of concurrent chronic conditions, median (range) 5 (2-12) 6 (2-15) 6 (2-14)

Number of unplanned admissions in the last 12 months, median (range) 0 (0-11) 0 (0-7) 0 (0-6)


Cardiovascular Disease (%) 46.6 52.7 51.6

Diabetes Mellitus (%) 8.2 2.2 1.1

Chronic Obstructive Airways Disease (%) 13.7 8.8 11.8

Infection (%) 13.7 14.3 15.6

Other (%) 17.8 22 19.9

% of the admissions that were planned 2.7 11 6.5

Length of stay, median (range) 4 (1-26) 4 (1-29) 4 (1-52)

PODs in hospital (%) 45.2 53.8 53.2

PODs at interview (%) 11.1 13.3 15.1

Clinical Pharmacist review prior to enrolment (%) 13.7 20.8 12.9


of 767

Carer assisting at interview (%) 16.4 15.4 10.2

Table 15 Some key baseline variables of patients within the three group split, including patients enrolled in Tasmania and Victoria only

Key Baseline Variables Full (n = 73)

Partial (n = 91)

Minimal (n = 186)

Regular medications on admission, median (range) 7 (3-19) 7 (3-18) 7 (3-15)

Regular medications on discharge, median (range) 7 (30-20) 7 (0-17) 7 (0-16)

Total number of initial medication chart discrepancies found on admission, (%) 69.9 65.6 62.4

Total number of drug related problems on admission, median (range) 5 (0-20) 5 (0-23) 5 (0-22)

Number of drug interactions on admission, median (range) 1 (0-7) 1 (0-6) 1 (0-7)

Number of self-reported drug-related issues on admission, median (range) 1 (0-5) 1 (0-7) 1 (0-9)

Baseline knowledge questionnaire score (%) 62 (0-100) 58 (0-100) 58 (0-100)

Baseline compliance questionnaire score (% fully compliant) 49.3 47.8 50.5

Baseline Quality of Life survey score, median (range) 31.4 (6.8) 31.8 (5.1) 31.4 (6.0)


of 767

9.2.1.5 Medication list discrepancy detection and resolution frequencies

9.2.1.5.1 Comparison of discrepancies found on admission and their origin

A reconciled list of medications that each patient was most likely to have been taking at

time of admission was constructed by comparing the following information:

• the CP’s six-month dispensing history for the patient and any supplementary

information provided,

• a comprehensive interview with the patient,

• review of the patient’s own medications brought into hospital,†

• the report of the Hospital Pharmacist reviewing the patient prior to

admission,‡

• information obtained from the General Practitioner,§ and

• the hospital RMO’s initial assessment of the patient’s medication history.

The final reconciled list was then compared with the RMO’s initial drug chart and

discrepancies between the two were recorded. Discrepancies relating to omissions of

medications, wrong drugs and dosing errors were included.

Table 16 and Table 17 summarise the number of medication chart discrepancies

identified at the time of admission to hospital. Upon discussion with medical staff (as

per trial process), some of the identified discrepancies were found to be legitimate. For

example, cessation of a medication that contributed to the patient’s condition, or

increasing a dose to treat the presenting condition. These discrepancies were then

removed from the count of discrepancies. Therefore, reported discrepancies in this

section are excluding legitimate therapeutic changes.

Data from all sites (487 patients [203 intervention, 284 control]) were used for analysis

of initial drug chart discrepancies and their resolution. Discrepancy resolution in

hospital has only been analysed using the two group split as patients in the sub groups

† Information obtained only when available ‡ Information obtained only when available § Information obtained only when available


of 767

received the same trial services (or not) whilst they were an inpatient. It was only

during the discharge process that four different levels of intervention were implemented.

The process of determining discrepancies was the same irrespective of the patient’s

allocated group. This is the project’s first intervention, and therefore at this point, there

is only a group who received the discrepancy review intervention and a group who did

not – namely the original control and intervention group. Therefore, for the analysis of

this outcome measure, the original control and intervention group split is used.

Table 16 Patients with medication discrepancies on admission

Group Patients with discrepancies (%)

Intervention 67.8

Control 65.0

All Patients 66.2

Approximately 66% of patients had at least one discrepancy between their reconciled

list of medications on admission and their initial drug chart. There was no significant

difference between the two groups (Chi-square = 1.1, df = 1, p = 0.3).

Table 17 Number of medication discrepancies on admission per patient (compared with reconciled list)

Group Median number of discrepancies (range)

Intervention 1 (0-8)

Control 1 (0-14)

All Patients 1 (0-14)

A median of one (range: 0-14) discrepancy was found per patient. Again, there was no

significant difference between the numbers of discrepancies found for the control and

intervention groups (Mann-Whitney test U = 28021.5, z = -0.5, p = 0.6).

9.2.1.5.2 Correction of discrepancies found on admission For those patients in the intervention group, the trial officer aimed to reach the RMO

with the list of identified discrepancies within 24 hours of admission to the ward, to

ensure discrepancies identified were corrected in a timely manner. Once the


of 767

discrepancies were presented to the RMO, ideally they would be corrected within the

next 24 hours (i.e., within 48 hours of admission to hospital). On average, the trial

officers reached the RMOs (or, for those wards with the service, the clinical pharmacists

caring for the patients) after a mean of 19.9 hours, SD = ± 6.2 hours. The uptake of

recommendations made was then monitored.

For the control patients, discrepancies were also identified, but not reported to the

hospital staff. Instead, they were “silently” monitored by the trial officers to allow for

comparison between the intervention group and usual hospital care (control group).

9.2.1.5.2.1 Median number of discrepancies on admission A median of 2 (range: 0-16) discrepancies were reported to the RMO per intervention

patient. This figure is higher than the number found on the RMO’s initial drug chart

because this figure includes those found, upon discussion with the RMO, to be

legitimate therapeutic changes. Approximately 86% of those discrepancies reported to

the RMO were in fact errors.

A significantly greater number of discrepancies per patient were resolved within the

first 48 hours of admission for the intervention group than for the control (Mann-

Whitney test U = 37340.0, z = -6.0, p < 0.0001). There was no difference found

between the groups for the number of discrepancies resolved after the initial 48 hours

(Mann-Whitney test U = 28739.5, z = -0.3, p = 0.7). Control patients had significantly

more discrepancies that were not resolved during their hospital stay (Mann-Whitney test

U = 39008.0, z = -7.1, p < 0.0001).


of 767

Table 18 Comparison of the resolution of discrepancies identified between the intervention and control groups over the hospital stay period, measured by the median number of discrepancies per patient

Discrepancy identified

Intervention Group, median (range)

Control Group, median (range)

Number of discrepancies found on the RMO’s initial drug chart, excluding those that are legitimate 1 (0-8) 1 (0-14)

Number of discrepancies reported to the RMO 2 (0-16) N/A

Number of discrepancies resolved within 48 hours 1 (0-8) 0 (0-6)

Number of discrepancies resolved after 48 hours 0 (0-6) 0 (0-5)

Number of discrepancies not resolved during admission 0 (0-4) 1 (0-12)

9.2.1.5.2.2 Presence of one or more discrepancies on admission 83.2% of control patients who had at least one discrepancy on their admission drug

chart had at least one discrepancy not resolved through usual care during their hospital

stay. This translates into 54% of all control patients having at least one admission drug

chart discrepancy not identified or addressed during their admission.

Discrepancies were reported to the RMO for 78.6% of intervention patients. Again, it is

to be noted that this includes those discrepancies found to be legitimate therapeutic

changes.

Significantly more intervention patients had at least one discrepancy resolved in the first

48 hours than control patients (Chi-square = 45.4, df = 1, p < 0.0001). There was no

significant difference in the percentage of patients who had at least one discrepancy

resolved after the initial 48 hour period (Chi-square = 0.5, df = 1, p = 0.5). Not

surprisingly there were a greater number of control patients who did not have at least

one of their discrepancies resolved in comparison to intervention patients. (Chi-square

= 61.3, df = 1, p < 0.0001).


of 767

Table 19 Comparison of the resolution of discrepancies identified between the intervention and control groups over the hospital stay period, measured by the percentage of patients with one or more discrepancies

Discrepancy identified Intervention

Group (%)

Control Group

(%) Percentage of patients whose initial drug chart had at least one discrepancy 67.8 65.0

Percentage of patients who had discrepancies reported to the RMO 78.6 N/A

Percentage of patients who had discrepancies resolved within 48 hours 78.1 36.5

Percentage of patients who had discrepancies resolved after 48 hours 30.7 28.7

Percentage of patients who did not have discrepancies resolved during admission 27.7 83.2

9.2.1.5.3 New discrepancies identified at discharge and comparison of resolution between the groups

Discharge prescriptions were checked for new discrepancies. For the intervention

group, any discrepancies identified were highlighted to the RMO or hospital pharmacist.

Resolution of identified discrepancies was recorded for both intervention and control

groups to allow for comparison.

For the analysis of discharge discrepancies, Perth intervention patients were excluded.

After careful investigation, for the majority of Perth intervention patients’ discharge

discrepancies were not discussed with the RMO at discharge, leaving only patients who

did not have any discrepancies at discharge available to analyse. It was felt that this

produced a skewed picture and therefore, only the data of Perth control patients was

included.

9.2.1.5.3.1 Median number of discrepancies on discharge There was a tendency for slightly more new discrepancies at discharge for control

patients than for intervention patients (Mann-Whitney test U = 23580.0, z = -1.8, p =

0.07).

Significantly more discrepancies were resolved prior to discharge for intervention

patients than for control patients (Mann-Whitney test U = 25237.5, z = -2.9, p = 0.04).

Control patients had significantly more new discrepancies not resolved by time of


of 767

discharge than intervention patients (Mann-Whitney test U = 27685.0, z = -4.78, p <

0.0001).

Table 20 Identification and resolution of new discrepancies at discharge


Group, median (range)

Control Group,

median (range)

Number of new discrepancies found at discharge 0 (0-13) 0 (0-12)

Number reported to RMO or hospital pharmacist, of those with new discrepancies at discharge 0 (0-13) N/A

Number resolved, of those reported 0 (0-7) 0 (0-5)

Number not resolved, of those reported 0 (0-2) 0 (0-12)

9.2.1.5.3.2 Presence of one or more discrepancies at discharge There was no statistically significant difference in percentage of patients who had new

discrepancies at discharge. Significantly more intervention patients had at least one

discrepancy resolved prior to discharge than control patients (Chi-square = 27.8, df = 1,

p <0.0001). Again, significantly more control patients had at least one discrepancy not

resolved by time of discharge than intervention patients (Chi-square = 30.4, df = 1, p

<0.0001).

Table 21 Identification and resolution of new discrepancies at discharge, measured by the percentage of patients with one or more discrepancies


Group, (%)

Control Group,

(%) Percentage of patients with new discrepancies found at discharge 27.9 35.6

Percentage of patients who had new discrepancies reported to the RMO or hospital pharmacist 22.5 N/A

Percentage of patients who had discrepancies resolved, of those reported 79.5 13.7

Percentage of patients who had discrepancies not resolved, of those recorded 12.96 87.0


of 767

9.2.1.5.4 Admission drug chart discrepancies not acted on in the first 48 hours of the hospital stay

When analysed including both those patients who had admission drug chart

discrepancies and those who did not, there was no significant difference in LOS found

for those patients who still had at least one admission discrepancy not resolved at 48

hours post admission versus those who had them all resolved (Mann-Whitney test U =

23985.0, z = -0.9, p = 0.3).

However, a weak, but significant correlation was found between the number of

admission drug chart discrepancies still not resolved at 48 hours post admission and

length of stay. LOS increased with number of discrepancies not resolved at 48 hours

(Spearman rank Correlation Rho = 0.1, n = 487, p = 0.01).

Table 22 Differences in length of stay between those who did or did not have all discrepancies on their admission drug chart resolved in the first 48 hours of admission for all patients

Discrepancy Length of Stay, median (range)

At least one admission drug chart discrepancy not resolved at 48 hours post admission 4 (0-50)

No admission drug chart discrepancies not resolved at 48 hours post admission 4 (0-52)

When measured for only those who had admission drug chart discrepancies, there was a

tendency for those who still had at least one admission drug chart discrepancy not

resolved at 48 hours post admission to have a longer LOS than those who had all their

admission discrepancies resolved in that time period (Mann-Whitney test U = 8538.0, z

= -1.6, p = 0.1). Again, a weak but significant correlation was found between LOS and

total number of discrepancies not acted on at 48 hours post admission. This suggested

that length of stay increased as the number of unresolved admission drug chart

discrepancies at 48 hours increased (Spearman rank Correlation Rho = 0.1, n = 284, p =

0.04).


of 767

Table 23 Differences in length of stay between those who did or did not have all discrepancies on their admission drug chart resolved in the first 48 hours of admission for those who had admission drug chart discrepancies


At least one admission drug chart discrepancy not resolved at 48 hours post admission 4 (0-50)

No admission drug chart discrepancies resolved at 48 hours post admission 4 (1-23)

9.2.1.5.5 Admission drug chart discrepancies not acted upon during the hospital stay.

There was no statistically significant difference between those who had at least one

admission drug chart discrepancy not resolved during the hospital stay and those who

did not (Mann-Whitney test U = 21833.5, z = -0.1, p = 0.9). However, a weak, but

significant correlation was found between number of discrepancies not acted on and

LOS. This suggested that LOS significantly increased with the number of admission

drug chart discrepancies not acted on during the hospital stay (Spearman rank

Correlation Rho = 0.1, n = 487, p = 0.008).

Table 24 Admission drug chart discrepancies not acted on during the hospital stay versus length of stay for all patients


At least one admission drug chart discrepancy not resolved during the hospital stay 4 (0-50)

No admission drug chart discrepancies not resolved during the hospital stay 4 (0-52)

When analysed for only those who had at least one admission drug chart discrepancy,

there was no significant difference in LOS between those who had at least one

admission drug chart discrepancy not resolved during the hospital stay and those who

did not (Mann-Whitney test U = 9209.0, z = -0.9, p = 0.4).

With this dataset, there was no significant correlation found between LOS and number

of discrepancies not resolved during the hospital stay (Spearman rank Correlation Rho =

-0.02, n = 284, p = 0.8).


of 767

Table 25 Admission drug chart discrepancies not acted on during the hospital stay versus length of stay for those patients who had admission drug chart discrepancies


At least one admission drug chart discrepancy not resolved during the hospital stay 4 (0-50)

No admission drug chart discrepancies not resolved during the hospital stay 4 (1-29)

9.2.1.5.6 Resolution of admission chart discrepancies effect on length of stay

No differences were found in LOS across the three different group splits. There were

no statistically significant differences in the length of stay between patients who did

have drug chart discrepancies (excluding legitimate therapeutic changes) on their initial

admission drug chart and those who did not.

Table 26 Differences in length of stay between those who did or did not have discrepancies on their admission drug chart


At least one discrepancy on admission drug chart 4 (0-50)

No discrepancies on admission drug chart 4 (0-52)

9.2.1.5.7 Discrepancies identified relating to the community pharmacy dispensing history

All community pharmacy dispensing histories received were used to assist the

construction of the final reconciled medication on admission list. The patient’s drug list

provided by the community pharmacy was then compared with the reconciled list to

identify any discrepancies including OTC medications.

As part of the process of collecting the dispensing histories, trial officers asked

Community Pharmacists to record, or recount during the phone call, any other useful

medication-related information they could share, including known OTC use.


of 767

Differences found on the community pharmacy dispensing history that were considered

as discrepancies included:

• previously ceased medications being dispensed as a current medication (i.e. patient had a current supply at home at time of admission according to dispensing date);

• wrong directions and/or doses, including ‘mdu’;

• new items having not been dispensed; and

• regular OTC items missing from the list.

The project team were interested in considering “if this piece of information was the

sole piece of information they had available to them at admission with which to compile

a medication list, how accurate would it be.”

Of the community pharmacy histories obtained, 58.7% had a least one dispensed

medication-related discrepancy and 49.6% of histories had a least one OTC-related

discrepancy.

Table 27 Community pharmacy originated discrepancies on admission, measured as percentage of patients with at least one discrepancy of this origin

Group CP’s dispensing history

discrepancies, (%)

OTC-related discrepancies,

(%)

Intervention 61.5 53.0

Control 56.7 47.2

All Patients 58.7 49.6

Overall, there was a median of one dispensed medication-related discrepancy per patient

and a median of one OTC-related discrepancy.

Table 28 Community pharmacy originated discrepancies on admission, measured as a median of discrepancies of this origin per patient

Group CP’s dispensing history

discrepancies, median (range)

OTC-related discrepancies,

median (range)

Intervention 1 (0-8) 1 (0-7)

Control 1 (0-16) 1 (0-6)

All Patients 1 (0-16) 1 (0-7)


of 767

9.2.1.6 Comparison of trial participant medication knowledge

The medication knowledge questionnaire analysis was performed using only the

patients from Tasmania and Victoria who were in the trial for the full period. The

Perth-based patients were not used in this analysis as discussed previously in Section

9.2.1.3.

Patients were found to improve in their knowledge over time (Wilcoxon signed rank test

Z = -7.9 p <0.0001). Each group was found to increase their knowledge significantly

over time (Wilcoxon signed rank test all p<0.001). There were no differences between

the three groups at baseline for knowledge scores. However, at 30 days after discharge

patients who received the full intervention had significantly higher drug knowledge than

minimal and partial intervention patients. (Kruskal-Wallis H = 7.1, df = 2, p = 0.03).

Table 29 Comparisons of patient knowledge of their medications, as inpatients and at thirty days post-discharge

Median Knowledge Questionnaire, Score (range) Group

Admission 30 days

Full Intervention 62 (0-100) 92 (0-100)

Partial Intervention 67 (0-100) 83 (0-100)

Minimal Intervention 67 (0-100) 75 (0-100)


of 767

9.2.1.7 Comparison of trial participant self-reported compliance with medication

The compliance survey analysis was performed using only the patients from Tasmania

and Victoria who were in the trial for the full period. The Perth-based patients were

excluded as discussed previously in Section 9.2.1.3.

There was no statistically significant difference between the patients when arranged into

the three group split for their baseline compliance questionnaire (Chi-square = 0.2, df =

2, p = 0.9). At thirty days post-discharge, a trend was found between the groups (Chi-

square = 4.6, df = 2, p = 0.1), suggesting the full intervention group were tending

towards better compliance than the other two groups by 30 days.

Neither the partial intervention nor minimal intervention groups improved over the

study period (Chi-square = 1.3, df = 1, p = 0.2 and Chi-square = 1.6, df = 2, p = 0.2

respectively). However, the full intervention group displayed a significant

improvement in their compliance over the 30 day post-discharge period (Chi-square =

5.1, df = 1, p = 0.02).

Table 30 Comparison of self-reported patient compliance as inpatients and at thirty days post-discharge

Compliance Questionnaire Score, (% fully compliant) Group

Inpatient 30 days

Full Intervention 49.3 71.2

Partial Intervention 48.0 54.0

Minimal Intervention 50.9 58.9


of 767

9.2.1.8 Comparison of actual or potential drug-related problems identified during the trial

Analyses relating to actual and potential drug-related problems (DRPs) were performed

using only the patients from Tasmania and Victoria who were in the trial for the full

period. Perth-based patients were excluded for reasons given in the Section 9.2.1.3.

Each parameter is analysed as both the median number of problems per patient and the

percentage of patients with at least one problem.

9.2.1.8.1 Changes in total number of actual or potential drug-related problems identified with time

DRPs were assessed at baseline (on admission), at discharge from hospital and at 30

days post-discharge. The measure of total drug-related problems comprises all actual or

potential drug-related problems identified for each patient’s medication; and includes:

• Actual or potential therapeutic issues identified through the Cognicare®

database,

• Actual or potential drug interactions identified through the Drug Interactions Facts database,

• Problems reported by the patient, as described in section 0

• Miscellaneous issues arisen from errors throughout the 30 days post-discharge period

9.2.1.8.1.1 Number of total actual or potential drug-related problems identified with time

No significant difference was found in the median number of total actual or potential

DRPs across the groups at admission, discharge or 30 days, (Kruskal-Wallis all p > 0.6).

All three groups experienced a significant increase in the median number of DRPs from

admission to discharge (Wilcoxon signed rank test all p < 0.01). All three groups also

experienced a significant decrease in total DRPs from the point of discharge to 30 days

post-discharge (Wilcoxon signed rank test all p < 0.01). However, over the entire study

period, from admission to 30 days post-discharge the minimal intervention group

displayed a significant increase in the median number of actual or potential DRPs

(Wilcoxon signed rank test T = 10, Z = -3.1, p < 0.002). The partial intervention group

and the full intervention group displayed no significant change over the study period

(Wilcoxon signed rank test all p > 0.2).


of 767

Table 31 Drug-related problems identified with time, measured by the median number of issues per patient

Drug-related issues identified per patient median (range) Group

Admission Discharge 30 days

Full Intervention 5 (0-22) 6 (0-21) 5 (0-21)

Partial Intervention 5 (1-23) 7 (2-30) 6 (0-27)

Minimal Intervention 5 (0-20) 7 (-024) 6 (0-31)

9.2.1.8.1.2 At least one actual or potential drug-related problem with time

Table 32 shows the changes in the numbers of total recorded drug-related problems over

time for the three different groups.

Between 97% and 98% of all patients had at least one DRP at each DRP analysis point.

No significant difference was found across the three groups: at admission (Chi-square =

1.3, df = 2, p = 0.5), discharge (Chi-square = 1.0, df = 2, p = 0.06) or at 30 days (Chi-

square = 0.3, df = 2, p = 0.9).

Table 32 Drug-related problems identified with time, measured by percentage of patients with at least one or more issue

Percentage of patients with at least one drug-related problem, (%) Group


Full Intervention 97.3 98.6 97.2

Partial Intervention 100 100 96.0

Minimal Intervention 97.5 98.1 97.5


of 767

9.2.1.8.2 Changes in total numbers of drug-related issues identified through Cognicare® with time

9.2.1.8.2.1 Median number of Cognicare® problems with time There was no significant difference between the three groups for total number of

significant and moderate DRPs per patient at admission, discharge or 30 days (Kruskal-

Wallis all p > 0.2).

The full intervention group and the partial intervention group experienced a significant

increase in the total number of significant and moderate DRPs per patient from

admission to discharge (Wilcoxon signed rank test T = 4, z = -3.8, p = 0.0002) and

(Wilcoxon signed rank test T = 5, z = -3.1, p = 0.002) respectively. The minimal

intervention group did not experience a significant change over the same time period

(Wilcoxon signed rank test T = 6, z = -1.1, p = 0.3).

During the peri-discharge period (discharge to 30 days post-discharge), the full

intervention group and the partial intervention group experienced a significant decrease

in the total number of significant and moderate DRPs per patient (Wilcoxon signed rank

test T = 4, z = -2.2, p = 0.03) and (Wilcoxon signed rank test T = 4, z = -3.0, p = 0.002)

respectively. In contrast, the minimal intervention group did not experience an

improvement of any significance over the same time period (Wilcoxon signed rank test

T = 6, z = -1.2, p = 0.2).

Overall, the three groups did not significantly change in total number of significant and

moderate DRPs per patient from admission to 30 days post-discharge.

Table 33 Drug-related issues classified as significant or moderate per Cognicare®

identified with time, measured by median number of issues per patient

Significant and Moderate DRPs identified per patient, median (range) Group




Minimal Intervention 3 (0-14) 3 (0-15) 3 (0-21)


of 767

9.2.1.8.2.2 At least one Cognicare® problem with time There was no significant difference across the three groups in percentage of patients

with at least one significant or moderate DRP at admission, discharge or 30 days (Chi-

square = 1.1, df = 2, p = 0.6), (Chi-square = 1.1, df = 2, p = 0.6) and (Chi-square = 0.2,

df = 2, p = 0.9) respectively.

It was also found that there were no changes experienced for any of the groups from

admission to discharge, from discharge to 30 days, or overall from admission to 30 days

post-discharge.

Table 34 Drug-related problems classified as significant or moderate per Cognicare®

identified with time

Percentage of patients with at least one Significant or Moderate DRP, (%) Group



Partial Intervention 84.0 92.0 88.0


9.2.1.8.3 Changes in total numbers of drug related issues identified through Drug Interactions Facts® with time

The number of level 1 or level 2 interactions as identified by the Drug Interaction

Facts® (DIF®) program was recorded for each patient.

9.2.1.8.3.1 Median number of DIF® drug interactions with time There was no significant difference in the number of drug interactions per patient across

the three groups at admission, discharge or 30 days post-discharge (Kruskal-Wallis H =

1.9, df = 2, p = 0.4), (Kruskal-Wallis H = 3.4, df = 2, p = 0.2) and (Kruskal-Wallis H =

2.3, df = 2, p = 0.3) respectively.

All three groups displayed a significant increase in number of drug interactions per

patient from admission to discharge (Wilcoxon signed rank test all p ≤ 0.02). Over the

peri-discharge period, from point of discharge to 30 days, the full intervention group

experienced a significant decrease in the number of drug interactions per patient

(Wilcoxon signed rank test T = 3, z = -2.9, p = 0.004). Over the same time there were

no significant changes for the partial intervention group (Wilcoxon signed rank test T =


of 767

3, z = -0.6, p = 0.6) and the minimal intervention group (Wilcoxon signed rank test T =

3, z = -0.9, p = 0.4).

Over the full study period (admission to 30 days post-discharge) the full intervention

patients did not have an increase in the total number of drug interactions identified

(Wilcoxon signed rank test T = 2, Z = -1, p = .3). Over the same time there was a

significant increase in the number of drug interactions identified for the minimal

intervention group (Wilcoxon signed rank test T = 4, Z =-3.7, p < 0.001) and the partial

intervention group (Wilcoxon signed rank test T – 3, Z = -2.7, p = 0.007).

Table 35 Drug interactions identified using DIF® (level 1 & 2) identified with time

Drug interactions identified per patient, median (range) Group





9.2.1.8.3.2 At least one DIF® drug interaction with time No significant difference across the groups was found for percentage of patients who

experienced at least one actual or potential drug interaction (level 1 or level 2 as

identified by DIF®) at admission (Chi-square = 1.4, df = 2, p = 0.5), discharge (Chi-

square = 2.9, df = 2, p = 0.2) or 30 days post-discharge (Chi-square = 1.0, df = 2, p =

0.6).

All groups displayed a slight increase in percentage of patients with at least one drug

interaction from admission to discharge and then a slight decrease was displayed by the

full and minimal intervention group from discharge to 30 days. The partial intervention

group experienced a slight increase again over that same time period. The full

intervention group displayed the greatest decrease from discharge to 30 days. However,

the decrease in drug interactions was not statistically significant. Overall, the three

groups all experienced a slight increase in percentage of patients who experienced at

least one actual or potential drug interaction from admission to 30 days post-discharge,

but these changes were not found to be significant.


of 767

Table 36 Drug interactions identified using DIF® (level one & 2) identified within the three group split at the three critical assessment points

Percentage of patients with at least one drug interaction identified per patient, (%) Group





9.2.1.8.4 DRPs identified by patients with time Patients were interviewed on admission and at 30 days post-discharge and during both

interviews, were asked if they were experiencing any issues with their medications. The

types of issues reported included issues such as:

• Suspected side effects,

• Dosage form and usage concerns,

• Compliance -related issues,

• Supply-related issues,

• Dissatisfaction with medications,

• Confusion regarding medication management.

Patients were asked open ended questions and not prompted to report specific issues.

At discharge, all reported issues that had not clearly been addressed during the

admission were recorded as issues continuing from admission.

9.2.1.8.4.1 Number of DRPs identified by patient with time No significant difference across the groups was found for number of self-reported DRPs

per patient at admission and at discharge (Kruskal-Wallis H = 1.5, df = 2, p = 0.5) and

(Kruskal-Wallis H = 0.5, df = 2, p = 0.8). At 30 days post-discharge, the minimal

intervention group reported slightly more DRPS than the partial intervention group who

in turn, reported slightly more DRPS than the full intervention group. These differences

in number of self-reported DRPs between the groups were found to be not significant

(Kruskal-Wallis H = 4.4, df = 2, p = 0.1).


of 767

All three groups displayed a significant decrease in self-reported DRPs from admission

to discharge (Wilcoxon rank signed tests all p < 0.004). The full intervention and

partial intervention groups displayed a slight decrease in the number of self-reported

DRPs from discharge to 30 days, but this was found to be not significant (Wilcoxon

signed rank test T = 3, z = -1.0, p = 0.3) and (Wilcoxon signed rank test T = 3, z = -0.4,

p = 0.7) respectively. Conversely, the minimal intervention group significantly

increased in the number of self-reported DRPs at 30 days compared to the number that

they reported at discharge (Wilcoxon signed rank test T = 4, z = -2.3, p = 0.02).

Overall, from admission to 30 days the minimal intervention group reported slightly

more DRPs at 30 days than they did at admission, but this was not found to be

significant (Wilcoxon signed rank test T = 4, z = -0.7, p = 0.5). The partial intervention

group reported slightly fewer DRPs at 30 days than at admission, but again, this was not

found to be significant (Wilcoxon signed rank test T = 4, z = -1.5, p = 0.1). However,

the full intervention group reported significantly fewer DRPs at 30 days than they did at

admission (Wilcoxon signed rank test T = 4, z = -2.8, p = 0.005).

Table 37 Drug-related problems identified by the patient, measured by median number of issues per patient

Patient reported DRPs identified per patient, median (range) Group





9.2.1.8.4.2 At least one DRP identified by patient The groups were similar in percentages of patients with at least one DRP at admission

and discharge (Chi-square = 0.5, df = 2, p = 0.8) and (Chi-square = 0.7, df = 2, p = 0.7)

respectively. At 30 days post-discharge, the minimal intervention group had more

patients reporting at least one DRP than the partial or full intervention groups, however

this difference was not significant (Chi-square = 5.1, df = 2, p = 0.08).

For all three groups, there was no change in percentage of patients who reported at least

one DRP from admission to discharge and from discharge to 30 days. However, across


of 767

the full trial period (admission to 30 days post-discharge) the minimal intervention

group had a slight increase in the percentage of patients who reported at least one DRP,

but this was not a significant change (Chi-square = 0.1, df = 1, p = 0.7). Conversely, the

partial intervention group displayed a slight decrease over the same period, but again,

this was found to not be a significant change (Chi-square = 1.1, df = 1, p = 0.3). The

full intervention group also displayed a decrease in the percentage of patients who

reported at least one DRP from admission to 30 days. This was found to be almost

significant (Chi-square = 3.6, df = 1, p = 0.06).

Table 38 At least one drug-related problem identified by the patient

Patient reported DRPs identified per patient, (%) Group





9.2.1.8.5 Acutual or potential DRPs by DOCUMENT category with time

DRPs were categorised using the DOCUMENT categorisation system.104 To review,

the DOCUMENT system is a categorisation scheme for drug-related problems,

including:

D – Drug Selection,

O – Over or under dose prescribed,

C – Compliance,

U – Untreated Indications,

M – Monitoring,

E – Education or Information,

N – Non-clinical, and

T – Toxicity or Adverse Reaction.

It was found for the setting of this project, the majority of issues fell into the D, C, U

and T categories and details of these DRPs were analysed.


of 767

9.2.1.8.5.1 ‘D’ – Drug Selection Category ‘D’ problems are described as ‘problems related to the choice of drug

prescribed or taken’.

Changes in category ‘D’ DRPs were analysed for statistical significance over time from

admission to discharge to 30 days after discharge.

9.2.1.8.5.1.1 Median number of ‘D’ category DRPs

There were no significant differences in total recorded ‘D’ DRPs per patient for the

three groups at admission (Kruskal-Wallis H = 0.144, df = 2, p = 0.6), discharge

(Kruskal-Wallis H = 2.1, df = 2, p = 0.3) and 30 days (Kruskal-Wallis H = 0.5, df = 2, p

= 0.8).

From admission to discharge, all three groups experienced a significant increase in total

recorded ‘D’ DRPs per patient (Wilcoxon signed rank test all p <0.02). From discharge

to 30 days post-discharge, the full intervention group patients experienced a significant

decrease in number of ‘D’ DRPs (Wilcoxon signed rank test T = 4, z = -3.4, p =

0.0007), whereas the partial intervention (Wilcoxon signed rank test T = 3, z = -0.5, p =

0.6) and minimal intervention group (Wilcoxon signed rank test T = 1, z = -0.5, p =

0.2) patients experienced no change in numbers of ‘D’ DRPs over the same time period.

Over the full duration of the study, from admission to 30 days, all patients experienced

an increase in total recorded ‘D’ DRPs. However, for the minimal intervention group,

this was found to be a significant increase (Wilcoxon signed rank test T = 6, z = -2.3, p

= 0.02) and for the partial intervention group, almost significant (Wilcoxon signed rank

test T = 4, z = -1.8, p = 0.07). Whereas, for the full intervention group, this increase

was not found to be significant (Wilcoxon signed rank test T = 5, z = -1.6, p = 0.1).

Table 39 Category ‘D’ DRPs identified per patient, measured by median number of problems per patient

Category ‘D’ DRPs identified per patient, median (range) Group



Partial Intervention 2 (0-12) 3 (0-15) 2.5 (0-10)



of 767

9.2.1.8.5.1.2 At least one ‘D’ category DRP

On admission no significant differences were found for the percentage of patients who

had at least one ‘D’ DRP (Chi-square = 0.885, df = 2, p = 0.6423). At discharge, the

full intervention group had significantly more patients with at least one ‘D’ DRP than

the other two groups (Chi-square = 5.824, df = 2, p = 0.0544). By 30 days post-

discharge, there was no significant difference between the three groups (Chi-square =

2.212, df = 2, p = 0.3308).

There were no significant changes in percentage of patients with at least one ‘D’ DRP

from admission to discharge, from discharge to 30 days and from admission to 30 days.

Table 40 Percentage of patients who had at least one category ‘D’ DRP(s) identified.

Percentage of patients who had at least one category ‘D’ DRP(s) identified, (%) Group





9.2.1.8.5.2 ‘C’ – Compliance Category ‘C’ problems are described as ‘problems related to the way the patient takes

the medication.

Changes in category ‘C’ DRPs were analysed for statistical significance over time from


9.2.1.8.5.2.1 Median number of ‘C’ category DRPs

No significant difference in total recorded ‘C’ DRPs per patient were found at

admission (Kruskal-Wallis H = 1.02, df = 2, p = 0.6) or at discharge (Kruskal-Wallis H

= 0.5, df = 2, p = 0.8). At 30 days post-discharge, the full intervention group had fewer

recorded ‘C’ DRPs per patient than the partial intervention group, who in turn, had

fewer than the minimal intervention group. These differences were not found to be

significant (Kruskal-Wallis H = 4.7, df = 2, p = 0.1).

During the peri-discharge period, from discharge to 30 days, the full and partial

intervention patients displayed a decrease in the number of recorded ‘C’ DRPs, but this


of 767

decrease was only significant for the full intervention group (Wilcoxon signed rank test

T = 2, z = -2.4, p = 0.02) and not for the partial intervention group (Wilcoxon signed

rank test T = 2, z = -0.7, p = 0.5). The minimal intervention group displayed a slight

increase in total recorded ‘C’ DRPs, but this increase was not significant (Wilcoxon

signed rank test T = 2, z = -0.9, p = 0.3).

Overall, from admission to 30 days post-discharge, the full intervention group displayed

a significant decrease in total number of recorded ‘C’ DRPs (Wilcoxon signed rank test

T = 2, z = -3.0, p = 0.003). The partial intervention group also displayed a decrease over

this time period, but this was not found to be significant (Wilcoxon signed rank test T =

2, z = -1.7, p = 0.09). In comparison, the minimal intervention group did not display

any significant change in total recorded ‘C’ DRPs from admission to 30 days (Wilcoxon

signed rank test T = 2, z = -0.2, p = 0.8).

Table 41 Category ‘C’ DRPs identified per patient, measured by median number of problems per patient

Category ‘C’ DRPs identified per patient, median (range) Group





9.2.1.8.5.2.2 At least one ‘C’ category DRP

There was no significant difference found across the three groups for percentage of

patients with at least one ‘C’ DRP at admission or at discharge (Chi-square = 0.9, df =

2, p = 0.6) and (Chi-square = 0.6, df = 2, p = 0.8) respectively. At 30 days post-

discharge, the full intervention group had fewer patients with at least one ‘C’ DRP than

the partial intervention group, who in turn had fewer than the minimal intervention

group. These differences were found to be almost significant (Chi-square = 5.6, df = 2,

p = 0.06).

Neither the partial intervention group nor the minimal intervention group displayed any

changes in percentage of patients with at least one ‘C’ DRP from admission to

discharge, from discharge to 30 days or from admission to 30 days. In comparison, the


of 767

full intervention group displayed no change from admission to discharge (Chi-square =

0.3, df = 1, p = 0.6), but from discharge to 30 days post-discharge, the percentage of

full intervention patients with at least one ‘C’ DRP decreased significantly (Chi-square

= 3.9, df =1, p = 0.05). Overall, from admission to 30 days post-discharge, the full

intervention group displayed a significant decrease in the percentage of patients with at

least one ‘C’ DRP (Chi-square = 6.2, df = 1, p = 0.01).

Table 42 Percentage of patients who had at least one category ‘C’ DRP(s) identified.

Percentage of patients who had at least one category ‘C’ DRP(s) identified (%) Group





9.2.1.8.5.3 ‘U’ - Untreated indications Category ‘U’ problems are described as ‘problems related to actual or potential

conditions that require management’.

Changes in category ‘U’ DRPs were analysed for statistical significance over time from


9.2.1.8.5.3.1 Median number of ‘U’ category DRPs

There were no significant differences found between the groups for total number of

recorded ‘U’ DRPs per patient at admission (Kruskal-Wallis H = 0.163, df = 2, p =

0.9216). From admission to discharge, all three groups experienced a significant

increase in number of ‘U’ DRPs per patient (Wilcoxon signed rank test all p < 0.02) but

at the point of discharge, the full intervention group had less than the partial

intervention group, who in turn had less than the minimal intervention group. These

differences at this point were found to be significant (Kruskal-Wallis H = 20.438, df =

2, p <0.0001). From discharge to 30 days, all groups experienced a decrease in total ‘U’

DRPs per patient, however this was only found to be significant for the partial and

minimal intervention groups (Wilcoxon signed rank test T = 2, z = -2.411, p = 0.0159)

and (Wilcoxon signed rank test T = 5, z = -5.706, p < 0.0001) respectively. The


of 767

decrease for the full intervention group was found to be almost significant (Wilcoxon

signed rank test T = 2, z = -1.789, p = 0.0736), however, like the intervention – PDMR

model in the four group split, this group had the least number of ‘U’ DRPs per patient at

discharge.

At 30 days post-discharge, the full intervention group had less ‘U’ DRPs per patient

than the partial intervention group, who in turn had less than the minimal intervention

group. These differences were found to be significant (Kruskal-Wallis H = 6.225, df =

2, p = 0.0445). Overall, from admission to 30 days post-discharge the partial

intervention group and the minimal intervention group experienced a significant

increase in the total number of ‘U’ DRPs per patient (Wilcoxon signed rank test T = 1, z

= -2.201, p = 0.0277) and (Wilcoxon signed rank test T = 5, z = -4.284, p < 0.0001)

respectively. In comparison, the full intervention group patients experienced no

significant change in the number of ‘U’ DRPs over the same period (Wilcoxon signed

rank test T = 1, z = -0.630, p = 0.5286).

Table 43 Category ‘U’ DRPs identified per patient, measured by median number of problems per patient

Category ‘U’ DRPs identified per patient, median (range) Group





9.2.1.8.5.3.2 At least one ‘U’ category DRP

There was no difference across the groups for percentage of patients with at least one

‘U’ DRP on admission (Chi-square = 1.215, df = 2, p = 0.5447).

From admission to discharge all three groups experienced a significant increase in

percentage of patients who had at least one ‘U’ DRP (χ2 all p < 0.05), but at the point of

discharge, the full intervention group had fewer patients with at least one ‘U’ DRP than

the partial intervention group, who in turn had fewer than the minimal intervention

group. These differences were found to be significant (Chi-square = 24.930, df = 2, p <

0.0001).


of 767

From discharge to 30 days, the full intervention group experienced a small decrease in

percentage of patients with at least one ‘U’ DRP, but this decrease was not significant

(Chi-square = 2.527, df = 1, p = 0.1119) and the partial and the minimal intervention

groups did experience a significant decrease over the same time period (Chi-square =

4.000, df = 1, p = 0.0455) and (Chi-square = 16.565, df = 1, p < 0.0001) respectively.

At 30 days after discharge the full intervention group had fewer patients with at least

one ‘U’ DRP than the partial intervention group, who in turn had fewer than the

minimal intervention group (Chi-square = 13.605, df = 2, p = 0.0011).

Overall, from admission to 30 days post-discharge, the full intervention group were

found to not experience any significant change for the percentage of patients with at

least one ‘U’ DRP the full intervention group had fewer patients with at least one ‘U’

DRP than the partial intervention group, who in turn had fewer than the minimal

intervention group. These differences were found to be not significant (Chi-square =

0.529, df = 1, p = 0.4670). In comparison, the partial intervention group displayed a

significant increase in percentage of patients with at least one ‘U’ DRP over the

complete study period (Chi-square = 3.840, df = 1, p = 0.05). The minimal intervention

group were found to also display an increase in percentage of patients with at least one

‘U’ DRP, but this was found to not quite be significant (Chi-square = 2.700, df = 1, p =

0.1003).

Table 44 Percentage of patients who had at least one category ‘U’ DRP(s) identified

Percentage of patients who had at least one category ‘U’ DRP(s) identified, (%) Group





9.2.1.8.5.4 ‘T’ – Toxicity or Adverse Reaction Category ‘T’ problems are classed are described as ‘problems related to the presence of

signs or symptoms which are suspected to be related to an adverse effect of the drug’.

Changes in category ‘T’ DRPs were analysed for statistical significance over time from



of 767

9.2.1.8.5.4.1 Median number of ‘T’ category DRPs

No significant difference was found across the three groups for total number of ‘T’

DRPs at admission, discharge or 30 days (Kruskal-Wallis H = 0.335, df = 2, p =

0.8457), (Kruskal-Wallis H = 2.350, df = 2, p = 0.3088) and (Kruskal-Wallis H = 0.078,

df = 2, p = 0.9620) respectively.

From admission to discharge, both the full intervention group and the partial

intervention group experienced an increase in total number of ‘T’ DRPs from admission

to discharge, however, this was only found to be a significant increase for the partial

intervention group. (Wilcoxon signed rank test T = 3, z = -1.647, p = 0.0996 and

(Wilcoxon signed rank test T = 4, z = -2.519, p = 0.0118) respectively. The minimal

intervention group displayed no change over the same time period (Wilcoxon signed

rank test T = 4, z = -0.044, p = 0.9652).

Conversely, from discharge to 30 days, both the full intervention group and the partial

intervention group experienced a decrease in total ‘T’ DRPs per patients, but again, this

was only a significant decrease for the partial intervention group (Wilcoxon signed rank

test T = 4, z = -0.401, p = 0.6886) and (Wilcoxon signed rank test T = 4, z = -2.188, p =

0.0287) respectively. The minimal intervention group displayed no change over this

time period either (Wilcoxon signed rank test T = 5, z = -0.094, p = 0.9250).

For the duration of the trial, all three groups did not experience a change of any

significance in total number of recorded ‘T’ DRPs from admission to 30 days.

Table 45 Category ‘T’ DRPs identified per patient, measured by median number of problems per patient

Category ‘T’ DRPs identified per patient, median (range) Group






of 767

9.2.1.8.5.4.2 At least one ‘T’ category DRP

No significant differences in percentage of patients with at least one ‘T’ DRP were

found between the three groups at admission (Chi-square = 0.802, df = 2, p = 0.6696),

discharge (Chi-square = 0.973, df = 2, p = 0.6149) and 30 days (Chi-square = 0.869, df

= 2, p = 0.6475).

All groups experienced a slight increase in percentage of patients who had at least one

‘T’ DRP from admission to 30 days. However, no significant changes were found for

each of the three groups from admission to discharge, from discharge to 30 days or from

admission to 30 days.

Table 46 Percentage on patients with at least one category ‘T’ DRP(s) identified.

Percentage of patients who had at least one category ‘T’ DRP(s) identified, (%) Group






of 767

9.2.1.9 PDMR/HMR related-results

9.2.1.9.1 Uptake Sixty-seven (96%) of a possible 71 PDMRs were conducted within 30 days after

hospital discharge for the PDMR group as displayed in Table 47. The uptake of HMRs

in the streamlined referral group was 16 (22%) from 73 referrals. There were two

HMRs conducted in the control group, one for each of the no recommendation and

passive recommendation group.

Table 47 Uptake of each model of medication reviews

Group Number of patients in each group

Patients who received a HMR during the 30 day period,

n (%) Intervention

PDMR model 71 67 (96%)

Streamlined HMR recommendation 73 16 (22%)

Control

HMR recommendation (passive) 123 1 (0.9%)

No HMR recommendation 115 1 (0.8%)

The median time to the initial pharmacist visit to conduct the patient interview was 6

days in the PDMR group and 21 days in the streamlined HMR recommendation (P

=0.0001, U=205.5). All of the streamlined HMRs were coordinated by the patients’

community pharmacy. Fourty eight (72%) of the PDMRs were conducted by accredited

pharmacists within the community without the assistance of the trial office.

Table 48 Time after discharge to conduct home visit

Group Days since discharge review performed, median (range)

Intervention

PDMR model 6 (1-32)

Streamlined HMR recommendation 21 (2-50)

Control

HMR recommendation 31 (only one patient)

No HMR recommendation 5 (only one patient)


of 767

9.2.1.10 Low uptake of PDMR/HMR recommendations During the trial process, concerns mounted surrounding the low uptake of

recommendations made in the trial PDMRs/HMRs. One thought was that the manner in

which the recommendations were communicated in some of the reviews may have

contributed to this. It was decided to perform a simple subjective analysis of the

PDMR/HMRs received to try to establish an idea of the quality of the reviews received.

Although it was beyond the scope of the request for tender, an independent analysis of

the reviews was performed by a recognised expert. The consultant’s report is presented

in Appendix XXXV.

9.2.1.11 Key Findings The Med eSupport trial and its evaluation is a substantial piece of work covering many

disciplines each with many outcome measures. Additionally, outcome measures were

investigated in detail. Consequently, the volume of results can be overwhelming for the

reader. To consolidate the clinical evaluation chapter, a table of key findings are

presented in Table 49.


of 767

Table 49 Key findings of the Med eSupport trial

Section Finding 9.2.1.5.2.1 A significantly greater number of discrepancies per patient were resolved

within the first 48 hours of admission for the intervention group than for the control.

9.2.1.5.3.1 Significantly more discrepancies were resolved prior to discharge for intervention patients than for control patients.

9.2.1.5.3.2 LOS increased with number of discrepancies not resolved at 48 hours.

9.2.1.6 Generally, all patients were found to improve their knowledge over time. However, at 30 days after discharge patients who received the full intervention had significantly higher drug knowledge than minimal and partial intervention patients.

9.2.1.7 The full intervention group displayed a significant improvement in their compliance over the 30 day post-discharge period.

9.2.1.8.2.1 During the peri-discharge period (discharge to 30 days post-discharge), the full intervention group and the partial intervention group experienced a significant decrease in the total number of significant and moderate DRPs per patient.

9.2.1.8.3.1 Over the full study period (admission to 30 days post-discharge) the full intervention patients did not have an increase in the total number of drug interactions identified. In comparison, over the same period the minimal intervention and partial intervention patients did have a significant increase in drug interactions identified by Drug Interaction Facts® software.

9.2.1.8.4.1 Patient identified drug-related problems reported by the full intervention group were significantly fewer than the other groups over the period from admission to 30 days post-discharge.

9.2.1.8.5.1.1 Generally, all patients experienced an increase in drug selection DRPs over the period from admission to 30 days post-discharge. However, in the full intervention group of patients this increase was not significant, where it was in the other groups.

9.2.1.8.5.2.1 Overall, from admission to 30 days post-discharge, the full intervention group displayed a significant decrease in total number of recorded compliance DRPs.

9.2.1.8.5.3.1 Generally, all patients experienced an increase in untreated indications DRPs over the period from admission to 30 days post-discharge. However, in the full intervention group of patients this increase was not significant, where it was in the other groups.


of 767

9.2.2 Results of anonymous participant surveys

9.2.2.1 Comparison of results for the anonymous participant satisfaction surveys

After patients received a 30 day phone call they were sent an anonymous survey. Of the

408 surveys sent out, 279 were returned giving an overall response rate of 68%. As the

surveys were anonymous, their results could not be excluded later if the patient was

withdrawn from the trial for another reason (Section 9.2.1.3). Patients assigned to the

control and intervention groups returned 171 (70%), 108 (65%) surveys respectively.

A breakdown of the number of respondents in each study arm for each location is

presented for patients and providers in Table 50 and Table 51 respectively.

Three different surveys were sent to patients enrolled in the trial depending on the group

they were originally allocated to. Results for the patient respondents are presented in

Table 52. The results of the General Practitioner and Community Pharmacist

respondents are presented in Table 53 and Table 54 respectively.

9.2.2.2 Key Findings

9.2.2.2.1 Patient survey responses • Patients who had a PDMR were more likely to want a home visit by a

pharmacist to be available in the future. (χ2 = 12.043, p = 0.002, df = 2). It is likely that exposure to a service increases future uptake.

• When asked how much money they felt they would pay for a home visit by an accredited pharmacist most replied they would pay less than $20. Common reasons for not wanting to pay was the perception they could get the same information from their GP or Community Pharmacist.

• Patients in the PDMR group were more likely to feel confident about their medications after discharge (χ2 = 11.59 p = 0.021 df = 4)

9.2.2.2.2 General Practitioner survey responses • Most GP respondents thought that the medication summary was provided

within an adequate timeframe and they strongly agreed that receiving discharge medication information in the future would be valuable. There were no statistical differences across groups for the timeframe that General Practitioners received information.

• General Practitioners who had patients in the PDMR trial were more likely to think that Med eSupport gave them a clearer picture of their patient’s medication on discharge (Mann-Whitney U = 323.50 Z = -2.25 p =0.024)

• The General Practitioners who had patients in the PDMR trial arm were more likely to use the website (χ2 = 5.449, p < 0.02, df = 1). However, there


of 767

was a trend for the group of General Practitioners who had patients in the PDMR trial arm to state the website was more difficult to use. (Mann Whitney U = 22, Z = -1.5, p = 0.1)

• There was a very positive response when General Practitioners were asked if the PDMR/HMR assisted them in the medication management of their patient.

• When General Practitioners were asked if they would like to see an automatic PDMR for their patients in the future 74% responded that they wanted the service. There were no significant differences between the study groups. (χ2 = 3.8, p = 0.15 df = 2)

• There was a positive response from GPs when asked if they thought that the study benefited them in optimising the patient’s medication management through improved communication of medication related information.

9.2.2.2.3 Community Pharmacists results • Community pharmacists whose patients received a PDMR were more likely

to use the website (χ2 = 5.7, p = 0.02, df =1) than those whose patients were in the streamlined HMR recommendation group.

• Community pharmacists were asked to respond how useful the PDMR/HMR outcomes were in the medication management of the patient using a likert scale (0 = Strongly Disagree, 100 = Strongly Agree) Range = 7 – 100, median 74). There were no differences between study arms (Mann Whitney U = 215.5, Z = -1.3, p = 0.17)

• Patient group allocation did not influence the opinion of community pharmacists regarding whether or not an automatic PDMR should be a regular service in the future. (χ2 = 2.79, p = 0.24 df = 2)

• Community pharmacists rated the value of receiving a discharge medication summary in giving them a clearer understanding of the patient’s medication management at discharge more highly than general practitioners. (Wilcoxon signed rank test T = 17, z = -3.4, p < 0.001)


of 767

9.2.2.3 Results of the participating patient anonymous satisfaction surveys

Table 50 Number of patient surveys returned from each trial site

Trial Site Control n (%)

Intervention HMR n (%)

Intervention PDMR n (%)

RHH 111 (65%) 43 (78%) 36 (68%)

LGH 12 (7%) 1 (2%) 3 (5%)

Bendigo 5 (3%) 0 (0%) 2 (4%)

SCGH 33 (19%) 10 (18%) 11 (21%)

HPH 10 (6%) 1 (2%) 1 (2%)

Total 171 (100%) 55 (100%) 53 (100%)

Table 51 Provider responses by trial site

Trial Site GP HMR n (%)

GP PDMR n (%)

CP HMR n (%)

CP PDMR n (%)

RHH 21 (81 %) 28 (72%) 28 (74%) 26 (67%)

LGH 1 (4%) 1 (3%) 2 (5%) 3 (7%)

Bendigo 0 (0%) 2 (5%) 0 (0%) 1 (3%)

SCGH 4 (15%) 8 (20%) 6 (16%) 9 (23%)

HPH 0 (0%) 0 (0%) 2 (5%) 0 (0%)

Total 26 (100%) 39 (100%) 38 (100%) 39 (100%)

Please note: responses given in the following tables represent the raw data received.

Anomalies reflect the data collected.


of 767

Table 52 Patient survey responses

Question

Response

Control

n (%)

Intervention HMR, n (%)

Intervention PDMR, n (%)

How would you rate the quality of information you received about your medications when you left hospital?

Excellent 83 (49%) 30 (56%) 30 (56%)

Good 70 (42%) 20 (38%) 20 (38%)

Fair 10 (6%) 1 (2%) 2 (4%)

Poor 5 (3%) 2 (4%) 1 (2%)

Did you experience any difficulties with your medications after you left hospital?

Yes 24 (14%) 8 (15%) 10 (19%)

No 146 (86%) 44 (85%) 42 (81%)

If yes, were the difficulties fixed?

Yes 23 (74%) 5 (63%) 9 (90%)

No 8 (26%) 3 (37%) 1 (10%)

If they were fixed, how? (more than one answer was possible)

Visit to GP 20 (61%) 6 (46%) 3 (23%)

Visit to Community Pharmacist 2 (6%) 2 (15%) 1 (7%)

Home visit by Pharmacist 1 (3%) 2 (15%) 8 (63%)

By myself or my carer 2 (6%) 3 (24%) 1 (7%)

By contacting the hospital 8 (25%) 0 (0%) 0 (0%)

Did you receive a home visit from a Pharmacist to discuss your medications?

Yes 10 (6%) 13 (25%) (100%)

No 149 (93%) 39 (73%) (0%)

A visit has been planned 1 (1%) 1 (2%) (0%)

If you did receive a home visit from a Pharmacist, did you find it useful?

Yes, it helped 10 (84%) 11 (92%) 44 (90%)

It didn’t make much difference 2 (16%) 1 (8%) 5 (10%)

No, it made it worse 0 (0%) 0 (0%) 0 (0%)

Would you pay for a home visit from a Pharmacist in the future?

Yes 7 (15%) 3 (14%) 10 (20%)

No 41 (85%) 19 (86%) 41 (80%)


of 767

If yes, mark how much you would be prepared to pay:

$1 - $10 5 (72%) 2 (67%) 6 (60%)

$11 - $20 0 (0%) 1 (33%) 2 (20%)

$21 - $30 1 (14%) 0 (0%) 2 (20%)

$31 - $40 0 (0%) 0 (0%) 0 (0%)

$41 - $50 1 (14%) 0 (0%) 0 (0%)

> $50 0 (0%) 0 (0%) 0 (0%)

If no, why not?

(More than one answer possible)

I don’t think it is worth it 3 (7%) 1 (5%) 0 (0%)

I think it should be a free service 7 (17%) 8 (42%) 11 (26%)

I could get the same information from visiting my Pharmacist 19 (48%) 6 (32%) 23 (55%)

I could get the same information from visiting my GP 27 (68%) 10 (53%) 20 (48%)

The government should pay 5 (13%) 2 (5%) 6 (14%)

Other… 1 (3%) 0 (0%) 2 (5%)

Do you feel more confident about your medications than you did when you went into hospital

Yes, I feel more confident 60 (36%) 25 (47%) 31 (61%)

I don’t feel any different 102 (61%) 25 (47%) 19 (37%)

No, I feel less confident 6 (3%) 3 (6%) 1 (2%)

Do you think there should be an ongoing medication support service available to people after they leave hospital?

Yes 118 (72%) 40 (77%) 39 (74%)

Unsure 30 (18%) 11 (21%) 10 (19%)

No 16 (10%) 1 (2%) 4 (7%)

If yes, which of the following would you like to see available in the future?

(more than one answer possible)

Provision of a medication information sheet when discharged from hospital 93 (71%) 35 (79%) 32 (80%)

Communication of discharge medication information to your GP 95 (73%) 26 (59%) 30 (75%)

Communication of discharge medication information to your Pharmacist 51 (39%) 15 (34%) 22 (55%)

Access to personal, up-to-date medication information on a secure website

7 (5%) 5 (11%) 8 (20%)


of 767

A home visit from a Pharmacist shortly after discharge from hospital 26 (20%) 12 (27%) 19 (48%)

Other…. 2 (2%) 1 (5%) 4 (10%)

Is there any information or services you would have liked about your medications but didn’t receive?

Yes 17 (12%) 3 (7%) 4 (9%)

Not sure 20 (14%) 5 (11%) 1 (2%)

No 108 (74%) 38 (82%) 39 (89%)


of 767

9.2.2.4 Results of the participating Community Health Professional anonymous satisfaction survey

Table 53 General Practitioner survey responses indicating median and range

Receiving the discharge medication summary through Med eSupport provided me with a clearer picture of my patient’s discharge medication

Intervention HMR

Strongly Disagree Strongly Agree

Intervention PDMR


The discharge medication summary was received within an adequate time frame

Intervention HMR


Intervention PDMR


I believe receiving discharge medication information in the future would be valuable

Intervention HMR


Intervention PDMR


Question

Response



If so, which format would you prefer to receive them (more than one answer allowed)

Fax 14 (58%) 21 (54%)

Secure email 7 (29%) 17 (44%)

Password-protected website 1 (4%) 2 (5%)

Post 6 (25%) 11 (28%)

Other 3 (13%) 2 (5%)


of 767

Did you access the Med eSupport Website regarding this patient’s medication management?

Yes 2 (8%) 13 (33%)

No 23 (92%) 26 (67%)

If not, why not?

I do not know how 5 (25%) 2 (10%)

I do not have a computer 2 (10%) 0 (0%)

I have not had time 7 (35%) 4 (20%)

I have not felt I needed to view it 6 (30%) 8 (40%)

I am not interested in viewing my patient’s information this way 3 (15%) 2 (10%)

I do not believe this is a useful service 2 (10%) 0 (0%)

There was a technical error 0 (0%) 4 (20 %)

Other…. 1 (5%) 5 (25%)

How did you find the Med eSupport website to navigate

Intervention HMR

Difficult Easy

Intervention PDMR

Difficult Easy

Question

Response/Group



What aspects of the Med eSupport website did you find useful? (more than one answer possible)

Access to dispensed medication history 0 (0%) 5 (45%)

Current medication list 1 (50%) 7 (64%)

Discharge medication information 1 (50%) 7 (64%)

Ability to maintain an up to date medication list for this patient 0 (0%) 0 (0%)

Ability to create up to date medication counselling sheets 0 (0%) 0 (0%)

Ability to create up to date weekly checklist 0 (0%) 0 (0%)

Access to related links 0 (0%) 0 (0%)

Other… 0 (0%) 0 (0%)


of 767

Did you find there were any errors or inconsistencies in the patient’s initial medication information provided on the website?

Yes 0 (0%) 1 (8%)

Unsure 1 (50%) 3 (23%)

No 1 (50%) 9 (69%)

Did you refer this patient for a HMR using the trial pre-filled referral form?

(Question not asked to PDMR respondents)

Yes 6 (25%) N/A

I used another method of referral 0 (0%) N/A

No 18 (75%) N/A

If you did refer the patient using the pre-filled trial referral form, did you find the process user friendly?

(Question not asked to PDMR respondents)

Yes, it made it easier 4 (80%) N/A

Not much different to the current system 1 (20%) N/A

No, it was more difficult 0 (0%) N/A

The outcomes of the PDMR/HMR assisted me in the medication management of this patient

Intervention HMR


Intervention PDMR


I believe this patient feels they have benefited from the PDMR/HMR

Intervention HMR


Intervention PDMR



of 767

Question

Response/Group

Intervention HMR n (%)


Do you think there should be an “automatic” post-discharge medication review process for patients with risk factors for medication misadventure (eg elderly, multiple medications) post-discharge?

Yes 12 (60%) 31 (82%)

Unsure 5 (25%) 3 (8%)

No 3 (15%) 4 (10%)

I believe my patient now has a greater understanding of their medication after being involved in this study

Intervention HMR


Intervention PDMR


I believe this study benefited me in optimising this patient’s medication management through improved communication of medication related information.

Intervention HMR


Intervention PDMR



of 767

Table 54 Community Pharmacist survey responses indicating median and range

Question

Response/Group



Receiving the discharge medication summary through Med eSupport provided me with a clearer picture of my patient’s discharge medication

Intervention HMR


Intervention PDMR


I was able to further improve this patient’s medication management as a result of receiving this summary

Intervention HMR


Intervention PDMR


The potential issues to follow-up, listed to me from the hospital, facilitated my assistance in the medication management of this patient

Intervention HMR


Intervention PDMR


The discharge medication summary was received within an adequate time frame

Intervention HMR


Intervention PDMR



of 767

If so, which format would you prefer to receive them?

(more than answer possible)

Fax 24 (65%) 27 (69%)

Secure email 9 (24%) 16 (41%)


Post 7 (19%) 6 (15%)

Other 1 (3%) 0 (0%)

Did you access the Med eSupport Website regarding this patient’s medication management?

Yes 13 (34%) 24 (62%)

No 25 (66%) 15 (38%)

If not, why not ?


I do not know how 0 (0%) 1 (10%)




I am not interested in viewing my patients information this way 0 (0%) 0 (0%)


How did you find the Med eSupport website to navigate?

Intervention HMR

Difficult Easy

Intervention PDMR

Difficult Easy What aspects of the Med eSupport website did you find useful?








of 767


Other……. 1 (7%) 0 (0%)

Did you find there were any errors or inconsistencies in the patient’s initial medication information provided on the website?

Yes 3 (23%) 4 (20%)

Not Sure 2 (15%) 6 (30%)

No 8 (62%) 10 (50%)

Was your Pharmacy involved in performing the PDMR/HMR for this patient?

Yes 9 (100%) 34 (87%)

No 0 (0%) 5 (13%)

The outcomes of the PDMR/HMR assisted me in the medication management of this patient

Intervention HMR


Intervention PDMR

Strongly Disagree Strongly Agree I believe this patient feels they have benefited from the PDMR/HMR

Intervention HMR


Intervention PDMR

Strongly Disagree Strongly Agree Do you think there should be an “automatic” post-discharge medication review process for patients with risk factors for medication misadventure (eg elderly, multiple medications) post-discharge?

Yes 12 (74%) 18 (90%)

Unsure 2 (13%) 0 (0%)

No 2 (13%) 2 (10%)

I believe my patient now has a greater understanding of their medication after being involved in this study

Intervention HMR



of 767

Intervention PDMR


I believe this study benefited me in optimising this patient’s medication management

Intervention HMR


Intervention PDMR


I would like to see services such as this continuing in the future

Intervention HMR


Intervention PDMR


Do you think involvement in the transfer of patients to and from hospital is an important role for community pharmacy?

Intervention HMR

No importance Extremely important

Intervention PDMR

No importance Extremely important


of 767

9.3 Discussion

9.3.1 Discussion of the Med eSupport trial The discussion of the clinical evaluation is presented in section 13. Being the key

outcome measures, they are best presented at the end of the report as part of the overall

discussion.

9.3.2 Discussion of anonymous participant surveys Respondents were not asked a general question about their overall impressions or

perceived effectiveness of the Med eSupport project as it was thought that most

participants would not have a good understanding of the complete trial and its

processes. Further to this, many patients were only exposed to elements of the project.

Consequently, questions were structured in such a way that participants were asked

about the specific aspects of the trial individually.

In general, all patient respondents thought the quality and level of the information that

they received on discharge was good or excellent; they did not want any additional

information or services; and they thought there should be ongoing medication support

available to people after they leave hospital. When asked what they would like to

receive when they left hospital in the future, the most common forms of information

that patients requested were a counselling sheet and their discharge summary sent to

their GP.

Patients who had a PDMR generally felt more confident about their medications post-

discharge. Perhaps this is because they received timely education post-discharge in the

form of a pharmacist visitng them in their home within a week of their discharge and

reviewing their medication management with them.

General Practitioners generally find patient discharge medication information to be

useful, and their preferred method of receiving this information is fax. Next most

popular method was secure email.

Community Pharmacist respondents were more inclined than GPs to agree that

participation in Med eSupport gave them a clearer picture of their patients’ discharge

medication. It is common practice that a discharge summary is sent to a patient’s GP

and not the Community Pharmacist. The Med eSupport trial process also sent

information to the patient’s Community Pharmacist. Anecdotally, feedback from


of 767

Community Pharmacists was positive, especially for patients who had weekly dose

administration aids.

The feedback from the Community Pharmacist respondents was positive. Many

respondents agreed that the medication summary arrived within an adequate time. The

most popular methods of receiving discharge summary information were either by fax

or secure emails. Community Pharmacist respondents agreed they thought their

patients’ knowledge of their medication increased and they would like to see similar

services in the future. Community Pharmacists generally thought their involvement in

the transfer of patients to and from hospital was important. A great majority of the

patients enrolled used a single community pharmacy to have all their medications

dispensed. Although this was one of the selection criteria, as previously described, very

few patients had to be excluded due to them not having a regular pharmacy.

Community pharmacists are well-placed to provide medication information to hospitals

and furthermore see their role in this area as important.

ICT systems evaluation

of 767

10 ICT systems evaluation 10.1 Methodology An overview of ICT study methodology will be presented in the following section

discussing project planning, implementation and methods.

10.1.1 Planning stage The plan was reported in the tender document, section 2.A.1. The team conceived a

model where an electronic communication pathway would be developed to facilitate

transfer of a patient’s medication history from community pharmacy to hospital. The

proposed communication model was two directional, so that current medication

information would be available to the patient’s primary health care providers at the time

of discharge from hospital. It was intended that this model was to work with the

Phoenix Rex® and PCA-Nu systems Winifred® dispensing systems, with the potential for

other dispensing systems to be integrated at a later date.

The transfer of information was to comply with PGA Privacy standards, with

appropriate security standards in place. It was intended to use HL7 messaging and SSL

Secured 128 bit “Bank Level” encryption.

It was planned to use the Pharmcare® package which provides patients at the RHH with

a counselling document on discharge, and to integrate Pharmcare® into the Med

eSupport package.

Upon a patient’s discharge from hospital, information for GPs was to be automatically

faxed to their surgery.


of 767

Patient

General Practitioner

Community Pharmacist

Hospital Pharmacist

Hospital firewall

PharmCare Sheet

Web ServerSituated at University of Tasmania

Figure 17 Proposed network model

10.1.2 Implementation Table 55 shows a time-line of the work done with regards to ICT.

Table 55 Timeline of work done with regards to ICT

Date Work done November 2003

Request for tender submitted to the PGA where the following was proposed: An electronic communication pathway for medication profiles between

community and hospital pharmacies. Routine supply of a comprehensive medication information sheet to the

patient/carer prior to discharge from hospital. Automated faxing or encrypted emailing of the medication information

sheet to the patient’s GP at discharge. A model whereby suitable patients are referred for a HMR after

discharge from hospital, incorporating an electronic alerting mechanism.

March 2004 Project commences.


of 767

March-May 2004

Planning time: Wrote specifications for program use. Considered different models of ICT architecture. Instigated writing an interface for Rex® and Winifred® community pharmacy dispensary systems. Met with various stakeholders to determine requirements and proposed functionality of program.

July 2004 It was determined that the Pharmcare® package did not use standardised data and was not able to be integrated with other software without significant modifications. This obstacle was overcome by out-sourcing the creation of a web-based program that created counselling sheets that were incorporated into the Med eSupport web site. Furthermore using a web-based solution to generate counselling sheets would allow sites away from the RHH better access to the ICT solutions provided. However this delay cost the project approximately three months.

August 2004

Visit to Perth and Bendigo sites, in part to establish ICT requirements and the scope of ICT solutions required.

30th August 2004

Work on the drug linking table commences. The project team had to overcome incompatibilities with drug information stored in various computer programs. Linking tables had to be constructed to allow different software products to exchange information. The final drug linking table had over 34,500 entries.

September 2004

The Med eSupport website was due to go online

1st October 2004

IT Coordinator (JE) left the project.

October 2004

Technical issues with a firewall at the RHH prevented hospital pharmacists accessing the Med eSupport web site. Resolution of this issue involved liaising between the trial staff, ICT staff at the university and ICT staff at the RHH. This issue took two weeks to resolve.

15th November 2004

The launch of the website was postponed to December 2004. This was proposed to be a fully working system, without extra auditing and HL7 (security) protocols. A second stage of rollout was proposed for late January 2005 that included these features.

30th November 2004

Submission for a project extension to September 2005. This was in part caused by delays in delivery of the ICT components.

6th December 2004

The first patient’s details were added to the Web site. Over the course of the next few months, minor changes were made to the website interface as problems were discovered.

7th December 2004

The drug linking table for Winifred® and Rex® dispensing systems are complete.

January 2005

Revised delivery date for full HL7 functionality to be complete, requested by software vendor, passed, with no HL7 functionality achieved.

18th February 2005

First recorded Rex® automatic upload occurs. After a brief period of activity, automatic uploads from Rex® pharmacies were sporadic for the next three months.


of 767

1st March 2005

Problems with proxy server at SCGH were resolved.

18th March 2005

The first step in the auditing package, which was resolving the Client’s IP address, was added to the transactions table.

March 2005 Dispute with Software vendor (Phoenix) halted work temporarily due to non delivery of contracted items.

3rd May 2005

Interim report gives summary of work to date and requests further extension of timeline.

10th May 2005

The drug linking table is finalised.

16th May 2005

Expected date of PCA/Nu systems delivering automatic uploads from Winifred® Dispensing software to the Med eSupport web site.

1st July 2005

Two staff were sent from Hobart to give staff training at Bendigo. This included resolving ICT issues as they arose.

7th July 2005

Winifred® automatic uploading works successfully for the first time.

31st August 2005

Patient recruiting stops in project.

16th September 2005

HL7 auditing implemented.

There were significant delays creating the software to automatically upload from

Winifred® and Rex® Pharmacy dispensing software. This delay required manual entry of

the prescription data, a process that took trial officers on average forty minutes per

patient. The automatic uploading for both types of dispensing systems became fully

functional in a limited number of pharmacies during July 2005.

All transfers on the website were to be logged to aid auditing and reporting on the use of

the web site. After the first deadlines to deliver the software expired it was agreed to

forgo detailed logging in order to let the developers concentrate their efforts on getting

the core functionality of the software working. The detailed logging initially agreed

upon as part of the initial contract with the primary software vendor (Phoenix), was

provided after the active phase of recruiting.

Fully automatic faxing of discharge information to patient’s GPs was not realised.

However, reports and letters to patient’s GPs were automatically generated, and then

manually faxed.


of 767

10.1.2.1 Review of the original model In the original tender document it was proposed that the patient’s medication history

was transferred from the patient’s community pharmacist to the website for viewing by

the hospital pharmacist. (See Figure 17) After the patient was discharged from

hospital they would receive a Pharmcare® counselling sheet, and the contents of the

Pharmcare® sheet were to be sent to the website where the community pharmacist could

view the sheet. In addition, the community pharmacy would be automatically faxed or

emailed with the discharge counselling documents. The patient’s GP could also use the

web to view patient medication history; however the preferred method of

communication with the GP was either fax or secure emails. The patient’s data on the

website was to be de-identified. Patients were not to be given access to their

information on the web server. The EAG supported rolling out and testing the “full”

ICT solution in Tasmania, and seeking the best possible solution at each of the other

sites. It was considered that deployment of the full system in Tasmania alone would

provide a model for analysis.

Significantly, this model required the Pharmcare® sheet as a counselling document to

be given on discharge. Pharmcare® counselling sheets are only standard practice at the

RHH. Other hospitals use their own in-house solutions to provide patients with

counselling sheets and were reluctant to change their procedures. When it was

determined that the Pharmcare® counselling sheets could not be integrated with other

systems without significant modifications a new solution was developed (see Figure 18

below)

10.1.2.2 Refined network model The project team contracted the services of ICS, a web design company to make a web-

based solution that would generate counselling sheets, weekly drug check lists, hospital

discharge information, and letters to the patient’s GP and community pharmacist.

Using the Web based model, other hospitals could issue Med eSupport counselling

sheets without conflicting with their established procedures on patient counselling.

Figure 18 shows the University of Tasmania firewall. A firewall is a set of programs

that act to protect an internal network from potential ‘attacks’ from an external network.

Since the Med eSupport Server was located at the University, the University firewall

had to be programmed to allow network traffic to the server. The new model also

allowed provision for the patient to be given access to their medication information.


of 767

The new model proposed had a framework that was more scalable to a national roll-out

through its greater use of the web-based functionality and, importantly offered the same

model of ICT to all sites, thereby providing a greater level of functionality than required

in the deed of grant.

Proxy Server/firewallSituated at University of Tasmania

Database

Web Server

General Practitioner

Hospital firewallHospital Pharmacist

Community Pharmacist

PatientCounselling

Sheet

Figure 18 Med eSupport Network system architecture

10.1.3 Procedural impact The Med eSupport program was purposely designed to not interrupt the work flow of

the community pharmacists and in general, requests for information took pharmacists

around five minutes to complete.

In pharmacies that did not have automatic uploads enabled, the trial officer contacted

the patient’s community pharmacy with the patient’s consent and requested a six month

drug history for the patient. This report was faxed to the trial officer. In the trial

protocol, the community pharmacist received a payment of $5 for providing this

service. The trial officer then manually entered this information into the web site, a

process that on average took around 40 minutes per patient.


of 767

10.1.3.1 Using the website A detailed step-by-step guide to the website is presented in Appendix XXIII on page

587.

10.1.3.2 Vendors’ documentation and source code Program definitions, source code and vendors’ documentation has been supplied on the

accompanying CD.

10.1.3.3 Discharge procedures On discharge from the hospital, the trial or hospital pharmacist could use the website to

generate a discharge medication summary, including admission and discharge dates,

name of RMO and hospital pharmacist caring for the patient while in hospital, discharge

diagnosis, changed/ceased medications and a current list of medications with

counselling information. This sheet is similar in content to the Medprofs® and

Pharmcare® systems currently used by some hospitals. It was also possible for the

patient and their primary providers to access this information as required, via the

website.

10.1.3.4 Automatic uploads The objective was to use ICT to facilitate the transfer of patient’s medication

information ‘seamlessly’ between health care providers engaged in the patient’s care

with the patient’s consent. This transfer complied with PGA standards on privacy,

using appropriate and secure technologies. The prescription information was sent from

the patient’s community pharmacy to a central server that allowed access to the patient

and authorised health care professionals.

Firstly, an upgrade package provided by the vendors was installed on to the community

pharmacists’ dispensing computer. This would take an experienced operator

approximately one hour to complete under normal circumstances. During installation

the community pharmacy’s dispensing software would be unavailable for periods of

around five to ten minutes – similar to a monthly update.

Upon enrolment, a patient’s consent was obtained to get prescription information from

their community pharmacy, their personal details were entered into the Med eSupport

web site, and providers (doctors and community pharmacists) were assigned. Typically

this would take a trial officer or hospital pharmacist five to ten minutes to complete.


of 767

Figure 19 Automatic upload screenshot from the Rex® dispensing system for a test patient

Figure 20 Automatic upload screenshot from the Winifred® dispensing system for a test patient


of 767

The trial officer or hospital pharmacist would contact the patient’s community

pharmacy, and tell the community pharmacist how to change the patient’s profile on

their dispensing program to allow information to be sent to the server. This call would

take around five minutes. At this point the community pharmacy’s computer would

establish a secure connection with the Med eSupport web site and, using HL7

messaging, send six months encrypted patient’s history. Two models of secure

communication were available at this point. One model was intended for one time only

use and another was intended as a long term solution. For the one time use, the

patient’s user name and password was used to authenticate the session. For longer term

use, the community pharmacy’s PKI key along with their PBS approval number were

used in combination. This transfer would take place without further community

pharmacist input, and would not interfere with their work flow. The time taken to

transfer data depended on many factors, including the number of items to transfer,

internet speed and Med eSupport server load, however typically it would take five to ten

minutes.

At this stage the nominated providers, hospital and trial pharmacists would have access

to the information on the www.medesupport.com.au web site.

Afterwards, as a scheduled daily task the community pharmacist’s dispensing computer

checks for new items dispensed for the patients enrolled in the trial, and automatically

sends this information to the server. This stage does not require any pharmacist input.

After an interview with the patient, a complete record of the patient’s medications on

admission was finalised and the information could be given to the hospital doctors using

normal channels.

10.1.3.5 Website utilisation Initially in the project design, Phoenix were to provide an audit trail that would provide

a detailed analysis of website usage. However, due to delays in development, it was

agreed to sacrifice detailed logging to get the core functionality of the system working

more quickly. Consequently the logging system provided by Phoenix for this project is

the minimum needed to get usage statistics. After the trial had ended a more detailed

logging system was provided by Phoenix, however its results were not analysed, as no

usable data from the trial period was available, due to the delivery time.


of 767

10.1.3.6 Users of the program The server database was accessed using Microsoft Access® via a secure read-only

ODBC link to the Med eSupport SQL server.

Providers used for testing, along with trial officers were omitted from the analysis.

Data from patients who withdrew their consent after enrolment (n = 3) were excluded

from analysis, along with patient accounts used in testing.

The list of active patients was determined by omitting withdrawn, ‘test’ and ‘not active’

patients (SQL 1 on page 619)

The website logged successful and unsuccessful logins. These data were stored in the

[transactions] and [exceptions] tables on the server respectively.

Patients were surveyed by telephone at thirty days after discharge. In the course of this

interview, intervention patients were asked questions about the use of the web site.

These data were entered into a Microsoft Access® database and analysed using

Statview.®

10.1.3.7 Automatic uploads Script data automatically uploaded to the website from participating community

pharmacies were determined by examining the [MEDICATION RECORD] table. For

uploaded items, the data of the field [item number] did not start with the text ‘INUM’.

The difference between Rex® and Winifred® Pharmacies was determined by the length

of the script number. Rex® Pharmacies use a script number that is 20 characters in

length, whereas Winifred® Pharmacies use six characters for their script number. The

query to present the results is given (SQL 10 on page 620).

A monthly breakdown of the number of automatic uploads was also calculated (SQL

11).

10.1.4 Patient telephone interview at thirty days after discharge.

Responses for the 30 day telephone interview were entered into a Microsoft Access®

Database, and exported into the statistical program Statview® for analysis. As the trial

progressed it became evident that a significant number of patients were not using the

web site, so in the last few months of data collection the reasons why intervention

patients did not use the website were recorded.


of 767

10.1.5 Comparison of results for the anonymous patient satisfaction survey

Data were entered from the questionnaires into the Statview® program for analysis.

10.1.6 Community Health Professional anonymous satisfaction survey

Data were entered from the questionnaires into the Statview® program for analysis.

10.1.7 Data extraction techniques Data were stored in two main databases. Data relating to the website were stored on an

SQL server. This database was queried using Microsoft Access®. All information

transcribed from the patient data collection sheet was entered into a Microsoft Access®

database that was written and maintained by trial officers. Patient admission episodes

for Tasmanian patients were also imported into this database. These data were queried

and exported into the statistical package Statview® for analysis.

10.2 Results

10.2.1 Analysis of website utilisation - including usability and usefulness

General observations by the users of the website are presented below. The total number

of logins per user type is presented in Table 56. The number of users per user type is

given in Table 57. Monthly breakdowns are given to provide an overview of the

timeline for the website usage by login and user type, and are presented in Table 58 and

Table 59 respectively. A graph displaying the number of users versus the number of

logins is presented in Figure 21.

Table 56 Number of logins per user group

User Group Number of logins Community Pharmacists 196

General Practitioners 66

Patients 67


of 767

Table 57 Number of users per user group

User type Number of Users

Total Active accounts Proportion (%)

Community Pharmacists 63 138 (45%)

General Practitioners 36 191 (19%)

Patients 28 240** (12%)

Table 58 Monthly breakdown of website use by number of logins from 1/12/2004 to 5/10/2005 for each user type

Date (year: month) Community Pharmacists

General Practitioners Patients

2004:12 20 4 4

2005:01 14 13 0

2005:02 6 5 0

2005:03 14 4 6

2005:04 30 6 15

2005:05 14 8 3

2005:06 16 3 3

2005:07 47 20 17

2005:08 28 2 11

2005:09 7 1 5

2005:10 0 0 3

Totals 196 66 67

** Note that accounts were assigned on admission to the trial, and the active accounts

include patients who were later withdrawn from other analyses because of

difficulties implementing trial protocols.


of 767

Table 59 Monthly breakdown of website use by number of users from 1/12/2004 to 5/10/2005

Date (year: month) Community Pharmacists

General Practitioners Patients

2004:12 5 3 3

2005:01 10 4 Nil

2005:02 5 5 Nil

2005:03 7 4 3

2005:04 14 4 8

2005:05 11 7 3

2005:06 8 3 2

2005:07 20 14 10

2005:08 13 2 8

2005:09 5 1 5

2005:10 Nil Nil 2

Totals 98 47 44

Note: total not presented because same user may log in at different times. Total users

given in Table 57

Figure 21 Number of users vs number of logins grouped by user type

0

5

10

15

20

25

30

1 2 3 4 5 6 7 8 9 10 11 12 13

Num ber of logins

Num

ber o

f use

rs

Community Pharmacist General Practioner Patient


of 767

10.2.1.1 Automatic uploads Table 60 Dispensing script item automatic uploads by dispensing system from 1/12/2004

to 5/10/2005

Date (year: month) Rex® Winifred®††

2005:02 50 -

2005:03 104 -

2005:04 7 -

2005:05 3 -

2005:06 4 -

2005:07 167 454

2005:08 63 78

2005:09 33 106

2005:10 6 0

Total 437 638

Initially twelve sites were contracted to work with the Winifred® dispensing system.

Two sites were not suitable because of internal hardware problems and a fire. Seven of

the pharmacies using the Winifred® system and four pharmacies using the Rex® system

automatically uploaded patient’s prescription details to the server. There were a total of

23 patients whose details were automatically uploaded.

10.2.2 Results of the patient telephone interview at thirty days after discharge

Of the 159‡‡ intervention patients called at thirty days after discharge, 12 reported they

had used the website; eight reported they did so alone. Three patients reported they

printed a counselling sheet from the website and one patient reported they printed a

weekly checklist from the website.

Table 61 presents the reasons telephone respondents gave for not using the website. It

should be noted that patients were only asked why they did not use the website during

the latter stages of the trial after it became evident that a significant number of patients

were not using the website.

†† Winifred®systems began uploading data in July 2005.

‡‡ Note that failure to receive interventions required by the trial protocol was not an exclusion criterion for these data.


of 767

Table 61 Reasons given by patients at phone interview for not using the website

Reason Given Number (%) No access to the internet 45 (68%)

No reason given 2 (3%)

Lack of time 2 (3%)

Poor Health 2 (3%)

Lack of relevance 3 (5%)

Technological Gap 10 (15%)

Total 66 (100%)

The number of patients reporting they had used the website was low, reducing the

statistical power of any analysis.


of 767

10.2.3 Comparison of results for the anonymous patient satisfaction surveys

After patients received a 30 day phone call they were sent an anonymous survey. Of the

408 surveys sent out, 279 were returned giving an overall response rate of 68%. As the

surveys were anonymous, their results could not be excluded later if the patient was

withdrawn from the trial for another reason. Patients assigned to the control and

intervention groups returned 171 (70%), and 108 (65%) surveys respectively.

10.2.3.1 Control Patients When patients assigned to the control group were asked if they wanted an ongoing

medication support service the most common services requested were discharge

information being sent to their GP (72%, n = 95) and being given a counselling sheet on

discharge (71% n = 93). When control patients were asked if they would like access to

a secure internet website 5% ( n = 7) indicated they would.

10.2.3.2 Intervention Patients All patients enrolled into the intervention group were given access to the web site; 12%

used the website (Table 57). There was no significant difference between the two

intervention sub-groups when patients were asked to report if they had used the website

(χ2 = 180 df = 1 p = 0.67). The results from both the HMR and PDMR sub-groups of

the intervention patients were grouped for analysis.

Eleven intervention respondents (10%) reported that they used the web site. This

compares well to the actual number of patients who did log in (12% from Table 57)

Patient’s reported ease of use of the website are presented in Table 62.


of 767

Table 62 Patient’s response on website ease of use

Response Number (%)

Difficult 2 (18%)

Easy 6 (55%)

OK 3 (27%)

Total 11 (100%)

The patient’s opinions of the website as reported from the questionnaire are presented in

Table 63. Only eleven patients responded that they used the web site, so their results

are presented and further statistical analysis was not performed.

Table 63 Patient’s response of website services used and their usefulness

Information useful?

Information on web site

Accessed? n (% of question

total) Yes it

Helped Not

Particularly No it made

it worse

Current medication summary 9 (90%) 8 - -

My hospital stay information 7 (88%) - - -

Medication counselling sheet 5 (62%) 2 1 1

Weekly checklist 3 (37%) 1 - 1

Links to related websites 3 (37%) 1 - -

Additional notes about my medications 6 (67%) 4 - -

Consumer medication information 3 (37%) 1 - -

Other… 1 (12%) - - 1

Respondents could nominate more than one option so the total can be greater than

100%.

Of the intervention respondents who did not use the web site, their reasons are presented

in Table 64.


of 767

Table 64 Reasons why people did not use the web site

Question Number of respondents (%) I do not know how 4 (6%)

I do not have a computer 48 (72%)

I have not had time 1 (1.5%)

I have not felt I needed to view it 10 (15%)

I am not interested in viewing my information this way 10 (15%)

I do not believe this is a useful service 0 (0%)

Other… 4 (6%)


100%.

10.2.4 The participating Community Health Professional anonymous satisfaction surveys

Of the 332 surveys sent to health care providers of intervention patients, 141 were

returned giving a response rate of 42%. A health provider was sent a survey each time a

patient received a 30 day phone call, and some providers were sent multiple surveys.

The surveys were anonymous, so results from patients who were later excluded from

the trial were included in these analyses.

Table 65 Responses for which form respondent would like to receive discharge medication information in future

Delivery form of discharge information

Community Pharmacists,

n (%)

General Practitioners,

n (%) Fax 51 (67%) 35 (55%)

Secure email 25 (32%) 24 (38%)


Post 13 (17%) 17 (27%)

Other 1 (1%) 5 (8%)


100%.

Community Pharmacists were more likely to use the website than GPs (χ2 = 9.097 df =

1 p = 0.0026) and GPs were more likely to want another delivery form of discharge

information than Community Pharmacists (χ2 = 3.656 df = 1 p = 0.0559)


of 767

The respondents were asked why they did not use the website on the questionnaire, and

the responses are summarised below in Table 66.

Table 66 Reasons providers did not use the web site

Reasons why website was not used Community Pharmacists,

n (%)


n (%) I do not know how 1 (4%) 7 (18%)




I am not interested in viewing my patient’s information this way 0 (0%) 5 (13%)


Other… 7 (25%) 4 (10%)


100%.

When providers were asked to indicate how easy the website was to use on a Likert

scale (0 = difficult, 100 = easy) the median was 73.5, (range 12 to 100). However

community pharmacists were more likely to respond that the website was easy to use,

(Mann Whitney U = 168, z = -2.212 p = 0.0270)

The aspects of the Med eSupport website that respondents found useful are summarised

in Table 67.

Errors and inconsistencies in patient information are presented in Table 68.


of 767

Table 67 Aspects of Med eSupport website that respondents found useful

Aspect of Med eSupport website that was useful

Community Pharmacists,

n (%)


n (%) Access to dispensed medication history 28 (75%) 5 (38%)







Other……. 1 (3%) 0 (0%)


100%.

Table 68 Did you find there were any errors or inconsistencies in the patient’s initial medication information provided on the website?

Response Community Pharmacists,

n (%)


n (%) Yes 7 (21%) 0 (0%)

Not Sure 8 (24%) 0 (0%)

No 18 (55%) 13 (100%)

Respondents agreed when asked on a likert scale (0 = strongly disagree, 100 = strongly

agree) that receiving discharge information in the future would be valuable (median 90,

range 20 to 100, n =112)


of 767

10.3 Discussion and interpretation

10.3.1 Website utilisation The initial specifications given to the software vendor requested full auditing of the web

site. As this was not available until after the trial finished, only login data were

available. However, given the relatively small numbers of logins to the web site, it is

doubtful if detailed auditing would have been useful.

10.3.1.1 Website usage Figure 21 shows the number of website users is inversely proportional to the number of

times they logged in. This can illustrate the function of the website and the nature of its

use. The people interested in using the website would probably do so to “try it out”

once or twice. The information stored on the website is relatively static, so patients

would not have to view it regularly. Other times people would probably use the website

on an as needed basis, often in response to some other event, such as a clinic

appointment, hospital visit etc. The other factor that may influence the number of

website logins is the time the user had access (how long they were enrolled before the

website was closed down). This information was not presented as the date providers

were entered onto the system was not stored. Providers may have had more than one

patient in the system, so if they accessed the details of more patients, their number of

logins would likely be greater.

Of the 240 patient accounts made, 193 patients were enrolled in the trial at the point of

discharge. Of these patients 30 (15%) used the website at least once. The Australian

Bureau of Statistics reported in 2005 that the number of elderly Australians (aged 65

years and older) using the internet in 2002 was 13 per cent.105 In comparison, our data

shows that of the patients given access to the website who were over 65 years of age, 12

(7%) did so. The most common reason people gave for not using the web site, was not

having access to a computer. Other reasons people gave for not using the website

include technology gap, lack of interest, failure to care about their medications, and

faith in health care providers.

10.3.1.2 Automatic uploads One of the outcome measures of the project was to provide seamless integration of

information technology.


of 767

Part of this was to enable prescription data dispensed in community pharmacies to be

transferred securely and automatically for consented patients from community

pharmacies to the central data repository at the University of Tasmania.

The number of monthly uploads are presented in Table 60. Table 60 demonstrates the

delays that the project experienced waiting for ICT vendors to deliver the automatic

uploading functionality.

When a user was registered to send prescription data from the community pharmacy to

the central repository, six month’s history was sent. During July, 2005 some project

team members visited participating community pharmacies to install the application for

the automatic uploads for both the Rex® and Winifred® dispensing packages, and the

surge of results reflect this.

10.3.2 Subjective data Patients were asked their opinions of the ICT at the 30-day post-discharge phone call.

Patients, pharmacists and GPs were anonymously surveyed after the 30 day phone call

to gauge their opinions of the project. Data from these surveys is presented in section 0.

10.3.3 Barriers to implementation Widespread implementation of Med eSupport would require software manufacturers to

have software that could communicate with the Med eSupport server. Currently only

Rex® and Winifred® have this functionality. Modules for each dispensing system would

have to be written and installed into each pharmacy. In addition, drug linking tables for

each software manufacturer would have to be created.

There is reluctance in ICT circles to use the central repository model used in this

project. The favoured model, adopted by healthConnect, proposes direct

communication between each stake-holder (peer-to-peer). Each system has its merits,

and the technology developed for Med eSupport can be adapted to the peer-to-peer

model for communications between hospital and community pharmacists.

The website is functional, but slow to use. The website data entry would have to be

streamlined for mainstream implementation.

10.3.4 Project team insights The initial focus of the project was to use technology to deliver seamless integration of

diverse computer systems and to transfer information between these systems.


of 767

However, with ICT vendors not producing these solutions in a timely manner it became

necessary for the trial officers to rely on more manual methods to achieve the project’s

goals. Where we had hoped to automatically upload most of the dispensing data from

community pharmacies we had to receive faxed history print-outs and then manually

type the details in to the web site.

The Med eSupport Website was slow to use when manually entering patient’s

medication history. Form controls that could have been populated automatically from

available data were not, and the data had to be retyped (eg doctor’s name, pharmacist’s

name etc). Information such as counselling was available for around one third of

available drugs. When not available, counselling information had to be obtained from

standardised sources and manually typed in. If errors were made in the entry process

they were difficult to correct and at times needed to be completely re-entered – it was

possible they would appear twice, obviously causing confusion for patients.

The average time it took to manually enter a patient’s medication history was around 40

minutes (for an average of 28 prescription entries). The time delays created by entering

this information were sufficient to employ part-time staff for this task.

The website was primarily designed to be used online; however the website produced

printed drug lists and counselling sheets that were poorly formatted on paper.

The software vendor also did not believe maintaining the drug table in the Med

eSupport database was part of their original agreement, and consequently no new drugs

were added or modified during the trial. All newly released drugs had to be entered

using a complicated interface each time they were used.

Many of the problems encountered were technically easy to solve during the course of

the trial. However, they were not, and the project team members were burdened with

performing repetitive manual entry tasks.

10.3.5 Ongoing maintenance To keep the program functional as it stands, any new drugs and providers have to be

added to the system. The website has a feedback mechanism that allows users to report

problems to trial officers for correction.


of 767

10.4 Conclusions

10.4.1 Interpretation As can be expected with such ambitious projects, there were obstacles along the way.

The project team often had to seek alternative methods to keep the trial progressing.

The core functionality of the project’s Information and Communications Technology

(ICT) was realised towards the end of project’s timetable – prescription information was

sent from community pharmacies to the data repository using secure, encrypted

technologies. Throughout the trial the patient’s primary health providers were given

access to patient’s discharge information via both the web and via fax, and patients were

given access to their information on a secure interactive web site.

10.4.2 Software vendors In dealing with ICT vendors, the operation of the required software should be specified

in advance in minutia detail. Not doing so may lead to misinterpretation by a

programmer who may know very little about healthcare. Moreover, new features

sought after initial development may require changes to the core program framework

and significantly increase development time. In retrospect, some of the ICT issues

encountered may have resolved more quickly if the project team had addressed these

requirements more thoroughly.

Another issue that was difficult to resolve during the project timetable were delays that

were caused by the number of software vendors involved. Delays caused by one

vendor had a knock-on effect to the work of another, causing further delays to the

project. Working with multiple software vendors on the project was unavoidable and

this emphasises the need for vendors to complete their work to the timetable agreed

upon in the contract.

Software developers contracted should be able to demonstrate a history of completing

projects on time, to budget and to the specifications detailed.

In future projects it is recommended the software vendor’s ability to perform the work

required in the time specified should be contractually assured with the inclusion of

pecuniary damages. This is of course reliant on providing exact technical specifications

to the vendor in advance.


of 767

10.4.3 Drug linking table The drug linking table, a fundamental tool for integrating drug databases from different

vendors should be expanded in scope and kept up to date to enable future projects to

share data. To address this and other issues of standards the National E-Health

Transition Authority (NEHTA) was formed. NEHTA Limited is a not-for-profit

company limited by guarantee, with continuing responsibility for developing national

health IM&ICT standards and specifications. NEHTA Limited is jointly funded by

Australian state, territorial and national governments, and the Board of NEHTA Limited

is comprised of chief executives from health departments within these jurisdictions.106 It

is hoped that with a national standard for drug tables, software vendors can use a

common-key that will allow exchange with other vendors’ products.

Economic and financial analysis of Med eSupport

of 767

11 Economic and financial analysis of the Med eSupport trial

11.1 Introduction The objectives of the economic analysis were to:

• Estimate the value of pharmacists’ interventions at admission and post-discharge for patients enrolled in the Med eSupport project,

• Assess the impact on the health care system of the pharmacists’ actions, and

• Assess sustainability and expansion of the program.

Initially, changes in AQoL and readmission rates were to be examined for economic

implications; however, largely due to the reduced sample size impacting on the

statistical power of the analysis, significant differences in these outcomes were not

found and a second method for analysing the economic value of Med eSupport was

undertaken.

An economic analysis of the Med eSupport project was primarily performed to assist in

the evaluation of the future sustainability of the project. This analysis was separately

applied to the two key components of the intervention program:

• the medication list reconciliation process undertaken at the time of hospital admission and discharge, and

• performance of post-discharge medication reviews (PDMRs).

The analysis was conducted according to internationally accepted criteria for

pharmacoeconomic analyses107-109 and was methodologically based on recently

published evaluations of pharmaceutical care that appear in the international

literature.108, 110-114

11.2 Methodology A number of techniques were utilised to assess the economic value of the interventions

performed within the Med eSupport project. An experienced Consultant Health

Economist, Brita Pekarsky, was closely involved in this process. The costs of the

program were calculated, based on those services provided by the trial pharmacists

which extended beyond those offered by normal hospital procedures, and balanced

against the estimated cost-savings achieved.

The estimation of the value of the services provided in the Med eSupport trial, included:


of 767

• an assessment of the time taken to perform the hospital-based interventions (with an extrapolation to all cases),

• a clinical evaluation of the admission and post-discharge interventions in a random sample of interventions by a clinical panel (with an extrapolation to all cases).

11.2.1 Quality of Life Due to the difficulties associated with quantifying the benefits of an intervention

program, a common unit of measurement for benefit, quality of life, was utilised. Med

eSupport measured quality of life using the Assessment of Quality of Life (AQoL)

questionnaire, made up of 15 questions (Appendix XXXIII, page 751). The

questionnaire is divided into five dimensions (each containing three questions). The

dimensions are illness, independent living, social relationships, physical senses and

psychological well-being. For each question, there were four answers to choose from.

The answer suggesting the best quality of life attracts a score of ‘1’, the answer

suggesting the worst quality of life attracts a score of ‘4’. The responses to the first two

questions of the AQoL were easily obtained from the patient’s medication account. As

a result these were not asked directly by the trial officer. The scores for all 15 questions

were tallied to produce a final result. The lower the score, the better the patient’s self-

perceived quality of life. The final score could range between 15 and 60.

The questionnaire was carried out with each patient twice. At admission the interview

was conducted in person, and at the 30-day follow-up all information was collected over

the telephone. The data was then evaluated to determine if there was a difference in

QoL from the time patients were inpatients in hospital to the 30-day follow-up

telephone call.

The results for the patient’s admission and 30-day AQoL questionnaires were entered

into a Microsoft Access® database. The AQoL results were exported from the Access

database into SPSS where the algorithm supplied by Hawthorne1 (with permission) was

used to generate the AQoL Scores. The data were then analysed in Statview® using

ANOVA.

11.2.2 Readmission costs and rates of readmission for patients within thirty days of discharge

At the time of the 30-day follow-up phone call, the patient was asked about

readmissions and frequency of readmissions. Additionally, the readmission rates were


of 767

obtained from medical records for patients enrolled in the trial at the Tasmanian and

Victorian sites (RHH, LGH, and BHCGH).

11.2.3 Costs of activities, determined through time trials

The economic evaluation was conducted from the perspective of the health care payer.

Only direct health service costs have been considered; therefore, variables such as time

spent away from work, relatives’ time and costs, and transport costs, have not been

included. The cost of patient self-care has been assumed to be zero. PBS costs were

not studied. The costs that were included in the analysis are detailed below.

Costs associated with additional time spent by hospital and community pharmacists on

the four core activities were measured, as follows:

• Using the electronic communication pathway for medication profiles between community and hospital pharmacies,

• Routine supply of a comprehensive medication information sheet to the patient/carer prior to discharge from hospital.

• Faxing or emailing of the medication information sheet to the patient’s general practitioner at discharge, and

• Promoting the appropriate use of PDMRs. [Wage rates as per the Health Departments and Pharmacy Guild]

The times spent on these activities were determined by observation within the study

period. Time trials were performed at the RHH during the final months of the patient

enrolment process. Time trials were carried out over 15 days and were performed on 20

control and 20 intervention patients. The trial officers recorded a start and finish time

for each individual task performed for each patient. These times were documented on a

separate form to the main data collection sheet. (Appendix XXXII, page 750).

The time trials included informed consent and trial specific tasks, which would not be

required in the event that rollout took place. The tasks that should already be routinely

conducted by hospital pharmacists (as per the SHPA guidelines115) were also subtracted

from the total time. The time trials were used to determine the potential employment

needs to provide the service on a continuing basis. Once the personnel requirements

were established, it was possible to assign a monetary value which could be directly

compared to the possible cost savings of the service.


of 767

11.2.4 The clinical panel An independent clinical panel was utilised to assess the significance of the interventions

performed when identifying medication discrepancies and conducting post-discharge

medication reviews, and the likelihood of misadventure for the patient if the

interventions had not occurred. This could then be used to reflect financial costs to the

health system prevented in the form of hospital bed days avoided, days of adverse

health impact avoided and consultations with GPs or specialists avoided.

The panel members included one consultant physician, one hospital medical registrar,

one GP, one clinical pharmacist and one community pharmacist. The five panel

members were invited via email (Appendix XXVIII, page 662) to assess a sample of 40

random cases (20 relating to medication discrepancies and 20 to medication reviews),

expected to take 5 hours to complete. They were allowed a total of two weeks to

complete the assessment in their own time. To encourage participation each panel

member was offered remuneration for their time. All panel members invited to take

part took up the offer.

11.2.4.1 Construction of case studies A different template for each type of case study was created. (Appendix XXIX, page

666) The discrepancy case studies included information regarding the patient’s age,

reason for admission, past medical history, dispensing summary, relevant laboratory

data and an outline of the discrepancy. The post-discharge medication review

recommendations focussed more on the key suggestions made by the accredited

pharmacist, but provided a similar background history to those in the discrepancy case

studies. The outcome at follow-up of the patient, whether it was positive or negative,

was provided in each of the cases. If the outcome was not known this was also stated in

the case.

The instructions for marking were included on each of the cases (Appendix XXX: Panel

Discrepancies on page 668 and Appendix XXXI: HMR Recommendation, page 709).

The instructions were slightly different for the discrepancies and post-discharge

medication review recommendations. The panel members were provided with a space

to assign a numeral to reflect the percentage likelihood of that consequence resulting in

either a mild, moderate or severe outcome, as defined in Table 69 on page 181. The


of 767

panel were also asked to consider the situation before and after the pharmacist’s

intervention.

A summary of the panel methodology and ensuing economic analysis follows. This

consists of a six-step process:

1. Estimate of the economic value of a typical recommendation or intervention identified as part of a home medication or discrepancy review,

2. Adjust the above estimate for the rate of recommendations per review and the likely uptake of the recommendation by the GP, hospital medical officer and/or patient,

3. Estimate the time taken to perform a review,

4. Estimate the per review, per site and national costs of a program,

5. Estimate the costs and outcomes of a national program, including consideration of the reviews that would otherwise have occurred, and

6. Estimate the economic value of a national program.

11.2.5 Panel methodology and economic analysis

11.2.5.1 Step One: Estimate of the economic value of a typical recommendation or intervention identified as part of a home medication or discrepancy review.

A random selection of 20 discrepancy reviews and 20 post-discharge medication

reviews (PDMRs) were identified and the pharmacists’ recommendations were

reviewed in detail. The patient files from the RHH were the only available and ready

for assessment at the time the case studies were to be constructed. The RHH

intervention patient files were reviewed to identify all patients who had a discrepancy

identified at admission or discharge. A discrepancy was considered to be the omission

of a drug, a drug without indication or an incorrect drug dose or frequency. The

discrepancies did not include compliance issues or clinical assessments. Each case that

had a discrepancy was included in the final sample regardless of the magnitude of the

discrepancy (i.e. lubricating eye drops not charted, through to a sub-therapeutic dose of

an antiarrhythmic agent). The final sample did not include any cases that did not have a

discrepancy, nor patients who were subsequently withdrawn from the trial. The control

patients could not be used as the assessment required the panel member to consider the

severity of the consequence before and after the pharmacist’s intervention. If the patient

had more than one discrepancy at admission or discharge, two trial officers then agreed

on the most significant discrepancy. A random number sequence list, generated using


of 767

the Excel® random number generating tool, was then used to randomly select the 20

cases to be reviewed by the panel.

The sample of 20 cases was taken from the 95 available RHH intervention patients with

discrepancies at admission or discharge. One of the selected cases was not well

described and was removed from the analysis, leaving 19 discrepancy cases. A similar

process was used to identify PDMR recommendations for review by the clinical panel.

The sample of 20 cases was taken from the 125 available RHH intervention patients

who had received a PDMR/HMR. Any reviews in which the accredited pharmacist did

not provide a recommendation (i.e. suggested a change in drug therapy, whether it be

addition or cessation of a drug or an altered dosage) were excluded, as were the

withdrawn patients. Any reviews that had not been received at the time of sampling

were also excluded. If the patient had more than one recommendation within the

medication review, two trial officers then agreed on the most significant

recommendation.

The clinical panel were presented with a description of each of the 40 scenarios and

asked to assign probabilities of severe, moderate or mild consequences occurring, in the

absence and presence of the intervention. Each member was provided with a copy of

the case studies and requested to independently assign a percentage severity to the

suggested possible consequences.

Table 69 Marking tool provided to guide the clinical panel

Significance rating Description Specific example

Mild Minor Symptoms could occur / specialist unit consultation required in hospital.

Codeine prescribed: provide prophylactic stool softeners.

Moderate May end up at Dr/Refer to Dr / if already in hospital, short addition of hospital bed days (2-5 days).

Patient develops mild toxicity due to doubled dose.

Severe May be readmitted into hospital/ admitted to higher care unit (ICU/HDU) and prolonged extra hospital stay.

Serious drug interaction with warfarin which could result in a major bleed.

The panel was given up to two potential clinical outcomes to consider for each situation,

and an assessment of every scenario was required. They were also given the

opportunity to recommend another consequence and assign a percentage of likelihood

of that occurring.


of 767

The PROMISe Project Team116 had already developed a clinical assessment method

which takes into account the probability of a consequence occurring (with the

intervention and also without the intervention) and also the “attributability” of the

intervention to the pharmacist. This method was easily adapted for use in the

assessment of Med eSupport economic outcomes.

For example, if a patient was taking two different non-steroidal anti-inflammatory

agents (NSAIDs) and the accredited pharmacist intervened to suggest cessation of one

of the agents, one possible prevented outcome would be a gastrointestinal bleed. The

probability that a bleed would have actually occurred, however, would depend on a

range of other factors that may or may not be present in the particular situation being

considered. In the assessment system that the PROMISe team members developed, the

clinical expert assessor assigns their estimate of the probability of a gastrointestinal

bleed before the intervention (that is, while the person was taking both NSAIDs), and

also assigns the probability that a gastrointestinal bleed would occur after the

intervention (that is, once the person went back to just taking one NSAID, which does

still carry some risk of a gastrointestinal bleed). The expert’s assessments would take

into account whatever other factors are known to be present in the patient, and the

potential benefit of the intervention can be considered in terms of the reduced

probability of a gastrointestinal bleed.

There is, however, another aspect to consider in the assessment of interventions, which

is the severity of the outcome. To continue with the example above, while there is some

probability that the patient may have a gastrointestinal bleed requiring admission to

hospital, blood transfusion and endoscopy, the probability of a less severe manifestation

of the same pathological process may be greater (for example, a mild sub-clinical bleed

that would not require immediate medical management). Thus, it is necessary to

consider the probability of different levels of severity (severe, moderate and mild levels)

of any consequence. In this example, the expert assessor would be asked to assign the

before and after probability of a severe gastrointestinal bleed, a moderate

gastrointestinal bleed and a mild gastrointestinal bleed. The overall potential clinical

significance of the intervention would therefore be considered in terms of the changes in

probability of the different levels of severity of the different potential consequences of

the intervention.


of 767

Previous estimates of the health, financial and service consequences were used to

determine the average economic value of the recommendations made in the reviews,

using the probabilities assigned by the panel. The health, financial and service

implications of clinical consequences used were those developed in the PROMISe

project.116 In brief, a set of clinical consequences for interventions were developed.

Consequences were common diseases or signs or symptoms that were impacted on by

interventions. The consequences were grouped into main diagnostic categories (MDCs)

and definitions were prepared to describe the severe, moderate and mild levels for each

consequence. Examples of two clinical consequences (hypertension and gastrointestinal

bleeding) and their definitions are shown in Table 70.

Table 70 Examples of clinical consequences and their severity descriptions, derived from the PROMISe project

MDC Code

MDC heading

Sub-group Code

Subgroup Sub-Group Severity Code

Subgroup Severity Description

5 Circulatory System 05.02 Hypertension 05.02Mild

Mild signs or symptoms which resolve without intervention

5 Circulatory System 05.02 Hypertension 05.02Moderate

Moderate elevation of blood pressure requiring modification of or commencement of medical management

5 Circulatory System 05.02 Hypertension 05.02Severe

Acute injury to target organs (e.g. renal, ocular or cerebral) requiring prompt medical management

6 Digestive System 06.01 GI bleeding 06.01Mild

Occult gastrointestinal bleeding likely to require medical management only if persistent

6 Digestive System 06.01 GI bleeding 06.01Moderate

Overt gastrointestinal bleeding requiring medical management

6 Digestive System 06.01 GI bleeding 06.01Severe

Overt gastrointestinal bleeding with haemodynamic consequences requiring admission to hospital and prompt medical management


of 767

Each consequence of an intervention (for example, a severe gastrointestinal bleed) has a

level of disability and expense associated with it. The consequences table was

expanded to include a number of different economic and non-economic parameters.

The different parameters were:

1. Impact on Health Status

• This was a scale of impact based on the severity of the particular consequence, with 1 being mild impact on health and 3 being a severe impact on health. We have used a measure that is relative (loss relative to health status that would otherwise have occurred) rather than absolute (the absolute health status that occurred). By using a relative health status, we account for the range of possible initial health states of patients.

2. Duration of Health Status Impact

• This was a value in days of the duration of the health impact. For chronic conditions, a one year timeframe was considered.

3. Duration of Admission

• The duration in days of any admission associated with the consequence. Where the consequence definition matched that of an existing AR-DG definition, the information was obtained from the National Hospital Cost Data Collection Cost Weights for AR-DG Version 4.2, Round 7 (2002-2003). Where no matching definition existed, the average duration of admission was used.

4. Cost of Admission

• A value in dollars for any admission associated with the consequence. This was determined from the same information as the duration of admission.

5. Number of General Practitioner consultations

• The number of community based General Practitioner consultations required to manage the particular consequence, as per the clinical panel.

6. Cost of General Practitioner Consultations

• The total cost of the General Practitioner consultations, based on an average of 3:1 Level B (Item 23) to Level C (Item 26) consultations as per the 2005 Medicare Schedule.

7. Number of Specialist Consultations

• The number of specialist consultations required to manage the particular consequence, as per the clinical panel, and

8. Cost of Specialist Consultations


of 767

• The total cost of the specialist consultations, based on an initial consultation cost according to MBS Item 110 and subsequent consultation costs according to MBS Item 26.

9. Investigation and Pathology Costs

• The costs of typically required investigation or pathology tests required in the management of the particular consequence, as per the clinical panel. These were based on the schedule fee for the appropriate item.

The initial estimates for these parameters were made by the PROMISe Project

Team,116,and then reviewed and modified by a consensus group process that included a

physician, a general practitioner and two experienced clinical pharmacists. The values

assigned to the examples outlined in Table 70 are shown in Table 71 (on page 186).

Using a table closely related to the consequence table used for economic evaluation of

the PROMISe trial,116, we were able to give an estimated total direct cost of the

different consequences considered by each of our panel members in Med eSupport.

These costs were derived from the national hospital cost data for 2002-2003 and the

AR-DG. This consensus-based costing took into account the duration of admission,

cost of admission, number of GP consultations, cost of GP consultations, number of

specialist consultations, cost of specialist consultations and also investigational or other

costs (if not included in the DG cost).


of 767

Table 71 Examples of values assigned to consequences, derived from the PROMISe project116

Subgroup

Hea

lth S

tatu

s Im

pact

Dur

atio

n of

Hea

lth S

tatu

s Im

pact

(day

s)

Dur

atio

n of

Adm

issi

on

(day

s)

Cos

t of A

dmis

sion

No.

of G

P C

onsu

ltatio

ns

Cos

t of G

P C

onsu

ltatio

ns

Num

ber

of S

peci

alis

t C

onsu

ltatio

ns

Cos

t of S

peci

alis

t C

onsu

ltatio

ns

Inve

stig

atio

n C

osts

Hypertension 05.02Mild

1 360 0.00 $0 3 $113

Hypertension 05.02Moderate

2 360 0.00 $0 8 $302 $85

Hypertension 05.02Severe

3 90 3.65 $2,381 4 $151 4 $320 $85

GI bleeding 06.01Mild

1 180 0.00 $0 2 $76 $35

GI bleeding 06.01Moderate

2 60 1.68 $1,199 1 $38 1 $128 $1,847

GI bleeding 06.01Severe

3 90 5.62 $3,881 2 $76 2 $192

11.2.5.2 Step Two: Adjustment process Adjust the above estimate for the rate of recommendations per review and the likely

uptake of the recommendation by the GP, hospital medical officer and/or patient.

Estimates of the actual rate of recommendation per review from the study were used to

determine the average value per review performed (rather than the average value per

recommendation intervention). For instance, only 87% of the study’s medication

reviews had therapy recommendations (i.e. 115 reviews have to be performed to

generate 100 recommendations).

Estimates of the actual uptake of the recommendations from the study were utilised.

Uptake of the pharmacists’ recommendations by doctors within the hospital could be

determined as each patient had discrepancies highlighted at admission. These

discrepancies were followed by a trial officer (review of inpatient drug chart and

progress notes) and they were documented as being acted on within 48 hours, after 48


of 767

hours, or not at all. Recommendation uptake by GPs in the community was determined

by an assessor who systematically reviewed each of the PDMR reviews. The patient’s

medications (from the community pharmacy report and the patient’s account) at 30 days

were recorded. From this it was possible to determine if a recommendation made

earlier by the pharmacist had been implemented by the GP, had not been implemented

or whether it was unclear.

11.2.5.3 Step Three: Estimate the time taken to perform a PDMR

The participating pharmacists recorded estimates of the time taken to perform the actual

review, including travel and report preparation (the average value was 180 minutes

(section 11.4.1).

11.2.5.4 Step Four: Estimate the per review, per site and national costs of a program

Payments made by the Medicare Benefits Scheme (MBS) and costs incurred by

hospitals as part of discrepancy reviews or PDMRs were identified.

The costs of implementation at site and national levels were identified. The number of

sites at which the program could be implemented in an initial rollout was estimated to

be 50. A site is essentially a combination of a Division of General Practice and one or

two mid-size (approximately 400-bed) general public hospitals.

11.2.5.5 Step Five: Estimate the costs and outcomes of a national program, including consideration of the reviews that would otherwise have occurred

It was considered whether costs would otherwise have been incurred – for example, if a

PDMR or discrepancy review would have occurred irrespective of the study, then the

total costs and additional costs were assessed separately. The results in the study were

used to determine the rate at which reviews would otherwise have occurred.

11.2.5.6 Step Six: Estimate the economic value of a national program

Comparison of the costs and the effect of the program were carried out.

The economic outcomes were determined and the total costs and additional costs per

review were estimated. Days of impacted health status that would be prevented were

also measured.


of 767

The additional costs were adjusted by the estimated financial savings to the system.

The savings are those made as a result of the implementation of the reviewing

pharmacist’s recommendations.


of 767

11.3 Results

11.3.1 Quality of Life The data from the Tasmanian sites (RHH and LGH) and Bendigo were included in the

analysis for Quality of Life. Reasons for not including data from the Perth-based sites

are discussed in section 9.2.1.3. The AQoL score was subdivided for the service level

each patient received. A detailed description of the service level received is discussed

in section 9.2.1.2

Table 72 Comparison of patient Quality of Life survey (Tas and Bendigo only) as inpatients and at 30 days post-discharge for each service level received

Quality of Life Survey Score, mean (SD) Group

Admission (n=343) 30 days (n=277)

Minimal Intervention 31.4 (6.0) 28.7 (5.0)

Partial Intervention 31.8 (5.0) 29.5 (5.4)

Full Intervention 31.4 (6.8) 29 (6.5)

Figure 22 Comparison of AQoL scores at admission and 30 days post-discharge

26

28

30

32

34

36

Inpatient 30 daysSurvey Time

Mea

n Q

OL

Scor

e

Discharge counselling or nil Discrepancy check & counselling Discrepancy check & HMR


of 767

The groups all displayed a significant improvement in AQoL from baseline to 30-day

follow up. (See Table 72 and Figure 22) (ANOVA test F = 23, df = 1 and 762,

p<0.0001). There was no significant difference found between the three groups for the

AQoL score, either as inpatients or at 30 days post-discharge. (ANOVA test F = 0.51,

df = 2 and 336, p = 0.6).

The patients without 30-day data were removed from the data set and a second analysis

was carried out.(See Table 73 and Figure 23) Of the 343 patients included in the AQoL

analysis, 66 patients were unable to complete the questionnaire at the 30-day follow-up,

mostly due to being uncontactable. The second analysis gave very similar results to that

which included the missing patients at 30 days.

Table 73 Comparison of patient Quality of Life survey (Tas and Bendigo patients with both survey results) as inpatients and at thirty days post-discharge

Quality of Life Survey Score, mean (SD) Group

Admission (n=277) 30 days (n=277)

Minimal Intervention 30.8 (5.4) 28.7 (5.0)

Partial Intervention 31.9 (4.7) 29.5 (5.4)

Full Intervention 31.1 (6.7) 29 (6.5)


of 767

Figure 23 Comparison of AQoL scores (only patients with thirty day data)

Again, there was no significant difference found between the three groups for the AQoL

score, as inpatients or at 30 days post-discharge (ANOVA test F = 1.12, df = 2 and 691,

p = 0.33). The groups all displayed a significant improvement in AQoL from baseline

to the 30-day follow-up (ANOVA test F = 17.0, df = 1 and 521, p<0.0001).

The illness dimension of the AQoL was also separately examined as above, producing

very similar statistical results as the total AQoL scores.

11.3.2 Readmission costs/rates within thirty days of discharge

11.3.2.1 Patient reported readmissions (all sites) All patients were asked during their 30-day follow-up telephone call whether they had

been readmitted to hospital since their initial admission. However, it is the experience of

the project team, that this is not always a reliable way of collecting readmission data.

Therefore, where possible (in Tasmania and Bendigo), trial officers also consulted

medical records to identify readmissions for trial patients over the 30-day post-

discharge period. A readmission was defined as spending at least one night in hospital.

26

28

30

32

34

36

Inpatient 30 daysSurvey Time

Mea

n Q

OL

Scor

e

Discharge counselling or nil Discrepancy check & counselling Discrepancy check & HMR


of 767

The following analysis included all patients who could be contacted at the 30-day

follow-up phone call point from all sites (n=378).

There was no statistically significant difference in readmissions across the groups (χ2

=0.18, df = 2, p = 0.92).

Percentage of patients that reported a hospital readmission:

• 18% Full intervention

• 19% Partial Intervention

• 17% Minimal intervention

Most (82%) of the readmissions were seemingly unplanned.

It was interesting to note that of the 9 patients who reported they were readmitted within

5 days of initial discharge (“rebound readmission”), 8 were control patients and only 1

was an intervention patient. Unfortunately, due to the small numbers, this was not

found to be statistically significant, but a trend was seen (χ2 = 2.6, df = 1, p = 0.11).

Readmissions within 90 days of initial discharge were also examined from medical

records for Tasmanian patients. Again, differences across the groups were not

statistically significant.

11.3.3 Additional health care costs for trial patients The following results include a sample of patients from the Tasmanian sites (RHH and

LGH) and Bendigo with 30-day follow-up data (n=158). The percentage of patients

that visited their GP within 30 days of discharge was similar across the groups (χ2 = 1.9,

df = 2, p = 0.39).

The percentage of patients that visited a specialist within 30 days of discharge was also

similar across the three groups (χ2 = 8.13, df = 6, p = 0.23).

Table 74 Proportion of patients from each group who visited the indicated practitioner within thirty days post-discharge

Group GP visits, (%) Specialist visits, (%) Full intervention 92% 36%

Partial intervention 81% 37%

Minimal intervention 85% 34%


of 767

The average number of medical consultations and their associated costs were slightly

higher in the full intervention group. This was not surprising as the conduct of the

medication review would generally have necessitated at least one GP visit.

Table 75 Medical consultations for each level of service

Minimal Intervention

Partial Intervention

Full Intervention

Specialist Mean no. visits 1.01 0.8 1.1

Cost of standard visit $72.00 $72.00 $72.00

Mean cost per patient visits $72.70 $57.60 $79.20

General Practitioner Mean no. visits 2.6 3.4 2.9

Cost of standard visit $21.00 $21.00 $21.00

Mean cost per patient visits $54.60 $71.40 $60.90

Total Costs GP + Specialist visits $127.30 $129.00 $140.10

11.3.4 Time trials The time trial conducted at the RHH enabled the costs of the human resources

component of the trial to be established. The time taken to conduct the additional

program tasks per patient was estimated to be 30 minutes; 20 minutes of this could have

been reduced had a fully functional ICT system been in place. From this we can

suggest a hospital-based liaison pharmacist would need to be employed to optimally

implement the services in a national rollout.

The rate of pay chosen for pharmacists for the purpose of this economic analysis was

$40/hr. The project team is aware that there are a number of different awards and pay

rates for pharmacists across Australia but chose this as a typical working figure.

Therefore, approximately 30 minutes in additional time required would result in $20

labour cost to provide the service per patient. Most of the additional costs are the

manual tasks, which could be eliminated if the ICT supported the original model.


of 767

11.3.5 Clinical panel and economic analysis For the discrepancy review and PDMR panels, the attributability to the pharmacist was

considered to be 64% and 85%, respectively. For instance, it was estimated that 64% of

the time, in the absence of the trial pharmacist’s recommendations, no recommendation

or change would have been made in the hospital with medication discrepancies

(attribution was derived from the trial’s control patients’ actual number of discrepancies

corrected within 48 hours).

As indicated below, an average discrepancy would prevent 41 days of health loss and

save $205 in financial savings to the health sector. Similarly, just one randomly

selected PDMR recommendation would prevent 46 days of health loss and $206 in

financial savings to the health sector. It was estimated that 85% of the time another

health professional would not have intervened to correct the issues found throughout the

PDMR, had the pharmacist not intervened.

11.3.5.1 Step One: The economic value of interventions (trial results)

The economic value of a typical intervention recommended as a result of a discrepancy

review or PDMR was estimated.

The average economic value of a discrepancy review is:

• 41 days of health loss prevented,

• 0.19 days in hospital prevented (discrepancy reviews as assessed by the clinical panel with reference to the consequences table led to benefits in terms of reduced average days in hospital from 0.31 per discrepancy to 0.12),

• 0.67 medical consultations prevented,

• $205 financial savings to the health sector, and

It was found that 64% of the time, another health professional would not have

intervened to correct the discrepancies, had the trial pharmacist not intervened.

This means that the average economic value of a discrepancy review performed as a

result of Med eSupport is:


• 0.12 days in hospital prevented,

• 0.43 medical consultations prevented, and

• $131 financial savings to the health sector


of 767

The average economic value of a post-discharge Home Medication Review is:





It was estimated that 85% of the time another health professional would not have

intervened to correct the major issue identified, had the pharmacist not intervened.

This means that the average economic value of a post-discharge Home Medication

Review performed as a result of Med eSupport is:




• $175 financial savings to the health sector.

Table 76 presents the results of the clinical panel’s assessment on the value of

discrepancy reviews at admission.


of 767

Table 76 The value of discrepancy reviews - 19 reviewed cases

Prior to intervention Average Total Days in level 3 loss 9.38 178.2

Days in level 2 loss 24.92 473.4


Days in hospital 0.31 5.87

Medical Consultations (number) 1.11 21.13

Total costs $343.74 $6,531.11

After intervention Average Total Days in level 3 loss 2.70 51.30





Total costs $138.41 $2,629.85

Effect (prior – after) Average Total Days in level 3 loss prevented 6.68 126.90

Days in level 2 loss prevented 15.11 287.10


Days in hospital prevented 0.19 3.58

Medical Consultations prevented (number) 0.67 12.79

Financial Savings $205.33 $3,901.26

Effect adjusted for attribution, (64% attributability§§)

Average Total






Financial savings adjusted for attribution $131.41 $2,496.81

§§ Attribution was derived from control patients’ actual number of discrepancies corrected within 48 hours


of 767

From the clinical panel analysis of the selected HMRs the impact of the review can be

seen in Table 77. Each review performed led to savings in the order of $175 to the

health system.

Table 77 presents the results of the clinical panel’s assessment on the value of

PDMRs/HMRs.

Table 77 The effect of HMRs – 20 reviewed cases

Prior to intervention Average Total (20 reviews) Days in level 3 loss 12.15 242.97


Days in level 1 loss 51.13 1,022.61



Total costs $520.86 $10,417.11

After intervention Average Total (20 reviews) Days in level 3 loss 7.56 151.28





Total costs $314.90 $6,298.06

Effect (prior – after) Average Total (20 reviews) Days in level 3 loss prevented 4.58 91.69





Financial Savings $205.95 $4,119.05

Effect adjusted for attribution, (85% attributability)***

Average Total (20 reviews)




*** Attribution was derived from control patients’ actual number of discrepancies corrected within 48 hours


of 767



Financial savings adjusted for attribution $175.06 $3,501.19

As indicated in Step Two, recommendations were suggested 66% of the time by the

pharmacist after review of the initial drug chart. PDMRs/HMRs contained

recommendations 87% of the time. However, not all of the recommendations made by

pharmacists were implemented by prescribers. Discrepancy review recommendations,

which occurred in the hospital, were implemented 78% of the time. However, post-

discharge medication review recommendations, which occurred in the community

setting, were implemented only 29% of the time.

11.3.5.1.1 Discrepancy recommendations At the time of admission, the trial officer compiled a list of potential discrepancies and

the recommendations they made for each patient and discussed these with the hospital

doctor, within 24 hours of that patient’s admission. The trial officer then recorded if,

for those discrepancies found to be legitimate, the recommendation had been acted on

within 48 hours of admission (24 hours from when it was provided) or if it was not

acted on for the duration of patient’s stay. It was then possible to ascertain the

percentage of patients who had a recommendation acted on by the hospital medical

practitioners.

11.3.5.1.2 PDMR/HMR recommendations An assessor reviewed each of the PDMR/HMR recommendations made for each patient

who received a PDMR/HMR throughout the trial. The assessor reviewed the patient’s

medication list at the time of the 30-day post-discharge follow-up (provided by the

community pharmacy and the patient). A comparison revealed if the recommendations

had been taken up by the GP (e.g. Patient commenced on therapy).

11.3.5.2 Step Two: Rate of recommendations per review and uptake of recommendations

For discrepancy reviews:

• Discrepancies were identified on 66% of patient’s drug charts at admission and recommendations were made to the hospital doctors to correct these, and

• 78% of the time these discrepancies were corrected by hospital doctors.


of 767

For PDMRs/HMRs:

• 87% of home medication reviews contained recommendations, and

• At the 30 days post-discharge point, these recommendations were implemented only 29% of the time.

The average value of a review was found to be reduced for two reasons. Firstly, the

average value of a review is reduced because not all reviews are associated with a

recommendation. (Refer to Scenario 1 and 2 in Table 78 and Table 79). Secondly,

because not all of the recommendations were acted upon, the full benefits were not

achieved. (Scenario 2, Table 79). For instance, the average financial savings associated

with each post-discharge medication review drops from $175 to $51 with the relatively

low implementation rate observed. The following tables illustrate the effects of full or

partial implementation of the pharmacists’ recommendations.


of 767

Table 78 Scenario 1 – full implementation of the recommendations

Reviewed cases HMR, (n=20)

Discrepancy, (n=19)

Average days in level 3 loss prevented 3.9 4.3



Average number of bed days prevented 0.16 0.12

Average number of medical consultations prevented 0.6 0.4

Estimated financial savings $175.00 $131.00

Other Cases, (Cases with discrepancies but not reviewed)

HMR, (n=105)

Discrepancy, (n=76)

Total days in level 3 loss prevented 409 325

Total days in level 2 loss prevented 1,471 735


Total number of bed days prevented 16.64 9.16

Total number of medical consultations prevented 67 33

Estimated financial savings $18,381.00 $9,987.00

Total Cases (Reviewed + Other) HMR,

(n=125) Discrepancy,

(n=95) Total days in level 3 loss prevented 487 406


Total days in level 1 loss prevented 2,627 1,154


Total number of medical consultations prevented 80 33.76


Intervention Rate HMR Discrepancy 87% 66%


of 767

Table 79 Scenario 2 Adjusted for uptake of recommendations by prescriber

Reviewed cases PDMR/HMR, (n=20)

Discrepancy, (n=19)




Average number of bed days prevented 0.05 0.09

Average number of medical consultations prevented 0.19 0.34

Estimated financial savings $51.00 $103.00

Other Cases (Cases with discrepancies but not reviewed)

PDMR/HMR, (n=105)

Discrepancy, (n=76)







Total Cases (Reviewed + Other) PDMR/HMR, (n=125)

Discrepancy, (n=95)







PDMR/HMR Discrepancy Percentage uptake of recommendations 29 % 78%

11.3.5.3 Step Three: Time taken for reviews For discrepancy reviews:

• The time taken to conduct the additional program tasks was estimated to be 30 minutes; 20 minutes of this could have been reduced had a full ICT system been in place.

PDMR reviews:

• The average review was estimated to take 180 minutes, although the time varied substantially.


of 767

The additional activities associated with the discrepancy review included the following:

• Identification of discrepancies (at admission),

• Compilation of a reconciled prescribed medication list (on admission),

• Reconciled list discussed with the ward pharmacist/hospital doctor, and

• Discrepancies at discharge were recorded and discussed with the ward pharmacist/hospital doctor.

It is expected that a full ICT solution, in the form of a secure website or database, would

have the potential to prevent tasks such as those listed below and save around 20

minutes per discrepancy review per patient.

• Printing and faxing, by community pharmacist, of patient medication dispensing history to trial officer (in hospital) at admission,

• Manual upload of medication history to the website, and

• Creation, printing and faxing of patient medication discharge information to GPs and community pharmacists.

11.3.5.4 Step Four: Estimate the per review, per site and national costs

Per review costs:

Cost per discrepancy review

• Cost to program: $5 (payment to community pharmacy for upload of medication history), and

• Cost to hospital for information provided to GP and community pharmacist at discharge: $20 without ICT and $6.67 with ICT.

Cost per PDMR/HMR

• Cost to Community Pharmacy Agreeement: $140 payment to community pharmacies

• Cost to MBS: $120 to GP, and

• Cost to hospital: $0.

Per site costs and activity:

• Cost to program (site overhead costs): $32,500 per annum, and

• It is estimated that 4,500 discrepancy reviews and 4,500 HMRs per site would be conducted.

National costs:

• National overhead costs: $180,000 per annum.

In this proposed model, the hospital would be expected to cover the costs of the

discrepancy review (most hospital pharmacists already provide a similar service, but


of 767

without obtaining community pharmacy data as a six-month dispensing history).

However, it would be assumed that the community pharmacy would be paid a nominal

fee for provision of the patient’s medication dispensing summary to the hospital at

admission.

Table 80 Costs of program: payments

Cost of programs: Per review costs

Payments to community pharmacist and GP

Discrepancy review

By program to community pharmacy $5.00

PDMR/HMR

By Community Pharmacy Agreeement to community pharmacy $140.00

By MBS to GP $120.00

Hospital based pharmacist costs

Cost per pharmacist per hour $40.00

Cost per Discrepancy

With ICT $6.67

No ICT in place $20.00

Cost per PDMR/HMR

With ICT $0.00

No ICT in place $0.00


of 767

Table 81 Costs of program: program fixed costs at sites and nationally

Costs of program: Program fixed costs at sites and National Per site

Training costs per annum

Pharmacists $2,000.00

Other $500.00

Administration

Administration per site $20,000.00

Software and hardware

Software and hardware per site $5,000.00

Other $5,000.00

Total site overhead costs $32,500.00Nationally

Administration (Divisions of General Practice/PGA Branches) $150,000.00

Software $30,000.00

Total national overhead costs $180,000.00

We characterised a site, as including a hospital of the size similar to the Royal Hobart

Hospital (400 beds), with 5623 medical admissions per annum (4498 patients age 50

years plus, with 3 or more medications and 2 chronic disease states) and perhaps a local

Division of General Practice.

The site training costs could be utilised to host a half-day workshop (4 hours) for

approximately 18 pharmacists. These pharmacists would be both hospital-based and

accredited pharmacists. The training would focus on the processes only, and aid to

initiate and improve communication between the two groups at the outset. Assuming a

fair knowledge and experience with computers, the introduction to the website and a

how to use session would only need to be short. Any additional training, (e.g.

medication review quality and clinical skills revision), would also be assumed to be

separate and standard knowledge. The other training costs would enable a promotional

session for GPs, community and accredited pharmacists and hospital doctors at the

launch of the program. This training would be aimed mainly at improving awareness

about the service and to encourage its use. A quick overview of the website, including


of 767

the advantages and ease of use, security and a question/answer session would be

sufficient training if a telephone support line was set up for the program users.

The administration could cover aspects such as provision and maintenance of up to date

contact details of the key health care providers involved in the program. It might

provide a telephone support line for website users (health professionals and patients). It

could also be used to supervise and co-ordinate the information transfer between the

hospital and the community. A staff member (at approximately 0.4 FTE) would need to

be available to co-ordinate accredited pharmacists for the post-discharge medication

reviews. Ideally, this role would be fulfilled by existing MMR facilitators. Circulation

and installation of software updates would be necessary for each site on at least a 6-

monthly basis.

The software and hardware budget would provide for a server per site and possibly two

additional computers for storage and transfer of information over the

hospital/community interface. The national estimated costs for 50 sites would fund a

national coordinator and staff to train each state/site representative.


of 767

Table 82 Costs of program: activity per site

Costs of program: activity per site Number of sites in initial rollout 50

Discrepancy

Total discrepancy reviews 4,500

% that would otherwise have occurred 36%

Number of additional reviews due to Med eSupport 2,880

Intervention rate 66%

Number of cases of intervention 2,970

Number of cases of additional interventions 1,901

PDMR/HMRs

Total PDMR/HMRs 4,500

% that would otherwise have occurred 10%

Additional reviews due to Med eSupport 4,050



Number of cases of additional interventions due to Med eSupport 3,524

All reviews

Total reviews (discrepancy reviews + PDMR/HMRs) 9,000

Additional reviews due to Med eSupport 6,930



Number of additional interventions due to Med eSupport 5,424

11.3.5.5 Step Five: The cost and effect of a national program Number of Sites

• Assumed number of sites in initial rollout: 50 sites

Cost per annum:

• $63M in total costs and $57M in additional costs (as some PDMRs/HMRs and discrepancy reviews would otherwise have occurred)

Reviews per annum:

• 450,000 reviews and 288,000 additional reviews

Days of health loss prevented:


of 767

• 11.6M (additional) to 14.6M (total) days of health loss prevented if all recommendations are implemented and between 5.2M (additional) and 7.1M (total) days of health loss prevented if adjusted for current rate of uptake.

Financial savings per annum

• Between $54M (additional) and $69M (total) in financial savings to sector if all recommendations are implemented and between $25M (additional) and $34M (total) in financial savings with partial uptake of recommendations.

The total and additional cost of the program if implemented at the expected number of

sites is presented in Table 83 and the outcomes in terms of days of health loss

prevented, service use prevented and financial savings are presented in Table 84.


of 767

Table 83 Costs of program: total costs per annum of a national program

Cost of program: Total costs per annum Additional costs

due to Med eSupport

Total costs

Site costs – overheads $32,500.00 $32,500.00

Reviews Discrepancy

Number of reviews 2880 4500

Costs to Hospitals (assuming ICT in place) $19,200.00 $30,000.00

Payments to Community Pharmacists $22,500.00 $22,500.00

PDMRs

Number of PDMRs 4050 4,500

Payments to Community pharmacists (Community Pharmacy Agreeement) $567,000.00 $630,000.00

Payments to GPs (MBS) $486,000.00 $540,000.00

All reviews

Site overhead costs $32,500.00 $32,500.00

Reviews (number) 6,930 9,000

Payments to Hospitals $19,200.00 $30,000.00

Program payments to community pharmacy $22,500.00 $22,500.00

MBS payments to GPs and Community Pharmacy Agreeement payments to accredited pharmacists $1,053,000.00 $1,170,000.00

Cost per site Hospital $19,200.00 $30,000.00

Program $22,500.00 $22,500.00

MBS and Community Pharmacy Agreeement $1,053,000.00 $1,170,000.00

Cost all sites (n=50 Number of sites 50 50

Site overhead costs $1,625,000.00 $1,625,000.00

Hospital $960,000.00 $1,500,000.00

Program $1,125,000.00 $1,125,000.00


Total $56,360,000.00 $62,750,000.00

National costs Program Overhead costs $180,000.00 $180,000.00

Total costs Site overhead costs $162,500.00 $162,500.00


of 767

Hospital $960,000.00 $1,500,000.00

Program + program overhead costs $1,305,000.00 $1,305,000.00


Total $56,540,000.00 $62,930,000.00

Table 84 Outcomes per site, two scenarios

Outcomes: per site (Two scenarios)

Outcome Scenario 1: Full uptake of recommendations

Scenario 2: Likely uptake of recommendations

Total reviews Additional

reviews due to Med eSupport

Total reviews Additional

reviews due to Med eSupport

Discrepancy reviews per site Reviews 4,500 2,880 4,500 2,880

Number of days of improved health state

117,413 75,144 91,582 58,613

Service use change due to discrepancy reviews – per site Days in hospital prevented

543 345 423 271

Consultations prevented 1,939 1,241 1,512 968

Financial savings $591,349.00 $378,463.00 $461,252.00 $295,201.00

PDMRs - per site Reviews 4500 4050 4500 4,050


175,145 157,630 50,792 45,713

Service use change due to PDMRs – per site Days in hospital prevented

713 642 207 186

Consultations prevented 2,875 2,587 834 750

Financial savings $787,769.00 $708,992.00 $228,453.00 $205,608.00

Total reviews - per site


of 767

Reviews 9,000 6,930 9,000 6,930


292,558 232,774 142,374 104,325

Total service use change (due to PDMR and discrepancy reviews) – per site Total days in hospital prevented

1,256 987 630 457

Total consultations prevented

4,814 3,828 2,346 1,718

Total financial savings $1,379,118.00 $1,087,455.00 $689,705.00 $500,809.00

Total - all sites

Reviews 450,000

346,500

450,000 346,500


14,627,879.00 11,638,721 7,118,705 5,216,266

Total service use change – all sites Total days in hospital prevented

62,788 59,222 31,502 22,850

Total consultations prevented

240,677 191,406 117,294 85,906

Total financial savings - all sites

$68,955,876.00 $54,372,752.00 $34,485,254.00 $25,040,450.00

11.3.5.6 Step Six: Estimate the economic value of the program

Estimate of the national economic value of the overall program, includes cost of

medication history reconciliation at hospital and post-discharge medication reviews.

The ultimate measure of the economic value of a program is defined as the expected

additional costs compared to the expected additional benefits.

We looked at the expected total and additional costs compared to the total and

additional number of days of health loss prevented, excluding the net of the expected

financial savings.


of 767

Assume full compliance with pharmacists’ recommendations:

• The program would be close to cost-neutral to offer. The additional cost per additional day of health loss prevented is $4.86, and $0.19 net of expected financial savings.

Assume partial compliance with recommendations:

• The program would cost less than $30M per annum more than the financial benefits to offer. The additional cost per day of health loss prevented is $10.84, and $6.04 net of expected financial savings.

Is the program value for money?

• Even assuming partial compliance and excluding the expected financial savings, the program represents value for money at an additional $10.84 per day of health loss prevented.


of 767

Table 85 Economic value of a national program

Value for Money: Two scenarios All Sites

Value for money Scenario 1 Scenario 2

All reviews Additional reviews All reviews Additional

reviews

Cost per annum Site overhead costs $1,625,000.00 $1,625,000.00 $1,625,000.00 $1,625,000.00

Hospital $1,500,000.00 $960,000.00 $1,500,000.00 $960,000.00

program $1,305,000.00 $1,305,000.00 $1,305,000.00 $1,305,000.00

MBS and Community Pharmacy Agreeement

$58,500,000.00 $52,650,000.00 $58,500,000.00 $52,650,000.00

Total additional costs $62,930,000.00 $56,540,000.00 $62,930,000.00 $56,540,000.00

Financial savings $68,955,876.00 $54,372,752.00 $34,485,254.00 $25,040,450.00

Additional costs net of savings -$6,025,876.00 $2,167,248.00 $28,444,746.00 $31,499,550.00

Reviews

All reviews 450,000 346,500 450,000 346,500

Cost per review $139.84 $163.17 $139.84 $163.17

Cost per review (net of Financial savings) -$13.39 $6.25 $63.21 $90.91

Additional days of health loss prevented

Days of health loss prevented 14,627,879 11,638,721 7,118,705 5,216,266

Cost per day of health loss prevented $4.30 $4.86 $8.84 $10.84

Net cost per day of health loss prevented -$0.41 $0.19 $4.00 $6.04


of 767

11.4 Discussion

11.4.1 Time trials As anticipated, there was a difference between the median times spent on the two

patient groups. This was mainly due to the fact that a number of services were not

provided to the control patients. It was considered a mean time of 30 minutes was a fair

estimate of the time necessary to provide the services, and 10 minutes if ICT

(particularly in terms of uploading medication data from community pharmacy

dispensing software other than Winifred® or Rex® and from hospital systems) was fully

supportive of the proposed model. Cornish et al suggested in their results that the

median time for the entire process of medical chart review, interview and follow up on

discrepancies was 24 minutes (inter-quartile range, 20-30 mins).35

Assumptions were made in regards to the services currently provided by hospital

pharmacists in regards to discrepancy review type services at admission into hospital.

The level of services assumed to be already provided were based on the SHPA

standards of practice.115 This was incorporated into the costs of providing the service

on a site and national level.

For the purpose of the economic evaluation it was estimated a typical PDMR/HMR

would take 180 minutes. There was substantial variation in the time taken for

PDMR/HMRs to be conducted. This occurred for a number of reasons. One should not

rule out geographical distances travelled, availability of the patient and reviewer or

other external influences. The group of accredited pharmacists included in the trial had

varying levels of experience at conducting PDMR/HMRs. Skills such as time

management, clinical knowledge and confidence on the reviewer’s part, all influence

the time taken to produce a PDMR/HMR report.

11.4.2 Clinical Panel The clinical panel assessment was closely modelled on that used in the PROMISe

study.116 There were two reasons why PROMISe used this additional assessment to rate

the clinical significance of the interventions. Firstly, the PROMISe pilot study data

indicated that the recording pharmacists had a tendency to overestimate the clinical

significance of the interventions. This meant that many of the interventions classified

as being of severe clinical significance, when evaluated by other health professionals


of 767

were not considered so. This could also be applied to an individual trial officer’s

perception of an intervention made in the Med eSupport trial.

Secondly, and more importantly, the PROMISe Project Team felt that the assessment of

the clinical significance of the intervention required more than a unilateral estimate of

its severity. Although the potential severity of a particular outcome or consequence

from the intervention was an important factor, there were other factors that should be

considered in the assessment of the potential consequences of an intervention.

The factors to be considered were:

• The likelihood that any particular event or consequence would actually occur after an intervention,

• The likelihood that the same consequence may occur despite the intervention,

• The likelihood of a less or more severe manifestation of the same consequence, and

• The likelihood that another health professional may detect and resolve the problem.

The panel was a useful way to determine the economic value of interventions. This

method of analysis had been used successfully in the community setting for the

PROMISe study.116 In the process, costing tables had been established to estimate the

savings that may be associated with prevention of hospitalisation and use of other

medical services, as a result of pharmacist interventions. This placed patients in the

community pharmacy setting only as opposed to the hospital setting. It was possible to

apply the PROMISe estimate tables to the Med eSupport case studies. However, it was

necessary for the Med eSupport panel to consider patients in the community setting

(PDMR/HMR) and also in the hospital setting (discrepancy reviews). Additionally,

Med eSupport had outcomes for each of the case studies which made it easier for the

panel to give a more accurate response to each situation.

The panel consisted of clinicians from a variety of backgrounds. Despite this, each

panel member gave very similar answers for each of the cases which suggests they all

had a common interpretation of the information provided.

11.4.3 The economic value of interventions The savings associated with an intervention either in the hospital setting (discrepancy

review) or in the community setting (PDMR/HMR review) identified in this economic


of 767

analysis are supportive of continuation of a medication management service across this

care interface. There is a very strong argument for continuation of services when the

savings to the health sector are estimated at $131 for each discrepancy picked up on a

drug chart at admission and it is possible to prevent the patient from having 26 days of

health loss per corrected discrepancy.

The estimated savings to the health sector of $175 and the 39 days of health loss

prevented from having just one recommendation made in a PDMR/HMR review also

implies that the service represents value for money.

It should be noted that the suggested savings seen in the panel results in Table 78 and

Table 79 are conservative in manner. Patients enrolled in the Med eSupport trial were

found to have a median of one discrepancy (range 0 – 14) per drug chart at admission

(from Table 17). The panel results reflect the costs associated with just one discrepancy

occurring per patient at admission, and as a result the savings are conservative in nature.

Cornish et al reported on a study that found 60% of patients admitted to hospital will

have at least one discrepancy in their admission medication history and 6% of these

patients will have an inadvertent drug discontinuation of a serious nature on admission

to hospital.35

Also, a review of the medication reviews found a median of two therapy

recommendations per review. However, the panel was again asked to consider only one

recommendation in each of the cases/reviews. Sorensen et al reported that an average

of 5.5 problems was identified per home medication review and an average of 6.8

(range 1-17) recommendations was suggested for each review. The reviewing

pharmacist sometimes suggested a range of possible recommendations for one problem,

which explains the larger number of recommendations than problems identified.117

The warfarin focused aspect of the Med eSupport trial; a specific 'high-risk' example

of 767

12 The warfarin focused aspect of the Med eSupport trial; a specific 'high-risk' example.

12.1 Background Adverse events from warfarin use in Australia in 1992 were estimated to cost over $100

million per annum in direct hospital costs alone.118 As noted by Murray in 2003 there

needs to be more emphasis placed on assisting the elderly cope with existing

medications.119

“Many major improvements in medication use among older adults will also depend on closing the gap between knowledge and practice and increasing the ability of older adults to manage their medications.” 119

The major complication of anticoagulant therapy is bleeding.120-122 Based on estimates

from randomised trials, the average annual frequencies of fatal, major, and total

bleeding during long-term warfarin therapy are 0.6%, 3.0%, and 9.6%, respectively;

these frequencies are approximately five times those expected without warfarin

therapy.120-122 Increased variation in the INR is associated with an increased frequency

of haemorrhage independent of the mean INR.123, 124 This effect is probably attributable

to increased frequency and degree of marked elevations in the INR.

Bleeding complications with anticoagulant drugs appear to occur more frequently in

older patients than in younger individuals.121, 122, 125-127 Most individuals who receive

oral anticoagulant therapy are elderly patients with non-valvular atrial fibrillation (AF)

and acute or recurrent venous thromboembolism (VTE). Anticoagulation in the elderly

patients poses unique challenges because they are simultaneously at higher risk for

recurrent thromboembolism and major bleeding, including catastrophic intracranial

haemorrhage. Older patients have characteristics that may place them at higher risk for

anticoagulant-related bleeding, but they also have characteristics that make them more

likely to benefit from the therapy. 121, 126

A number of studies have reported that the risk of bleeding associated with warfarin is

highest early in the course of therapy.120, 122, 124, 128, 129 In one of these studies, for

example, the frequency of major bleeding decreased from 3.0% during the first month

of outpatient warfarin therapy to 0.8% per month during the rest of the first year of

therapy and to 0.3% per month thereafter.130 The intensity of anticoagulant effect is


of 767

probably the most important risk factor for intracranial haemorrhage, independent of the

indication for therapy, with the risk increasing dramatically with an INR > 4.0.131

Patients discharged from hospital are at high-risk of experiencing adverse effects,132, 133

alterations to a patients medication may occur in hospital, such as initiation of warfarin

and patients may not take dosages appropriately after discharge. The initiation of

warfarin is traditionally difficult and the trend to early discharge of patients with

thromboembolic disorders makes the transfer of information to GPs very important.

The safety and efficacy of warfarin depends on maintaining the INR within the

therapeutic range. Analysis of primary prevention trials for stroke prevention in AF,

found that a large number of thromboembolic and bleeding events occurred when the

INR was outside of the therapeutic range.131 Analysis of other cohort studies have

shown a sharp increase in the risk of bleeding when the INR is higher than the upper

limit of the therapeutic range134-136 and the risk of thromboembolism is increased when

the INR falls below 2.0.137, 138 A study published in the British Medical Journal in 2003

showed an excess mortality from all causes with increased INR.139

A validated bleeding index showed that the risk of major bleeding is also related to age

> 65 years, a history of stroke or gastrointestinal bleeding, and co morbid conditions

such as renal insufficiency or anaemia.140, 141 Importantly, these risk factors are

additive; patients with 2 or 3 risk factors have a much higher incidence of warfarin

associated bleeding than those with none or one. The cumulative incidence of major

bleeding at 48 months was 53% in high-risk patients (three or four risk factors), 12% in

middle-risk patients (one or two risk factors), and 3% in low-risk patients (no risk

factors). Kuijer et al142 also developed a prediction model based on age, gender, and the

presence of malignancy. In patients classified at high, middle, and low risk, the

frequency of major bleeding was 7%, 4%, and 1%, respectively after 3 months of

therapy.

Patient adherence has been implicated in 21.1% of preventable adverse events143 and

studies have generally shown a relationship between patient knowledge and adverse

outcomes of warfarin therapy. Good outcomes have been recorded where patients have

had increased participation in their care and encouraged to communicate more

effectively with doctors and other health professionals about drug interactions and

changes in lifestyle or diet.144 Compliance with warfarin is essential to maintain good

anticoagulant control and to prevent unnecessary dosage changes. Pharmacy managed


of 767

anticoagulation clinics have reported increased compliance after instituting education

programs.145

A study by Tang et al146 evaluated patients’ knowledge of warfarin and anticoagulant

control. They concluded that patients’ warfarin knowledge was poor. Less than 50% of

patients knew the strength of their warfarin tablets, the reason for taking warfarin and its

effect on the body. They went on to report that;

“Their deficiencies in knowledge were even more obvious with respect to the possible consequences of under- or over-anticoagulation, drugs and medicated oils that might interact with warfarin and the management of a missed dose”

Patients who read the anticoagulation booklet on warfarin had better knowledge than

those who hadn’t. Most importantly from this study was a positive correlation between

patients’ warfarin knowledge and the number of INR values that were in the therapeutic

range.146

The aim of this project was to evaluate whether home-based follow-up, by a pharmacist,

(incorporating, POC INR monitoring, anticoagulant education and medication review)

in the management of targeted ‘high-risk’ patients transitioning from hospital to

community care:

• Resulted in safer and more effective initiation of anticoagulation,

• Reduced anticoagulant-related bleeds,

• Reduced unplanned readmissions related to anticoagulation (thromboembolic or haemorrhagic complications) within 90 days of initial discharge from hospital, and

• Was valued and welcomed by patients and their general practitioners.

The underlying objective was to improve the safety and efficacy of anticoagulant drug

therapy in clinical practice. In a previous study with similar methodology (Ref Jackson

et al) 128 patients were randomised (First phase of recruitment). The aim of this further

recruitment was to increase the sample size and hence, statistical power, in order to

improve the assessment of the clinical and economic outcomes associated with the

intervention (second phase).


of 767

12.2 Methodology

12.2.1 Recruitment procedure Patients were recruited from the Royal Hobart Hospital (RHH). The RHH is a 450-bed

acute care teaching hospital and the only major public hospital in the southern region of

Tasmania. Inpatients initiated on warfarin at the RHH, between February 2002 and June

2003, were prospectively identified by the project pharmacist and/or ward pharmacists;

128 patients were recruited in this first phase.147 Recruitment by a part-time trial officer

was initiated for a second phase of this project, which commenced in May 2004 and

finished in April 2005. Recruitment was stopped at a total of 163 enrolled patients due

to the overwhelming benefit of the study in April 2005.

Those patients who provided informed consent were allocated to either the intervention

(Post-discharge INR monitoring; PDINR) or control (Usual Care; UC) group using a

computer-generated random number sequence. Patients were informed to which group

they were randomised. All patients received regular medical, nursing and pharmacy care

during their hospitalisation.

12.2.2 Post-discharge PDINR monitoring procedures General Practitioners for patients in the PDINR group were telephoned at discharge to

inform them that their patient had consented to be involved in the project. PDINR

patients received a home-visit by the project pharmacist on alternate days on four

occasions, with an initial visit two days after discharge from hospital. The project

pharmacist using a Point-of-Care (POC) INR monitoring device, (CoaguChek S, Roche

Diagnostics, Australia), tested the INR and discussed a number of important educational

points regarding anticoagulant therapy such as goals, adverse effects and interacting

medications with warfarin This was achieved in a consistent format by using the

pharmacist checklist for counselling, shown below in Figure 24.


of 767

Figure 24 Pharmacist's checklist for warfarin counselling.

All patients were given the one-page guide to warfarin treatment (see Figure 25) in

addition to the warfarin educational booklet they were all given on discharge from

hospital. This, combined with the educational warfarin booklet was used as the basis for

patient information materials. The pharmacist liaised with family members and

community pharmacists where necessary to ensure follow-up education and support was

provided.

Checklist for warfarin patient counselling:

Reason for treatment Mechanism of action Explanation of INR, target range and regular testing Compliance (maintaining a diary of INRs, doses) Possible effects of poor control of anticoagulation

Bleeding or severe bruising Recurrence of thromboembolism

Appropriate action if excessive bleeding or bruising occurs Appropriate action if diarrhoea or vomiting occurs Starting a new treatment or changing a dose of current

treatment Common OTC medication interactions, such as aspirin,

NSAIDs, paracetamol, complementary therapies and laxatives. Role of vitamin K, and the importance of consistency in regards

to vitamin K rich foods in the diet, rather than avoidance. Alcohol intake Minimise high risk activities associated with the risk of

physical trauma Medic Alert bracelet/necklace and warfarin ID card


of 767

Figure 25 One page guide to warfarin treatment for patients.

For the first phase, major drug-related problems were discussed with the GP (normally

at the first visit) although formal reports were not sent to the GP. For the second phase

the pharmacist performed a clinical medication review (at the first visit), problems were

One-page guide to warfarin treatment:

1 Warfarin belongs to a class of medications called anticoagulants (“blood thinners”)

2 Warfarin keeps blood clots from forming or getting larger.

3 Many medications can change the way warfarin works. Ask your doctor or pharmacist about using any other medication, including over-the counter medications, vitamins and herbal products

4 Make sure your doctor or pharmacist know if you are taking aspirin or aspirin-like medications, such as medications for pain relief and the common cold

5 Avoid drinking large amounts of alcohol

6 Certain foods will change the way warfarin works. Do not change your diet while taking warfarin. Foods that contain vitamin K (such as lettuce, spinach, broccoli, cabbage, cauliflower or liver) decrease the anti-clotting effect of warfarin. If you eat foods that have vitamin K, do not change the amount of these foods that you normally eat per week. The main point regarding diet is to eat a consistent amount of foods per week that contain vitamin K.

7 It is very important to have regular blood tests while taking warfarin. The test is called an INR, and it measures how thin your blood is compared to normal.

8 You should carry an identification card that shows you are taking warfarin.

9 Make sure your doctor or dentist knows you are taking warfarin before you have any surgery or dental work.

10 You should report the following to your doctor immediately

o Bleeding from the gums or nose o Coughing up blood o Red or black bowel motions o Red or dark-brown coloured urine o Unusually heavy menstrual bleeding o Heavy bleeding from cuts or wounds that does not stop o Easy bruising o Severe headache

If you miss a dose: Take the missed dose as soon as possible. If you do not remember until the next day, skip the missed dose. Only take your usual dose for the day. You should not use two doses at the same time.

*Adapted from the institute for clinical systems improvement (www.icsi.org)


of 767

documented and written recommendations from this review were sent to the GP (this

was generally discussed with the GP at the second or third visit). The main components

of the intervention comprised anticoagulant and other medication education, INR

monitoring, and medication review. The protocol for the PDINR is outlined in the

flowchart below (

Figure 26) and examples can be found in Appendix XXXIV, page 755.

Figure 26 Protocol for Post-discharge INR Monitoring Group

Patients’ GPs were telephoned with each INR result during the four visits and

subsequent dosage changes, if considered necessary, were discussed. All GPs were sent

a personalised letter and information sheet when the patient was discharged, indicating

the group that the patients were randomised to and what follow-up they would receive.

GPs with patients assigned to the PDINR group received a personalised letter after the

fourth visit containing INR results and doses received during the follow-up, and a GP

survey was also sent with the below letter and a reply paid envelope.

Post-discharge INR Monitoring Group

Day 2INR monitoring and medication review

GP called with INR and medication review report sent

Day 4INR monitoring

GP called with INR and medication review report discussed (if received)

Day 6INR monitoring

GP called with INR and medication review report discussed

Day 8INR monitoring

GP called with INR

Post-discharge INR Monitoring Group

Day 2INR monitoring and medication review

GP called with INR and medication review report sent

Day 4INR monitoring

GP called with INR and medication review report discussed (if received)

Day 6INR monitoring

GP called with INR and medication review report discussed

Day 8INR monitoring

GP called with INR


of 767

12.2.3 Usual care group procedures Instructions in the letter to GPs caring for UC patients indicated that this project was not

responsible for the monitoring of their patients and patients would receive a visit from

the project pharmacist eight days after discharge to determine anticoagulant control.

GPs caring for UC patients were telephoned on day eight if an adverse trend was

identified or if the INR was out of the therapeutic range.

12.2.4 Data collection Baseline demographics, indication for anticoagulation, current medications and

presence of contraindications to warfarin were assessed after consent for enrolment had

been given. Alternate day INRs for four visits after discharge were recorded for the

PDINR group; duration of the visits and outcomes were also assessed. The UC group

had their INR measured at day eight post-discharge. The number of pathology tests

taken by the GP (patient and/or GP reported) and INR results obtained by the GP

(patient and/or GP reported) were also recorded.

A multiple-choice ‘satisfaction questionnaire’ was given to each PDINR group patient,

to be completed anonymously after the fourth visit, along with a reply-paid envelope. A

systematic step-wise approach to assessing bleeding (major and minor) and embolic

complications, adapted from Heidinger et al,148 was used. All patients were interviewed

at 90 days after discharge to assess these complications, the types and frequency of

these events. The event rates were assessed through a combination of self-reported

events and medical record notes.

Patient knowledge was assessed using a self-administered anonymous questionnaire

with a reply paid envelope by all patients at 90 days after initial discharge using a

composite questionnaire from previous studies.146, 149, 150

12.2.5 Data analysis A number of outcomes were assessed, including the achievement of a therapeutic INR

value on day eight after discharge. Therapeutic ranges according to RHH

anticoagulation protocols were used. The therapeutic range for atrial fibrillation (AF),

venous thromboembolism (VTE), mural thrombus and biological heart valves were 2.0-

3.0. The therapeutic range for mechanical heart valve was 2.5-3.5.


of 767

Primary outcomes were total, major and minor bleeding complications, and unplanned

readmissions due to anticoagulant-related complications. Bleeding was defined as major

if it was clinically overt and associated with either a decrease in the haemoglobin level

of at least 20g per litre or the need for a transfusion of two or more units of red cells; if

it was retroperitoneal or intracranial; if it warranted the permanent discontinuation of

warfarin or if it required hospital admission. All other bleeding complications were

assessed as minor.

The medical records of all patients who reported symptoms of embolic complications

(TIA, stroke, AMI, complications of DVT) were examined and GPs contacted if

necessary to confirm these events. Secondary outcomes were unplanned readmissions

from any cause, death (cause determined from death certificates), proportion of patients

remaining on anticoagulant therapy and warfarin knowledge assessed 90 days after

initial discharge. Follow-up was complete through patient interview and medical record

review for all patients.

Rates of outcomes were expressed as number and percent of events with 95% CI.

Responses between groups were compared by Mann-Whitney tests and Kruskal-Wallis

tests for non-parametric responses. Changes within groups for numerical variables were

compared by Wilcoxon-signed rank tests. Categorical variables were compared by χ2

tests. General Practitioners were anonymously surveyed using a 10-point Likert scale

and PDINR patients were anonymously surveyed after the fourth visit to assess their

opinions of the service. The INR values during hospital admission and after discharge

were plotted with median and inter-quartile ranges marked at 10, 25, 75 and 90

percentiles. The following colouring was applied to tables to assist in readability of

outcomes at discharge, day eight and day 90 after discharge.

Sub-therapeutic

Therapeutic

Supra-therapeutic

12.2.6 Exclusions Exclusions comprised patients who were not being discharged home in Southern

Tasmania, who had dementia and were unable to answer basic questions about their

therapy or who were entering the Patients Acute Treatment and Care in the Home


of 767

(PATCH) program. The PATCH program is comparable with hospital in the home

treatment programs utilised in other hospitals. Patients are still classed as inpatients of

the hospital but are treated at home. Patients who were having warfarin re-initiated after

surgery were also excluded. The exclusion criteria were kept to a minimum to allow the

data to be relevant to a large proportion of patients who are initiated on warfarin in

hospital and discharged to community care.

12.2.7 Cost-effectiveness analysis Based on previous studies151, 152 that have indicated approximately 50% of major

bleeding complications are GI related, 20% are intracranial in origin and 30% are

classified as other major bleeding, the individual costs for each of these types of bleeds

was estimated. Data for estimates of costs was obtained from diagnosis related grouping

(DRG) codes from public hospitals in Australia for the years 2002-2003.153 The cost of

the major bleeding episodes also took into account the following parameters.

• Impact on Health Status

• This was a scale of impact based on the severity of the particular consequence, with ‘1’ being mild impact on health and ‘3’ being a severe impact on health.

• Duration of health status impact

• This was a value in days of the duration of the health impact. For chronic conditions, a one year timeframe was considered.

• Duration of Admission

• The duration in days of any admission associated with the consequence. Where the consequence definition matched that of an existing AR-DRG definition, the information was obtained from the National Hospital Cost Data Collection Cost Weights for ARE-DRG Version 4.2, Round 7 (2002-2003). Where no matching definition existed, the average duration of admission was used.

• Cost of Admission

• A value in dollars for any admission associated with the consequence. This was determined from the same information as the duration of admission.

• Number of General Practitioner Consultations

• The number of community based general practitioner consultations required to manage the particular consequence.

• Cost of General Practitioner Consultations

• The total cost of the general practitioner consultations, based on an average of 3:1 Level B (Item 23) to Level C (Item 26) consultations as per the 2005 Medicare Schedule.


of 767

• Number of Specialist Consultations

• The number of specialist consultations required to manage the particular consequence

• Cost of Specialist Consultations

• The total cost of the specialist consultations, based on an initial consultation cost according to MBF Item 110 and subsequent consultation costs according to MBF Item 26.

• Investigation and Pathology Costs

• The costs of typically required investigation or pathology tests required in the management of the particular consequence. These were based on the schedule fee for the appropriate item.

The average cost event was estimated at $7,030, $4149 & $3402 for intracranial, GI and

other related major bleeds respectively (Table 86). Given these proportions of major

bleed subtypes, the overall estimated admission cost of a major bleed was $4501.


of 767

Table 86 Costs associated with major bleeding sub-types

Subgroup Severity Description

Hea

lth S

tatu

s Im

pact

Dur

atio

n of

Hea

lth S

tatu

s Im

pact

Dur

atio

n of

Adm

issi

on

Cos

t of A

dmis

sion

Num

ber

of G

P C

onsu

lts

Cos

t of G

P C

onsu

lt

Num

ber

of S

peci

alis

t Con

sults

Cos

t of S

peci

alis

t Con

sults

Inve

stig

atio

n or

Oth

er C

ost (

if no

t in

clud

ed in

DR

G c

ost)

Tot

al D

irec

t Cos

ts

Resulting in severe symptoms and signs requiring hospitalisation and medical management (e.g. stroke)

3 360 8.14 $6,257 12 $453 4 $320 - $7,030

Overt gastrointestinal bleeding with haemodynamic consequences requiring prompt medical management

3 90 5.62 $3,881 2 $76 2 $192 - $4,149

Severe bleeding requiring hospitalisation, blood product and/or haemodynamic support

3 30 3.12 $2,952 2 $76 4 $320 $54 $3,402


of 767

12.3 Results

12.3.1 Recruitment A total of 166 patients were approached for their involvement in the study. Three

patients refused participation and two patients enrolled in the PDINR group were

discharged under the care of the PATCH program. Recruitment is displayed in the

recruitment flowchart.

Figure 27 Patient recruitment flowchart

The second phase of patient recruitment was terminated early when 160 patients had

completed the 90-day follow-up due to the overwhelming benefit conferred by the

PDINR group compared to usual care. This decision was made in conjunction with the

Chief investigator, project manager and the project pharmacist. As described later in the

n = 166 approached to enter study

n = 3 patients refused

n = 163 consented and randomised

Post-discharge INR monitoring Group Usual Care Group

Day 8 follow-up

n = 86 Usual care provided –

standard care delivered by GP and home visit received by project

pharmacist on day 8 to obtain INR

n = 75 Medication review including INR

monitoring - received alternate day INR monitoring by project pharmacist

for 4 visits with patient focused education and medication review

n = 1 excluded

Day 90 follow-up

n = 74Post-discharge INR monitoring Group

n = 86Usual Care Group

n = 0 excluded

n = 2 excluded n = 0 excluded

n = 166 approached to enter study

n = 3 patients refused

n = 163 consented and randomised

Post-discharge INR monitoring Group Usual Care Group

Day 8 follow-up

n = 86 Usual care provided –

standard care delivered by GP and home visit received by project

pharmacist on day 8 to obtain INR

n = 75 Medication review including INR

monitoring - received alternate day INR monitoring by project pharmacist

for 4 visits with patient focused education and medication review

n = 1 excluded

Day 90 follow-up

n = 74Post-discharge INR monitoring Group

n = 86Usual Care Group

n = 0 excluded

n = 2 excluded n = 0 excluded


of 767

results over one-quarter of UC patients had a high INR at day eight after discharge

compared to the PDINR group (P<0.0008). Also, this was translated into a significant

difference in major bleeding complications assessed at 90 days after discharge (11% vs

1%) in the UC and PDINR groups respectively, P=0.018. Table 86 summarises the

clinical characteristics of the 160 patients who completed the study. The groups were

well matched with regard to baseline characteristics.

Table 87 Baseline characteristics of trial participants

Characteristic PDINR, n (%)

Usual Care, n (%)

Participants 74 (46) 86 (54)

Age-median (range) 70 (19-94) 71 (20-91)

Age > 75 years 25 (34) 31 (37)

Female 34 (46) 40 (47)

Lives alone 25 (34) 19 (22)

Cognitive deficit 2 (2) 5 (6)

History of documented falls 3 (4) 0 (0)

Cardiovascular admission 49 (66) 55 (64)

Previous stroke or TIA 12 (16) 11 (13)

Hypertension 39 (53) 37 (43)

Diabetes 11 (15) 15 (17)

Ischaemic heart disease 25 (34) 30 (35)

Previous AMI 14 (19) 7 (8)

CCF 17 (23) 16 (19)

Contra-indications to warfarin 12 (17) 10 (12)

Previous warfarin use 11 (15) 11 (13)

Amiodarone use at discharge 19 (26) 13 (15)

Antibiotic use at discharge 15 (21) 19 (22)

Initial bed stay median (range) 8 (3-72) 8 (1-65)

Co morbidities median (range) 4 (1-12) 4 (1-9)

Chronic drugs median (range) 6 (1-12) 6 (1-15)

Table 88 describes the reasons for initiation of anticoagulation, with no significant

differences between the groups. The most common indication for initiation of warfarin

was stroke prevention in AF.


of 767

Table 88 Reason for initiation of warfarin

Reason for use PDINR, n (%)

Usual Care, n (%)

AF 33 (45) 36 (42)

VTE 23 (31) 29 (34)

Valve replacement 15 (20) 16 (19)

Mural thrombus 3 (4) 5 (6)

Table 89 displays the quality of anticoagulation for all patients on discharge from the

hospital. There were no significant differences between the groups. The median INR at

discharge in the PDINR and UC groups was 1.9 (1.0-3.6) and 2.1 (1.0-4.0), respectively

(P = 0.30, U = 3044). The following shading scheme is applied to the tables to assist in

readability.

Sub-therapeutic

Therapeutic

Supra-therapeutic

Table 89 Anticoagulant control at discharge

PDINR, n (%)

Usual Care, n (%)

Sub-therapeutic 40 (55) 39 (46)

Therapeutic 28 (39) 41 (48)

Supra-therapeutic 5 (7) 5 (6)

P=0.46 for differences between PDINR and UC groups

Table 90 displays the level of anticoagulation for all patients at day eight after

discharge. The median INR at day 8 in the PDINR and UC groups was 2.5 (1.3-5.8) and

2.6 (1.0-7.8), respectively (P = 0.87, U 2631).


of 767

Table 90 Anticoagulant control at day eight after discharge

PDINR, n (%)

Usual Care, n (%)




P<0.0008 for differences between PDINR and UC groups

Table 91 Shows a significant trend for UC patients discharged with a sub-therapeutic

INR to have a poorer outcome with respect to quality of anticoagulation at day eight

compared with the PDINR group (P = 0.005). There were no other significant

differences between the PDINR and UC groups with regards to level of anticoagulation

at discharge and INR control at day 8. There was no relationship between INR control at

discharge and INR at day eight for the PDINR group.

Sub-therapeutic

Therapeutic

Supra-therapeutic

Table 91 Anticoagulant control at day eight by discharge INR range and follow-up type

Day eight post-discharge INR

Discharge INR PDINR n (%)

Usual Care n (%)

P value

11 (30) 12 (32)


3 (8) 14 (37)

0.005, χ2 = 11.3

5 (20) 13 (35)


2 (8) 6 (16)

0.19, χ2 = 3.4

1 (33) 3 (60)


0 1 (20)

0.37, χ2 = 2.0

P value 0.72, χ2 = 0.65 0.21, χ2 = 3.1


of 767

Table 92 displays a brief description of anticoagulant-related problems that were noted

at day eight after discharge in the UC group. This list is not exhaustive.

Table 92 Examples of anticoagulant-related problems noted at day eight in the UC group

Description of problem and resolution Mr. JG found to be taking 2.3mg daily at day eight after discharge. He was confusing his warfarin dose with his INR. His dose was meant to be 2.5mg daily, thus no adverse outcome was reported

Mrs. PE was discharged on 3mg daily with an INR of 2.0. She had an INR of 1.8 two days after discharge and her GP indicated for her to “alternate 3 and 4mg”. Mrs. PE assumes this to mean 3mg in the morning and 4mg at night. Her INR was checked on Day 5 after discharge and an INR of > 8 was obtained, this was complicated by the GP not being able to contact Mrs. PE and she took her evening dose of 4mg and 3mg morning dose on day 6. Her warfarin doses were withheld for two days and she was to resume taking warfarin on day eight at 3mg per day. Her INR was checked by the project pharmacist on day eight and it was 7.6. The GP was contacted and he suggested withholding the dose for another two days, with Mrs. PE to follow-up with her GP before resuming dosing.

Mrs. IP aged 88, was discharged on warfarin 5mg daily after receiving two doses in hospital. She continued on 5mg until day 4 after discharge and her INR was 4.8. Warfarin was withheld for two days and her INR at day 6 was 4.0. Warfarin was withheld for another two days and when visited by the project pharmacist at day eight after discharge her INR was 1.2

Mrs. MW had received 27mg of warfarin over 5 days in hospital with a discharge dose of 2mg daily and INR on discharge of 2.6. Her INR on day 3 was 2.0 and she was told to continue on 2mg daily. Her INR when visited by the project pharmacist at day eight was 1.5; she had received no further testing by her GP.

Mr. SH was discharged on 3mg daily with INR of 2.9, when in fact he should have been on 1mg daily (communication error between doctor, pharmacist and patient whilst in hospital). His INR on day 3 was 5.0 and Warfarin was withheld for one day, and restarted at 1.5mg daily. His INR when visited by the project pharmacist at day eight was 1.9; he had received no further testing by his GP.

Mrs. BR was discharged on 3mg daily with an INR of 3.2. Her INR on day 1 was 3.2 and her dose was reduced to 2mg daily. Her INR on Day 2 was 2.7 and she was told to continue on 2mg daily. Her INR on day 4 was 2.2 and she was told to continue on 2mg daily. Her INR when visited by the project pharmacist at day eight was 1.4; She had received no further testing by her GP.

Mr. AF was discharged on 5mg daily with INR of 2.5. His INR on day 2 was 3.9 and the dose was reduced to 4mg daily. His INR on day 4 was 5.5 and warfarin was withheld for one day and resumed at 3mg daily. His INR when visited by the project pharmacist at day eight was 1.9; he had received no further testing by his GP.

Mrs. NM was discharged on 6mg daily with an INR of 1.2. Her INR on day 6 was 3.1 and warfarin was withheld. Her INR on day 7 was 2.3 and was withheld again. Her INR when visited by the project pharmacist at day eight was 1.5.


of 767

Mr. RP was discharged on 2mg daily with INR of 2.2 and whilst in hospital had received 27mg over 6 days. His INR on day 2 was 1.6 and his dose was increased to 2.5mg daily. His INR when visited by the project pharmacist at day eight was 1.0; he had received no further testing by his GP.

Mr. MN was initiated on 5mg of warfarin on the day of discharge. His INR on day 2 was 1.4 and the dose was increased to 7.5mg daily. His INR when visited by the project pharmacist at day eight was 4.8; he had received no further testing by his GP.

Mr. AT was discharged with an INR 4.0 on 1mg daily and was instructed to follow-up with his GP on day 3. He saw his GP on day 3 and no INR test was obtained. His INR when visited by the project pharmacist at day eight was 1.2 and he was still on 1mg. He had received no further testing by his GP.

Mr. EM was discharged on 1mg daily with INR of 3.8. He was non-compliant with medications and did not present to his GP for INR testing. He indicated to the project pharmacist that he had ceased warfarin on discharge. A pill count indicated he had been taking 1mg daily. His INR on day eight by the project pharmacist was 5.0.

Mrs. MS was discharged on 5mg daily with INR of 1.6. She was non-compliant with medication and ceased warfarin on discharge from hospital. Her INR was obtained by her GP on day 7 and the GP was intending to discuss warfarin with the patient. Her INR when visited by the project pharmacist at day eight was 1.0.

Mr. RP was discharged with an INR of 2.3 on 4mg daily. His INR on day one was 2.3 and his GP increased his dose to 5mg daily. His INR when visited by the project pharmacist at day eight was 5.4. He had received no further testing by his GP.

Mr. RS was discharged with an INR of 1.7 on 5mg daily. His INR on day one was 1.8 and his GP increased his dose to 6mg daily. His INR on day 4 was 3.2 and the dose was reduced to 5mg daily. His INR when visited by the project pharmacist at day eight was 4.8. He had received no further testing by his GP.

Table 93 displays the level of anticoagulation for the PDINR group at each visit after

discharge, with two-thirds of patients in the PDINR group having a therapeutic INR at

day eight after discharge.

Table 93 Anticoagulant control by day of follow-up for the Post-discharge INR monitoring group

Anticoagulant control Day 2 n (%)

Day 4 n (%)

Day 6 n (%)

Day 8 n (%)

Sub-therapeutic 46 (62) 21 (42) 28 (40) 17 (26)

Therapeutic 25 (34) 31 (42) 34 (49) 44 (67)

Supra-therapeutic 3 (4) 12 (16) 8 (11) 5 (7)

The median duration of pharmacist visits in the intervention group at each visit after

discharge are displayed in Table 94. Attempts to contact GPs were made in the majority


of 767

of cases, but some home visits arose on weekends when GPs were non contactable. If

this was reasonably foreseeable, an after hours contact number was obtained or dose

adjustment plans were made in advance.

Table 94 Median time taken for home visits in minutes (range)

Day 2 Day 4 Day 6 Day 8 Patient visit at home 30 (15-80) 25 (10-90) 20 (10-40) 20 (10-60)

Discussion with GP 1 (0-6) 1 (0-5) 1 (0-3) 1 (0-5)

The range is 0 in some cases, as the GP was not contactable.

After the day eight visit by the pharmacist to UC patients, 23% of patients required dose

modification as a result of this visit. Eleven percent had a subsequent dose increase and

12% had a dose decrease.

Only 32 patients (31%) of UC patients were using their warfarin booklet to record their

INR results compared to 100% of PDINR patients (P<0.0001, χ2 = 66.6) when visited at

day eight after discharge. At 90 days after initial discharge 22 (31%) UC patients

compared with 35 (53%) PDINR patients were using their warfarin books to record INR

results (P=0.007, χ2 = 7.17).

Table 95 displays number and rates of adverse events occurring up to 90 days after

discharge for all patients (n=160). There was a significant reduction in total bleeding

complications (14% vs 24%) between the PDINR and UC groups. There was also a

significant reduction in major and minor bleeding complications between the PDINR

and UC groups, (1% vs 11% for major bleeds and 12% vs 30% for minor bleeds

respectively). For patients who were readmitted to hospital during the 90-day follow-up,

there were no significant differences in the median readmission INR (the first

readmission INR) between the UC and PDINR groups, 2.6 (1.3-9.9) and 2.6 (1.6-7.6)

respectively.


of 767

Table 95 Adverse outcomes occurring up to ninety days after discharge

PDINR Usual Care Outcomes

n (%) [95% CI] n (%) [95% CI] P value

Total bleeding 10 (14) [7-24] 28 (34) [24-45] 0.004, χ2=8.5

Major bleeding 1 (1) [0-7] 9 (11) [5-19] 0.02, χ2=5.6

Minor bleeding 9 (12) [6-22] 24 (30) [20-40] 0.01, χ2=6.4

Embolic complication 5 (7) [2-15] 8 (9) [4-18] 0.56

Unplanned readmission 16 (22) [13-33] 22 (26) [17-37] 0.56

Death 4 (6) [2-13] 6 (7) [3-15] 0.68

For fifteen patients in the PDINR arm of the trial that were enrolled in the second phase

of recruitment, sixteen drug-related problems (DRPs) were identified in six patients

with a further nine patients having no DRPs identified. The classifications of the

identified DRPs are displayed in Table 96.

Table 96 Classification of identified drug-related problems in the PDINR group

Type of drug-related problem Number identified

Drug Selection (problems related to the choice of drug prescribed or taken)

3

D1 Duplication of drug

D2 Duplication of therapeutic class

D3 Drug interaction 1

D4 Wrong drug

D5 Incorrect or inappropriate dosage form

D6 Existing drug allergy

D7 Drug given for no clinical reason 1

D8 Therapeutic or other contraindication

D9 Other drug selection problem 1

Over or under dose prescribed (problems related to the prescribed dose or schedule of the drug)

1

O1 Dose too high

O2 Dose too low 1

O0 Other dose problem


of 767

Compliance (problems related to the way the patient takes the medication)

3

C1 Taking too little

C2 Taking too much

C3 Intentional drug misuse

C4 Difficulty using dosage form

C5 Patient taking discontinued medication

C6 Patient taking out of date medication

C7 Unwillingness to take drug

C0 Other compliance problem 3

Untreated Indications (problems related to actual or potential conditions that require management)

6

U1 Drug not given when clinically needed 4

U2 Preventive therapy required 1

U0 Other untreated indication problem 1

Monitoring (problems related to monitoring the efficacy or adverse effects of a drug)

3

M1 Laboratory monitoring

M2 Non-laboratory monitoring 2

M0 Other monitoring problem 1

Non-clinical (problems related to administrative aspects of the prescription)

N1 Dispensed incorrect drug or strength

N2 Patient hoarding medication

N3 Difficulty obtaining further supply from GP or CP

N0 Other administrative problem

Toxicity or Adverse Reaction (problems related to the presence of signs or symptoms which are suspected to be related to an adverse effect of the drug)

T1 Adverse reaction due to drug/disease interaction

T2 Adverse reaction due to dosing problem

T3 Toxicity evident

T0 Other toxicity/adverse effect problem


of 767

12.3.1.1 General practitioner survey Of the 74 GP evaluations sent after the completion of the day eight protocols, 52 were

returned giving a response rate of 70%. The results are shown in Table 97 and

unsolicited comments are displayed in Table 98

Table 97 GP responses to questionnaire

GP responses to the evaluation questionnaire Medians, with range lines plotted at the 10th and 90th percentiles 1. I found this to be a valuable service provided to my patient(s).

Strongly Agree

Strongly Disagree

2. I would feel more confident in initiating or managing newly initiated patients on warfarin if this was a regular service.

Strongly Agree

Strongly Disagree

3. I received adequate feedback from the pharmacist.

Strongly Agree

Strongly Disagree

4. I believe that more patients would benefit from this type of service.

Strongly Agree

Strongly Disagree

5. I found the suggestions made by the pharmacist to be useful.

Strongly Agree

Strongly Disagree

6. I believe that my patient(s) found this to be a worthwhile service.

Strongly Agree

Strongly Disagree


of 767

Table 98 Comments about the monitoring from general practitioners

Comments Didn’t go for long enough

This is a fairly accurate way of measuring INR, hopefully it will be used more in general practice

Period of monitoring should vary according to patient. 4 visits still but a difficult patient according to individual circumstances

Especially new patients starting but also older patients as the warfarin dose can be confusing

Need to compare INR result with that of laboratory

I have no trouble doing INRs without a pharmacist but they were very nice

Accuracy & cost

I feel this service is irrelevant. We have many patients on INRs for many years for varied conditions and feel quite confident in managing them

Reliability of the INR test compared to lab testing, if is reliable why aren’t GPs allowed to use and claim for these tests in their surgeries, POC testing should be funded by the commonwealth under the Medicare scheme

Project Pharmacist’s name I have been away from the practice till today, therefore I can’t really comment

The money involved could provide me with a monitor to more efficiently manage my bourgeoning INR load

Improved safety factor for patient

I am used to monitoring my own but for elderly it is useful. Is there some reason why GPs could not do the same thing

This is like teaching us GPs how to suck eggs

How soon can we start and keep up the good work

Patient & Dr confidence in dealing with warfarin, especially since patient has had bleeding on warfarin previously. Please keep this service going it is excellent

Not sure about accuracy?

Better liaison at start, I doubled up with INRs as I didn’t know this service was happening

Great because in patients home

Sorry Project Pharmacist’s name, but I don't think a pharmacist needs to be involved. The concept of the INR monitor is good though

There would be a place for more flexibility e.g. daily INRs I'm happy with the service. In this particular case it seems inappropriate to have commenced the patient on 3mg this meant that the INR as very slow to rise and the INR was far from therapeutic at day 8

This patient required other blood tests during the week and it would be good if somehow that could be coordinated with the home visit for INR

Pleased to participate, with thanks

It will be very useful for doctor if this service continues


of 767

LW was scheduled to have an operation ~ 1 week after hospital discharge. Why not use Clexane to tide him over that week then start warfarin after the procedure. Note as well that he had vitamin K in hospital so that his initial INRs are lower than expected

Patient was confused when pharmacist output ceased, who was giving instructions? Where was Project Pharmacist’s name? That nice young man!

am confident managing/initiating warfarin therapy and felt additional service was duplication

would be worthwhile if blue warfarin book, patient details, indication, duration etc is also filled out also

The patients find this service very helpful in coming to terms with being on warfarin

Obviously a good service provided by Project Pharmacist’s name. The quality may be dependent on the standard of the operator

I felt that pharmacist involvement post hospital discharge is very useful and not just for follow up of warfarin therapy

Of the 66 satisfaction surveys given to the PDINR patients on completion of the day

eight protocol (8 did not complete the day eight follow-up), 54 surveys were returned

completed giving a response rate of 82%, and the results are shown in Table 99. All of

the patients who indicated that they were quite dissatisfied with the contact

subsequently indicated that the information they had received helped them a great deal.

Unsolicited comments about the anticoagulant monitoring and education are displayed

in Table 100.


of 767

Table 99 Responses to patient satisfaction survey from patients in the PDINR group

Reponses to questions n (%) How satisfied are you with the amount of contact you had with the pharmacist?

Quite dissatisfied 3 (6)

Indifferent or mildly dissatisfied 0

Mostly satisfied 1 (2)

Very satisfied 50 (93)

Has the information and other services provided by the pharmacist helped you to deal more effectively with your new medication warfarin?

Yes, they helped a great deal 53 (98)

Yes, they helped somewhat 0

No, they didn’t really help 0

No, they seemed to make things worse 1 (2)

Did you get the kind of information and other services you wanted from the pharmacist?

No, definitely not 1 (2)

No, not really 0

Yes, generally 7 (13)

Yes, definitely 46 (85)

Is there other information you need, or would like, about warfarin and have not received?

Yes, there definitely is 3 (6)

Yes, I think there is 3 (6)

No, I don’t think there is 32 (60)

No, there definitely is not 15 (28)

Overall, how would you rate the quality of the service that you received from the pharmacist?

Excellent 53 (98)

Good 1 (2)

Fair 0

Poor 0


of 767

Did you find the regular warfarin (INR) monitoring*

Painful 1 (2)

Informative 29 (54)

Motivating 5 (9)

A waste of time 0

Interesting 18 (33)

Annoying 0

Too frequent 0

Beneficial 37 (69)

Do you think this service would be best provided in your home or at your local pharmacy?

Home 48 (91)

Local pharmacy 5 (9)

Do you think this service should be available to all patients commencing warfarin therapy?

Yes 51 (100)

No 0

If this were a regular service would you be prepared to pay for it?

Yes 43 (83)

No 9 (17)

If you answered yes to the previous question, how much would you be prepared to pay per visit?

$1-$5 21 (54)

$6-$10 11 (28)

$11-$15 5 (13)

$16+ 2 (5)

Totals are less than 54 for some questions, due to non-completion.

*Totals more than 54 responses as patients could indicate more than one response.


of 767

Table 100 Unsolicited comments from patients

Comments My family and I appreciated the courtesy of Project Pharmacist’s name at all times.

Very punctual & helpful.

A thoughtful efficient and friendly service had been provided.

A very helpful young man willing to help in any way and I wish him the very best for the future, thank you Project Pharmacist’s name.

Was very pleased with the service provided by Project Pharmacist’s name.

Definitely available to everybody, I couldn’t afford it.

As a daughter of the elderly patient on warfarin I found Project Pharmacist’s names’s help and advice of great benefit to assisting me to help and explain to my mum. Should be an ongoing regular service.

Project Pharmacist’s name is the right person for this type of help as he is most informative with his detail for taking the prescribed dosage. Thank you Project Pharmacist’s name, God bless you in your life.

Mum would like to say she was very satisfied and very grateful for Project Pharmacist’s name’s service and care also for the extra bits he did with her medication as her family we really appreciated his care and communication, thank you.

Friendly, helpful service. Young man very helpful and informative.

The treatment I received from the pharmacist Project Pharmacist’s name was excellent.

Project Pharmacist’s name’s visits were very informative and pleasant.

This appears to me to be a quite essential service for new patients starting warfarin. If this were to be implemented for all there must be a reduction in complications/failures of warfarin therapy. One needs at least 2-3 sessions with the pharmacist before one feels comfortable with understanding warfarin.

Project Pharmacist’s name was the most helpful person we have dealt with, ever.

I think it is very important that the patient is kept informed of what going on. How to fill the warfarin book and all to see if the correct tablets are taken. Four visits is great benefit to the patient.

As my wife does not drive we have appreciated the visits from the pharmacist.

Project Pharmacist’s name very helpful.

Reference to question 10, $5.00 per visit for pensioners.

Project Pharmacist’s name done his job very well and answered all the questions we asked and some we did not think of.

I found the pharmacist, Project Pharmacist’s name very helpful and explained every detail about warfarin clearly.

Visits always very friendly and seemed immediately to put you at ease.

Warfarin is a good tablet but it has its side effects for me. No more than one I can handle as it upsets me tummie.


of 767

I am extremely supportive of the provision of the pharmacy service and commend those who have had the foresight and initiative to implement such a worthwhile and convenient service. It is an uncomplicated informative and cost effective service. Project Pharmacist’s name was very helpful, has a good manner, polite and sits and listens and answers any questions we had.

Project Pharmacist’s name was very helpful, he even bought me a tablet organiser which is helpful as I only have to refill once per week. Project Pharmacist’s name was very helpful at a time when I was a little confused. I am on 10 tablets per day.

Project Pharmacist’s name the pharmacist who attended me when I came home from hospital was wonderful. He explained everything so that we could understand, not all at once but a little each time so we could really understand my medication by the time he finished with us. Thank you again.

I found the first 4 home visits to me were most beneficial and necessary. It was always done in a very friendly atmosphere. To help continue this programme some contribution relevant to the patient's circumstances could be made.

Home visits were always on time - excellent pharmacist was also most friendly - again an excellent interpersonal skill.

Project Pharmacist’s name is a very pleasant young man, seems to know what he is talking about and does his best to work in with the patient.

12.3.1.2 Patient knowledge questionnaire Patient knowledge was assessed 90 days after initial discharge, and the responses are

shown in Table 101

Fifty-six of 81 eligible patients in the UC group and 33 of 69 eligible patients in the

PDINR group returned the knowledge questionnaire – a response rate of 69% and 48%,

respectively (p = 0.01 χ2 = 6.16).

Table 101 Patient knowledge ninety days after discharge

Responses PDINR, n (%)

Usual Care, n (%)

Since starting warfarin would you say that your general health has

Improved 23 (62) 22 (43)

Worsened 2 (5) 5 (10)

Stayed the same

P=0.21, χ2=3.1

12 (33) 24 (47)

Correct reason for using warfarin

Yes 37 (95) 46 (85)

No P=0.14, χ2=2.2

2 (5) 8 (15)


of 767

Do you worry about warfarin treatment?

A lot 3 (8) 6 (11)

A little 13 (33) 21 (39)

Not at all

P=0.67, χ2=0.8

23 (59) 27 (50)

When you left the Royal Hobart Hospital (RHH) were you handed a “warfarin booklet”?

Yes 37 (95) 52 (93)

No 2 (5) 4 (7)

Not Sure

P=0.69, χ2=0.16

0 0

Were you told how warfarin works, and was this clear?

Yes, and clear 31 (84) 43 (80)

Yes, but not clear 5 (14) 8 (15)

No

P=0.78, χ2=0.48

1 (3) 3 (5)

Could you briefly explain in your own words how warfarin works

Thins the blood 23 (59) 34 (60)

Prevents clots 10 (26) 8 (14)

No written response

P=0.21, χ2=4.4

6 (15) 14 (25)

Were you told of the possible problems with warfarin treatment, and was this clear?

Yes, and clear 32 (84) 38 (68)


No

P=0.18, χ2=3.4

1 (3) 5 (9)

Were you told what to do if you have a nosebleed or bruising and was this clear?

Yes, and clear 31 (80) 31 (56)


No

P=0.06, χ2=5.5

4 (10) 10 (18)

Were you told what drugs to avoid and was this clear?

Yes, and clear 27 (77) 28 (50)


No

P=0.06, χ2=5.8

4 (10) 14 (25)

Were you given advice on drinking alcohol and was this clear to you?

Yes, and clear 38 (100) 37 (73)

Yes, but not clear 0 6 (12)

No

P=0.002, χ2=12.4

0 8 (15)

Could starting a new treatment or any other preparation affect your warfarin treatment?

Yes 26 (68) 29 (53)

No 1 (3) 5 (9)

Don’t know

P=0.22, χ2=3.0

11 (29) 21 (38)


of 767

The following are statements about any patient drinking alcohol while receiving warfarin treatment Alcohol can affect anticoagulant treatment

Yes 28 (93) 35 (78)

No P=0.07, χ2=2.9

2 (7) 10 (22)

Alcohol must be totally avoided

Yes 7 (25) 14 (30)

No P=0.61, χ2=0.25

21 (75) 32 (70)

Eight units of alcohol a night is OK (for example eight glasses of beer or wine)

Yes 2 (7) 3 (7)

No P=0.90, χ2=0.14

25 (93) 42 (93)

One unit of alcohol a night is OK (for example 1 glass of beer or wine)

Yes 22 (76) 38 (83)

No P=0.48, χ2=0.51

7 (24) 8 (17)


of 767

Table 102 Statistics comparing PDINR and UC groups

Responses PDINR, n (%)

Usual Care, n (%)

Of the list below indicate which of the following could interfere with your warfarin therapy?†††

Aspirin P=0.70, χ2=0.15 30 (79) 46 (82)

Weather conditions P=0.80, χ2=0.07 1 (3) 2 (4)

Coffee P=0.32, χ2=0.96 4 (11) 10 (18)

Herbal Remedies P=0.45, χ2=0.58 20 (53) 25 (45)

Panadol P=0.13, χ2=2.2 11 (30) 9 (16)

Some illnesses P=0.30, χ2=1.1 19 (50) 22 (39)

Missed dose of warfarin P=0.19, χ2=1.7 28 (74) 34 (61)

Nurofen P=0.48, χ2=0.50 9 (24) 17 (30)

Antacids P=0.14, χ2=1.9 7 (18) 18 (32)

Some foods P=0.40, χ2=0.71 25 (66) 32 (57)

Of the list below, which of the following could be side effects of taking the wrong (too little or too much) warfarin?‡‡‡

Blood in stools P=0.69, χ2=0.15 22 (60) 31 (55)

Nausea P=0.47, χ2=0.51 4 (11) 9 (16)

Blood in the urine P=0.70, χ2=0.14 19 (51) 31 (55)

Nervousness P=0.14, χ2=2.2 3 (8) 1 (2)

Blood clots P=0.81, χ2=0.06 22 (58) 31 (55)

High blood pressure P=0.54, χ2=0.38 8 (21) 9 (16)

Weakness P=0.27, χ2=1.2 8 (21) 7 (13)

Ringing in the ears P=0.56, χ2=0.33 4 (11) 4 (7)

Nose bleeds P=0.41, χ2=0.67 30 (79) 40 (71)

Sleeplessness P=0.99, χ2=0.003 2 (5) 3 (5)

Prolonged bleeding after cuts P=0.49, χ2=0.47 32 (84) 44 (79)

Bruising without injury P=0.24, χ2=1.4 20 (79) 38 (68)

Loss of appetite P=0.97, χ2=0.001 6 (16) 9 (16)

Responses do not total the number of patients returning surveys in all cases as some

patients did not answer all questions.

††† Proportion of patients responding “YES” to each question ‡‡‡ Proportion of patients responding “YES” to each question


of 767

12.3.2 Cost-effectiveness The money saved in healthcare related costs by instituting this intervention across the

country would amount to over $A9 million dollars per annum in reduced bleeding costs

as displayed in Table 103. From data obtained in this study, it is estimated that

approximately 20,000 patients are initiated on warfarin in public hospitals per annum in

Australia.

Table 103 Cost savings associated with home monitoring compared with usual care

Cost savings of post-discharge INR monitoring compared with usual care

No. Patients initiated on warfarin per annum 20000

Major bleed in the first three months of warfarin treatment (n) UC-11% 2200

Major bleed in the first three months of warfarin treatment (n) PDINR-1% 200

Major bleeds "saved" from instituting PDINR (first three months of treatment) 2000

Costs saved from major bleeds per annum (000s) ($)§§§ $9,002

The approximate time spent with the patient over the course of the intervention neared

two hours in total. The approximate travelling time for this type of intervention is likely

to be similar, therefore, the total time costs would approximate 4 hours, likely to be paid

at $40 per hour (total time costs of $160). Additional costs such as test strips, lancets,

telephone calls and travel costs would likely bring costs in the order of approximately

$200 per patient. The program would cost approximately $A4 million for 20,000

patients per annum. It is therefore likely that this program is highly cost-effective,

saving over $A6.7 million if the costs of bleeding are solely compared with the costs of

the program. Performing a sensitivity analysis, if the difference in major bleeding risk

was reduced to 8% from 11% in the UC group, the program would save $A6.3 million.

§§§ Average bleed cost is $5366


of 767

12.4 Discussion This service was initially analysed in a separate published study; the study was designed

to be extended to involve 250 patients in total which would enable sufficient statistical

power to examine thromboembolic outcomes and perform robust analyses. However, at

163 patients, it was apparent that there was an overwhelming benefit of the study

intervention. Although, the sample size was initially going to be larger, the authors did

not feel that it was ethically appropriate to continue to randomise patients, when the

benefit of the study was strikingly clear. This study examined the effect of POC

monitoring of anticoagulant therapy, with patient-focused education by a pharmacist,

and medication review amongst a population of newly initiated warfarin patients who

were discharged from hospital to GP care. This study has clearly shown that control of

anticoagulation after discharge from hospital is sub-optimal, with some patients at high-

risk of anticoagulant-related complications because of lack of appropriate testing and

dosing, education, and review. We found that this comprehensive program reduced the

frequency of bleeding in patients randomly assigned to the PDINR group at the start of

anticoagulant therapy. A multifaceted program delivered by a pharmacist, that

comprises POC INR monitoring, medication education (particularly focussing on

anticoagulant education) and medication review significantly improves the outcomes

associated with warfarin therapy. This type of program is an enhanced use of available

resources and applies pharmacists’ skills in drug and disease state management.

Essentially, pharmacists could perform INR monitoring within a framework that

ensured patients were appropriately referred and monitored after discharge from

hospital.

The majority of the study population represented those chronically ill older patients,

who were initiated on warfarin for the treatment of chronic conditions. The two groups

were well matched in baseline characteristics, except for a significantly higher incidence

of previous AMI in the PDINR group. We are unaware of any lifestyle differences

between the two groups that could have been responsible for this difference. However,

if a number of baseline demographics are reported the risk of random chance

contributing to a significant difference in these variables increases.

The study population represents those patients who are at high risk of anticoagulant

misadventure if not adequately monitored and educated. Elderly patients are at an


of 767

increased risk of bleeding complications but also have a higher benefit with

anticoagulant therapy compared to younger populations.121, 126 In fact nearly half of the

patients had warfarin initiated for stroke prevention for AF.

There is clear evidence that the trend to early discharge for hospital patients is putting

strain on the primary care system. This study, which commenced in 2002, has shown

that 41% of all patients had a therapeutic INR at discharge, compared to 63% in a

previous audit of anticoagulant therapy at the same hospital, completed in 1994.154 The

median days of initiation of warfarin in the previous study was eight days compared to

six in this study.

This study shows the actual practice of initiation of warfarin in a teaching hospital and

subsequent discharge of patients to community care may result in poor outcomes,

compared with follow-up after discharge. The combination of shorter periods of

hospitalisation and increasing usage of warfarin is placing stress on general practice

health services to care for newly anticoagulated patients.

This project meets a number of the principles of the APAC national guidelines to

achieve the continuum of quality use of medicines between hospital and community.60

It appears that standard care for discharge of newly initiated anticoagulated patients is

not fulfilling key roles of the national guidelines.

“It is the responsibility of the admitting institution to ensure the development and coordination of a medication discharge plan for each patient”

“Information to the patient’s health care providers should include details of the medication management during the hospital stay……………. And any specific needs with respect to drug management”

The PDINR group had two-thirds of patients in the therapeutic range at day 8, which

had increased from 39% at discharge. Conversely, the proportion of patients in the UC

group with therapeutic INR at day eight was less than the proportion therapeutic at

discharge, (38% and 48% respectively). The alarming part of the follow-up at day eight

in the UC group was the high proportion of supra-therapeutic INR. More than one-

quarter of patients in the UC group had a supra-therapeutic INR, and the risk of

bleeding in this group of patients is elevated with increasing INR. It is evident that

follow-up conducted in the PDINR group increased the proportion of therapeutic

patients at day eight compared to discharge and also reduced the supra-therapeutic INR

that occurred in the UC group.


of 767

A similar proportion of patients in both groups were sub-therapeutic at discharge and at

day 8, suggesting that the follow-up conducted by the pharmacist is most beneficial at

reducing preventable elevations in the INR. It also appears that supra-therapeutic INR at

day eight occur most frequently in patients with sub-therapeutic INR at discharge,

reinforcing the benefit of regular monitoring of patients who are not therapeutic or

stable.

A number of anecdotal cases point to similar root causes of unstable INR after

discharge. These cases indicate a lack of sufficient communication between the hospital

and GPs, and GPs extending the interval of monitoring too early, and/or increasing

dosages too quickly. The study clearly implicates the trend to early discharge as a

potential cause for poor communication and outcomes in the usual care group within the

study. It is well documented in the literature that many patients have a poorly planned

discharge and the GP is not fully informed of their patient’s admission or discharge.155-

158

This project is an example of a systems solution to improve the management of

anticoagulation, which looks to improve the processes of care rather than individual

behaviour. When processes of care are examined, a common root cause of medication

errors occurs at the time when decisions about therapy are made.159, 160 Failure to obtain

sufficient information about the patient or about the pharmaceutical agent has

contributed to medication errors.161 This intervention aimed to provide patients and GPs

with information regarding dose decisions at the POC, and has shown an improvement

compared to usual care.

A number of systems could be implemented to adhere to the APAC guidelines for

continuity of care between hospital and the community for warfarin initiation. These

systems include this type of program, improving transfer of information regarding

warfarin doses and INR whilst in hospital to the GP, and the utilisation of existing

services such as Home Medicines Review (HMR) to improve patient knowledge of

anticoagulant therapy. Pharmacists should take a key role in reinforcing knowledge

regarding anticoagulation to reduce the risk of complications of anticoagulant therapy.

The impact that the pharmacist had on the UC group at day eight is difficult to measure,

with almost one quarter of the UC group having a dose change after the visit on day

eight. It is likely that the visit by the project pharmacist on day eight may have

prevented some type of adverse event in some of the patients visited and, although this


of 767

is speculative, it is highly likely and difficult to quantify. Data for anticoagulant control

between day eight and day 90 after discharge was unable to be obtained for patients.

Therefore, the contribution of the PDINR to anticoagulant control between these time

intervals was unable to be assessed. However, a higher proportion of patients in the UC

group with a supra-therapeutic INR at day eight subsequently developed embolic

complications assessed at 90 days after discharge. This may have reflected poor control

at day eight as a marker for erratic control up to 90 days after discharge, potentially

contributing to an increased risk of embolic complications.

This study showed a reduction in all bleeding complications in the PDINR group

compared to UC. It is likely a combination of improved monitoring post-discharge and

education resulted in better outcomes compared to usual care. Total bleeding was

reduced in the PDINR group compared to UC, 14% and 34%, respectively. Importantly,

major bleeding was reduced in the PDINR group compared to UC, 1% compared to

11% respectively. This has significant impact on costs associated with warfarin, and

also doctors’ perception of anticoagulant therapy.

Despite low rates of warfarin-related bleeding reported in randomised trials,162-166 which

included strict exclusion criteria, higher rates of major bleeding have been reported in

many studies of warfarin used in clinical practice.124, 151, 167 In this study, the sample

of patients was assembled with few exclusion criteria, and the bleeding rates in the UC

group were similar to those observed in other clinical practice studies. The PDINR

group achieved a frequency of major bleeding that was similar to the lower rates

achieved in previous randomised trials assessing the efficacy of warfarin.

Hylek et al168 studied a group of patients in an anticoagulation clinic who obtained an

INR greater than 6.0. Nearly 10% of these patients sought medical attention for

abnormal bleeding and 5% of these experienced a major haemorrhage within 14 days of

follow-up. Interestingly, they noted that none of the patients had clinically important

bleeding at the time of the INR measurement. This raises the issue that fluctuations in

the INR need to be kept to a minimum to reduce the potential for life-threatening

bleeding complications.

A study by Beyth et al144 evaluated an intervention that consisted of patient education

about warfarin, training to increase patient participation, self-monitoring of INR, and

guideline-based management of warfarin dosing compared to normal care for six

months. The cumulative incidence of major bleeding in the usual care group was similar


of 767

to our study, 12% and the incidence of major bleeding was reduced to 5.6% in the

intervention group. Importantly, they noted that after 6 months, major bleeding occurred

with similar frequencies in the intervention and usual care groups.

GP evaluations were generally positive, and they could see the benefit of providing this

type of service for patients newly initiated on warfarin. They saw this program as a

valuable service to patients; they also thought that more patients could benefit, and that

their patients individually benefited from the program. From the evaluation

questionnaire it appears that this service has mixed effects in regards to feeling more

confident in initiating warfarin, and perhaps this reflects the difficulty of initiation of

warfarin, irrespective of the type of follow-up.

The pharmacist home visits were well received by patients, and they were very positive

in their feedback and evaluation of the service. Most patients found it informative,

beneficial and interesting. Over 80% of patients indicated that they would be willing to

pay, with most indicating in the range of $1-$5 or $6-10 per visit. It must be

acknowledged that most of the group was elderly and receiving social security benefits,

which makes the willingness to pay more significant.

Nowadays, it is not enough to show that programs reduce hospital admissions or

composite end-points; we must show that programs are cost-effective or at least cost-

neutral. This program was estimated to save over $A9 million dollars per annum in

direct hospital costs if rolled out across the country for patients commenced on warfarin

in the hospital setting. The program costs associated with this are likely to add up to

$A4million dollars, with a likely cost saving of over $A5 million dollars. This type of

program needs to be integrated into existing structures to lower the program costs

associated with it, thus making it more cost effective. A model with accredited

pharmacists conducting testing, education and medication review would have ongoing

sustainability after initial training. Previous studies illustrate the potential health and

economic benefits of organised care management approaches and POC monitors in the

management of patients receiving warfarin therapy.169 Although the generalisability of

this program remain to be demonstrated, these findings support the premise that efforts

to reduce the likelihood of major bleeding will lead to safe and effective use of warfarin

therapy in older patients.


of 767

“Many major improvements in medication use among older adults will also depend on closing the gap between knowledge and practice and increasing the ability of older adults to manage their medications.” 119

In terms of monitoring, some of the best evidence for improvements in primary care

relate to the monitoring of warfarin.170-174 The evidence suggests that nurse led

monitoring clinics,173 computerised decision support systemsm,172, 173 patient education

and involvement,171 may help improve control through improved monitoring. In terms

of patient adherence, a number of studies have shown that improved education175 and

approaches that provide greater involvement of patients in decision making175, 176

improve patient adherence and may reduce drug related admissions.

“Interventions focused on improving patient adherence with prescribed regimens and monitoring of prescribed medications also may be beneficial.”143

It is difficult to attribute the reduced bleeding complications in the PDINR group, to any

single part of the intervention. It is important to view the success of the program as a

sum of its components. The study was not designed to determine the relative importance

of the components of the intervention. It is likely that the combination of improved

monitoring post-discharge, medication review and patient-focused education program

has resulted in a reduction in the number of bleeding complications. It has been reported

by a number of studies that adherence to anticoagulant treatment is enhanced by

knowledge and understanding of the drug, its benefits and side effects.177-179 In a large

observational study investigating reasons for drug related admissions, they were mainly

attributed to problems with prescribing, monitoring and patient adherence.180 The

APAC guidelines again refer to follow-up education of patients after discharge

“Factors influencing the patients’ knowledge about the medication and the ability to comply with medication regimens should be identified……………….Implementation of drug therapy should be accompanied by the use of appropriate education programs”60

The patient-focused education program employed in this project significantly improved

knowledge in some areas, and there were clear trends to improved knowledge in other

areas. It is possible that components of increased knowledge were responsible for

reduced bleeding complications. There may be some non-response sample bias

associated with the knowledge assessment, as significantly more patients in the UC

group returned their knowledge questionnaire. However, over 60% of patients in the


of 767

PDINR group returned the survey, therefore this may be unlikely to have caused a

significant difference.

Roddie and Pollock181 showed that 85% of patients with a good understanding of

warfarin had a well controlled and stable INR, compared to only 63% in the poor-

understanding group. Generally, patient’s knowledge, drug compliance and

anticoagulant control all improve after patient education became part of a structured

management program.144, 145, 181, 182 The Newcastle Anticoagulation Study Group150

found no relationship between knowledge and INR level, but found a positive

relationship between education level and knowledge. Importantly, they noted

“knowledge was generally poor” and 24% of patients answered less than half of the

questions correctly.

The knowledge evaluation is limited by its reliance on a questionnaire-based survey. A

reason for lower knowledge in the UC group could be due to poor counselling and

information giving by health professionals. With regards to written material, Estrada et

al found that some of the patient information on anticoagulant therapy was above the

comprehension level of most patients.183 More emphasis should be given to adequate

education of patients on anticoagulant treatment post-discharge, with special emphasis

on high-risk groups of patients, such as the elderly and poly-pharmacy users. Warfarin

education should be tailored to the level of education and age of the patient.146, 150

Education of elderly and illiterate patients may require special consideration and include

the use of visual aids.146

There are some limitations to this randomised controlled trial. The trial was non-blinded

to allocation concealment and outcome assessment. Allocation concealment would have

been difficult to conduct in this type of project. The use of objective criteria for

assessment of outcomes at eight and 90 days, and strict criteria for major bleeding

assessment, minimises the risk of blinding status exerting a major influence on trial

findings.

The true effect of the intervention may have been understated because patients in the

UC group could have been more likely to take an active interest in their anticoagulant

treatment due to the Hawthorne effect, which may have underestimated the effect of the

intervention. The potential for underestimation of effect is very likely because of the

impact the pharmacist had on the UC group when they were seen at day eight after

discharge. Almost one-quarter of patients in the UC group required dose changes as a


of 767

result of the visit at day eight. It is therefore very likely that the intervention also had a

positive impact on the UC group and may have prevented subsequent bleeds that may

have occurred if patients’ GPs were not contacted at day eight after discharge and dose

changes instituted.

12.5 Recommendations 1. All patients who are initiated on warfarin in the hospital setting should receive a

HMR after discharge, with a focus on reinforcing key education points regarding

anticoagulant therapy. All patients who are anticoagulated with warfarin should

receive an annual HMR.

2. Negotiations should be undertaken by the Pharmacy Guild of Australia to

implement a program delivered by pharmacists that incorporates POC INR

monitoring, patient-focused anticoagulant education and medication review for

all patients commenced on warfarin.




4. Appropriate systems should be developed for the transfer of information to the

general practitioner regarding the initiation of anticoagulation in hospital. This

information should also be provided to the patients’ community pharmacy to

ensure appropriate information is reinforced by community pharmacists.

Overall Discussion

of 767

13 Overall Discussion 13.1 Key findings A major contributor to inadvertent polypharmacy and drug-related problems in the

elderly appears to be hospitalisation and the consequent changes in medication (drugs or

brands of drugs) at the transition from outpatient to inpatient care and back.184

Forster et al132 recently demonstrated in their study of 400 consecutive general medical

patients (mean age 57 years) that 19% experienced adverse events after hospital

discharge and that adverse drug events were the most common (66%). Furthermore,

they concluded that many of these events could have been prevented or ameliorated

with simple strategies.

To provide a patient and cross-national perspective on health care, especially during

transitions, the 2005 Commonwealth Fund International Health Policy Survey

interviewed adults in six countries who had recently been hospitalised, had surgery, or

reported health problems. The eighth in a series of cross-national surveys, the 2005

survey for the first time included Germany in addition to Australia, Canada, New

Zealand, the United Kingdom, and the United States. Conducted in countries with

distinct insurance and care delivery arrangements, the study examined country systems’

performance, with a focus on safety, coordination, access, and chronic disease

management.185

The final study included 700–750 adults in Australia, Canada, and New Zealand and

1,500 or more in the United Kingdom, United States, and Germany. In each country,

sizable shares of adults who had recently been hospitalised indicated deficiencies in

care. Failures to coordinate care during discharge emerged in all countries. At least one-

third of patients in each country said that they did not receive instructions about

symptoms to watch for, did not know whom to contact with questions, or left without

arrangements for follow-up care. Half of German patients said that there had been no

follow-up arrangements.

Concerns about transitional care extended to medications. Patients in all countries were

often given a new medication when discharged, with U.S. patients the most likely to

report new medications. Yet in all countries but Germany, at least one in four patients

Overall Discussion

of 767

said that nobody had reviewed the medications they were taking before their

hospitalisation.

Despite the evidence of often complex medication regimens, sizable majorities of

chronically ill patients in all countries said that their physicians had not always

reviewed all of their medications during the past year, and one-third or more reported

infrequent reviews (sometimes, rarely, or never). Patients’ reports also indicate sizable

gaps in physicians’ explanations about side effects. Among those taking multiple

medications, this lack of review raises the risk of adverse drug interactions as well as

potentially undermining the effectiveness of care.

Lack of review and attention to medications emerged in all countries. With chronically

ill patients frequently taking multiple medications, failure to review medications,

including when discharged, puts both patients at increased risk of adverse drug reactions

and signals coordination concerns.

“Investment in information technology, including integrated electronic medical records, also offers potential new tools to support patients and physicians with expanded infrastructure capacity to manage care. As these initiatives evolve, there are opportunities to evaluate and learn from failures as well as successes.” 185

Med eSupport examined medication management amongst older, chronically ill patients

entering and recently discharged from hospital, and found clear evidence of problems

relating to sub-optimal use of prescribed medications, particularly with regard to

discrepancies in medication histories at the community-hospital interface. The Med

eSupport program, which was directed at improving medication history reconciliation,

patient education, provider communication and the provision of a post-discharge

medication review, has resulted in a significant reduction in DRPs in high-risk medical

patients.


first 48 hours of admission for the intervention group than for the control. There

appeared to be a weak relationship between increased length of stay and the number of

discrepancies not resolved at 48 hours. Significantly more discharge prescription

discrepancies were also resolved prior to departure from hospital for intervention

patients than for control patients.

Overall Discussion

of 767

The full intervention group displayed a significant improvement in their compliance and

drug knowledge over the 30-day post-discharge period, along with a significant

decrease in the total number of major and moderate DRPs per patient. Patient-identified

DRPs reported by the full intervention group were also significantly fewer than the

other groups over the period from admission to 30 days post-discharge. Although the

numbers were small, it was interesting to note that 3% of control group patients vs.

0.6% of intervention group patients reported being readmitted to hospital within 5 days

of initial discharge (“rebound readmissions”). Also, 44% of the patients who had such a

rebound admission had left hospital with apparent medication discrepancies at

discharge, compared with an overall figure of 24% of all the study patients.

Patients who had a medication review were more likely to feel confident about their

medications after discharge. The Med eSupport program was also welcomed by the

patients’ general practitioners and community pharmacists. This study shows that

relatively simple investigation and interventions by a pharmacist working closely and

liaising with patients and caregivers can improve the efficacy and safety of drug use in

the elderly.

Our economic analysis, which is based on conservative assumptions, indicates that Med

eSupport can save the health sector between $54M (additional) and $69M (total) in

financial savings annually on a national level (50 sites initially), assuming full

implementation of the key PDMR/HMR recommendations. With the current rate of

pharmacists’ recommendation uptake being only partial, the sector can save $25M

(additional) and $34M (total) annually at a national level. Even assuming partial

compliance and excluding the expected financial savings, the program represents value

for money at an additional $10.84 per day of health loss prevented.

It should be emphasised that our economic analyses were intentionally conservative.

For instance, we only asked the clinical panel to examine one major medication-related

issue per patient, when most patients had more than one medication discrepancy in

hospital or DRP at the post-discharge medication review. We have also assumed that

10% of older, medical patients currently receive a home medicines review after

discharge from hospital, and this has diluted the effect of implementing an automated

review scheme to some extent. The true figure is likely to be closer to that observed in

our control group (1%).

Overall Discussion

of 767

13.2 Project complications The Med eSupport Project Team, after being awarded the QUM tender from the

Pharmacy Guild of Australia, commenced the planning process in March 2004.

Although the project started to take shape early on, a number of obstructing events

occurred in the early months that delayed the planned commencement of patient

recruitment from September 2004 to December 2004. In the early months, along with

the chief investigator, only two full-time team members were involved. After many

delays and unavoidable changes in direction for the ICT aspect of the project (discussed

in more detail in the ICT section of the discussion (see section 13.2.1)) one of these

team members chose to leave the project during September 2004. At this point, the

majority of the ICT planning had been finalised, but not completed, which led to the

decision to delay commencement of patient recruitment. The three interstate sites had

been visited to collect information on the hospital protocols and local settings and

establish ICT requirements, as the Team’s goal was to implement a full ICT solution at

all sites and not only in Tasmania, as required in the project deed.

Two new staff members were recruited to the team in November 2004. During this time

the trial design was finalised and the project had been promoted to GPs, community

pharmacists and hospitals across Tasmania. During November 2004, a trial officer was

employed to enrol at both of the Western Australia sites, and a 179 page trial protocol

manual (Appendix XIV) was written to facilitate uniform data collection processes were

followed across the five sites. Despite advertisement campaigns, the Project Team were

not successful in finding a trial officer for the LGH site by this stage. In Bendigo at this

point, local Project Team members were working on achieving local ethics approval for

the project. An extension of the project was sought until September 2005 in order to

compensate for these delays.

Finally, on December 2nd 2004, the first patient was enrolled into the trial at the RHH

site. Data collection commenced in Perth on December 12th 2004. Two team members

flew to Perth beforehand to provide training to the local trial officer and ensure the ICT

requirements could be met (further discussed in the ICT section of this chapter). As time

went on, and a trial officer had not been found for LGH and the ethics approval process

had still not been completed in Bendigo, the primary Project Team in Hobart employed

more local staff in an attempt to maximise the enrolment rates at RHH. A second trial

officer was employed, on a part time basis, to assist in Perth.

Overall Discussion

of 767

By April 2005, it was decided to apply for a final 2-month extension. At this point it

was apparent that Bendigo would finally achieve ethics approval and so the Project

Team was keen to continue and collect maximum number of patients from as many sites

as possible. At the time of applying for this extension, it was believed that LGH would

not be a viable site due to attempts to date to find a local trial officer being unsuccessful.

However, shortly after the submission of the interim report, the chief investigator was

approached by a Pharmacologist from Melbourne requesting short-term work in

Launceston. As the hospital was essentially part of the same ICT system as the RHH,

set up of the project at that site was relatively simple and data collection commenced at

the start of June 2005. At the start of July, a Project Team member from Tasmania flew

to Bendigo to facilitate the roll out of the trial and train the local trial officer.

Unfortunately however, this project officer only recruited 10 patients in the two month

period, which was very disappointing. The proposed Nursing Home Patients arm of the

trial, on further investigation, proved difficult to design and implement. In the end, it

was implemented for a short period at the RHH site only. Data collection was wrapped

up at all sites on 31st of August 2005, meaning RHH, SCGH and HPH collected data

for approximately 9 months, LGH for 3 months and BHCGH for 2 months.

13.2.1 ICT issues As detailed in the ICT section, the development of the ICT components of Med

eSupport was an arduous, time-consuming and expensive process that significantly

prolonged the project and ultimately hindered the recruitment of larger numbers of

subjects. Delays caused by one vendor had a knock-on effect to the work of another,

causing further delays to the project. Working with several software vendors on the

project was unavoidable and this emphasises the need for vendors to complete their

work to the timetable initially agreed.

Many impediments surfaced - for instance when installing the system into the trial site

hospitals. It is our belief that many of these impediments should have been foreseen by

our principal ICT subcontractor. It is unfortunate that considerable time and effort was

spent in attempting to resolve disputes with the subcontractor, even though a contract

with clear timelines and deliverables existed. In future projects it is recommended the

software vendor’s ability to perform the work required in the time specified should be

contractually assured with the inclusion of pecuniary damages.

Overall Discussion

of 767

The Project Team, having now worked with a number of ICT companies across several

large projects, would recommend that the Guild be very selective in commissioning ICT

companies to work on health informatics projects. As previously mentioned, software

developers contracted should be able to demonstrate a history of successfully

completing projects on time and to budget, and to the funder’s satisfaction.

The Project Team exceeded the ICT requirements for the project in that the contract

required rolling out and testing the “full” ICT solution in Tasmania, and sought the best

possible solution at each of the other sites. Instead, the Project Team developed a

system that was more scalable to a national roll-out through its greater use of the web-

based functionality and, importantly offered the same model of ICT to all sites, thereby

providing a greater level of functionality than was required in the deed of grant.

The team successfully developed and implemented a model where an electronic

communication pathway was developed to facilitate secure transfer of a patient’s

medication history from community pharmacy to hospital. The communication model

was bi-directional, so that current medication information would be available to the

patient’s primary health care providers at discharge from hospital. This model works

with the Phoenix Rex® and PCA-Nu systems Winifred® dispensing systems with the

potential for other dispensing systems to be integrated at a later date. The Med eSupport

program was purposely designed to not interrupt the work flow of the community

pharmacists and in general, requests for information took pharmacists around five

minutes to complete.

Unfortunately, because of delays with the primary ICT subcontractor, the automatic

uploading for both types of dispensing systems only became fully functional during July

2005, corresponding to approximately six weeks of patient recruitment. Throughout the

trial, primary health providers were given access to patients’ discharge information via

both the web and via fax/mail, and patients were given access to their information on a

secure interactive web site.

13.2.2 Patient Exclusions Patients were eligible for enrolment if they were 50 years or older, had at least 2 chronic

medical conditions, where at least one was diabetes mellitus, cardiovascular disease or

chronic obstructive airways disease and were taking at least 3 chronic medications. For

trial process purposes, they also needed to be able to nominate a regular GP and

Overall Discussion

of 767

community pharmacist. It was decided, from a practical point of view that up to three

of each could be named, but it was felt that more than that would cause difficulties in

ensuring the intervention process ran smoothly and efficiently. Only those patients

admitted to hospital within the last 24 hours were considered for enrolment, as part of

the design of the admission intervention was to ensure that it was completed within the

first 24 hours of admission to achieve timely intervention and reconciliation of

medication histories.

It was initially anticipated that a total of 750 patients would be enrolled into the trial,

with ideally 150 from each of the five sites. It was unclear in the early stages of the

project, whether LGH or BHCGH would be able to recruit trial officer. To compensate

for this potential shortfall in patient recruitment, the recruitment rate at RHH and the

two Perth sites was increased. In total, 4176 medical patients across all sites were

screened for eligibility; 3361 were ineligible for inclusion. The most common reasons

identified for exclusion were;

• identified greater then 24 hours since admission

• too unwell to provide informed consent

• not meeting all selection criteria.

At the time of trial design, it was felt that early intervention regarding admission drug

chart discrepancies was essential to assist in ensuring patients did not experience serious

medication misadventure. However, this particular selection criterion did limit the

enrolment process both from a practical and sensitivity point of view. The trial officer

who might have identified a number of eligible patients upon first arriving at the

hospital, might find after having enrolled two or three patients, others on the list had

moved into the greater than 24 hours time frame. The other problem commonly

encountered was that the first 24 hours of an admission to hospital for an individual is

more often than not a traumatic time. The individual will often feel exceptionally unwell

and/or distracted, and this often leads them to not feeling open to new ideas or

answering questions they may not feel are imminently critical to their well-being. In

previous studies performed by the local project team, where patients could be enrolled

at any time through the hospital stay, it was found that enrolling them closer to their

time of discharge meant that they were feeling a lot better and ready to think about

Overall Discussion

of 767

things beyond their immediate condition. Therefore, it was felt that this aspect of the

enrolment process was very limiting, but essential in meeting the project’s objectives.

The selection criteria used, to ensure patients were considered ‘high-risk’ for medication

misadventure, did inevitably result in a large number of patients who may have

benefited from the trial or aspects of the trial not being eligible due to not meeting one

or more aspects of the criteria. It was essential to adhere to these criteria to ensure a

similar sample was being collected for the purposes of data analysis, but in the event of

a wider rollout of a program such as Med eSupport, patient groups outside those

specifically targeted in this trial may potentially benefit.

These exclusions left 815 patients identified as eligible for enrolment. Of these, 253

refused to enrol. Trial officers reported that common reasons for refusal related to a

belief that they would not benefit from the program, a perception that they did not have

time to participate and concerns with the privacy of their information. However, the

most common reason was the feeling at the time that they were too unwell and often

patients would request trial staff to come back later in the stay (and then placing the 24

hours time frame requirement at risk) . As discussed previously, refusals of this nature

further highlighted the limiting nature of the essential time frame restriction aspect of

the enrolment process.

13.2.3 Withdrawals In total, 562 patients were successfully recruited into the trial. However, withdrawals

resulted in 184 patients being removed at different points in the trial. Of the 75 who

were completely removed from the trial, the most common reason related to incomplete

intervention being performed at admission which rendered the patients unable to remain

in the trial for any part of the data analysis. This occurred for a variety of reasons

across the sites, mostly relating to a breakdown in the communication chain at some

point. However, of the 32 patients removed for this reason, 25 patients were from the

Perth sites.

Despite training involving two site visits, one initial and one follow-up, and the use of a

comprehensive trial manual and constant telephone communication, when it came to

compiling the data in Hobart at the end of the trial, it was found that a large number of

the intervention patients enrolled from the Perth sites did not actually receive the full

intervention at admission. Most commonly, reconciled medication lists were found to

Overall Discussion

of 767

be incomplete and hence, not all discrepancies had been reported to the RMO. This

meant that the data collection on resolution of discrepancies was not accurate. It was

felt that this data could not be combined with the full data set.

Another common reason for complete withdrawal was discharge prior to completion of

enrolment. This is something that was not possible to control as, although patients who

were clearly about to be sent home were excluded, sometimes quick decisions were

made by hospital staff and patients would be discharged at short notice. At the point of

discharge from hospital, 49 patients were removed from the trial. The most common

reason for this was incomplete intervention at discharge. This was experienced across

the sites and most commonly occurred because patients were discharged over a

weekend or late at night, when no trial or hospital staff were available to complete the

necessary discharge counselling or other aspects of the discharge intervention process.

Problems relating to the Perth data were also experienced at this point. Fifteen patients

from the Perth sites were removed due to incomplete interventions similar to those

found at admission. It became apparent, despite initial assurances to the contrary, that

the hospital process at both the Perth sites relating to discharge did not integrally

involve pharmacists, which appeared to result in discharge summaries with many

medication related errors and discharges occurring at short notice without the

knowledge of the pharmacy staff. These issues did not affect control patients as no

intervention was being made; however, data was sought from medical records for a

number of the patients to ensure there was accurate information available for analysis of

the resolution of medication list discrepancies.

The other common reasons for patients being withdrawn at the point of discharge were

death in hospital and transfer to another facility, such as a nursing home or a low grade

care facility. It was felt that considering the ‘high-risk’ nature of the patients being

enrolled; losses of this nature were inevitable.

Finally, 59 patients had to be removed from the trial at some point during the 30 days

post-discharge. The most common reason for withdrawal in this group of patients was

an inability to contact them for their 30-day phone call. When patients were enrolled,

every effort was made to secure follow-up details, including where appropriate, details

of carers and/or relatives and confirming preferred times to be contacted. However, 17

patients across the sites were not able to be contacted. Also, some patients were found

to be in another health care facility, such as long-term respite, at the time of the call.

Overall Discussion

of 767

Finally, readmission within five days of discharge occurred occasionally. It was felt

that for those patients in the intervention – PDMR model group, being readmitted

within 5 days of discharge invariably meant they had not received the full intervention

before they were readmitted, as the PDMR was ideally scheduled for 5-7 days post-

discharge. This rule was then applied to all other groups to maintain consistency.

Readmissions within 5 days, or ‘bounce backs,’ have been measured as a separate

indicative parameter in parts of the analysis. As mentioned, 89% of the patients with a

rebound readmission were from the control group.

In summary, 487 patients remained enrolled in the trial and available for initial data

analysis, 427 for analysis at the point of discharge and 378 by the end point.

13.2.4 Removal of data collected outside of trial protocol specifications.

As previously identified in Section 9.2.1.3, other issues were identified in relation to the

collection of the data from Perth. After the recruitment phase was completed, analysis

of the data from Perth brought into question the integrity of the data collection process.

It was concluded that the trial protocol had not been followed, particularly for the

implementation of the various surveys, namely knowledge, compliance, AQoL and the

detection of DRPs.

Additionally, statistical analysis of baseline demographic and clinical data showed

unexpected significant differences between Perth and the remaining sites, suggesting

bias in patient recruitment in Perth – despite the protocol employing a blocked

allocation concealment randomisation process. After careful consideration, it was

decided to keep only the following data of patients from Perth:

• admission / baseline data for control patients,

• admission discrepancy data for those intervention patients where it appeared that

the trial protocol had been followed,

• discharge discrepancy data for control patients,

• all anonymous patient and provider satisfaction surveys,

• all website usage and ICT data, and

• quality of PDMR/HMR report assessment.

Overall Discussion

of 767

Effectively, intervention patient discharge data and all 30-day follow-up data from Perth

were omitted from these analyses.

13.2.5 Grouping study patients by services received An alternate split of the data focusing on which interventions patients actually received

was outlined in the section 9.2.1.2. During data analysis it was found that, despite initial

randomisation to one of the study groups (intervention – PDMR model; Intervention –

Streamlined HMR recommendation; Control – HMR recommendation; Control – No

HMR recommendation), some patients may or may not have received the two primary

interventions: reporting of drug chart discrepancies in hospital and a home medicines

review (PDMR/HMR) post-discharge. Therefore, it was decided that the Project Team

would also analyse the data by regrouping patients based on the services that they

actually received. The new grouping system comprised of three groups:


• Comprised of patients who either did not receive any intervention or only

received discharge counselling. This group comprised, for the large part,

control patients who received no intervention and control patients who

received discharge counselling as part of usual care.

2. Partial Intervention

• The partial intervention group consists of patients who received discharge

counselling and had their admission and discharge drug chart discrepancies

reported to the RMO. This group was made up of intervention streamlined

patients who did not receive a HMR post-discharge.

3. Full Intervention

• The full intervention group is patients who received all three intervention

elements and consisted of intervention PDMR patients and intervention

streamlined HMR recommendation patients who did receive a PDMR/HMR.

This split was used when analysing outcome measures from discharge to 30 days and/or

admission to 30 days. It was felt it better defined the difference in impact of the

different levels of intervention and, in summary, compared the effectiveness of

reporting of drug chart discrepancies in hospital, discharge counselling and a home

medicines review (PDMR/HMR) post-discharge.

Overall Discussion

of 767

13.3 Medication chart discrepancies in the hospital

13.3.1 RMO drug chart discrepancies in general Like the review of published studies by Tam et al,34 our study revealed that medication

history errors at the time of hospital admission are very common and potentially

harmful to patients. As recommended by Tam, the main comparator when ascertaining

discrepancies should be a comprehensive medication history that includes an interview,

inspection of medication containers or lists, or both, and contact with community

pharmacies or family doctors.

The Med eSupport program incorporated two of the recommendations made by Tam et

al34 to ensure the acquisition of accurate medication histories at the time of hospital

admission.

“Pharmacists could be routinely involved in ensuring accurate medication histories at the time of admission, with particular attention to high-risk groups (e.g. patients with cognitive impairment using multiple medications).

Integrated community pharmacy databases accessible to hospital staff could also enhance the accuracy of medication histories.”

It was found that approximately 66% of initial hospital drug charts had at least one

error. This highlighted the current insufficiencies in the process of attaining an accurate

history of what patients are taking at the time of admission to hospital. It was generally

found that patient interviews were not a reliable means of determining what patients

were taking, especially when conducted by medical officers. This has previously been

identified in other studies.34 Factors such as memory, confusion, definition of a

medication and patient distress over their current condition are all factors potentially

contributing to this lack of reliability. A lot of older patients on many medications also

rely on dosage administration aids and personal carers or community nurses to manage

their medications, and they may have little knowledge of their own medications or what

they are being used for.

Across all trial sites, usual care procedures sometimes involved RMOs contacting GPs

for an account of the patient’s medications, but in the majority of cases, this came in the

form of an automatically generated list from the Medical Director® software. These

lists tended to include many medications that were no longer being taken and did not

include OTC medications unless known and recorded by the GP. It was found that the

community pharmacy was rarely contacted in the usual care process. In contrast, the

Overall Discussion

of 767

trial officers contacted the GP and community pharmacist, and interviewed the patient

to compile a reconciled medication list.

Discrepancies were reported to the RMO within 24 hours of admission for 79% of all

the intervention patients. There was a median of 2 discrepancies reported per patient.

These figures included all discrepancies found on initial review of the data. In the

process of discussing the issues with the RMO, it was found that 14% of discrepancies

were legitimate due to changes in therapy since admission of the patient, for therapeutic

reasons such as cessation of a drug or reduction of a dose. Discrepancies were identified

for control patients using the same method but were not discussed with the RMO.

Medication management was monitored by trial officers for the duration of each control

patient’s stay and resolution and the legitimacy of discrepancies was ascertained using

alternate methods so as not to actively intervene.

Trial officers were asked to avoid intervening in control patients’ discrepancies

wherever possible unless they felt the patient was at great risk of harm and they were

compromising their duty of care as pharmacists. Such intervention occurred on two

occasions. The first patient had suffered myalgia previously while taking atorvastatin,

and was consequently changed to ezetimibe and was taking this on admission. During

their hospital stay, pravastatin was added to their medications. The intervention was

made for this control patient when the trial officer thought it necessary to inform the

ward pharmacist of the patient’s previous reaction to a drug in the same class as the

newly commenced pravastatin. The ward pharmacist followed up on this issue with the

RMO. The second intervention that was made for a control patient occurred when the

patient was admitted with anaemia due to a severe gastrointestinal bleeding ulcer. The

doctors were unaware the patient was also taking indomethacin and prednisolone. The

trial officer contacted the ward pharmacist to alert them to this. This ward pharmacist

contacted the RMO and these medications were subsequently ceased.

13.3.2 Early resolution It was hypothesised by the Project Team that early intervention was needed to achieve

the greatest reduction in harm during the hospital stay due to prompt identification of

medication chart discrepancies. Trial officers identified and recorded how many

discrepancies had been addressed at 48 hours post admission, for all patients. It was

found that at least one discrepancy was acted on for 78% of the intervention patients

Overall Discussion

of 767

with admission drug chart discrepancies during this time period. In comparison, only

37% of the control patients with discrepancies had at least one discrepancy identified

and resolved through the usual care processes. This difference was found to be

significant, and suggests that admission drug chart errors are only identified for

approximately one-third of patients in usual practice.

13.3.3 Any resolution All patients were followed for the duration of their stay. Resolution of the discrepancies

found on admission (at any time throughout their stay) was recorded. Over the rest of

their stay, 30% of all patients with discrepancies on admission had at least one

discrepancy resolved after 48 hours. As it appeared that the majority of the

discrepancies found on admission for the intervention group were resolved in the first

48 hours, the lack of difference in numbers resolved across the groups after 48 hours

was not considered. This was reflected in the findings relating to admission drug chart

discrepancies not resolved at any point during the hospital stay. Many (83%) of the

control patients with identified admission drug chart discrepancies went home with at

least one discrepancy still not resolved (median 1, range 0-12). This translates into 54%

of all patients undergoing usual care going home from hospital with at least one

admission drug chart discrepancy still not resolved and hence, being carried over in to

the community setting. In many cases, such discrepancies may be interpreted as

changes made in hospital and not resolved, leaving the patient open to potential harm.

It is apparent that identifying a more accurate list of medications being taken at the time

of admission, with the aid of the community pharmacy dispensing history, and

discussing discrepancies found with the RMO leads to more accurate inpatient

medication charts and, hence, discharge summaries.

13.3.4 Admission discrepancy resolution impact on length of stay

Lengths of stay for patients in the trial ranged from 0 to 52 nights (median 4 nights).

Length of stay was not clearly related to the presence or absence of drug chart

discrepancies on admission. Interestingly, it was found that length of stay was weakly,

but significantly, impacted by the resolution of discrepancies in the first 48 hours of

admission. This correlation suggested that length of stay was directly proportional to

the number of discrepancies not resolved in the first 48 hours. Using only those who

Overall Discussion

of 767

had discrepancies identified on admission to measure this correlation, the same type of

weak, but statistically significant, correlation was found. It is possible that a larger

sample size would strengthen this correlation, but current data certainly suggests that

early rectification of admission drug chart discrepancies may result in a slight decrease

in length of stay.

13.3.5 New discharge discrepancies All discharge medication prescriptions were reviewed by the trial officer for new

discrepancies. These were identified using a combination of the inpatient drug chart

and the medical progress notes. Discrepancies found were reported to the RMO prior to

discharge for the intervention group and monitored for those in the control group. In

part due to the rapid turnover of patients, the discharge process implemented in both of

the Perth hospitals was considerably different to the other trial sites. This involved the

RMOs preparing the discharge medication list at the time of admission, with the

intention to update it at the time of discharge. We found that there were considerably

more discrepancies in the discharge summaries from these hospitals.

The overall result across all sites indicated slightly more control patients with at least

one new discharge discrepancy than the intervention patients. Generally, it was found

that approximately 30% of patients had at least one new discrepancy on the discharge

summary. These included omissions in therapy, wrong drugs, drugs ceased earlier in

the hospital stay being prescribed and wrong doses.

Of the intervention patients who had at least one new discrepancy at discharge, it was

found that, similarly to admission, 80% were resolved prior to discharge and only 13%

had at least one new discrepancy unresolved by the time they left hospital. In contrast,

87% of the control patients with new discrepancies went home with at least one

unresolved. Although 30% of control patients had at least one new discrepancy on

discharge, only 23% of these discrepancies were reported to the RMO by the ward

pharmacist. Overall, 24% of all the study patients left hospital with apparent medication

discrepancies at discharge.

Although resolution of discharge discrepancies is essentially a hospital issue, the Project

Team felt that provision of information at discharge and the planned post-discharge

interventions would be less than effective if the information they were based on had

errors. For the few intervention patients who did not have discrepancies resolved at

Overall Discussion

of 767

discharge, trial staff highlighted these on the discharge summary produced on the Med

eSupport website and during the faxing and phoning process to the GP and community

pharmacist, to minimise the chances of these errors continuing post-discharge. This

was not a difficult process for the trial team because production of a discharge summary

on the Med eSupport website and one-on-one communication with the community

health providers was part of the trial protocol. However, if a system of this nature were

to be rolled out and automated in some way, the possibility of errors in the information

and best processes to address them should be considered.

13.3.6 The value of the community pharmacist dispensing history

It was recognised from the beginning that no single source of information regarding a

patient’s medications was perfect. Overall, it was found that the community pharmacist

dispensing history was critical in producing the final reconciled list. A six-month

dispensing history was obtained for each patient and this provided extra information

unavailable from traditional sources, such as:

• compliance assessed via collection dates;

• evidence of medications taken intermittently or sporadically; and

• confirmation of medication taken as opposed to medication prescribed.

However, this source of information was also not necessarily complete. Non-specific

directions in histories, such as “Take as directed”, confounded the trial officer’s efforts

to establish patterns of medication usage. Similarly, patients choosing to use up old

prescriptions and self-adjusting the dose accordingly was also common place.

Generally, patients in the study had one regular pharmacy, but one-off collection from

other pharmacies occurred frequently and for a variety of reasons including price,

Section 100 accessibility, participation in a clinical drug trial, convenience, holidays

and collection of the medication by a friend/family member. These issues all contribute

to the difficulty experienced in attempting to create a reconciled medication list.

The discrepancies found between the reconciled medication list on admission and the

community pharmacy dispensing history were measured. Discrepancies relating to

OTCs were measured as a separate parameter. It was found that 59% of community

pharmacy dispensing histories had at least one prescription medication-related

discrepancy and for 50% of patients, at least one piece of information was missing

regarding OTCs. Currently, it is not common practice for the community pharmacy to

Overall Discussion

of 767

record or document OTC purchases. In the project setting, trial officers did ask

community pharmacists to include in the information they provided any known use of

OTCs for each patient. The recording of OTC medications continues to be a problem

area with no easy solution. In the event of further rollout of the Med eSupport system,

the transfer of information regarding OTC and complementary medicine use should be

addressed if an accurate medication history is to be produced as an end result. The

project Team does however acknowledge that this is not currently a legal requirement

and is unlikely to become one in the immediate future.

In summary, it is apparent that provision of a six-month community pharmacy

dispensing history at the time of hospital admission is an important addition to ensure

an accurate medication chart and discharge medication summary is compiled. The

community pharmacy dispensing history does not necessarily give an accurate or

complete picture of the patients’ medication usage as a single entity, but it certainly is a

large and invaluable piece of the jigsaw. Resolution of errors found on the inpatient

medication chart as a result of this additional piece of information may result in a

decreased length of stay. However, careful consideration is needed regarding how this

information is transferred and how OTC medications and complementary medicines

may be included before future rollout of the program occurs.

13.3.7 Economic analysis Some cases illustrating the benefits of obtaining a comprehensive medication history

and reconciling prescribed drug lists follow.

Case 1: 79 year old male in the intervention group:

• community pharmacy fills dose administration aid

• was taking Asasantin®, ONE tablet TWICE DAILY at home, but was charted for Asasantin® ONE tablet DAILY while in hospital

• was taking metoprolol XL 47.5mg (extended release) DAILY at home, but charted for immediate-release metoprolol 50mg DAILY

• both discrepancies were reported to medical staff at the hospital and corrected within 24 hours

Case 2: 71 year old male in control group:

• Charted and given a number of doses of incorrect insulin formulation. Patient usually on Novomix 30® and had been suffering a lot from hypoglycaemic events prior to admission. Whilst in hospital, patient was charted for Mixtard 30/70®, not Novomix 30® and given a number of doses

Overall Discussion

of 767

until the patient commented that it looked like a different insulin to what he was usually on

• Also was charted for double his usual dose of Seretide®, an incorrect dose of pancrealipase, and recommenced on iron tablets (which had been ceased previously by GP). In addition to that, he was not charted for his tiotropium or salbutamol inhalers. Finally, he was charted for quinine for nocturnal leg cramps even though the TGA no longer recommend it be used for this indication.

• At discharge the insulin was corrected, but the Seretide, pancrelipase, iron tablets, tiotropium and salbutamol were not corrected.

Case 3: 81 year old female in control group:

• Prior to admission, patient was taking enalapril 20mg TWICE daily, but only charted in hospital for ONCE daily

• Patient’s usual dose of omeprazole was 20mg ONCE daily but patient was charted for double this dosage

• GP recently commenced patient on oral isosorbide mononitrate at night in place of nitroglycerin patch for prevention of angina; neither was charted whilst in hospital

• These discrepancies were not corrected until discharge, the patient received incorrect doses and therapies for the duration of her stay in hospital

Case 4: 61 year old female in intervention group:

• On admission to hospital this patient was charted for double her usual dose of perindopril and half her usual dose of fluoxetine

• Prior to discharge from hospital this patient was commenced on ibuprofen in combination with her usual dose of meloxicam

• Both issues were rectified due to Med eSupport interventions

The clinical panel and economic analysis of a medication discrepancy review at hospital

admission or discharge indicated that a single review results in:





Given that 64% of the time, another health professional would not have intervened to

correct the discrepancies had the trial pharmacist not intervened, this means that the

average economic value of a medication discrepancy review performed solely as a result

of the Med eSupport program was:

• 26 days of health loss prevented

Overall Discussion

of 767

• 0.12 days in hospital prevented



13.4 Post-discharge medication review The design of the Med eSupport post-discharge medication review trial arms allowed

the comparison of different methods to promote medication reviews post-discharge and

to determine which model was likely to have the most success in achieving the desired

result of a medication review for the high-risk patient. The PDMR arm of the trial

achieved a high performance (96%) of medication reviews within 30 days after

discharge. This suggests that a program that is not GP-dependent for the medication

review referral, but uses standard criteria for identifying patients at risk of medication

misadventure, can be implemented with success. In fact, 74% of GPs exposed to the

PDMR process agreed to the statement that an automatic PDMR should be undertaken

for patients at risk of medication misadventure. This indicates a relatively high

satisfaction with the process and the service. The GP evaluations were generally

positive, and they could see the benefit of providing medication reviews to patients at

high-risk of medication misadventure after discharge from hospital.

Provider respondents who had a patient enrolled in the PDMR trial arm were more

inclined to subsequently want, in the future, an automatic PDMR for patients with risk

factors being discharged from hospital - exposure to a service increases the perception

of its worth. Importantly, the GPs agreed that this type of program should continue in

the future. Pre-conceived ideas that doctors would not welcome advice or suggestions

were not borne out in this study and it was felt that most GPs appreciated the interest of

another professional in the care of their patient.

13.4.1 Uptake of the models for promoting conventional HMRs post-discharge

It was pleasing to see that nearly one-quarter of patients in the streamlined HMR

recommendation group received a home medication review within 30 days after

discharge. This intervention involved a pre-filled referral form given to the patient, the

GP and the community pharmacy, and telephone contact with the GP and community

pharmacy to explain the reasons for the referral. The Medication Alert Project94 186

funded by the Department of Human Services of Victoria endeavoured to promote

Overall Discussion

of 767

Home Medicines Review (HMR) after discharge from hospital for high-risk patients.

The process used was similar to ours, with the uptake of HMR referrals within three

months of discharge for high-risk patients being only 9%. They have noted that the

reluctance of GPs to instigate HMRs was frustrating, despite close working

relationships between the hospital and HMR facilitators. One of the key differences in

relation to instigation of referrals is that the trial officers in the medication alert project

did not have a discussion with the GP or community pharmacist; we believe this is a

critical component of the overall medication review process and our higher uptake

results supports this belief.

The Med eSupport trial has demonstrated that a passive intervention of placing a sticker

on the hospital discharge summary to indicate that the patient may benefit from a HMR

simply does not work. There was the same proportion of patients who received a

medication review after discharge from hospital in both control groups (i.e. no HMR

recommendation versus passive recommendation). Anecdotally, a number of hospitals

have similar schemes to promote HMRs after discharge from hospital, such as ticking a

box on the discharge summary that says “HMR suggested or recommended”. The Med

eSupport trial has shown that these types of passive recommendation do not work and

hospitals should seek more active ways of promoting medication reviews after

discharge from hospital.

Post-discharge medication reviews are a key intervention to assist in the adherence to

the APAC guidelines for continuity of care in relation to medication management. The

APAC guidelines refer to the following:

“Factors influencing the patients’ knowledge about the medication and the ability to comply with medication regimens should be identified……………….Implementation of drug therapy should be accompanied by the use of appropriate education programs”187

A HMR helps hospitals ensure that patients are adhering to their intended regimens and

also that they understand their medications.

13.4.2 Timeliness The Med eSupport trial was specifically designed to assess if community pharmacists

were capable of performing timely medication reviews after discharge from hospital.

Importantly, all of the streamlined HMRs were conducted by community pharmacists

and the median time for the patient interview to take place was 21 days. The length of

Overall Discussion

of 767

time it took to conduct the patient interview is a reflection on the HMR process, which

requires patient consent (patient needed to see the GP after discharge), receipt of the

referral by the community pharmacy and organisation of a time to conduct the patient

interview. In comparison, the median time to patient interview from discharge in the

PDMR group was only six days (P < 0.0001). This also reflects the simplicity of the

PDMR design; a “high-risk” patient’s community pharmacy was contacted as well as

the GP informing them that a medication review should be performed.

It has been suggested that the most appropriate time to visit patients after hospital

discharge is approximately one week later. This allows the patient to identify any issues

surrounding their medication use and also allows the reviewing pharmacist to identify

drug-related problems and compliance issues that have manifested but have not yet

caused serious consequences. It should be noted that GPs were more likely to feel that a

PDMR helped them in the medication management of their patients, when compared

with a HMR (p = 0.07). However, the conduct of the medication reviews per se was not

significantly different between PDMRs and the HMRs. The only general difference was

the time to conduct the patient interview and this timeliness of the medication review

appears to have assisted the GP more than the standard HMR.

Stewart and Pearson visited patients at one week after discharge and studied 342

chronically ill patients discharged from acute care at the Queen Elizabeth Hospital,

Adelaide.188 At one-week post-discharge a home visit was performed by a nurse and a

pharmacist, during which medication management (including compliance and

medication-related knowledge) was assessed. During the majority of home visits, at

least one medication-related problem was detected and approximately half of the cohort

was found to be poorly compliant. Other previously unknown problems detected during

the home visit included hoarding of previously prescribed medication (35%) and

reducing medication intake to minimise costs (21%).

Two other South Australian studies by Stewart et al have also examined interventions to

improve outcomes following hospitalisation.189, 190 The first study looked at the effect

of a home-based intervention on readmission and death among a small sample of ‘high-

risk’ patients with congestive heart failure discharged home from acute hospital care.188,

189 Home-based intervention comprised a single home visit (by a nurse and pharmacist)

at one week after discharge to optimise medication management, identify early clinical

deterioration, and intensify medical follow-up and caregiver vigilance as needed. The

Overall Discussion

of 767

home-based intervention was associated with reduced frequency of unplanned

readmissions plus out-of-hospital deaths within 6 months of discharge from hospital.

A study by Spurling et al191 evaluated the effectiveness of a medication liaison service

to reduce medication incidents for patients discharged from hospital. There was an

intervention and a control group and the study found that the intervention group that

received the liaison service (home visit within 48 hours of discharge) had fewer

medication-related problems six weeks after discharge from hospital. It would appear

that post-discharge medication review studies that have shown significant benefit have

generally visited patients with one week of hospital discharge.

The delay that often occurs in the time to patient interview causes significant concern if

the current system of instigating HMRs is the only system that remains. The current

process of HMRs does not take into account the need for timely patient visits after

discharge from hospital, and another model of HMRs needs to be implemented. A

referral for a PDMR could be made by the hospital specialist or registrar, with hospital

pharmacists acting as key identifiers of patients at high-risk. A referral could then be

sent to the patient’s community pharmacy from the hospital with the suggestion that this

be conducted within one week of hospital discharge. The HMR information for the

community pharmacist and accredited pharmacist could be the same information located

in the discharge summaries or discharge letters. The accredited pharmacist could then

perform the HMR, with a report sent to the patient’s GP.

Nearly three-quarters of PDMRs were co-ordinated by the patients’ regular community

pharmacy, with a little over one-quarter requiring assistance from the research team to

perform these in a timely manner. This demonstrates that the majority of community

pharmacies, even at this relatively early stage in the delivery of professional services,

are able to deliver a timely PDMR to patients discharged from hospital. There did not

appear to be any great differences amongst the perceptions from community

pharmacists whose patients received a PDMR or HMR. However, it appeared that

community pharmacists for PDMR patients indicated that the medication review

assisted them in their patients’ medication management more than the streamlined HMR

recommendation group.

The reduction in the number of patients readmitted within one week of discharge

(“bounce-back”) between the control and intervention groups may be explained by the

timeliness of medication reviews being delivered after discharge. Importantly,

Overall Discussion

of 767

approximately 90% of patients who received a home visit by a pharmacist found it

useful and significantly more PDMR patients reported that they felt more confident

about their medications. Also, significantly more patients in the PDMR group compared

to the intervention-HMR and control groups indicated that they would like to see a

home visit by a pharmacist available in the future.

13.4.3 Economic value The clinical panel and economic analysis of the Post-discharge Home Medication

Reviews indicated that a review results in an average of:





It was estimated by the panel that 85% of the time another health professional would not

have intervened to correct the major issue identified had the pharmacist not intervened.

This means that the average economic value of a Post-discharge Home Medication

Review performed solely as a result of Med eSupport is:





Two simple cases illustrating the benefits of obtaining a post-discharge medication

review follow:

82 year old female in the intervention group:

• During this patient’s admission to hospital she mentioned to the trial officer that she was having trouble halving her metoprolol tablets (Lopresor® 50mg, not scored). She reported experiencing dizziness and light-headedness if she had inadvertently taken a bigger dose of her metoprolol due to inaccurately halving the heart-shaped tablet. During her hospital stay, she was give ½ a Minax® 50mg tablet twice daily. Despite this change being made in hospital, when she received an automated Home Medication Review (PDMR) from an accredited pharmacist within seven days of leaving hospital, she had returned to the Lopresor® brand and was continuing to experience the problem. The accredited pharmacist recommended that she be supplied with Minax® brand of metoprolol rather than Lopresor®. This

Overall Discussion

of 767

illustrates how a simple recommendation can make a big difference to a patient’s therapy.

• This patient was also running out of isosorbide mononitrate tablets following hospital discharge and decided to decrease the dosage. This was remedied during the medication review.

It has recently been demonstrated that simple medication-related risk factors identified

in the home are associated with poor health outcomes.192 The medication-related risk

factors that are related to poor health outcomes and therefore important to identify at

home visits include lack of any medication administration routine, therapeutic

duplication, hoarding, confusion between generic and trade names and multiple storage

locations, many of which can be addressed by further intervention or education. 192 It is

important to conduct a home visit, because most of the risk factors cannot be detected

by other methods, e.g. by ‘brown bag’ interviews.192

13.4.4 Uptake of PDMR/HMR recommendations It is clear that the potential savings associated with providing an intervention program

like Med eSupport are promising. These savings are reduced significantly when there is

poor uptake of pharmacist recommendations. The uptake of recommendations in the

hospital setting was exceptional. Recommendations made by the trial officer regarding

discrepancies on the drug chart at admission were implemented 78% of the time. This

did not include recommendations of a highly clinical nature and were mainly issues

such as the omission of a drug from the chart.

The uptake in the community setting, however, was low. Recommendations to GPs

made by the accredited pharmacist were implemented only 29% of the time (at least by

the 30-day follow-up period). This result is not unique to the Med eSupport trial; there

are several studies revealing a low rate of acceptance and implementation of

recommendations made by pharmacists from medication management reviews. Sellors

et al193 found that there was a high rate of ‘acceptance’ of home medication review

recommendations, but the actual implementation of the recommendations was much

lower than would have been expected. After 5 months, the physicians had succeeded in

fully implementing 46.3% (506/1093) and partially implementing 9.3% (102/1093) of

the recommendations. For another 16.7% (182/1093) of the recommendations,

implementation had been attempted but was not successful. Bonner and Carr found that

pharmacists raised clinical issues in 50% (25/50) of patients with doctors acting upon

16% (8/50) of the clinical issues and changing medications in 8% (4/50) of patients.194

Overall Discussion

of 767

In an intervention trial by Allard et al,195 the multidisciplinary team made a total of 147

recommendations, of which 37 were accepted by physicians. This equates to a rate of

just 25.2% acceptance of recommendations from medication management reviews.

This is not dissimilar to the 29% uptake from Med eSupport.

However, several studies show a high rate of acceptance and uptake of

recommendations from medication management reviews performed by pharmacists. In

a study by Martin et al,196 96% (811/844) recommendations were accepted, and 99% of

these were deemed significant, very significant or extremely significant. In another

study by Hawksworth et al,197 it was found that GPs accepted 82% (1234/1503) of

clinical pharmacy intervention.

In an attempt to better understand the reasons behind the low uptake of PDMR/HMR

recommendations, the quality of the medication review reports in Med eSupport was

independently assessed by one expert clinical pharmacist (Appendix XXXV). Overall,

there appeared to be considerable room for improvement in the standard of the reviews.

The reviewing pharmacists identified 236 drug-related problems, but the assessor

identified a further 66 problems using the same information as was available to the

reviewing pharmacists.

Approximately half of the reports were adequate when given an overall appropriateness

score that involved assessing nature, description and appropriateness of the problems

identified.

From this assessment, it would appear that an improvement in the standard of

medication reviews would potentially produce an increased uptake of review

recommendations, and direct benefits for patients and the health care system. However,

the Project Team only explored one of the possible contributing factors to low uptake of

the recommendations and this assessment was a subjective one, undertaken by only one

assessor. Not only would it be advantageous for pharmacists to further examine the

quality of their reports with the aim of continual improvement and refinement of the

process, but an investigation of GP attitudes and perceived barriers would likely provide

invaluable insight into the reasons behind the current low uptake.

The Project Team suggest that further investigation around the barriers to HMR

recommendation uptake and ways in which we can overcome them is crucial.

Overall Discussion

of 767

Across the five Med eSupport sites there were a number of different pharmacists who

provided medication reviews. It was evident that the pharmacists were often able to

identify the drug-related problems or issues associated with the patients they reviewed,

but had difficulty offering a clear recommendation. The accredited pharmacists were

asked to perform the HMR as they would normally. They were not given any guidance

or procedures to follow. As a result it was seen that in a lot of cases the communication

with the GP ceased following the provision of a letter with the recommendations. The

issues discussed above may help to explain the low uptake rate seen in our study.

In studies with low rates of acceptance or implementation of recommendations,

physicians cited that reasons for not fully implementing the recommendations

included:193

• Patient reluctance,

• Previous attempt and failure using the same strategy as recommended by the

pharmacist, and

• The inability to deal with a recommendation within a reasonable period because

other more urgent issues had arisen with the patient.

But there are also a number of factors which the pharmacist has direct control over.

One of the most critical is the way in which the pharmacist makes contact with the

general practitioner. Successful educational strategies include face-to-face contact

(academic detailing), but this feedback must involve not only a description of current

practice, but also include specific recommendations for changes in the use of

medications and how this can improve the patient’s management.198, 199 Letters to

physicians regarding patient therapy have been found to have a variable rate of

success.195, 200-202 Some studies show improvements to patient therapy after a letter has

been sent to the patients’ physician,201 but conversely, some studies show no difference

or indeed a negative impact to patients’ therapy.195, 200, 202 A combination of both

academic detailing and written material has also been shown to have a positive effect on

patient therapy.203

The layout and readability of some of the home medication reviews conducted

throughout the Med eSupport trial may have also contributed to the poor rate uptake.

The style of the letter that is sent to the physicians in this case may be an influence for

whether or not a recommendation is implemented. Collaboration between the

Overall Discussion

of 767

pharmacist and physician could benefit by being more like academic detailing, which

has been found to have higher rates of success for the implementation of pharmacist

recommendations.204-207

There may also be a difference between what the pharmacist deems as significant and

what the physician does in regards to patient therapy and risk factors. In a trial by

Atthobari,201 pharmacists recommended that patients with cardiovascular risk factors

such as a history of smoking, past myocardial infarction, past cerebrovascular attacks,

or a family history of cardiovascular disease, as well as elevated lipid or blood pressure

levels, be placed onto a suitable drug. However, therapeutic advice was only followed

in a third of patients with hypertension and a quarter of patients with hyperlipidaemia,

with physicians being influenced by the level of risk factor itself, not the presence of

other cardiovascular risk factors.201

There is scope in the future for programs and support services to improve the quality

and therefore uptake of home medication reviews. In the long run this can only result in

greater economic gains in the prevention of medication misadventure. Pharmacists need

to be made aware of strategies which will help them in being more successful in

performing medication reviews. Through doing this, and by learning the ways that are

most effective to communicate with physicians, patients should benefit, as collaboration

between pharmacists and physicians is able to improve patient outcomes.208-211

Differences in the mode of conduct and quality of the post-discharge medication review

process is probably the major distinguishing feature between Med eSupport and a

previous trial of pharmacist-conducted home visits and medication reviews following

hospitalisation by the Project Team.93 In the latter randomised, controlled study,

medical patients admitted to hospital and fulfilling high-risk criteria (including aged 60

years or older and taking four or more regular medications) were randomly assigned to

an intervention or control group. Intervention group patients were visited at home by a

pharmacist 5 days after discharge. The pharmacist educated patients on their drug

therapy, promoted compliance with therapy, assessed patients for drug-related

problems, intervened when appropriate and communicated all relevant findings to

community-based health professionals. Patients in the control group were only visited

90 days after initial discharge to assess outcomes.

A total of 121 patients completed the study. A median of 3 drug-related problems were

identified in each intervention group patient at the 5-day home visit. After 90 days this

Overall Discussion

of 767

had declined to 1, compared to 2 for the control group patients (p < 0.0001).

Compliance with drug therapy improved in the intervention group and was significantly

higher than for the control group after 90 days (p < 0.0001). Forty-five percent of

patients in the control group had unplanned readmissions to hospital during the 90 days

following discharge compared to 28% of patients in the intervention group (p = 0.05).

The intervention program was well received by the patients and their general

practitioners. Recommendations from the pharmacist in the report following the five-

day visit were implemented by general practitioners in 79% of cases – a much higher

rate of acceptance of suggestions than in Med eSupport. We propose that 3 factors were

involved in producing this higher uptake of pharmacist recommendations following a

medication review (and consequently larger effect on preventing hospital readmissions).

The pharmacist in the first study was reasonably well known locally by GPs and the

pharmacist contacted all GPs before and after performing the review and discussed

findings with them.

The potential financial savings per annum of $54M (additional) to $69M (total) to the

health sector based on 50 sites provides a strong argument for the value of rollout of an

intervention program based on the Med eSupport model. These expected savings only

apply, however, if all of the recommendations made by pharmacists were implemented.

If the current rate of implementation was to continue the savings nationally each year

would be a lot less and expected to be between $25M (additional) and $34M (total).

13.5 Patient specific variables Improvements in transitional care are needed on more than one level. For patients, an

improvement in their knowledge and understanding of their medications and

compliance with medications is vital in minimising the risk of patient-originated ADEs.

Rosenow212 stated “the problem of patients not understanding their disease(s) or its

importance and their non-compliance with their physicians’ recommendations is one of

gigantic proportions, in both morbidity and mortality as well as in costs.”212

13.5.1 Medication knowledge Providing adequate information to patients about their medications is an essential

principle of rational pharmacotherapy, as the level of a patient’s medication knowledge

is highly associated with outcomes of medication therapy.213 Gilbert et al found, when

trialling a medication management service involving pharmacists visiting people in their

Overall Discussion

of 767

homes, that 20% of all medication issues identified were due to patient knowledge and

skills.214

Patient knowledge was assessed during the initial interview and again during the 30-day

follow-up phone call. Med eSupport found that all patients improved their knowledge

about medications in this time frame. Those who received interventions displayed a

significantly higher knowledge score at 30 days post-discharge than those who received

little to no intervention. It would appear that a medication review, particularly when

done within one week post-discharge, contributed the most to the improvement in

knowledge. This is evidenced by the finding that the full intervention group (i.e. those

who received discrepancy intervention as inpatients, discharge medication counselling

and a PDMR/HMR post-discharge) had a significantly higher knowledge score at 30

days than those who had not received the PDMR/HMR service. Furthermore, of these

patients, PDMR patients displayed a significantly higher knowledge score than those

patients in the other groups.

13.5.2 Self-reported medication compliance It is well documented that patient compliance is difficult to accurately measure.215 For

example, in the study by Harder et al,216 following up elderly cardiovascular patients for

one year, only 8% of patients admitted to non-compliance,216 which in reality is highly

unlikely. The recently published study by MacLaughlin et al found that the most

reliable method for measuring compliance is through an interview involving open-

ended, non-threatening and non-judgemental questions. 215

Compliance of patients in this study was assessed using a four-question validated Self-

Reported Medication Compliance Scale designed by Morisky,217 shown on pages 9 and

31 of the patient data collection sheet in Appendix VII. The questions were asked in a

non-threatening, conversational interview style in order to ensure accurate responses

were given by the patient and/or carer.

Our results indicated there was an improvement in self-reported compliance using a

validated medication compliance scale for patients who received the full Med eSupport

intervention. In general, there were no statistical differences for the self-reported

compliance at the initial interview and at the 30-day follow-up phone call across the

three groups. However, results of all of the 30-day phone call compliance data indicated

Overall Discussion

of 767

there was a trend for the full-intervention patients to report a greater compliance than

the other two groups.

13.5.3 Self-reported drug-related problems It would appear that significant improvement in knowledge, understanding and

compliance with medications does result in improved medication management and

outcomes from a patient’s perspective. Patients were interviewed on admission and at

30 days post-discharge and during both interviews, were asked if they were

experiencing any issues with their medications. The types of issues reported included:

• suspected side effects,

• dosage form and usage concerns,

• compliance-related issues,

• supply-related issues,

• dissatisfaction with medications, and

• confusion regarding medication management.

Patients were asked open-ended questions and not prompted to report specific issues.

At admission there were no statistical differences in the number of self-reported DRPs

across the three groups. At discharge there were fewer self-reported DRPs reported, but

the difference between admission and discharge was not significant.

At 30 days post-discharge, it was found that there was a tendency for the minimal

intervention group patients to report more DRPs than the partial and full intervention

group, but this was not statistically significant. However, the minimal intervention

group was found to display a significant increase in the number of self-reported DRPs at

30 days compared with at discharge. In comparison, the partial and full intervention

groups displayed a slight decrease in number of DRPs reported over the same time

period. Overall, from the admission interview to the 30 days post-discharge interview,

the minimal intervention patients displayed a slight increase in self-reported DRPs. The

partial intervention group displayed a slight decrease in total number of DRPs, but the

full intervention patients were the only group who displayed a positive and significant

change, reporting significantly less DRPs at 30 days than at admission. This finding was

similarly reflected when analysed for the two and four group split and further

emphasises the very real and positive effects that the trial interventions had.

Overall Discussion

of 767

13.5.4 Quality of life Quality of Life (AQoL) is an assessment of the individual’s overall sense of well-being.

The difference in AQoL scores across the groups was not statistically significant at

baseline or follow-up. As might be expected, there was a statistically significant

improvement for all groups’ quality of life scores from the time of being an inpatient to

30 days post-discharge (over time). This can simply be explained by the patient’s

presence in hospital being associated with poorer health at the initial interview and a

resultant poorer perceived quality of life.

Similar results have been seen in a number of studies. Health related quality of life did

not differ between the intervention group and control group at baseline or at one year

follow-up in a management programme for patients with heart failure.209 The quality of

life was evaluated using a disease specific questionnaire (quality of life in heart failure

questionnaire), a generic questionnaire (the Nottingham health profile) and the patient’s

global self assessment.209

Sorensen et al117 measured health related quality of life using the SF-36 subscales and

the physical and mental component scores. The SF-36 was measured at baseline and at

the end of the trial and there were no differences in the scores between the intervention

(patients who received multidisciplinary review in the community) and control patients

(normal care). Krska et al210 found that there were no significant changes in any of the

scores between their groups. Intervention patients received a review by a pharmacist

and a care plan was drawn up and implemented, control patients received normal care.

None of the domains used in the SF-36 questionnaire showed any significant changes in

either group at follow-up. This lack of any discernible effect on health related quality of

life may be attributed to the heterogenous nature of the patients, the generic tool used

and a minimal effect of modifications to medicine use on quality of life.210

One might suggest that the pharmacist intervention and recommendations in the Med

eSupport trial also had little effect on a patient’s perceived quality of life - particularly

for those patients who had recommendations made that were never carried out. In

comparison, Lim et al had a significantly different change in AQoL scores between

baseline and one month follow-up. The intervention group (who received post-acute

care after discharge from hospital) had significantly greater improvements in

independent living (p = 0.002) and overall quality of life scores (p = 0.02) compared

with their control counterparts.218

Overall Discussion

of 767

Interestingly, Holland et al had a very different result; the QoL actually worsened with

time, rather than improved. Patients in the intervention group received two home visits

by a pharmacist within two and eight weeks of discharge. The study did not use the

same QoL scoring tool but had very similar methods. The EQ-5D score was used to

determine QoL in this study; both group’s scores decreased over the six-month follow-

up period, but the changes were not significantly different between the groups.211

In 2004, Furlong et al219 reviewed the factors that are important for selecting

appropriate instruments to measure health status and health related quality of life in a

particular context. The literature suggests that there may be important effects due to the

mode of data collection and therefore, mode of administration should be standardised

across subjects, assessors and assessment points. The nature of the trial design required

our first QoL interview to be performed face-to-face and the second to be conducted

using the telephone. It is possible that the unexpected result (not statistically significant)

of difference in QoL for our patients who did not receive services could be partly

explained by the mode of data collection. Additionally, the length of the follow-up

phone call for intervention patients was longer than their control counterparts. The

questions regarding ICT and HMR services were only asked of intervention patients.

The length of the call may have influenced the intervention patient’s attitude and

therefore answers to the AQoL questionnaire, which was carried out late in the phone

call. There are limits to the number and type of measures that respondents can be

expected to complete without getting tired or frustrated.220

One could also suggest that the QoL would have been better assessed at a number of

points in time. We only conducted the QoL at admission (baseline) and at 30 days from

discharge. The benefits experienced due to the pharmacist intervention would be likely

to peak at a later stage than one month. This is a limitation of the trial design and in

hindsight it may have been more useful to measure the QoL at one, three and six months

post-discharge.

Of the 352 patients included in the AQoL analysis, 73 patients were unable to complete

the questionnaire at the 30-day follow-up. They were not contactable for a number of

reasons. It was evident that there was a significant difference in the baseline AQoL

score for these patients compared to those patients who were contactable at 30 days.

This difference suggested the patients who were unable to be contacted at 30 days may

have been sicker and may have been more likely to benefit from the services. It was

Overall Discussion

of 767

determined, however, that across the 3 groups there was no statistical difference in the

baseline AQoL score. The baseline characteristics of these patients were also very

similar. The study by Lim et al218 was a retrospective study. Patients could only be

included in this study if they were expected to live at least one month post-discharge,

and were not admitted from or discharged to a nursing home or hostel. It would have

been difficult for us to exclude patients at the outset of our study based on all of these

criteria and as a result we had an unavoidable drop out at 30 days of these sicker

patients.

The first domain in the AQoL questionnaire is illness. It was suggested that although

the other dimensions are likely to be sensitive to a number of other factors it would be

fair to assume this dimension might reflect an improvement due to our intervention.

Again, however, the results showed no statistically significant difference across the

three groups.

The AQoL measurement tool may not have been sensitive enough for measuring quality

of life in such a variety of health conditions across the patient groups. Because of

variability it is highly desirable that users should include more than one generic

instrument in their study.102

Particular issues in the assessment of QoL in older patient populations include the

persistent finding of a poor relationship between QoL and disability/disease severity,

the dynamic nature of QoL, and the importance of valid proxy ratings for those unable

to make decisions or communicate for themselves.221

13.6 Drug Related Problems

13.6.1 Total actual or potential DRPs It was found that approximately 90% of all patients had at least one DRP identified at

admission, discharge and 30 days. DRPs were collated using a variety of sources. This

included running the patient’s medical and medication information through the

standardised decision support tool, Cognicare®, and selecting all those DRPs, actual or

potential, that were considered significant or moderate by the program and which the

analysing pharmacists considered feasible. Each patient’s medication list was also run

through Facts and Comparisons Drug Interactions Facts® (DIF®), and drug interactions

listed in the two most serious categories (Level 1: ‘Potentially severe or life-threatening

interaction; occurrence has been suspected, established or probable in well controlled

Overall Discussion

of 767

studies’ or level 2: ‘Interaction may cause deterioration in a patient’s clinical status;

occurrence suspected, established or probable in well controlled studies’) were recorded

as being potentially clinically relevant. These two pieces of information were then

combined with patient-reported DRPs and practical findings such as untreated

indications at discharge and medication issues contributing to conditions on admission.

In hindsight, it would appear that the measures used were not sensitive enough to show

changes in DRPs across the groups and with time. Using the Cognicare® and DIF®

databases ensured a scientific approach and uniformity for all patients based on a

standardised set of rules. However, databases generally are unable as yet to view the

patient and their circumstances as a whole and both actual and potential DRPs were

considered. It is possible that manual review by experienced pharmacists at admission,

discharge and 30 days may have produced a more realistic set of DRPs.

Some findings of note have been identified despite the concerns regarding sensitivity of

the measure. Overall, it was found that minimal intervention patients did display a

significant increase in the number of DRPs per patient from admission to 30 days,

compared to the partial and full intervention groups who displayed no significant

change. This same outcome was identified between the intervention and control groups,

with the control patients experiencing a significant increase in total number of DRPs per

patient. These findings suggest the intervention package prevented an escalation of

drug-related issues immediately after discharge as is so commonly seen in everyday

practice53 and was demonstrated by the control and minimal intervention groups.

13.6.2 Breakdown of the identified actual or potential DRPs

Total DRPs were made up of several different elements and so it was decided to

measure these elements individually to ascertain if some particular types of DRPs

changed across the groups or with time more than others did.

13.6.2.1 Cognicare®-identified DRPs Firstly, DRPs identified through Cognicare®, including only those classified as

significant and moderate, were measured individually. It was found that overall, 86%

of patients had a least one Cognicare®-identified DRP at admission and this increased to

90% at discharge and at 30 days. Such a high finding further illustrates the Project

Team’s concerns with the sensitivity of this analysis tool. Despite the high numbers of

Overall Discussion

of 767

patients with at least one Cognicare®-identified DRP, it was found that full and partial

intervention patients experienced a significant decrease from admission to discharge in

the number of Cognicare®-identified DRPs per patient. Although there was an increase

for these two groups from admission to discharge, at the point of discharge, there was

no difference across the three groups, which validates this post-discharge finding

further. In comparison, the minimal intervention group showed no change post-

discharge.

Therefore, despite the lack of sensitivity of the analysis tool, it would still appear that

those who received at least two of the trial interventions did display an improvement

over the peri-discharge period.

The Project Team did analyse for significant and moderate DRPs separately, but found

no changes of interest above and beyond those found when analysing both levels of

Cognicare-identified DRPs combined.

13.6.2.2 Drug interactions identified through DIF® It was found that all patients experienced a significant increase in drug interactions

identified through DIF® from admission to discharge. Again, concerns over the

sensitivity of this analysis tool are illustrated by the high proportion of patients with

drug interactions identified and lack of change across the groups over time. However,

like the Cognicare®-identified DRPs, it would appear that the full intervention group

significantly improved in number of drug interactions per patients from discharge to 30

days compared to the other two groups.

13.6.2.3 DRPs by the DOCUMENT categorisation tool As described in the results section, DRPs were categorised using the DOCUMENT

categorisation system.116 The eight DOCUMENT categories are:

D – Drug Selection,

O – Over or under dose prescribed,

C – Compliance,

U – Untreated Indications,

M – Monitoring,

E – Education or Information,

N – Non-clinical, and

T – Toxicity or Adverse Reaction.

Overall Discussion

of 767

It was found for the setting of this project, the majority of issues fell into the C, D, T or

U categories and details of theses DRPs were analysed. ‘C’ category problems are

described as ‘problems related to the way the patient takes the medication’ and are

loosely grouped as compliance related problems. Generally, there were a similar

percentage of patients across the groups with a similar number of ‘C’ DRPs per patient

at admission and at the point of discharge. However, by 30 days post-discharge, it was

found that the intervention group and the two intervention sub-groups had experienced a

significant reduction in ‘C’ DRPs per patient compared to the control group and sub-

groups. When analysed across the three groups, this difference was only found to be a

trend. However, over the duration of the trial, it was clear that those who received

some or all of the trial interventions did experience a reduction in ‘C’ DRPs compared

to those who received minimal to none of the trial interventions. This finding would

appear to support the results from the compliance survey that those who did experience

the new services did improve in their personal medication management.

Category ‘D’ problems are described as ‘problems related to the choice of drug

prescribed or taken’ and are loosely described as drug selection problems. These

involved, in part, those identified through Cognicare® and those identified through DIF

and therefore, the sensitivity of this group as a measure is again unclear. Findings for

this group closely mirrored those for the DIF®-identified and the Cognicare-identified

DRPs. In particular, it was found that the full intervention group experienced a

significant decrease in ‘D’ DRPs over the discharge to 30 day period compared to the

other groups.

Category ‘T’ problems are described as ‘problems related to the presence of signs or

symptoms which are suspected to be related to an adverse effect of the drug’ and are

loosely described as toxicity or adverse reaction type problems. The majority of the

DRPs in this group are identified through Cognicare® which is reflected in the high

percentages of patients with at least one ‘T’ DRP across the groups and the lack of a

perceptible change over the time periods.

Category ‘U’ problems are described as ‘problems related to actual or potential

conditions that require management’ and are loosely described as problems relating to

untreated conditions. The findings for this group appear to correlate well with the

findings for the reported discrepancies in hospital, as well as the impact of the

intervention post-discharge. Despite no differences across the groups at admission there

Overall Discussion

of 767

were significantly more of these DRPs at discharge for those in the control group and

minimal intervention groups than those in the full intervention group. This difference

was maintained at 30 days; despite all patients displaying a decrease in number of ‘U’

DRPs, the minimal intervention and control groups still had significantly more not

resolved at 30 days post-discharge than the full intervention group. Overall, it was

found that those in the full intervention group displayed no change in the number of ‘U’

DRPs over time compared to the other two groups, who both displayed a significant

increase due to their stay in hospital. This finding would indicate that the combination

of identifying and reporting drug chart discrepancies during hospital and providing a

PDMR/HMR shortly after discharge provides the best possible chance to prevent errors

occurring in the first place and to prevent those that do occur from perpetuating and

potentially causing further DRPs after discharge.

Overall, it would appear that, despite concerns over lack of sensitivity of the DRP

compilation process, the combination of the discrepancies identified and reported during

hospital and a PDMR/HMR shortly after discharge helped to reduce actual and/or

potential DRPs for patients post-discharge.

13.7 Annonymous participant satisfaction surveys Feedback on Med eSupport was very positive. Some of the findings are listed below.

Over 90% of the study patients who had a medication review reported that it had been

useful. Patients who had a PDMR were more likely to want a home visit by a

pharmacist to be available in the future.

Most GP respondents thought that the medication summary was presented within an

adequate timeframe and they strongly agreed that receiving discharge medication

information of this nature in the future would be valuable.

There was a positive response from both GPs and community pharmacists when asked

if they thought that the study benefited them in optimising the patient’s medication

management through improved communication of medication-related information.

General practitioners indicated that Med eSupport gave them a clearer picture of their

patient’s medication on discharge.

Overall Discussion

of 767

There was a very positive response when GPs were asked if the post-discharge home

medication review (PDMR) assisted them with the medication management of their

patient.

When GPs were asked if they would like to see an automatic PDMR for their patients in

the future, 74% responded that they wanted the service.

13.8 Website utilisation Usage of the Med eSupport website by trial patients and their carers to access

information on their medications was relatively low; 15% of the intervention patients

used the website at least once. This was not surprising, however, given the

demographics of the patients. Most stated that they did not have access to a computer,

let alone the internet. This should not necessarily deter from future applications of this

type as there is clearly an increasing uptake of the use of the Internet by elderly patients

to obtain health information.

Those patients who did use the Med eSupport website generally found it useful,

particularly for accessing a current medication summary. Community pharmacists were

more likely to use the website, and found it more useful than GPs.

13.9 Additional health care costs Although costs associated with GP and specialist visits were slightly higher for patients

who received more intervention services in our study, this could be interpreted as a

good outcome.

The patients who received more services were shown to have better knowledge and

compliance scores than those who received little to no extra services. One would

suggest that this could imply the patient has a greater awareness and understanding of

their health conditions and symptoms, and are more likely to seek help from their GP

earlier than a patient who receives usual care would be. Indeed, the intervention process

reinforced the need for patients to follow-up with the GP after discharge from hospital.

The intervention group patients also had a smaller rate of bounce-back (readmission

within 5 days of discharge) into hospital. One might suggest this increased rate of GP

visit seen in the intervention group may have prevented a bounce-back readmission, and

avoided the cost associated with an acute presentation. Both of these explanations may

Overall Discussion

of 767

help to explain the intervention patients having a greater number of visits to their GP

than the control group.

Similar results have also been seen in other studies. The UK Homer study (84

intervention, 81 control patients) found GPs carried out 204 home visits in the

intervention group and 125 in the control group, a difference of 43%. (rate ratio = 1.43,

P = 0.002) There was no statistically significant difference between the two groups in

attendance at general practices.211

Krska et al found there were slight increases in contacts with both practice nurses and

GPs for drug-related or therapy monitoring purposes in their intervention group

(patients who received a review and a care plan), which was not seen in the control

group.210

Although not statistically significant in Med eSupport, the patients who received a

medication review had the most visits to the GP of all the groups. Again, this is

inherent in the intervention process. At the time of discharge the patient automatically

received a HMR within 7 days of discharge. This in most cases lead to the patient

contacting their doctor for a follow-up consultation. It may have inadvertently served as

a prompt or reminder to the patient to see their GP post-hospital.

13.10 Readmission rates of patients within 30 days of discharge

There was no difference in unplanned hospital readmission rates between the control

and intervention groups found in this trial. This outcome is not unique to this study. In

2004, Hanlon et al222 performed a comprehensive literature review of randomised

controlled studies examining clinical pharmacy services. They found considerable

evidence that clinical pharmacy interventions reduced the occurrence of discrepancy

drug-related issues in the elderly, but at the same time, found limited evidence that such

interventions reduced morbidity, mortality and healthcare costs.222

Although not statistically significant, our results suggested there may have been slightly

higher rates of readmission for patients that received intervention services. A number of

other studies including the UK Homer study by Holland et al showed that the

intervention group had a higher rate of readmissions than the control group. (178

emergency readmissions occurred in the control group and 234 in the intervention

group). The Poisson model indicated 30% greater rate of readmission in the

Overall Discussion

of 767

intervention group (rate ratio = 1.30 and confidence interval 1.07 to 1.58; p < 0.01).223

This could be explained by improved knowledge and awareness of their condition and

better health seeking behaviour.

One other possible factor that may have influenced the increase in readmissions seen in

the intervention group may be the lack of compliance with the HMR recommendations

made by the accredited pharmacist. As discussed, there was a poor uptake by GPs and

their patients of recommendations made by the accredited pharmacist in the community.

The pharmacist may have provided the patient with information making them more

aware of their condition but the recommendations to solve or help their condition were

never carried out and in effect made them not unlike a patient who received no post-

discharge services.

Reasons for early unplanned readmissions to hospitals are varied and complex. In a

study of 133 elderly patients who had an unplanned readmission to a district general

hospital within 28 days of discharge, seven possible principal reasons for the

readmission were identified: relapse of original condition, development of a new

problem, carer problems, complications of the initial illness, need for terminal care,

problems with medication and problems with service,224 with medication complications

being just one in a list of many. A literature review on hospital readmissions performed

by Benbassat et al56 found that most readmissions are believed to be caused by patient

frailty and progression of chronic disease. However, they did find that from 9 to 48%

of all readmissions have been judged to be preventable because they were associated

with indications of substandard care during the initial hospitalisation, such as poor

resolution of the main problem, unstable therapy at discharge and inadequate post-

discharge care.56 Tierney et al55 also reviewed readmission rates of elderly patients and

concluded that readmission rates, as currently defined and recorded, are neither a

sensible nor sensitive indicator of the effectiveness of acute hospitals when elderly

patients are now among their main users.55

Data reliant on a patient’s account may also not always be 100% reliable. As a result, a

comparison was made between the patient-reported readmission data (Tasmanian

patients only) and compared with the matching data received from the medical records

departments at the hospitals. There was a fairly strong correlation between the two sets

of data, suggesting the patients reported fairly accurately. However, the two sets of data

did not match up exactly. Some likely explanations may include the patient having

Overall Discussion

of 767

been admitted to a private hospital or an outpatient centre and subsequently reporting

this as a readmission (both not likely to be picked up by a medical records report).

Alternatively, the patient may have recalled the information incorrectly (i.e. Included

their initial admission as a readmission or miscounted how many times they had been

readmitted in the last month).

It was shown that almost one in five patients were readmitted within one month from

initial discharge. Of those patients approximately 15% were readmitted within five

days. These “rebound” patients consisted of 89% control patients. It was interesting

that 44% of the patients who had a rebound admission had left hospital with apparent

medication discrepancies at discharge, compared with an overall figure of 24% of all the

study patients. It is also possible with increased trial participant numbers, a trend in

medication-related (documented as such in the medical record) readmission rate

improvement may have been found as two of the control patients in this cohort

experienced a readmission of this nature, compared to none of the intervention group

patients.

13.11 Future direction The project has proven the benefit of communication between community pharmacies

and the patient’s hospital in preventing medication discrepancies on admission. In

addition, when the system is automated using ICT the transfer of the medication

information can be realised efficiently. Consequently we consider this aspect of the trial

could be expanded to a national level.

To up-scale the Med eSupport trial to a national level it is recommended that the

equivalent of the Med eSupport server is located in each hospital. The installation and

administration of the server should be performed by the Hospital’s ICT department,

where experience to fix problems at a local level exist. An external company should be

contracted to provide updates to website design, drug and provider information to each

hospital by way of automatic updates.

The model shown in Figure 28 allows any Community Pharmacy to send patient’s

medication information with the patient’s consent to any hospital. The prescription

information could be stored at the hospital for a nominated period of time.

Overall Discussion

of 767

In the short-term an interim solution for transfer of medication information from

community pharmacies to hospital may be achieved by faxing six-month dispensing

histories to the hospital, with appropriate remuneration structures in place.

Figure 28 Proposed future model for the ICT delivery of Med eSupport

With staff shortages being common in public hospital pharmacy departments it might be

more sustainable to employ a community based liaison pharmacist to co-ordinate an

accredited pharmacist already in place in the community. Ideally, this role would be

fulfilled by existing MMR facilitators. Further, the MMR facilitators could also

coordinate and perform education of GPs, CPs and hospital staff regarding the PDMR

as needed. Liaison with the hospital on a regular basis to encourage uptake would also

be a vital role. At the outset, the MMR facilitators could also have a key role in setting

up the new services, including intial training and establishment of local databases of

accredited and community pharmacists to ensure PDMRs can be performed in a timely

manner.

One would suggest there are many hospitals/communities Australia-wide that could

take on the Med eSupport model without exorbitant economic outlay, as most already

have some of the services in place. Theoretically, there should not be a need to employ

Overall Discussion

of 767

more hospital pharmacists, as medication reconciliation on admission and medication

counselling on discharge are activities that, in accordance with APAC National

guidelines to achieve the continuum of quality use of medicines between hospital and

community60, should be part of current practice.

13.12 Project limitations It was disappointing for the Project Team that the goal of 750 patients was not achieved,

thereby reducing the statistical power of the study. Difficulties such as the timely

recruitment of patients within 24 hours of hospital admission and the drop-out of

patients following discharge have been discussed. These were not unexpected, but still

impacted significantly on the study numbers. A larger study would have enabled closer

examination of issues such as the bounce-back readmissions to hospital within 5 days of

discharge and the relationship between medication discrepancies on admission and

length of stay in hospital. The divergent implementation of Med eSupport at the Perth

sites was a major concern and has significantly reduced the study numbers.

Ideally, there would have been a longer follow-up of patients; in particular, 30 days is

inadequate to determine the full effects of a post-discharge medication review.

Parameters such as QoL would have been better assessed at a number of points in time

and for a longer period. A longer follow-up period would also have facilitated the

examination of outcomes in those patients who had pharmacists’ recommendations

implemented vs. those patients who did not. This would have been a very useful sub-

analysis. Also, it would have been beneficial to closely examine temporal changes in

the Pharmaceutical Benefits Scheme drug costs in the groups of patients.

The possibility of enrolment in the study influencing the care and conduct of control

patients is inevitable. It was not possible to blind patients and their health care

professionals from group allocation, or services and follow-up would not have been able

to be performed. Therefore, it is possible that patients enrolled in the study took more

care with their medication management as the process of their enrolment would have

alerted them to the issue. Also, health care professionals, both in the hospital and in the

community, who knew their patients were not receiving the various interventions may

have paid extra attention to their patient’s care. This phenomemnon is known as the

Hawthorne effect and although every effort was made to minimise it, to a large extent,

is unavoidable in a trial of this nature. Had the project allowed for complete blinding of

Overall Discussion

of 767

patients and health care professionals, it is possible that control patients would have

received less suitable care and would have cared for their medicines in a less

appropriate manner.

Conclusions and recommendations

of 767

14 Conclusions and recommendations This study examined medication management amongst older, chronically ill patients

entering and recently discharged from hospital, and found clear evidence of problems

relating to sub-optimal use of prescribed medications, particularly with regard to the

reconciliation of medication histories. The Med eSupport program, directed at

improving medication history reconciliation and the provision of a post-discharge

medication review (PDMR), has resulted in an improvement in patient compliance, and

less medication-related issues that could adversely affect a patient’s health. The

intervention group displayed a significant improvement in their compliance and drug

knowledge over the 30 day post-discharge period, along with a significant decrease in

the total number of major and moderate DRPs per patient. Patient-identified DRPs

reported by the full intervention group were also significantly fewer than the other

groups over the period from admission to 30 days post-discharge.


first 48 hours of hospital admission for the intervention group than for the control.

Significantly more discrepancies were also resolved prior to discharge for intervention

patients than for control patients.

A clinical and economic analysis has illustrated the benefits to the health care system of

medication reconciliation at hospital admission and discharge, and of an early post-

discharge medication review. The program was also welcomed by the patients, their

general practitioners and their community pharmacists. Although not surveyed, it was

noted by trial officers that anecdotally, hospital staff also found the extra services

provided valuable. This study shows that relatively simple information flow and

investigation and interventions by a pharmacist working closely and liaising with

patients and other caregivers can improve the efficacy and safety of drug use in the

elderly.

Implementing the trial at 5 different hospitals across 3 different states provided valuable

lessons regarding the logistics required to implement any type of new service on a

national scale. Each state, region and hospital, as well as individual health care

providers within them, had distinctly different protocols, procedures and health care

cultures. It is an important finding from this study that careful consideration, regarding

the logistics of rolling out a new program, needs to be undertaken for each individual


of 767

place and tailored to their needs. Once implemented, it is clear from the Project Team’s

experiences that a process of ongoing quality assurance should be developed and

implemented to ensure the new service or services are running to their maximum

capacity and achieving the goals they were originally designed to achieve.

The Med eSupport program has also clearly established the benefits of patients initiated

on warfarin in hospital subsequently receiving home visits by a pharmacist after

discharge from hospital. The pharmacist, using point-of-care testing, obtained INR

results and educated the patients regarding anticoagulant therapy. Control of

anticoagulation was significantly improved, and there was a significant lower incidence

of total, major and minor bleeding complications within 90 days in the intervention

patients. The program was highly cost-effective, and could save over $A10 million in

reduced bleeding costs per year if implemented across the country. The program was

well received by patients and doctors. Home follow-up incorporating INR monitoring,

can assist general practitioners in the management of high-risk patients on warfarin

making the transition from hospital to community care.









information to GP(s) and CP(s) regarding initiation of warfarin in hospitals is

one priority.






group.


of 767




reviews.


















significantly increased. The PDINR program should comprise POC INR


References

of 767

15 References 1. Hawthorne G. Assessment of Quality of Life (AQoL) Home. 3 ed. Melbourne;

1999.

2. [cited 13/1/2006]; Available from: http://www.icsmultimedia.com.au/

3. PCANU. [cited 13/1/2006]; Available from: http://www.pcanu.com.au/index.aspx

4. phoenix. [cited 13/1/2006]; Available from: http://www.phoenixcorp.com.au/

5. PKI. [cited 6/1/2006]; Available from: http://www.its.monash.edu.au/ staff/security/staff-only/certs//theory/pki-definitions.html

6. sqlref. [cited 13/1/2006]; Available from: http://en.wikipedia.org/wiki/SQL

7. Roberts MS, Stokes JA. Prescriptions, practitioners and pharmacists. Medical Journal of Australia. 1998;168(7):317-8.

8. Australian Council for Safety and Quality in Health Care. Second national report on patient safety: Improving medication safety. Canberra: Australian Council for Safety and Quality in Health Care; 2002.

9. Goldney R, Fisher L. Use of prescribed medications in a South Australian community sample. Medical Journal of Australia. 2005 September 2005;183(5):251-3.

10. Silverman J, Stapinski C, Churchill W, Neppl C, Bates D, Gandhi T. Multifaceted approach to reducing preventable adverse drug events. Am J Health Syst Pharm. 2003 Mar 15;60(6):582-6.

11. Smith J. Building a safer NHS for patients: Improving medication safety. London: British Department of Health; 2004.

12. Winterstein A, Sauer B, Hepler C, Poole C, Suarez E, Kaiser J. Preventable drug-related hospital admissions. Annals of Pharmacotherapy. 2002;36(7-8):1238-48.

13. Howard R, Avery A, Howard P, Partridge M. Investigation into the reasons for preventable drug related admissions to a medical admissions unit: observational study. Qual Saf Health Care. 2003 Aug;12(4):280-5.

14. Peyriere H, Cassan S, Floutard E, Riviere S, Blayac J, Hillaire-Buys D, et al. Adverse drug events associated with hospital admission. Annals of Pharmacotherapy. 2003;37(1):5-11.

15. Runciman W, Roughead E, Semple S, Adams R. Adverse drug events and medication errors in Australia. Int J Qual Health Care. 2003 Dec;15(Suppl 1):i49-59.

16. Wu A. Adverse drug events and near misses: Who's counting? American Journal of Medicine. 2000 Aug 1;109(2):166-8.

17. Hanlon J, Lindblad C, Hajjar E, McCarthy T. Update on drug-related problems in the elderly. American Journal Geriatric Pharmacotherapy. 2003 Sep;1(1):38-43.

18. Gurwitz J, Field T, Harrold L, Rothschild J, Debellis K, Seger A, et al. Incidence and preventability of adverse drug events among older persons in the ambulatory setting. Jama. 2003 Mar 5;289(9):1107-16.

References

of 767

19. Zhan C, Sangl J, Bierman A, Miller M, Friedman B, Wickizer S, et al. Potentially inappropriate medication use in the community-dwelling elderly: Findings from the 1996 Medical Expenditure Panel Survey. Journal of the American Medical Association. 2001;286(22):2823-9.

20. Merle L, Laroche M, Dantoine T, Charmes J. Predicting and preventing adverse drug reactions in the very old. Drugs & Aging. 2005;22(5):375-92.

21. Burgess CL, Holman CA, Satti AG. Adverse drug reactions in older Australians, 1981-2002. Medical Journal of Australia. 2005;182(6):267-70.

22. Hohl C, Dankoff J, Colacone A, Afilalo M. Polypharmacy, adverse drug-related events, and potential adverse drug interactions in elderly patients presenting to an emergency department. Annals of Emergency Medicine. 2001;38(6):666-71.

23. Chan M, Nicklason F, Vial J. Adverse drug events as a cause of hospital admission in the elderly. Intern Med J. 2001 2001 May-Jun;31(4):199-205.

24. Lowe C, Raynor D, Purvis J, Farrin A, Hudson J. Effects of a medicine review and education programme for older people in general practice. British Journal of Clinical Pharmacology. 2000;50(2):172-5.

25. Shah R. Drug Development and use in the elderly: search for the right dose and dosing regimen. British Journal of Clinical Pharmacology. 2004;58(5):452-69.

26. Vermeire E, Hearnshaw H, Van Royen P, Denekens J. Patient adherence to treatment: Three decades of research. A comprehensive review. Journal of Clinical Pharmacy & Therapeutics. 2001;26(5):331-42.

27. Rosenow III E. Patients' understanding of and compliance with medications: The sixth vital sign? Mayo Clinic Proceedings. 2005;80(8):983-7.

28. Beers MH, Sliwkowski J, Brooks J. Compliance with medication orders among the elderly after hospital discharge. Hospital Formulary. 1992;27(7):720-4.

29. Vik S, Maxwell C, Hogan D. Measurement, correlates, and health outcomes of medication adherence among seniors. Annals of Pharmacotherapy. 2004;38(2):303-12.

30. Kerzman H, Baron-Epel O, Toren O. What do discharged patients know about their medication? Patient Education & Counseling. 2005;56(3):276-82.

31. Kanjanarat P, Winterstein A, Johns T, Hatton R, Gonzalez-Rothi R, Segal R. Nature of preventable adverse drug events in hospitals: A literature review. American Journal of Health-System Pharmacy. 2003 Oct 2003;60(14):1750-9.

32. Brown BK, Earnhart J. Pharmacists and their effectiveness in ensuring the appropriateness of the chronic medication regimens of geriatric inpatients. Consultant Pharmacist. 2004;19(5):432-36.

33. von Laue N, Schwappach D, Koeck C. The epidemiology of preventable adverse drug events: a review of the literature. Wien Klin Wochenschr. 2003 Jul 15;115(12):407-15.

34. Tam V, Knowles S, Cornish P, Fine N, Marchesano R, Etchells E. Frequency, type and clinical importance of medication history errors at admission to hospital: a systematic review. CMAJ. 2005;173:510-5.

References

of 767

35. Cornish P, Knowles S, Marchesano R, Tam V, Shadowitz S, Juurlink D, et al. Unintended medication discrepancies at the time of hospital admission. Archives of Internal Medicine. 2005;165(4):424-9.

36. Stowasser D, McGuire T, Petrie G, Lauchlan R, Collins D. Information quality: a major consideration in the development of medication liaison services. Australian Journal of Hospital Pharmacy. 1997 Oct;27(5):362-6.

37. Cockayne N, Duguid M, Shenfield G. Health professionals rarely record history of complementary and alternative medicines. British Journal of Clinical Pharmacology. 2005;59(2):254-8.

38. Oborne C, Luzac M. Over-the-counter medicine use prior to and during hospitalization. Annals of Pharmacotherapy. 2005;39(2):268-73.

39. Kohn L, Corrigan J, Donaldson M, (Editors). To err is human: Building a safer health system. Committee on Quality of Health Care in America, Institute of Medicine. Washington, DC: National Academy Press; 1999.

40. Dean B, Schachter M, Vincent C, Barber N. Prescribing errors in hospital inpatients: Their incidence and clinical significance. Quality & Safety in Health Care. 2002;11(4):340-4.

41. Coombes I, Pillans P, Storie W, Radford J. Quality of medication ordering at a large teaching hospital. Australian Journal of Hospital Pharmacy. 2001;31(2):102-6.

42. Moore C, Wisnivesky J, Williams S, McGinn T. Medical errors related to discontinuity of care from an inpatient to an outpatient setting. Journal of General Internal Medicine. 2003;18(8):646-51.

43. Wheeler K, Crawford R, McAdams D, Benel S, Dunbar V, Caudle J, et al. Inpatient to outpatient transfer of care in urban patients with diabetes: patterns and determinants of immediate postdischarge follow-up. Archives of Internal Medicine. 2004;164(4):447-53.

44. Midlov P, Bergkvist A, Bondesson A, Eriksson T, Hoglund P. Medication errors when transferring elderly patients between primary health care and hospital care. Pharmacy World & Science. 2005;27(2):116-20.

45. Barnsteiner J. Medication reconciliation. Journal of Infusion Nursing. 2005;28(2 SUPPL):31-6.

46. Forster A, Murff H, Peterson J, Gandhi T, Bates D. Adverse drug events occurring following hospital discharge. Journal of General Internal Medicine. 2005;20(4):317-23.

47. Forster A, Murff H, Peterson J, Gandhi T, Bates D. The incidence and severity of adverse events affecting patients after discharge from the hospital. Annals of Internal Medicine. 2003 Feb 4;138(3):161-7.

48. Letrilliart L, Hanslik T, Biour M, Fagot J, Guiguet M, Flahault A. Postdischarge adverse drug reactions in primary care originating from hospital care in France: a nationwide prospective study. Drug Safety. 2001;24(10):781-92.

49. Coleman E, Smith J, Raha D, Min S-J. Post-hospital medication discrepancies: Prevalence and contributing factors. Archives of Internal Medicine. 2005;165:1842-7.

References

of 767

50. Coleman E, Berenson R. Lost in transition: Challenges and opportunities for improving the quality of transitional care. Annals of Internal Medicine. 2004;141(7):533-6.

51. Stuffken R, Egberts T. Discontinuities in drug use upon hospital discharge. Pharm World Sci. 2004 Oct;26(5):268-70.

52. Parke J, Howat C. Developing a form to enhance the continuum of medication care following hospital discharge. Journal of Pharmacy Practice & Research. 2003;33(4):268-71.

53. Gray S, Mahoney J, Blough D. Adverse drug events in elderly patients receiving home health services following hospital discharge. Annals of Pharmacotherapy. 1999 Nov;33(11):1147-53.

54. Duggan C. Good communication and discharge planning reduce medication errors. Pharmacy in Practice. 2004;14(3):81-3.

55. Tierney A, Worth A. Review: Readmission of elderly patients to hospital. Age & Ageing. 1995;24(2):163-6.

56. Benbassat J, Taragin M. Hospital readmissions as a measure of quality of health care: Advantages and limitations. Archives of Internal Medicine. 2000;160(8):1074-81.

57. Collins M, Peterson G. Follow-up of elderly medical patients discharged from the Royal Hobart Hospital. Society of Hospital Pharmacists of Australia, Federal Conference; 1999; Perth; 1999.

58. Frankl S, Breeling J, Goldman L. Preventability of emergent hospital readmission. American Journal of Medicine. 1991;90(6):667-74.

59. Coleman E. Falling through the cracks: Challenges and opportunities for improving transitional care for persons with continuous complex care needs. Journal of the American Geriatrics Society. 2003;51(4):549-55.

60. Australian Pharmaceutical Advisory Council. National guidelines to achieve the continuum of quality use of medicines between hospital and community. Canberra: Australian Pharmaceutical Advisory Council; 1998.

61. Mant A, Rotem W, Kehoe L, Kaye K. Compliance with guidelines for continuity of care in therapeutics from hospital to community. Medical Journal of Australia. 2001 Mar 19;174(6):277-80.

62. O'Leary K, Allinson Y, Stowasser D, Dowling H. Survey of implementation levels and pharmacy requirements for the Continuum of Quality Use of Medicines between Hospital and Community. Journal of Pharmacy Practice & Research. 2003;33(2):93-100.

63. Australian Pharmaceutical Advisory Council. Guiding principles to achieve continuity in medication management. Canberra: Australian Pharmaceutical Advisory Council; 2005.

64. Wilson S, Ruscoe W, Chapman M, Miller R. General practitioner-hospital communications: A review of discharge summaries. Journal of Quality in Clinical Practice. 2001;21(4):104-8.

References

of 767

65. Coleman A, Ricketts L, Teal S, Mitchell E, Silcock J, Wright DJ. An audit of hospital discharge and outpatient information from a primary care perspective. Journal of Social & Administrative Pharmacy. 2001;18(6):226-31.

66. Sexton J, Ho Y, Green C, Caldwell N. Ensuring seamless care at hospital discharge: a national survey. J Clin Pharm Ther. 2000 Oct;25(5):385-93.

67. Stowasser D, Collins D, Stowasser M. A randomised controlled trial of medication liaison services - Patient outcomes. Journal of Pharmacy Practice & Research. 2002;32(2):133-40.

68. Stowasser D, Collins D, Stowasser M. A randomised controlled trial of medication liaison services - Acceptance and use by health professionals. Journal of Pharmacy Practice & Research. 2002;32(3):221-6.

69. Van Walraven C, Rokosh E. What is necessary for high-quality discharge summaries? American Journal of Medical Quality. 1999 Jul/Aug 1999;14(4):160-9.

70. Norris C, Thomas V, Calvert P. An audit to evaluate the acceptability of a pharmacist electronically prescribing discharge medication and providing information to GPs. Pharmaceutical Journal. 2001;267(7178):857-59.

71. Paulino E, Bouvy M, Gastelurrutia M, Guerreiro M, Buurma H. Drug related problems identified by european community pharmacists in patients discharged from hospital. Pharm World Sci. 2004 June 2004;26:353-60.

72. Robinson M, Connolly J. Inpatients' satisfactions with clinical pharmacists' counselling. The Australian Journal of Hospital Pharmacy. 1999;29(3):162-5.

73. Paquette-Lamontagne N, McLean W, Besse L, Cusson J. Evaluation of a new integrated discharge prescription form. Annals of Pharmacotherapy. 2001;35(7-8):953-8.

74. Parker S, Peet S, McPherson A, Cannaby A, Abrams K, Baker R, et al. A systematic review of discharge arrangements for older people. Health Technology Assessment. 2002;6(4).

75. Al-Rashed S, Wright D, Roebuck N, Sunter W, Chrystyn H. The value of inpatient pharmaceutical counselling to elderly patients prior to discharge. British Journal of Clinical Pharmacology. 2002;54(6):657-64.

76. Louis-Simonet M, Kossovsky M, Sarasin F, Chopard P, Gabriel V, Perneger T, et al. Effects of a structured patient-centered discharge interview on patients' knowledge about their medications. American Journal of Medicine. 2004;117(8):563-8.

77. Fallis W, Scurrah D. Outpatient laparoscopic cholecystectomy: Home visit versus telephone follow-up. Canadian Journal of Surgery. 2001 2001;44(1):39-44.

78. Nelson JR. The importance of postdischarge telephone follow-up for hospitalists: a view from the trenches. Am J Med. 2001 Dec 21;111(9B):43S-4S.

79. Rosbe K, Jones D, Jalisi S, Bray M. Efficacy of postoperative follow-up telephone calls for patients who underwent adenotonsillectomy. Archives of Otolaryngology Head & Neck Surgery. 2000;126(6):718-21.

References

of 767

80. Dudas V, Bookwalter T, Kerr K, Pantilat S. The impact of follow-up telephone calls to patients after hospitalization. American Journal of Medicine. 2001 Dec 21;111(9B):26S-30S.

81. Lanser E. Fulfilling the promise: technology's role in improving patient safety. Healthc Exec. 2001 Sep-Oct;16(5):6-11.

82. Rosenthal M. The role of information in reducing medical error. Medical Journal of Australia. 2004;181(1):27-8.

83. Ros J, Cramers B, Prins J, Van Oirschot P, Keuning D, Berrens J. The contribution of informatics to medication safety: Business process redesign in the hospital. Pharmaceutisch Weekblad. 2002;137(44):1570-3.

84. Bates D. Using information technology to reduce rates of medication errors in hospitals. British Medical Journal. 2000 Mar 18;320(7237):788-91.

85. Bates D, Gawande A. Patient safety: Improving safety with information technology. New England Journal of Medicine. 2003 Jun 19;348(25):2526-34.

86. Calabretto J-P, Warren J, Darzanos K, Fry B. Building common ground for communication between pharmacists with an internet medicine cabinet. 35th Hawaii International Conference on System Sciences; 2002; Hawaii; 2002.

87. Calabretto J-P, Couper D, Mulley B, Nissen M, Siow S, Tuck J, et al. Agent support for patients and community pharmacists. 35th Hawaii International Conference on System Sciences; 2002; Hawaii; 2002.

88. Prady SL, Norris D, Lester JE, Hoch DB. Expanding the guidelines for electronic communication with patients: Application to a specific tool. Journal of the American Medical Informatics Association. 2001;8(4):344-8.

89. PatientGateway. [cited 11/1/2006]; Available from: http://www.patientgateway. org/ptgw/login.html

90. Taylor G, Peterson G. Controlled evaluation of home-based clinicalpharmacy follow-up of 'high-risk' patients discharged from hospital. Society of Hospital Pharmacists of Australia, Federal Conference; 2001; Hobart; 2001.

91. Naunton M, Peterson G. Evaluation of home-based follow-up of high-risk elderly patietns discharged from hospital. J Pharm Pract Res. 2003;33:176-82.

92. Australian Pharmaceutical Formulary and Handbook. 19 ed. Curtin, ACT: The Pharmaceutical Society of Australia; 2004.

93. Naunton M, Peterson G. Evaluation of home-based follow-up of high-risk elderly patients discharged from hospital. J Pharm Pract Res. 2003;33:176-82.

94. Woodgate J, Jones J, Ford D, Garett K, Morehen J. The Medication Alert Project. The Society of Hospital Pharmacists of Australia 27th Federal Conference; 2005 10-13 November; Brisbane Convention Centre, Brisbane; 2005.

95. Commonwealth Department of Health and Ageing. The National Strategy for Quality Use of Medicines. Canberra: Commonwealth Department of Health and Ageing; 2002.

96. Elkerton J, Taylor G. A global database to document clinical activities. Society of Hospital Pharmacists of Australia. Society of Hospital Pharmacists of Australia: Federal Conference; 1999 November 25-29; Perth; 1999.

References

of 767

97. Marcus P. Revised Security Architecture: The Pharmacy Guild of Australia; 2002 6 May 2002.

98. Naunton M, Peterson GM. Importance of hospital pharmacists' involvement in the provision of home-medication reviews following hospitalisation of "high-risk" patients. Society of Hospital Pharmacists of Australia: Federal Conference; 2003 November 13-16; Canberra, ACT; 2003.

99. Jackson S, Peterson G, Vial J, Jupe D, Marsden K. A randomised controlled trial of point of care oral anticoagulant management for patients commenced on warfarin. Society of Hospital Pharmacists of Australia, Federal Conference; 2003 November 13-16; Canberra ACT; 2003.

100. Jackson S, Peterson G, Vial J, Jupe D, Marsden K. Improving the initial management of anticoagulation in the transition from hospital to general practice care. Royal Australian College of General Practitioners (RACGP), 46th National Convention & Annual General Meeting; 2003 October 5-10; Hobart, Tasmania; 2003.

101. Jackson S, Peterson G, Vial J, Jupe D, Marsden K. Point of Care INR monitoring for patients commenced on warfarin and discharged to community care: A randomised controlled trial (Invited presentation). Ambulatory Care Australia (ACA) conference; 2003 July 24-25; Melbourne; 2003.

102. Hawthorne G, Richardson J, Day N. A Comparison of Five Multi Attribute Utility Instruments, Working Paper 140. Melbourne: Centre for Health Program Evaluation; 2003.

103. Boynton PM. Administering, analysing and reporting your questionnaire. Br Med J. 2004 5/6/2004;328(5 June):1372-5.

104. Peterson G, Tenni P. Identifying, prioritising and documenting drug-related problems. Aust Pharmacist. 2004;23:706-9.

105. ABS. Year Book Australia. Communications and information technology. Article - Use of information technology by older people. Canberra: Australian Bureau of Statistics; 2005 18 March 2005. Report No.: 1301.0-2005.

106. National E-Health Transition Authority. 2005 [cited 2005 16/11/2005]; Available from: http://www.nehta.gov.au/index.php?option=com_content&task=view&id=1 &Itemid=103

107. Briggs A, Gray AM. Handling uncertainty when performing economic evaluation of healthcare interventions. York; 1999.

108. Drummond MF, Stoddart, G.L., Torrance, G.W. Methods for the economic evaluation of health care programmes. Critical assessment of economic evaluation. Oxford: Oxford University Press; 1987. p. 18-38.

109. Freund DA, Dittus RS. Principles of pharmacoeconomic analysis of drug therapy. Pharmacoeconomics. 1992 Jan;1(1):20-31.

110. Crealey GE, Sturgess IK, McElnay JC, Hughes CM. Pharmaceutical care programmes for the elderly: economic issues. Pharmacoeconomics. 2003;21(7):455-65.

References

of 767

111. Willett MS, Bertch, K.E., Rich, D.S., Ereshefsky, L. Prospectus on the economic value of clinical pharmacy services. A position statement of the American College of Clinical Pharmacy. Pharmacotherapy. 1989;9:45-56.

112. McMullin ST, Hennenfent, J.A., Ritchie, D.J., Huey, W.Y., Lonergan, T.P., Schaiff, R.A.,Tonn, M.E., Bailey, T.C. A prospective, randomized trial to assess the cost impact of pharmacist-initiated interventions. Arch Intern Med. 1999;159:2306-9.

113. Singhal PK, Raisch, D.W., Gupchup, G.V. The impact of pharmaceutical services in community and ambulatory care settings: evidence and recommendations for future research. Ann Pharmacother. 1999;33:1336-55.

114. Plumridge RJ, Wojnar-Horton, R.E. A review of the pharmacoeconomics of pharmaceutical care. Pharmacoeconomics. 14. 1998:175-89.

115. [cited 11/1/2006]; Available from: http://www.shpa.org.au/docs/practicestandards.html

116. Peterson GM, Tenni PC, Hasan O, Liauw C, Reeve J, Turner P, et al. Report to the Pharmacy Guild of Australia (RFT 2003-2, Evaluation of Clinical Interventions in Community Pharmacies); 2005.

117. Sorensen L, Stokes JA, Purdie DM, Woodward M, Elliott R, Roberts MS. Medication reviews in the community: results of a randomized, controlled effectiveness trial. Br J Clin Pharmacol; 2004. p. 648-64.

118. Rigby K, Clark RB, Runciman WB. Adverse events in health care: Setting priorities based on economic evaluation. Journal of quality in clinical practice. 1999;19:7-12.

119. Murray MD, Callahan CM. Improving medication use for older adults: An integrated research agenda. Ann Intern Med. 2003 September 2, 2003;139(5_Part_2):425-9.

120. Landefeld CS, Beyth RJ. Anticoagulant-related bleeding - Clinical epidemiology, prediction, and prevention. Am J Med. 1993;95(3):315-28.

121. Beyth RJ. Hemorrhagic complications of oral anticoagulant therapy. Clin Geriatr Med. 2001 Feb;17(1):49-56.

122. Levine MN, Raskob G, Landefeld S, Kearon C. Hemorrhagic complications of anticoagulant treatment [Review]. Chest. 2001;119(1 Suppl S):108S-21S.

123. Chesebro JH, Wiebers DO, Holland AE, Bardsley WT, Litin SC, Meissner I, et al. Bleeding during antithrombotic therapy in patients with atrial fibrillation. Archives of Internal Medicine. 1996;156(4):409-16.

124. Fihn SD, McDonell M, Martin D, Henikoff J, Vermes D, Kent D, et al. Risk factors for complications of chronic anticoagulation - a multicenter study. Ann Intern Med. 1993;118(7):511-20.

125. Sebastian JL, Tresch DD. Use of oral anticoagulants in older patients. Drugs Aging. 2000 Jun;16(6):409-35.

126. Gage BF, Fihn SD, White RH. Warfarin therapy for an octogenarian who has atrial fibrillation. Ann Intern Med. 2001;134(6):465-74.

References

of 767

127. Beyth RJ, Shorr RI. Epidemiology of adverse drug reactions in the elderly by drug class. Drugs Aging. 1999 Mar;14(3):231-9.

128. Palareti G, Leali N, Coccheri S, Poggi M, Manotti C, Dangelo A, et al. Bleeding complications of oral anticoagulant treatment: An inception-cohort, prospective collaborative study (ISCOAT). Lancet. 1996;348(9025):423-8.

129. White RH, Beyth RJ, Zhou H, Romano PS. Major bleeding after hospitalization for deep-venous thrombosis. Am J Med. 1999 Nov;107(5):414-24.

130. Petitti DB, Strom BL, Melmon KL. Duration of warfarin anticoagulant therapy and the probabilities of recurrent thromboembolism and hemorrhage. Am J Med. 1986 Aug;81(2):255-9.

131. Hylek EM, Singer DE. Risk factors for intracranial hemorrhage in outpatients taking warfarin. Ann Intern Med. 1994 Jun 1;120(11):897-902.

132. Forster AJ, Murff HJ, Peterson JF, Gandhi TK, Bates DW. The incidence and severity of adverse events affecting patients after discharge from the hospital. Ann Intern Med. 2003 Feb 4;138(3):161-7.

133. Gray SL, Mahoney JE, Blough DK. Medication adherence in elderly patients receiving home health services following hospital discharge. Ann Pharmacother. 2001 May;35(5):539-45.

134. Franke CL, Koehler PJJ, Gorter JW, Kappelle LJ, Rinkel GJE, Tjeerdsma HC, et al. A randomized trial of anticoagulants versus aspirin after cerebral ischemia of presumed arterial origin. Ann Neurol. 1997 Dec;42(6):857-65.

135. Cannegieter SC, Rosendaal FR, Wintzen AR, Vandermeer FJM, Vandenbroucke JP, Briet E. Optimal oral anticoagulant therapy in patients with mechanical heart valves. N Engl J Med. 1995 Jul 6;333(1):11-7.

136. Mohr JP, Thompson JLP, Lazar RM, Levin B, Sacco RL, Furie KL, et al. A comparison of warfarin and aspirin for the prevention of recurrent ischemic stroke. N Engl J Med. 2001 November 15, 2001;345(20):1444-51.

137. Hylek EM, Skates SJ, Sheehan MA, Singer DE. An analysis of the lowest effective intensity of prophylactic anticoagulation for patients with nonrheumatic atrial fibrillation. N Engl J Med. 1996 Aug 22;335(8):540-6.

138. Blackshear JL, Baker VS, Rubino F, Safford R, Lane G, Flipse T, et al. Adjusted-dose warfarin versus low-intensity, fixed-dose warfarin plus aspirin for high-risk patients with atrial fibrillation - Stroke Prevention in Atrial Fibrillation III randomised clinical trial. Lancet. 1996;348(9028):633-8.

139. Oden A, Fahlen M. Oral anticoagulation and risk of death: A medical record linkage study. Br Med J. 2002 November 9, 2002;325(7372):1073-5.

140. Landefeld CS, Rosenblatt MW, Goldman L. Bleeding in outpatients treated with warfarin - Relation to the prothrombin time and important remediable lesions. Am J Med. 1989 Aug;87(2):153-9.

141. Landefeld CS, Anderson PA, Goodnough LT, Moir TW, Hom DL, Rosenblatt MW, et al. The Bleeding Severity Index - Validation and Comparison to Other Methods for Classifying Bleeding Complications of Medical Therapy. J Clin Epidemiol. 1989;42(8):711-8.

References

of 767

142. Kuijer PMM, Hutten BA, Prins MH, Buller HR. Prediction of the risk of bleeding during anticoagulant treatment for venous thromboembolism. Archives of Internal Medicine. 1999 Mar 8;159(5):457-60.

143. Gurwitz JH, Field TS, Harrold LR, Rothschild J, Debellis K, Seger AC, et al. Incidence and preventability of adverse drug events among older persons in the ambulatory setting. JAMA-J Am Med Assoc. 2003 Mar 5;289(9):1107-16.

144. Beyth RJ, Quinn L, Landefeld CS. A multicomponent intervention to prevent major bleeding complications in older patients receiving warfarin - A randomized, controlled trial. Ann Intern Med. 2000;133(9):687-95.

145. Ellis RF, Stephens MA, Sharp GB. Evaluation of a pharmacy-managed warfarin-monitoring service to coordinate inpatient and outpatient therapy. Am J Hosp Pharm. 1992;49(2):387-94.

146. Tang E, Lai CSM, Lee KKC, Wong RSM, Cheng G, Chan TYK. Relationship between patients' warfarin knowledge and anticoagulation control. Ann Pharmacother. 2003 Jan;37(1):34-9.

147. Jackson SL, Peterson GM, Vial JH, Jupe DML. Improving the outcomes of anticoagulation: an evaluation of home follow-up of warfarin initiation. J Intern Med. 2004 August 01, 2004;256(2):137-44.

148. Heidinger KS, Bernardo A, Taborski U, Muller-Berghaus G. Clinical outcome of self-management of oral anticoagulation in patients with atrial fibrillation or deep vein thrombosis. Thrombosis Research. 2000;98(4):287-93.

149. Nadar S, Begum N, Kaur B, Sandhu S, Lip GYH. Patients' understanding of anticoagulant therapy in a multiethnic population. J R Soc Med. 2003 Apr;96(4):175-9.

150. The Newcastle Anticoagulation Study Group. Effectiveness of anticoagulation among patients discharged from hospital on warfarin. Med J Aust. 1998;169(5):243-6.

151. Palareti G, Manotti C, Dangelo A, Pengo V, Erba N, Moia M, et al. Thrombotic events during oral anticoagulant treatment: Results of the inception-cohort, prospective, collaborative ISCOAT study. Thromb Haemost. 1997;78(6):1438-43.

152. Gullov AL, Koefoed BG, Petersen P. Bleeding during warfarin and aspirin therapy in patients with atrial fibrillation: the AFASAK 2 study. Atrial Fibrillation Aspirin and Anticoagulation. Archives of Internal Medicine. 1999;159(12):1322-8.

153. Australian Institute of Health and Welfare. Australian hospital statistics 2001-02.

154. Corbett NE, Peterson GM. Review of the initiation of anticoagulant therapy. J Clin Pharm Ther. 1995;20(4):221-4.

155. Balla JI, Jamieson WE. Improving the continuity of care between general practitioners and public hospitals. Med J Aust. 1994 Dec 5;161(11-12):656-9.

156. Pearson B, Skelly R, Wileman D, Masud T. Unplanned readmission to hospital: A comparison of the views of general practitioners and hospital staff. Age and Ageing. 2002;31(2):141-3.

157. Mant A, Rotem WC, Kehoe L, Kaye KI. Compliance with guidelines for continuity of care in therapeutics from hospital to community. Med J Aust. 2001 Mar 19;174(6):277-80.

References

of 767

158. Isaac DR, Gijsbers AJ, Wyman KT, Martyres RF, Garrow BA. The GP-hospital interface: Attitudes of general practitioners to tertiary teaching hospitals. Med J Aust. 1997 Jan 6;166(1):9-12.

159. Senst BL, Achusim LE, Genest RP, Cosentino LA, Ford CC, Little JA, et al. Practical approach to determining costs and frequency of adverse drug events in a health care network. Am J Health-Syst Pharm. 2001 Jun 15;58(12):1126-32.

160. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. Incidence of adverse drug events and potential adverse drug events - Implications for prevention. JAMA-J Am Med Assoc. 1995 Jul 5;274(1):29-34.

161. Kucukarslan SN, Peters M, Mlynarek M, Nafziger DA. Pharmacists on rounding teams reduce preventable adverse drug events in hospital general medicine units. Archives of Internal Medicine. 2003 September 22, 2003;163(17):2014-8.

162. McBride R. Stroke Prevention in Atrial Fibrillation study : Final results. Circulation. 1991 Aug;84(2):527-39.

163. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagen AFASAK study. Lancet. 1989;1(8631):175-9.

164. Ezekowitz MD, Bridgers SL, James KE, Carliner NH, Colling CL, Gornick CC, et al. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. N Engl J Med. 1992 Nov 12;327(20):1406-12.

165. Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C. Canadian Atrial Fibrillation Anticoagulation (CAFA) Study. J Am Coll Cardiol. 1991 Aug;18(2):349-55.

166. Kistler JP. The Risk of Embolic Stroke -- Another Piece of the Puzzle. N Engl J Med. 1994 December 1, 1994;331(22):1517-9.

167. Landefeld CS, McGuire E, Rosenblatt MW. A bleeding risk index for estimating the probability of major bleeding in hospitalized-patients starting anticoagulant therapy. Am J Med. 1990 Nov;89(5):569-78.

168. Hylek EM, Chang YC, Skates SJ, Hughes RA, Singer DE. Prospective study of the outcomes of ambulatory patients with excessive warfarin anticoagulation. Archives of Internal Medicine. 2000;160(11):1612-7.

169. Lafata JE, Martin SA, Kaatz S, Ward RE. The cost-effectiveness of different management strategies for patients on chronic warfarin therapy. Journal of General Internal Medicine. 2000;15(1):31-7.

170. Sidhu P, O'Kane HO. Self-managed anticoagulation: Results from a two-year prospective randomized trial with heart valve patients. Ann Thorac Surg. 2001 Nov;72(5):1523-7.

171. Cromheecke ME, Levi M, Colly LP, de Mol BJM, Prins MH, Hutten BA, et al. Oral anticoagulation self-management and management by a specialist anticoagulation clinic: A randomised cross-over comparison. Lancet. 2000 Jul 8;356(9224):97-102.

172. Fihn SD, McDonell MB, Vermes D, Henikoff JG, Martin DC, Callahan CM, et al. A computerized intervention to improve timing of outpatient follow-up - A

References

of 767

multicenter randomized trial in patients treated with warfarin. Journal of General Internal Medicine. 1994 Mar;9(3):131-9.

173. Fitzmaurice DA, Hobbs FD, Murray ET, Holder RL, Allan TF, Rose PE. Oral anticoagulation management in primary care with the use of computerized decision support and near-patient testing: a randomized, controlled trial. Archives of Internal Medicine. 2000;160(15):2343-8.

174. Sawicki PT. A structured teaching and self-management program for patients receiving oral anticoagulation - A randomized controlled trial. JAMA-J Am Med Assoc. 1999 Jan 13;281(2):145-50.

175. Dowell J, Jones A, Snadden D. Exploring medication use to seek concordance with 'non-adherent' patients: a qualitative study. Br J Gen Pract. 2002 Jan;52(474):24-32.

176. Lowe CJ, Raynor DK, Purvis J, Farrin A, Hudson J. Effects of a medicine review and education programme for older people in general practice. Br J Clin Pharmacol. 2000 Aug;50(2):172-5.

177. McCormack PME, Stinson JC, Hemeryck L, Feely J. Audit of an anticoagulant clinic: Doctor and patient knowledge. Irish Medical Journal. 1997 Aug-Sep;90(5):192-3.

178. Brigden ML, Kay C, Le A, Graydon C, McLeod B. Audit of the frequency and clinical response to excessive oral anticoagulation in an out-patient population. American Journal of Hematology. 1998 Sep;59(1):22-7.

179. Arnsten JH, Gelfand JM, Singer DE. Determinants of compliance with anticoagulation: A case-control study. Am J Med. 1997 Jul;103(1):11-7.

180. Howard RL, Avery AJ, Howard PD, Partridge M. Investigation into the reasons for preventable drug related admissions to a medical admissions unit: Observational study. Quality & Safety in Health Care. 2003 Aug;12(4):280-5.

181. Roddie A, Pollock AM. Therapeutic control of anticoagulation: How important is patient education? (letter). Clinical and Laboratory Haematology. 1988;10:109-12.

182. Yermiahu T, Arbelle JE, Shwartz D, Levy Y, Tractinsky N, Porath A. Quality assessment of oral anticoagulant treatment in the Beer-Sheba district. International Journal for Quality in Health Care. 2001;13(3):209-13.

183. Estrada CA, Hryniewicz MM, Higgs VB, Collins C, Byrd JC. Anticoagulant patient information material is written at high readability levels. Stroke. 2000 Dec;31(12):2966-70.

184. Peterson GM. Risks associated with hospitalisation of elderly patients. Aust Pharmacist. 2002;21:520-4.

185. Schoen C, Osborn R, Huynh P, Doty M, Zapert K, Peugh J, et al. Taking the pulse of health care systems: experiences of patients with health problems in six countries. Health Affairs. 2005 3/11/2005(3/11/2005):web exclusive.

186. Woodgate J, Jones J, Ford D, Garrett K, Morehen J. The medication alert project. Society of Hospital Pharmacists of Australia 27th Federal Conference; 2005; Brisbane, Queensland, Australia.; 2005.

References

of 767

187. Australian Pharmaceutical Advisory Council. National guidelines to achieve the continuum of quality use of medicines between the hospital and community. Canberra; 1998. Report No.: ISBN 0642272646.

188. Stewart S, Vandenbroek AJ, Pearson S, Horowitz JD. Prolonged beneficial effects of a home-based intervention on unplanned readmissions and mortality among patients with congestive heart failure. Arch Intern Med. 1999 Feb 8;159(3):257-61.

189. Stewart S, Pearson S, Horowitz JD. Effects of a home-based intervention among patients with congestive heart failure discharged from acute hospital care. Arch Intern Med. 1998:1067-72.

190. Stewart S, Pearson S, Luke C, Horowitz JD. Effects of home-based intervention on unplanned readmission and out-of-hospital deaths. J Am Geriatr Soc. 1998;46:174-80.

191. Spurling L, Gilbert A, Phillips P, Odgers C. A hospital medication discharge service: A novel approach utilising a community-based medication liaison pharmacaist. The Society of Hospital Pharmacacists of Australia Federal Conference; 2001; Hobart, Tasmania; 2001.

192. Sorensen L, Stokes J, Purdie D, Woodward M, Roberts M. Medication management at home: medication-related risk factors associated with poor health outcomes. Age & Ageing. 2005;34:626-32.

193. Sellors J, Kaczorowski J, Sellors C, Dolovich L, Woodward C, Willan A, et al. A Randomized Controlled Trial of a Pharmacist Consultation Program for Family Physcians and Their Elderly Patients. CMAJ. 2003;169(1):17-22.

194. Bonner C. CB. Medication Compliance Problems in General Practice: Detection and Intervention by Pharmacists and Doctors. Aust J Rural Health. 2002;10:33-8.

195. Allard J, Hebert R, Rioux M, Asselin J, Voyer L. Efficacy of a clinical medication review on the number of potentially inappropriate prescriptions prescribed for community-dwelling elderly people. CMAJ. 2001;164(9):1291-96.

196. Martin J, Zavala E. The Expanding Role of the Transplant Pharmacist in the Multidisciplinary Practice of Transplantation. Clin Transplant. 2004;18(Suppl 12):50-4.

197. Hawksworth GM, Corlett AJ, Wright DJ, Chrystyn H. Clinical pharmacy interventions by community pharmacists during the dispensing process. Br J Clin Pharmacol. 1999 Jun;47(6):695-700.

198. Avorn J, Soumerai SB. Improving drug-therapy decisions through educational outreach. A randomized controlled trial of academically based "detailing". N Engl J Med. 1983 Jun 16;308(24):1457-63.

199. Avorn J, Soumerai SB, Everitt DE, Ross-Degnan D, Beers MH, Sherman D, et al. A randomized trial of a program to reduce the use of psychoactive drugs in nursing homes. N Engl J Med. 1992 Jul 16;327(3):168-73.

200. Avorn J, Soumerai S. Improving Drug-Therapy Decisions Through Educational Outreach. A Randomized Controlled Trial of Academically Based "Detailing". N Engl J Med. 1983;308:1457-63.

References

of 767

201. Atthobari J, Monster T, de Jong P, de Jong-van den Berg L. The Effect of Hypertension and Hypercholesterolemia Screening with Subsequent Intervention Letter on the Use of Blood Pressure and Lipid Lowering Drugs. Br J Clin Pharmacol. 2003;57:328-36.

202. Roberts M, Stokes J, King M, Lynne T, Purdie D, PP G, et al. Outcomes of a Randomized Controlled Trial of a Clinical Pharmacy Intervention in 52 Nursing Homes. Br J Clin Pharmacol. 2001;51:257-65.

203. Hanlon J, Weinberger M, Samsa G, Schmader K, Uttech K, Lewis I, et al. A Randomized, Controlled Trial of a Clinical Pharmacist Intervention to Improve Inappropriate Prescribing in Elderly Outpatients With Polypharmacy. Am J Med. 1996;100:428-37.

204. Atthobari J, Monster TB, de Jong PE, de Jong-van den Berg LT. The effect of hypertension and hypercholesterolemia screening with subsequent intervention letter on the use of blood pressure and lipid lowering drugs. Br J Clin Pharmacol. 2004 Mar;57(3):328-36.

205. Granas G, Bates I. The Effect of Pharmaceutical Review of Repeat Prescriptions in General Practice. Int J Pharm Pract. 1999;7:264-75.

206. Krass I, Smith C. Impact of Medication Regimen Reviews Performed by Community Pharmacists for Ambulatory Patients Through Liaison with General Medical Practitioners. Int J Pharm Pract. 2000;8:111-20.

207. Leape LL, Cullen DJ, Clapp MD, Burdick E, Demonaco HJ, Erickson JI, et al. Pharmacist participation on physician rounds and adverse drug events in the intensive care unit. Jama. 1999 Jul 21;282(3):267-70.

208. Boudreau D, et al.,. Collaborative Care Model to Improve Outcomes in Major Depression. Ann Pharmacother. 2002;36:585-91.

209. Cline CMJ, Israelsson BYA, Willenheimer RB, Broms K, Erhardt LR. Cost effective management programme for heart failure reduces hospitalisation. Heart. 1998 November 1, 1998;80(5):442-6.

210. Krska J, Cromarty JA, Arris F, Jamieson D, Hansford D, Duffus PRS, et al. Pharmacist-led medication review in patients over 65: a randomized, controlled trial in primary care. Age Ageing. 2001 May 1, 2001;30(3):205-11.

211. Holland R, Lenaghan E, Harvey I, Smith R, Shepstone L, Lipp A, et al. Does home based medication review keep older people out of hospital? The HOMER randomised controlled trial. BMJ. 2005 February 5, 2005;330(7486):293-.

212. Rosenow E. Patients' understanding of and compliance with medications: The sixth vital sign? Mayo Clinic Proceedings. 2005;80(8):983-7.

213. Akici A, Kalaca S, Ugurlu M, Toklu H, Iskender E, Oktay S. Patient knowledge about drugs prescribed at primary healthcare facilities. Pharmacoepidemiology & Drug Safety. 2004;13(12):871-6.

214. Gilbert A, Roughead E, Beilby J, Mott K, Barratt J. Collaborative medication management services: Improving patient care. Med J Aust. 2002;177(4):189-92.

215. MacLaughlin E, Raehl C, Treadway A, Sterling T, Zoller D, Bond C. Assessing medication adherence in the elderly: which tools to use in clinical practice? Drugs Aging. 2005;22(3):231-55.

References

of 767

216. Harder S, Fischer P, Krause-Schafer M, Ostermann K, Helms G, Prinz H, et al. Structure and markers of appropriateness, quality and performance of drug treatment over a 1-year period after hospital discharge in a cohort of elderly patients with cardiovascular diseases from Germany. Eur J Clin Pharmacol. 2005 Jan;60(11):797-805.

217. Morisky D, Green L, Levine D. Concurrent and Predictive Validity of a Self-reported Measure of Medication Adherence. Medical Care. 1986;24(1):67-74.

218. Lim WK, Lambert SF, Gray LC. Effectiveness of case management and post-acute services in older people after hospital discharge. Med J Aust. 2003 Mar 17;178(6):262-6.

219. Furlong W, Barr RD, Feeny D, Yandow S. Patient-focused measures of functional health status and health-related quality of life in pediatric orthopedics: a case study in measurement selection. Health Qual Life Outcomes. 2005 Jan 12;3(1):3.

220. Furlong W, Barr RD, Feeny D, Yandow S. Patient-focused measures of functional health status and health-related quality of life in pediatric orthopedics: A case study in measurement selection. Health and Quality of Life Outcomes. 2005;3(3).

221. Hickey A, Barker M, McGee H, O'Boyle C. Measuring health-related quality of life in older patient populations - a review of current approaches. Pharmacoeconomics. 2005;23:971-93.

222. Hanlon J, Lindblad C, Gray S. Can clinical pharmacy services have a positive impact on drug-related problems and health outcomes in community-based older adults? American Journal Geriatric Pharmacotherapy. 2004 Mar;2(1):3-13.

223. Allard J, Hebert, R., Rioux, M., Asselin, J., Voyer, L. Efficacy of a clinical medication review on the number of potentially inappropriate prescriptions prescribed for community-dwelling elderly people. CMAJ. 2001;164(9):1291-96.

224. Williams E, Fitton F. Factors affecting early unplanned readmission of elderly patients to hospital. BMJ. 1988 Sep 24;297(6651):784-7.

facilitating quality use of medicines (qum) between hospital and community...

Documents