infusion safety: addressing harm with high-risk drug administration

Infusion Safety:

Addressing Harm with High-Risk DrugAdministration

Proceedings fromThe Third Conference

ALARIS® Center for Medication Safety and Clinical ImprovementNovember 7, 2003, San Diego, CA

Philip J. Schneider, MS, FASHP, Editor

Medication Safety and Harm

Improving Medication Safety

Assessing IV Medication Harm

Using an IV Medication Safety System

Roundtable Discussion

Conference Report Published by:www.alarismed.com/alariscenter

The ALARIS® Center for Medication Safety and Clinical ImprovementSan Diego, CA

2004

Supplement to Hospitals & Health Networks©2003-2004 ALARISMedical Systems, Inc. All rights reserved. 03/04

Inter-Professional Conference on InfusionSafety Technology:

Addressing Harm with High-Risk DrugAdministration

The third invitational conference at the ALARIS®

Center for Medication Safety and Clinical Improvement

in San Diego, held on November 7, 2003, brought

together a distinguished faculty from clinical practice,

academia, organizations and government. Philip J.

Schneider, MS, FASHP, Director of the Latiolais Leadership

Program and Clinical Professor at The Ohio State

University, chaired the conference and moderated the

roundtable discussion. Nationally recognized experts

from different health professions focused on the use of

an intravenous medication safety system that addresses

harm with high-risk intravenous drug administration and

provides actionable data for best practice improvements.

EDITORIALP2 Addressing Harm with High-Risk Medications

Philip J. Schneider, MS, FASHP, The Ohio State University

MEDICATION SAFETY AND HARM

P4 Conflicting Priorities for Addressing Medication SafetyMay Adra, PharmD, Tufts-New England Medical Center

P7 Focusing on Harm to Set the Patient and Medication Safety AgendaPeter Provonost, MD, PhD, The Johns Hopkins Medical Institutions

P10 Using Trigger Tool to Detect Potential Harm in Medication ManagementTerri Simmonds, RN, Institute for Healthcare Improvement

IMPROVING MEDICATION SAFETY

P12 Complexity, Standardization, and Medication SafetySteve Meisel, PharmD, Fairview Health Services

P15 Improving Drug Order Entry TurnaroundNancy Pratt, MSN, Sharp HealthCare

P18 Variability in IV Therapy: A 65-hospital Analysis of IV Best PracticesTim Vanderveen, PharmD, MS, ALARIS Medical Systems

P21 The JCAHO Focus on Limiting Concentrations of High Alert Medications: Opportunities and Potential Impact Christopher L. Shaffer, PharmD, BCPS, The Nebraska Medical Center

P24 Intravenous Medication Safety ErrorsDavid Bates, MD, Brigham and Women's Hospital

ASSESSING IV MEDICATION HARM

P26 Potential for Harm in IV System Programming: A 10-hospital Analysis Rick Crass, PharmD, ALARIS Medical Systems

P29 IV Medication Harm Index: Results of a National Consensus Conference Jacqueline Sullivan, PhD, RN, Hospital of the University of Pennsylvania

USING AN IV MEDICATION SAFETY SYSTEM

P32 Using IV Medication Safety System Logs: A New Tool for Identifying Averted Harm Ray Maddox, PharmD, St Joseph's/ Candler Health System

P35 A Medical Center's Experience Using an IV Medication Safety System to Manage Medication Administration Glenn Billman, MD, Children's Hospital

ROUNDTABLE

P38 IV Therapy: A Higher Priority Safety Issue

P38 IV Medication Safety System: Improve and Measure IV Safety

P39 Safety Culture: Antecedent to New Technology

P39 Infrastructure and Strategies to Improve Safety

P40 Business Case

CONTENTS

Infusion Safety Conference 1

Sharpening Our FocusMedication safety has been at

center stage for at least four yearssince the publication of “To Err isHuman” by the Institute ofMedicine.1 Much of the informationused in preparing this report camefrom the Harvard Medical PracticeStudy that was published in 1991.2

The report indicated that the med-ical treatment most commonlyassociated with adverse medicalevents was medication.3 Even beforethat, in the early 1960's, pharmacistshad found that approximately one inten doses of medication adminis-tered deviated from the prescribedtherapy. 4 While some improvementshave been documented in the past40 years, recent evidence suggeststhese have not been sustained.5 Whyis this?

Perhaps one of the reasons isthat earlier efforts focused on errors,not harm. Many of the deviationsfrom prescribed therapy, while com-mon, did not result in harm topatients. The unit dose and intra-venous (IV) admixture systems cre-ated to reduce these deviations wereapplied to all medications, not justto those that had the greatestpotential for harm. Extensive andexpensive systems were needed toprepare doses in the pharmacy thatwere ready to administer at the bed-side and distributed in quantities of24 hours or less. These systems wereeasiest to apply to low-risk medica-tions used in low-risk patients.Preparing sterile doses that have

less standardization and are neededmore immediately were difficult tofit into unit dose and IV admixturesystems– and are increasingly so.The increase in the acuity of care inmost hospitals has resulted in aneed to have medications morequickly accessible at the point ofcare. This has resulted in the prolif-eration of point of care dispensingcabinets and IV systems that can beactivated at the bedside.

This results in a second reasonfor an erosion in medication safe-ty–technology. While technology iswidely advocated to reduce adversedrug events,1 there are often unin-tended and unexpected new prob-lems ("side effects") that emerge.6

One of these "side effects" is theelimination of double checks withinthe medication use system whenmedications are so available inpatient care areas. Other "sideeffects" include the need to calcu-late and prepare the dose at thebedside– a task performed by healthcare professionals who do not ordi-narily perform this function, andare doing so in a hectic environ-ment.

The new JCAHO patient safetystandards require that health careprofessionals focus more attentionon managing risk, not just waitingfor errors to happen. One way to dothis is through the use of FailureMode and Effects Analysis (FMEA).This method focuses on threeaspects of a process: likelihood offailure, chances of failure resulting

in harm, and the likelihood of thefailure being undetected. Commonfailures that are likely to cause seri-ous harm and not be detected andprevented are the ones on whichimprovement efforts shouldfocus. Hospitals are required to doat least one FMEA per year, but wecan apply the thinking behind thistechnique to our everyday thinking.What failures are common? Whichones seem to have the potential tocause the greatest harm? How canfailures in a process be made morevisible, so that detection and correc-tion are increased?

Focusing on both high-riskmedications and high-risk methodsof administering these medicationscan narrow the scope of work.Cohen has an excellent chaptertitled "High-alert medications: safe-guarding against errors" in his textMedication Errors.7 Sixteen medica-tions or drug categories are listed,14 of which can be or are adminis-tered by the IV route. Kaushaul, et alfound that the IV route of adminis-tration was the most common inmedication errors detected in pedi-atric inpatients.8 In their annualreport, USP states that "the intra-venous route of administrationoften results in the most seriousmedication error outcomes," basedon the reports submitted to MED-MARXSM in the year 2002.9 We do notneed a formal FMEA to know that IVdrug administration is a high-riskarea of medication use, and needsmore of our attention.

EDITORIAL

Addressing Harm with High-Risk MedicationsPhilip J. Schneider, MS, FASHP, Director of Latiolais Leadership Program, Clinical Professor, The Ohio State University, Columbus, OH

2 Infusion Safety Conference

The purpose of this conferencewas to address harm that resultsfrom high-risk drug administration.The content of the meeting andthese proceedings outline theimperative to do this, the methodsfor identifying near misses and harmresulting from medications, emerg-ing standards for improving medica-tion use safety, and the role of newtechnology to improve the safety ofIV medication administration intoday's highly complex patient careenvironment. Participants reachedconsensus on the following points:

1. IV therapy needs to be a high-er priority as a patient safety issue.

2. IV medication safety system isan effective way to improve andmeasure IV drug administrationsafety, with both error-preventionand process-improvement data-collection capabilities.

3. Improving the safety culture(especially with CEOs and nursing) inan organization is an importantantecedent to adopting new tech-nology such as smart pumps.

4. The organizational infrastruc-ture to improve safety differs inmany organizations, and differentstrategies to improve medicationuse safety may be needed.

5. The business case for investingin improvements in medication usesafety may be intuitive in somecases and needed for quick deci-sions, but evidence to support theinvestment remains the ideal.

References1. Kohn LT, Corrigan JM and Donaldson MS,

eds. To Err Is Human: Building a Safer HealthSystem. Institute of Medicine. Washington DC: National Academy Press; 1999.

2. Brennan TA, LL Leape, NM Laird, et al. Incidence of adverse events and negligencein hospitalized patients : results of the Harvard Medical Practice Study I. New Eng J Med. 1991;327:370-6.

3. Leape LL, TA Brennan, NM Laird, et al. The nature of adverse drug events in hospital-ized patients : results of the Harvard Medical Practice Study II. New Eng J Med. 1991;327:377-84.

4. Allen EA and KN Barker. Fundamentals of medication error research. Am J Hosp Pharm , 1990;47:555-71.

5. Barker KN, EA Flynn, GA Pepper, et al. Medication errors observed in 36 health care facilities. Arch Intern Med.2002;162:1897-1903.

6. Patterson ES RI Cook, and ML Render. Improving medication safety by identifyingside effects from introducing bar coding inmedication administration. J Am Med Inform Assoc. 2002;9:540-53.

7. Cohen MR and CM Kilo. High-alert medica-tions: safeguarding against errors. In: Cohen, MR, ed. Medication Errors.Washington DC: American Pharmaceutical Association; 1999.

8. Kaushal R, DW Bates, C Landrigan, et al. Medication errors and adverse drug eventsin pediatric inpatients. JAMA. 2001;285:2114-20.

9. Hicks RW, DD Cousins, R Williams. Summary of information submitted to MEDMARXSM in the year 2002. The Quest for Quality. Rockville, MD: UPS Center for the Advancement of Patient Safety, 2003.

Addressing Harm with High-Risk Medications


Conflicting Priorities forAddressing Medication Safety

May Adra, BS, PharmD, Director of Drug Information/Medication Safety Coordinator,Department of Pharmacy, Tufts-New England Medical Center, Boston , MA

Patient safety has been a focus ofindividual practitioners and institu-tions. The release of the Institute of

Medicine's (IOM) first report,1 "To Err IsHuman: Building a Safer Health System,"and the media coverage of victims ofmedical errors have thrust patient safetyinto the public domain.

The IOM report estimated that deathsresulting from medical errors range from44,000 to 98,000 per year, placing medicalerrors as the 8th leading cause of deathsin the United States. Although the accura-cy of these estimates can be debated, theimpact of the IOM report in mobilizingnational patient safety efforts must beacknowledged.

Many organizations have publishedpatient safety mandates. This overviewidentifies where those mandates can beconsolidated, where they differ, and Tufts-New England Medical Center's (T-NEMC)plans to accomplish recommendedimprovements in patient safety.

Patient SafetyOrganizations

In response to the IOM report, manyorganizations were directed to formulate

action plans to improve patient safety.Federal, state, accreditation, professional,and private organizations directed theirfocus to patient safety. The Agency forHealthcare Research and Quality (AHRQ)was responsible for evaluating, prioritiz-ing, and disseminating information aboutpatient safety.2 The AHRQ commissionedthe Evidence-Based Center at theUniversity of California in San Franciscoto review the evidence. In total, 79patient safety practices were evaluatedand rated based on the potential impactof the practice, the strength of the evi-dence, and barriers to implementation.Eleven of the 79 practices were highlyrated; however, none of these practicesincluded computerized prescriber orderentry (CPOE), the use of automated dis-pensing machines, or bar coding. Criticismof the AHRQ recommendations includedits focus on individual practices versussystem-related practices, its focus on theprovision of optimal care versus the pre-vention of adverse events, and its failureto highly rate safety practices that wererecommended by other patient safetyorganizations.

At the state level, the MassachusettsHospital Association (MHA), in collabora-

tion with the Massachusetts Coalition ofthe Prevention of Medication Errors,has published several Best PracticeRecommendations.3 The first list of BestPractice Recommendations was publishedin 1999,3 before the release of the firstIOM report. These recommendationsfocused on hospital settings and includedtwo basic principles, eight short-term rec-ommendations, and four long-term rec-ommendations. The basic principles of therecommendations were the need for cre-ating a systems-oriented approach topatient safety and promoting non-puni-tive medication error reporting. The short-term recommendations consisted mostlyof operational and educational strategiessuch as developing special procedures forhigh-risk drugs and educating both clini-cians and patients about medication use.MHA's long-term recommendations weretechnology oriented and included imple-menting CPOE, adopting electronic med-ication administration records (eMARs),and initiating bar coding at point of care.

The 2003 Best Practices targetedreducing medication errors due to com-munication failures by reconciling med-ications.4 Reconciling medications is theprocess of comparing and resolving dis-crepancies between the patient's currentmedication list and that at time of admis-sion, transfer or discharge. MHA is in theprocess of developing Best PracticeRecommendations for the ambulatorysetting. These recommendations willfocus on obtaining a complete and accu-rate medical and medication history,reducing prescribing errors by encourag-ing the use of clinical decision support,and improving patient-provider commu-nication.

The Joint Commission on Accreditationof Healthcare Organizations (JCAHO) hasannounced its 2004 National PatientSafety Goals. 5 The medication-related

Infusion Safety Conference

Key Points:

• Many organizations, including the Agency for Healthcare Research and Quality, the Massachusetts Hospital Association, the Joint Commission on the Accreditation of Healthcare Organizations, the American Society of Health-System Pharmacists, and the Leapfrog Group have published patient safety mandates.

• Recommendations are often consistent but sometimes conflict.

• A plan can be designed to achieve dramatic improvements in patientsafety, despite the conflicting recommendations from various organizations.

PROCEEDINGS


goals include improving communicationamong healthcare providers by imple-menting a "read-back" process for takingverbal and telephone orders and by stan-dardizing the abbreviations and symbolsused throughout the institution. A secondmedication-related goal is improving thesafety of high-alert medications byremoving concentrated electrolytes frompatient care units and by standardizingand limiting the number of drug concen-trations available in the institution.

Professional organizations, includingthe American Society of Health-SystemPharmacists (ASHP), have also focusedattention on medication safety.6 ASHP hascreated a Center on Patient Safety withthe goal of fostering "fail-safe medicationuse in health systems through the leader-ship of pharmacists." It developed anextensive list of responsibilities of the"medication-use safety coordinator" andhas been active in supporting the adop-tion of bar coding and CPOE.

The Leapfrog Group, which alsoresponded to the IOM report, consists ofmore than 145 public and private organi-zations that provide health care benefitsto their employees. It will be evaluatinghospitals that provide care to theiremployees and referring patients to hos-pitals based on performance in threeareas: CPOE implementation, intensivecare unit staffing with intensivists, andlocalizing high-risk procedures to high-volume centers.7

There are many other groups thathave published patient safety mandates.Health care institutions and providers areunder t r emendous p res su re torespond to these mandates.

ConsolidatingRecommendations

Consolidating the recommendationsof many patient safety organizations canbe accomplished by grouping these rec-ommendations into four categories: 1)creating a culture of safety and providingeducation about medication safety; 2)building the necessary infrastructure tosupport patient safety; 3) implementingpractices that reduce medication errors;and 4) adopting technological solutionsfor improving patient safety (Table 1). Asrecommended by IOM and MHA, there is aneed to foster a nonpunitive approach formedication error reporting and to en-courage an open dialogue about the caus-es of errors. A multidisciplinary medica-tion safety committee that is responsiblefor prioritizing and coordinating safetypractices is one component of the infra-structure that needs to be built to supportpatient safety initiatives. Practices such asthe use of safety checks for high-alertmedication, a pharmacy-based intra-venous (IV) admixture program, and edu-cation of both patients and cliniciansabout medications need to be adopted.

Technology needs to be incorporated intoeach step in the medication-use process.

Conflicting PrioritiesDespite attempts to consolidate the

recommendations of the increasing num-ber of patient safety groups, some differand conflicting priorities remain (Table 2).Although the patient safety groups arewell meaning in their intentions, the basisfor their recommendations can differ.AHRQ uses scientific evidence as the basisfor its recommendations, whereas JCAHOand MHA use intuition as the basis fortheirs. Some recommendations highlightreducing errors of omission such as fail-ure to provide peri-operative antimicro-bial prophylaxis, whereas others focus onreducing errors of commission such asadministering the wrong drug. JCAHO andMHA center their emphasis on opera-tional issues, whereas others highlightclinical issues. However, the conflictbetween these groups' mandates andinstitutional goals is paramount.Reconciling these organizations' recom-mendations with institutional goals,specifically given the limited human andfinancial resources, is important.

AccomplishingInstitutional Change

Given the number of recommenda-tions published by patient safety organi-zations, healthcare institutions need to

Conflicting Priorities for Addressing Medication Safety

Infusion Safety 5

TABLE 1.

Consolidating MedicationSafety Goals

• Culture and Education• Infrastructure• Practices• Technology

TABLE 2.

Conflicting Priorities• Evidence-based versus intuition-based• Differing foci

— Errors of omission versus commission

— Operational versus clinical issues

• Patient safety recommendations versus institutional goals— Allocation of resources

TABLE 3.Accomplishing

Institutional Change• Build the necessary infrastructure • Gather institutional-specific data • Identify and prioritize patient safety

goals• Implement patient safety initiatives

— Monitor and sustain progress— Learn from both successes and failures— Avoid moving in too many directions

at once

identify where to begin and how to pro-ceed in implementing medication errorreduction strategies (Table 3). A health-care organization needs to build thenecessary infrastructure by obtaining themoral and financial support of its execu-tive group. This infrastructure oftenincludes a medication safety team con-sisting of representatives from manage-ment, clinical departments, laboratory,and information systems to identify andprioritize medication safety goals.Gathering institutional-specific data isalso necessary since practitioners aremore likely to respond to "local" ratherthan national data. Patient safety goalsneed to be identified, prioritized, andimplemented based on both institutionaland national information. Once improve-ment strategies are implemented,progress needs to be monitored. Bothsuccesses and failures need to be re-viewed and analyzed.

Tufts-NEMC's ApproachThe Medication Safety and Quality

Committee at T-NEMC used MHA's 1999Best Practice recommendations as thebasis for patient safety initiatives. An ini-tial goal was to improve medication errorreporting to identify institution-specificmedication error patterns. MHA's short-term recommendations, including stan-dardizing the prescribing and concentra-tions of IV heparin, were adopted. Thehospital's executive group has made afinancial commitment to support theimplementation of CPOE.

For 2003-2004, several new initia-tives have been identified, includingreducing medication errors that reach thepatient, meeting MHA's 2003 BestPractice Recommendations, and meetingJCAHO's 2004 National Patient SafetyGoals.

Drastic improvements in patientsafety need to occur and can be accom-plished despite the excessive number ofpatient safety recommendations. Eachhealth care organization should be famil-iar with these recommendations anddevelop a plan based on those that bestmatch their individual institutional needsand the needs of their patients.



2. Agency for Healthcare Research and Quality. Making health care safer: a criticalanalysis of patient safety practices. www.ahcpr.gov/clinic/ptsafety (accessed 2003 Nov 1).

3. The Massachusetts Coalition for the Prevention of Medication Errors. MHA bestpractice recommendations to reduce med-icationerrors. http://www.macoalition.org/ documents/ Best_Practice_Medication_Errors.pdf (accessed 2003 Nov 1).

4. The Massachusetts Coalition for the Prevention of Medication Errors. Reconciling medication recommended practices. http://www.macoalition.org/pub-lications.shtml (accessed 2003 Nov 1).

5. Joint Commission on Accreditation of Healthcare Organizations. 2004 national patient safety goals. http://www.jcaho.org/accredited+organizations/patient+safety/04+npsg/ 04_npsg.htm (accessed 2003 Nov 1)

6. The American Society of Health-System Pharmacists. Patient Safety Resource Center. http://www.ashp.org/patientsafety/index.cfm?cfid=21981551& CFToken=96558264 (accessed 2003 Nov 1).

7. The Leapfrog Group. Patient safety. www.leapfroggroup.org/safety.htm (accessed 2003 Nov 1).

Conflicting Priorities for Addressing Medication Safety



Despite all the discussion and pro-motion of patient safety in recentyears, frustration continues that

relative little progress had been made inimproving patient safety both locally andnationally. At The Johns Hopkins Hospital,many efforts have been initiated toimprove patient safety, but they were notstructured in a way that would allow theorganization to cross the "qualitychasm." Lessons learned from theseexperiences showed that improvement inteamwork among health care profes-sions was needed.

Frontline staff has to assume respon-sibility for quality and safety, and con-tribute to these aspects of care at thebedside every day. Policies and procedurescan be written to support these healthcare professionals, but safety and qualityare not solely the responsibility of aprocess improvement or risk managementdepartment. The challenge is to makethat concept of improved teamworkreal for frontline staff—not just pro-vide team training that is divorcedfrom bedside care.

Creating a Culture ofSafety: An Eight-stepProgram

The culture of an organizationchanges incrementally, not all at once.Large-scale efforts intended to makerapid changes in culture uniformly fail.Rather, one small unit at a time must bechanged— one heart, one mind —and thechange grows. Time and effort arerequired. The cultural changes must beconcrete, or they will not be practicedwhen a team returns to the frontline. Forinterventions such as team training orcrew resource management to work, theyneed to be goal-directed, so that peopleactually practice these tools in their dailywork, with a focus on the patient and onsafety. The results of these interventionsneed to be documented.

At Johns Hopkins, an eight-step pro-gram has been established that is evolvingdaily and is achieving dramatic results(Table 1).

1. Measure the culture of safety. Thefirst step is to measure the culture of

safety. At Johns Hopkins, a survey instru-ment was developed to assess the cultureof safety.

2. Educate staff. Next, the staff iseducated about the "science of safety." Togain understanding of the need to focuson improving systems, a story is toldabout a child who pulls weeds by pullingoff the tops, rather than by pulling out theroots. This does not eliminate the weeds.To improve performance, a new systemof work is needed. Helping the staffunderstand systems is important, becausethis type of thinking has not beenincluded in the education of health careprofessionals.

3. Identify safety concerns. A one-page survey asks, "Who was the lastpatient that would have been harmed byan error that you prevented?" This tool isspecifically designed to identify heroes—those who work hard daily in healthcareenvironments to improve safety. The nextquestion is, "How might the next patientbe harmed by an error, and what can wedo to prevent it?" Results are then sum-marized. Experience has shown that oftenabout a year's worth of effort can bedirected by the results of the survey.Incident reporting systems are important,but in healthcare the frontline staffknows of many things that are broken. Itis not necessary to wait for a rare incidentto emerge in an incident reporting sys-tem. Talking to caregivers will reveal whatneeds to be done.

4. Assign executives. The fourth stepis to assign a senior executive to a servicearea in what is called "adopt a work unit."Executives meet monthly with the staff oftheir units, review what people have saidis broken, and help decide what the staffwants to fix. The executives make sureresources are available to fix what the


Key Points:

• To improve patient safety, it is important to improve teamwork among health care professionals.

• Frontline staff has to assume responsibility for quality and safety, be educated to systems thinking, and commit together that "harm is untenable."

• Interventions require creating and using practical tools that are goal-directed.

• Hospital leadership needs to be actively involved.

• Results need to be documented and stories shared.

PROCEEDINGS

Focusing on Harm to Set the Patient andMedication Safety Agenda

Peter Provonost, MD, PhD, Associate Professor of Anesthesiology and Critical Care Medicine,The Johns Hopkins Medical Institutions, Baltimore, MD


staff has identified, and follow up everymonth. This experience has been verysuccessful.

5. Identify priority areas. Originallythe Veterans Administration model foridentifying potential failure modes wastried, using a grid with a severity scoreand the probability of occurrence to iden-tify priority areas for improvement.At Johns Hopkins, this approach wasnot successful, because reliability of theassessments was low. Instead, the unitmanagers, nursing leaders, physicianleaders and executives now do thefollowing:• Evaluate what the staff has said are

hazards to patient safety.• Identify three changes that do not

require any marginal resources and which can be implemented tomorrow, and focus on these changes.

• Select two or three changes that do require resources. These suggested changes are directed through the safety infrastructure, so that the staff knows what needs to be improved. Efforts can then be priority ranked so that the staff can be advocates for getting the resources needed to imple-ment the changes.

Most of the resource requests arequite small and can be implemented with-in existing budgets. In many cases, thefunds have already been allocated and

were just was not being used. 6. Implement improvements. Once

staff members decide what they want todo, they implement changes to makeimprovements. Every one of the changesneeds a way to measure performance. Inthe Johns Hopkins culture, a suggestionfor change will not be supported unlessimprovements can be documented.

7. Share stories. The seventh step isto share stories that demonstrate whathas been learned and to help spread orga-nizational learning.

8. Remeasure. The final step is to re-measure the culture to see if it hasimproved.

Team Factors: A PracticalFramework

Improving teamwork provides enor-mous leverage in improving patient safe-ty; however, much effort is required tomake teamwork really effective. One rea-son is that healthcare professionals arenot taught to work collaboratively butrather to work independently. Safety willonly improve when clinicians learn how tointeract with each other so that everyoneis free to communicate and raise con-cerns.

When teams are asked to really thinkabout preventing mistakes, they are givena very practical framework that includes

three components (Table 2): • First, create a culture of safety. • Second, eliminate complexity,

because the more steps there are in aprocess, the greater the likelihood it willfail (see Meisel, these Proceedings).

• Third, create independent redun-dancies such as checklists for processes.Make certain that an independent check-list is used to ensure that all necessarysteps have been taken before a procedureis begun. Tragic examples of errors result-ing in harm that are reported by the pressare often instances where there was notan independent check.

To make the idea of culture very realand not just jargon, care teams are askedto publicly commit together to the con-cept that "harm is untenable." Saying thateveryone's efforts are focused on patientsafety galvanizes the team. One of theprinciples of negotiation is to identifysome common areas to which everyonecan agree. In health care, there is nodoubt that such an area is patient safety.That becomes the centering ground foreveryone to focus their efforts. The teamthen works on assertiveness, teamwork,and situational awareness.

"Goal Sheet": Example ofSuccessful Tool

An example of a successful commu-nication tools has been the Goal Sheet.Communication has been identified as aproblem in more than 90% of responsesof units surveyed. Nurses do not knowwhat the physicians are doing, and physi-cians do not know what their residentsare doing. A one-page Goal Sheet wascreated that asks, "Do you understandwhat needs to happen for this patienttoday? What is this patient's safety risk,and how might we reduce it? What is yourcommunication and care plan?"

To measure the results, before theproject began caregivers were asked, "Doyou understand the goals for this patient

Focusing on Harm to Set the Patient and Medication Safety Agenda

TABLE 1.

Creating a Culture of Safety1. Measure the culture of safety 2. Educate staff 3. Identify safety concerns using

uncomplicated surveys4. Assign executives to "adopt a

work unit" 5. Identify priority areas6. Implement improvements 7. Share stories 8. Remeasure

TABLE 2.

Team Factors: A PracticalFramework

• First, create a culture of safety

• Second, eliminate complexity• Third, create independent

redundancies such as checklistsfor processes

for the day, and do you understand whatwork needs to be done to accomplishthese goals?" Using a five-point scale, lessthan 10% of the physicians and nursesknew the work plan for the day.

Physicians were making "provider-centered rounds"— talking about evi-dence-based medicine and pharmacologyand physiology but not asking, "Whatwork is needed to get this patient to thenext level of care?" Once this was docu-mented on the Goal Sheet the staff clear-ly understood the goals of therapy.Length of stay has decreased to an almostunprecedented level of one day in a surgi-cal ICU. This was accomplished by reduc-ing complications and by making sure allstaff members are working as a team.

Case Study: TransferOrders and MedicationErrors

These interventions were used toimprove a common type of medicationerror. The nurses had identified that med-ication errors were often associated withtransfer orders. An information-gatheringsheet was created to evaluate this prob-lem. A nurse leader in a patient care areawas asked to review charts to answerthree simple questions:

• Are the medications being adminis-tered after transfer the same as they werein the ICU?

• Are the patient's allergies listed thesame in both areas?

• Did the patient start their homemedications?

If any of these questions wasanswered "No," the nurses were instruct-ed to ask the physicians, "Did you intendto make this change?" The definition of adefect was quite simply, "Did the physi-cian change the order?" after having beenasked.

In the first two weeks after this proj-ect began, 94% of orders were changed.Medication reconciliation was made partof routine discharge and now is done forevery patient. Auditing error rates hasshown that in three ICUs this type ofdefect has virtually been eliminated bychanging the way transfer orders arehandled.

An additional result is that in twoICUs the nursing turnover decreased. Theorganization is doing a lot to improvenurse retention, which is a strategic issuefor every organization in this country.With this project, there has been animprovement in nurse retention.

ConclusionThe Johns Hopkins Hospital has

developed a practical way to improvepatient safety that makes cultural changegoal-directed. Leadership support ensuresthat new behaviors are supported andrealized. The program has been imple-mented one ICU at a time.

The process of building a medicationsafety agenda is still evolving and it isnowhere near finished. Sharing storiesstill requires work, and the training mod-ules for assertive communication andteamwork are being developed for theeducation program. Finally, an entire cur-riculum is being developed for medicaland nursing students to ensure that, at avery early stage, a culture of teamworkand safety are presented.

Focusing on Harm to Set the Patient and Medication Safety Agenda


Need for ImprovedAssessment of MedicationSafety

In January 2000, the Institute forHealthcare Improvement, a not-for-profitBoston-based organization, and Premier,a healthcare alliance for 1,600 of thenations' hospitals, convened a group ofexperts to develop a model for redesign-ing the medication system to achieve alevel of safety in medication administra-tion that would be tenfold greater thancurrently exists. The project was knownas the Idealized Design of the MedicationSystem (IDMS). Effective measurement ofsafety in medication use was critical toassessing the effectiveness of the newsystem in achieving this goal.

Assessing medication use safety hasbeen difficult because of reliance ontraditional voluntary reporting of ADEsand medication errors using incidentreporting systems. Organizations haveused voluntary reporting systems to pro-vide data as a measure of patient safety.Leape and others have found that volun-tary reporting is unreliable and, at best,probably captures only 10 to 20 percent

of actual errors.1 Most errors are inter-cepted before reaching the patient, do notresult in ADEs, and are not even perceivedby many to be worthy of report. In addi-tion, studies suggest that only a small per-centage of medication errors actually resultin harm, and those that do not are not likelyto be reported.

The redesign team required a betterdetection method, one that focused onADEs, not errors, to measure the effective-ness of the new system. Classen in SaltLake City used computerized screening ofpatient information using sentinel signalsor "triggers,"2-3 but broad application ofthis automated screening methodologywas unlikely to occur due to fiscal andtechnical constraints. The redesign teamdeveloped a "low-tech" version that couldbe widely applied in any institution. Thismethod has since been used extensivelyby hospitals working with IHI on improv-ing medication safety. Organizations nowcan use this Trigger Tool method to meas-ure, monitor, and manage the safety ofthe medication use process.

Trigger Tool MethodThe Trigger Tool method involves the

monthly examination of 20 randomlyselected patient records with a minimumtwo-day length of stay to screen for thepresence of triggers such as laboratoryvalues, interventions, and administrationof reversal agents (Table 1). If a trigger isfound, the chart is examined further forevidence of an ADE.

The degree of harm associated withthe ADE is classified using the severityindex of the National CoordinatingCouncil for Medication Error Reportingand Prevention (NCC MERP).4 The five NCCMERP categories (E-I) that involve actualharm to patients range from categoryE, defined as contributing to or resultingin temporary harm that required inter-vention, to the most serious category I,defined as contributing to or resulting inthe death of a patient (Table 2).

ResultsAs shown in Table 3, data collected

from 86 organizations using the TriggerTool revealed 720 ADEs in 2,837 patientrecords (24.9%). Total doses of medica-tions administered in this populationwere 268,796, resulting in a calculatedADE per 1,000 dose rate for all 86 organ-izations of 2.67. Nine of the 86 organiza-tions evaluated their traditional mecha-nisms for finding errors and ADEs. Of the274 ADEs identified by the trigger tool inthese organizations, only 5 (1.8%) wereelicited.5

ConclusionThe use of the Trigger Tool appears to

be more effective than traditional report-ing methods for detecting medication-related harm. The methodology is reason-ably inexpensive to institute and is spar-ing of quality- and safety-personnel time.The tool takes advantage of all types ofevents, including "near-miss" errors and


Key Points:• Assessing medication safety has relied on voluntary reporting, which

captures a small percentage of actual errors, not all of which result in harm. The Trigger Tool identified adverse drug events (ADEs) in 24.9% of patient records, compared with 1.9% found through conventional systems.

• The Trigger Tool methodology screens patient records for laboratory values, interventions, and administration of reversal agents to identify possible ADEs; harm is assessed using the NCC MERP severity index.

• The Trigger Tool method measures total harm, goes beyond error but does not exclude error, is easy to use for sampling over time, and measures the results of patient safety improvement efforts.

PROCEEDINGS

Using The Trigger Tool to Detect Potential Harmin Medication Management

Terri Simmonds, RN, Director, Critical Care and Patient Safety, Institute for Healthcare Improvement, Boston, MA



ADEs. The surveillance methodologyimplicit in the tool is more reliable thanspontaneous reporting, and if the chartselection is properly randomized, the dataover time generate statistically validinformation untarnished by the limita-tions of spontaneous reporting. The dataobtained are best used for internal com-parison and measuring improvement. As apoint of further development, over thepast two years IHI has developed a similartool for detecting ADEs in the intensivecare unit. This tool is currently being usedby more than 60 organizations.

Further information about the IHITrigger Tool for Measuring Adverse Drug

Events is available at the following website:www.QualityHealthCare.org.

References1. Cullen, DJ, Bates, DW, Small, et al.

The incident reporting system does not detect adverse drug events: a problem in quality assurance. Jt Com J Qual Improv. 1995;21:541-8

2. Classen DC, Pestotnik SL, Evans RS, et al. Computerized surveillance of adverse drugevents in hospital patients. JAMA 1991; 266:2847-51.

3. Classen DC, Pestotnik SL, Evans RS, et al. Using a hospital information system to assess the effects of adverse drug events. Proceedings of the Annual Symposium on

Computer Applications in Medical Care1993;161-5.

4. www.nccmerp.org/dangerousabbr.htm (accessed December 6, 2003).

5. Rozich, JD, Haraden, CR, Resar, RK Adverse drug event trigger tool: a practicalmethodology for measuring medication related harm. Qual Saf Health Care 2003;12:194-200.

Using The Trigger Tool to Detect Potential Harm in Medication Management

TABLE 2:

NCC MERP Index CategoriesE through I

• Category E: harm that contributed to or resulted in temporary harm to the patient and required intervention.

• Category F: harm that contributed to or resulted in temporary harm to the patient and required initial or prolonged hospitalization.

• Category G: harm that contributed to or resulted in permanent patient harm.

• Category H: harm that required intervention to sustain life.

• Category I: harm that contributed to or resulted in the death of a patient.

Trigger Process Identified

T1: Diphenhydramine T2: Vitamin KT3: Flumazenil T4: Droperidol T5: Naloxone T6: Antidiarrheals T7: Sodium polystyrene

T8: PTT >100 seconds T9: INR >6 T10: WBC <3000 × 106/µLT11: Serum glucose <50 mg/dL T12: Rising serum creatinine T13: Clostridium difficile positive

stool T14: Digoxin level >2 ng/mLT15: Lidocaine level >5 ng/mLT16: Gentamicin or tobramycin levels

peak >10 µg/ml, trough >2 µg/mLT17: Amikacin levels peak >30 µg/mL,

trough >10 µg/mL T18: Vancomycin level >26 µg/mL T19: Theophylline level >20 µg/mLT20: Oversedation, lethargy, falls T21: Rash T22: Abrupt medication stopT23: Transfer to higher level of care T24: Customized to individual

institution

Hypersensitivity reaction or drug effectOver-anticoagulation with warfarinOversedation with benzodiazepineNausea/emesis related to drug useOversedation with narcoticAdverse drug eventHyperkalemia related to renal impairment

or drug effectOver-anticoagulation with heparinOver-anticoagulation with warfarinNeutropenia related to drug or diseaseHypoglycemia related to insulin useRenal insufficiency related to drug useExposure to antibiotics

Toxic digoxin levelToxic lidocaine levelToxic levels of antibiotics

Toxic levels of antibiotics

Toxic levels of antibioticsToxic levels of drugRelated to overuse of medicationDrug related/adverse drug eventAdverse drug eventAdverse eventAdverse event

TABLE 1.

Triggers for Identification of Possible Adverse Drug Events

TABLE 3.

Multi-center TriggerReview

• 2837 charts reviewed using Trigger Tool

• 86 institutions• 720 ADEs found on reviews• 268,796 medications doses

administered • ADEs/1,000 doses = 2.67 • Admissions with ADEs = 24.9%


Complexity, Standardization, and Medication Safety

Steven Meisel, PharmD, Director of Medication Safety, Fairview Health Services, Minneapolis, MN

The medication use process isextremely complex, involving bothindividuals and technology in a

series of complicated steps. As organiza-tions seek to make this process safer,much can be learned from complexitytheory, human factors engineering (HFE),and findings from other industries suchas nuclear power and aerospace. This arti-cle describes how key elements of com-plexity theory and HFE can be applied tothe medication use process. Several casestudies illustrate how simplification andstandardization can reduce complexity toimprove medication safety.

Medication ManagementMedication management includes

seven core processes: evaluation of apatient, decision to use a medication,drug ordering, order transcription, drugdistribution, drug administration, andfinally patient monitoring. No one personis responsible for the entire overallprocess. While nurses are primarilyresponsible for drug administration,patients still self-administer, and physi-cians, families, respiratory therapists andothers may also administer medications.Similarly, physicians are primarily respon-sible for prescribing drugs, but pharma-cists can prescribe, as do nurse practition-ers , while patients will self-prescribe.

Creating an effective medicationprocess requires the participation ofmembers from every discipline involvedwith medication use. The medicationprocess needs to be examined as a whole,not as subprocesses, so that improve-ments made in one area, such as drugdistribution, will not adversely affectother areas, such as prescribing or drugadministration.

Complexity TheoryBerwick has pointed out that systems

produce precisely the outcomes they aredesigned to achieve. Understanding keyaspects of complexity theory can helpdirect efforts to redesign the system toimprove medication safety.

The overall failure rate of a complexsystem can be calculated based on theprobability of error in each step and thenumber of steps in the system (Table 1).1

If, for example, the one step in a one-stepprocess fails 1% of the time, the systemfails 1% of the time. If the same 1% fail-ure rate occurs in a system with 25 steps,the system will fail 5% of the time; with50 steps, 39%; and in a 100-step process,the system will fail 63% of the time. Thus,increasing complexity increases the likeli-hood of error.

At Fairview, the medication processhas been calculated to have between 40

and 70 steps.3 To improve the process,efforts could be made to reduce the fail-ure rate in every one of 50 steps from 1%to 0.1%, and thereby decrease the failurerate from 39% to 5%. However, reducingthe number of steps from 50 to 25achieves the same result. Simplification isa significant component of medicationsafety improvement.

Human Error RatesConsidering the nominal human

error rates in the general population forperforming particular tasks (Table 2)2 alsosuggests ways to improve medicationsafety. An error of commission, such asmisreading a label, occurs at a rate of0.3%, or three times out of a thousand.An error of omission without remindersoccurs at a rate of 1%; however, embed-ding the item in a procedure reduces theerror rate to 0.3%

The rate for simple math errors withself-checking is 3%. A monitor or inspec-tor fails to detect an error 10% of thetime. Personnel on different shifts will failto check hardware at a rate of 10%, ifthey are not required to use a checklist.Finally, the rate of general error in a high-stress situation where dangerous activi-ties are occurring rapidly –e.g., a typicalintensive care unit (ICU), operating room,or emergency department– is 25%, or onein four.

Error rates from the nuclear powerindustry further underscore the impor-tance of checklists and simplicity.3 Whenworking without a checklist, errors ofomission occur at a 5% rate. Use of acomplicated checklist with more than tenelements reduces the failure rate to 0.3%.However, use of a simple checklist reducesthe error rate by an additional two-thirds,to 0.1%.


Key Points:

• Medication use safety is a problem in large part because of the complexity of the medication use process and a lack ofstandardization within it.

• Improving medication safety can be accomplished by taking steps out of the process, standardizing procedures, or both.

• Integrating steps in the medication use process is one way both to take steps out of the process and to increase standardization.

PROCEEDINGS

Human Factors Principlesand Systems Design

Clergue has pointed out that the sin-gle limiting factor of human activity isthat the brain cannot have multiplesimultaneous foci of interest.4 Pilots in acockpit are trained to look at four piecesof data, because the human brain cannotprocess more information at a given time.An ICU nurse is expected to keep in mindfar more pieces of information, which isone of the reasons failures occur.

A system designed to overcome thislimitation would avoid reliance on mem-ory; simplify all tasks; standardizeprocesses; use forcing functions, proto-cols and checklists wisely; improveaccess to information; decrease relianceon human vigilance; reduce hand-offs;decrease look-alike elements; and useautomation carefully.

Frankel has said, "If there is a betterway to do something, we should all do itthat way because it's better. But if there is noknown best practice, we should settle on one,because the system and its players cannotexecute all of those practices without anunacceptably high failure rate."5

How many different sliding-scaleinsulin protocols are necessary in a hospi-tal, health system, or even a country?How many surgeon-specific "recipe cards"are necessary in an operating room?Different methods of marking the surgicalsite? Of dosing warfarin? How many dif-

ferent infusion pumps are necessary?How many different pediatric immuniza-tion protocols in the clinics? How manypost-operative pain regimens, cardioplegiaformulas, methods of administering pre-operative antibiotics, potassium replace-ment protocols, double-check policies andsystems? Patient safety is improved bylimiting the available options, as shownby the following case studies.

Case Study #1–StandardizingConcentrations

A three-year-old child undergoes aliver transplant at a major university hos-pital. Because of very small blood vessels,there is concern about potential clotting.Heparin is prepared in the operating roomand infused correctly. Some hours later onthe general floor, the heparin runs dry. Anew bag comes up from pharmacy, and inkeeping with a hospital standard, isadministered at the same rate that waspreviously programmed on the pump.Twelve hours later the discovery is madethat the solution prepared in the operat-ing room was ten times more dilute thanthe hospital standard, so that in thepatient care area, the child received a 10-fold overdose for 12 hours and subse-quently died.

In 1996, in one of the first Institutefor Healthcare Improvement (IHI)

Breakthrough Series, a California hospitalstudied the preparation of pediatric stan-dard dilutions.6 It was found that anes-thesia was preparing the drips during anoperation, then post-operatively thephysician would order a different concen-tration that pharmacy would prepare anddispense. The nurses would change the IVbags and administer the medication.Opportunities for error were numerous.

To simplify the procedure, key stepswere moved to the pre-operative setting.Before surgery, a physician uses a weight-based protocol to order medications.Pharmacy generates a worksheet and labelsbased on the order. Anesthesia still pre-pares the drips, but based off the pharmacyprotocol and preprinted labels. Postopera-tively, the physician only had to order thedose to be administered with everyoneusing the same standard concentration.

Case Study #2– IV PumpProgramming

A hospital uses IV medication safetysystem with dose calculation software.The system requires that a nurse selectthe name of the drug from a library, thenenter the amount of the drug in the bag, thevolume in the bag, patient weight, desireddose, and units. When programmingheparin, the nurse mistakenly enters2,500 units in 250 mL, instead of 25,000units in 250 mL, which results in a 10-foldoverdose.

IV medication safety system hasstandard concentrations and dose limitspreprogrammed. A nurse only has toselect the name of the drug, the weightand the desired dose. If the programmingis outside of those limits, an alarm isgiven. Simplification and standardizationimprove medication safety.

Case Study #3–PrescribingA physician handwrites sliding-

scale insulin orders, using his own param-eters. The orders are manually transcribedonto to the medication administration



TABLE 1.

Complex Systems:Probability of Error 2

Probability of Error, Each Step

# Steps 0.05 0.01 0.001 0.0001

1 0.05 0.01 0.001 0.0001

25 0.33 0.05 0.005 0.0002

50 0.92 0.39 0.05 0.005

100 0.99 0.63 0.10 0.01

TABLE 2.

Nominal Human Error Rates 4

Activity Human errorprobability

Error of commission (misreading a label) 0.003

Error of omission without reminders 0.01

Error of omission when items embeddedin a procedure

0.003

Simple math error with self-checking 0.03

Monitor or inspector fails to detect error 0.1

Personnel on different shifts fail to checkhardware unless required by checklist

0.1

General error in high stress when dangerousactivities occurring rapidly

0.25


record (MAR). However, the order is mis-read, and an order for 12 units of insulinis transcribed as 120 units, resulting insevere hypoglycemia.

What if the hospital had agreed on asingle sliding-scale protocol? What if theprotocol had been pre-typed? What ifthere were pre-typed medication admin-istration records to go along with it?What if a computerized prescriber orderentry (CPOE) system populated an elec-tronic MAR? All these actions would haveresulted in simplification, which wouldhave improved patient safety.

Case Study #4-Evaluationand Monitoring

A patient undergoes cardiac valvesurgery. His care is managed by the car-diac surgeon, cardiologist and internist.While in the hospital he suffers a tran-sient ischemic attack, is seen by a neurol-ogist and placed on warfarin. About aweek later the patient is discharged withmultiple prescriptions, including warfarin,but no firm visit is scheduled for an INR.

Each physician involved in the caseassumed one of the others was monitor-ing the INR. Two weeks later, the patientpresents to the emergency departmentwith an INR > 10 and what is later diag-nosed as a retroperitoneal bleed.

What if there had been an inpatientanticoagulation service; if referrals to thisservice had been automatic; if the servicehad had full accountability for evaluation,monitoring, dosing and scheduling follow-up; or if the patient had been referred toan outpatient anticoagulation clinic?

Case Study #5-DispensingA pediatric patient is admitted for

chemotherapy, enrolled in an investiga-tional protocol. The physician writesorders that are correct for the protocoland patient characteristics. However, theprotocol contains certain dilution andvolume specifications. The dose and dilu-tions are calculated incorrectly by thepharmacy, and the error is not caughtwhen the order is double checked. Thepharmacy dispenses a dose 50% higherthan prescribed, which results in neu-tropenia and sepsis.

What if the hospital settled on asingle, simple dilution method and thehospital requested and received a waiverfrom the researchers for these dilutions?And what if these dilutions were pre-loaded into the pharmacy computer?Calculations would not be necessary,mistakes would not be made and thesekinds of errors would not happen.

Case Study #6–CombinedOrder Form/MAR forHeparin

A hospital with a pharmacy-basedheparin dosing service had numerousproblems because of the number of tran-scriptions and handoffs that take placeafter a dose has been selected. To simplifythis procedure, a combined MAR-orderform document was designed, so thatwhen a pharmacist writes an order (e.g.,for a heparin bolus), it is written on thesame document that the nurse uses formedication administration and documen-tation. This approach, whereby a pre-scriber writes orders directly onto theMAR, has been used in England for years.As shown in the Figure, implementing thischange to simplify the process reducedthe error rate from 0.81/month to0.33/month.7

ConclusionSimplification and standardization

can improve medication safety.

References1. Communication with Tom Nolan in

December 1999. 2. Salvendy G (ed.). Handbook of Human

Factors and Ergonomics. New York, NY: John Wiley and Sons, Inc; 1997.

3. Philipe Garnerin. Thinking simple. Presentation at IHI 14th Annual National Forum, Orlando, Florida. Dec 8, 2002

4. Communication with Francois Clergue in December 2002

5. Communication with Allan Frankel in December 2002.

6. Leape LL, Kabcenell A, Berwick DM, and Roessner J. Breakthrough Series Guide: Reducing Adverse Drug Events. IHI. Boston MA 1998:24

7. Meisel S, Schultz J, and Graf K. An Interdisciplinary Model for Reducing Intravenous Heparin Errors. Solutions Conference, sponsored by the Joint Commission on the Accreditation of HealthCare Organizations and the National PatientSafety Foundation, Chicago, Illinois, October6, 2000.


FIGURE 3.Results 9

At the request of the Director ofCritical Care, a Six Sigma projectwas begun to improve the turn-

around time for medication order fulfill-ment between pharmacy and a surgicalintensive care unit (SICU). The project wasconducted at a 330-bed acute-care facili-ty located in an urban community in thesouthwestern United States. The SICU wasa 16-bed, multi-specialty unit that includ-ed cardiac surgery, trauma, and an activeventricular assist device program.

The turnaround issue had a long-standing history. Recent changes includedthe implementation of point of care med-ication storage cabinets. These cabinetswere configured in a way that preventednurses from obtaining medications beforethe prescription was reviewed by a phar-macist. Before this change was made,nurses easily could obtain the medica-tions prescribed from medication cabinetsthat were not linked to pharmacy.Creating a link to pharmacy, which pre-vented retrieval of medications before thepharmacist review, was not accomplishedeffectively, and order processing delaysresulted in delays in administration of

essential medications (Figure 1). Thisresulted in an evaluation of the medica-tion use process.

The first attempt to solve this prob-lem was to allow the nurse to "override"the safety check, so that urgent medica-tions could be obtained before the phar-macist review. There were still unaccept-able delays with other medications.

Because of the difference betweenan "open-cabinet," available medicationssystem and a "locked down" medicationssystem with a double check, one wouldexpect staff resistance to a change fromone to the other, because of the addedinconvenience of having to wait for thepharmacist review. In hindsight, the med-ication use process should have beenstudied before implementing the double-check system to ensure that medicationswere available to administer in a timelyfashion. During the first four monthsafter implementing the new restricted-access system, tension developedbetween nurses and pharmacists. TheCritical Care Nursing Director requestedassistance from corporate ClinicalEffectiveness, because the nurses were

"ready to quit." The pharmacists were sim-ilarly disgruntled.

Definition PhaseThe first step in the project was to

define the service delivery area, list themedications prescribed in the SICU, anddevelop a map of the medication useprocess. The SICU was the area with thestrongest complaints. Many variancereports documented medication delaysthat occurred on all days of the week andat all times of day. These delays resulted insignificant dissatisfaction among bothnurses and pharmacists. Adverse effectsresulting from these delays on thepatients were not measured, but anecdot-al evidence suggested there were somenegative clinical events.

Initial discussions with the pharmacystaff led to the development of a causeand effect ("fishbone") diagram (Figure 2)that revealed several procedural issues.These included problems with inboundcommunication (failed faxes, faxed ordersthat did not get processed), outboundcommunication (no way to let a nurseknow a medication was available), andproduction issues (a production line thatwas not maintained at all times).Pharmacists would often leave their workstation to do pain consultations, attendcodes, and eat lunch, but with no processin place to cover the unstaffed periods.

Measurement PhaseAll new medications ordered for

seven consecutive days were recorded.The data were collected from faxed ordersto the pharmacy and electronic files fromboth the pharmacy system and the pointof care system. The physician orders wereentered into the pharmacy system by



Key Points:

• Failure to evaluate the medication use process before implementing point of care medication storage cabinets that do not allow nurses to obtain medications before a pharmacist review led to unaccept-able delays in drug administration.

• A Six Sigma project led to the implementation of several changes, each of which contributed to significant improvement in service delivery, such as limiting a pharmacist to ICU activities and alerting nurses when and where medications were delivered.

• Evening-shift turnaround time was identified as particularly prob-lematic, and resistance had to be overcome to include experienced pharmacists on the evening shift.

PROCEEDINGS

Improving Drug Order Entry Turnaround

Nancy Pratt, MSN, Senior Vice President, Clinical Effectiveness, Sharp HealthCare, San Diego, CA


pharmacy staff after the orders werereceived via fax. The orders were alsoentered into a nursing point of care infor-mation system by the ward clerk or nurse.The nursing point of care informationsystem provided the medication adminis-tration record.

The medication orders were recordedand time intervals were determined usingfax time, order entry time, verificationtime, and administration time. PRN med-ications were not included in the study.Intravenous continuous infusions werealso not included, because they were doc-umented differently. With these exclu-sions, there were 154 usable orders. Thedata showed substantial delays in med-ication order processing and a high vari-ability in turnaround times.

Analysis PhaseThe turnaround time data were eval-

uated using analysis of variance (ANOVA).Results were significant for shift (p=0.007), with the evening shift demon-strating the most variance from the otherdata (Figures 3-4). The data were notsignificant for order status (stat, now,and routine), suggesting there was poordifferentiation of orders based on theirpriority.

Despite having worked with thepharmacy staff to describe the medicationuse process design, the data suggested

the actual process functioned differently.Rather than the pharmacist verifying thephysician order followed by the nurseadministering the drug, the data suggest-ed the process was reversed almost halfthe time. Medications were being admin-istered up to 500 minutes before thepharmacy reviewed the orders.

Individual and Moving Range (I andMR) control charts documented the errat-ic variability in the entire medication useprocess but demonstrated a ratherconsistent pharmacy processing time.Unfortunately, the mean pharmacy pro-cessing time was 69 minutes, whichexceeded the desired time to get "now"order medications to the nurse. The upperspecification limit for administering drugswas considered to be 15 minutes for"now" orders and two hours for routineorders. The number of medication ordersprocessed per hour was analyzed forseven days. There was a bimodal distribu-tion with peak order times between 0800to 1300 and 1700 to 2300 (Figure 5).Analysis of the results associated with theevening shift and the staffing patternidentified some issues.

While day-shift turnaround time wasinadequate, the evening shift was unac-ceptable. Staff meetings with the SICUand pharmacy staffs provided additionalinsight:

1. New staff members were consis-

tently scheduled to work the evening shift(the least popular shift).

2. The ICUs had a dedicated pharma-cist on days, but during the evening shiftone pharmacist covered the main phar-macy and the ICUs.

3. There was little awareness of theneed to maintain the production line.

4. Inbound communication of orderswas very unreliable.

5. Outbound communication wasnon-existent.

6. Medications delivered to the unitwere left in inconsistent places.

7. There was no interface betweenthe electronic information systems.

Improvement PhasePotent ia l changes intended to

improve medication turnaround timewere evaluated using in Impact/Risk/Cost analysis tool. A schedulefor implementation of the changesagreed upon was established. Theteam agreed to implement and test thefollowing changes:

1. Limit the ICU pharmacist to ICUactivities.

2. Schedule more experienced phar-macists to the ICUs.

3. Move an experienced pharmacytechnician to the evening shift.

4. Provide a "status board" to alert a


FIGURE 3.

Fax Time to AdministrationTime versus Shift

FIGURE 2.

Cause and Effect Diagram

FIGURE 1.

Drug Order Entry Turnaround:Actual Process

nurse when and where medication hadbeen delivered.

5. Fax a list of RN assignments andcell phone numbers to the pharmacy.

6. Install a fax server.7. Move the ICU pharmacist away

from the front pharmacy window toreduce distractions.

Control PhaseThe changes were implemented and

turnaround time was measured using thesame methodology. A significantimprovement (p=0.001) was noted. Themean turnaround time and standard devi-ation decreased significantly (meandecreased from 81.6 to 39.7 and standarddeviation from 132.4 to 50.5); however,the evening shift remained significantlydifferent (p=0.015). During a subsequent

team session with a mixed group of SICUstaff and pharmacists, it was discoveredthat all of the proposed changes wereaccomplished with the exception ofstaffing the evening shift from 1730-2300. The staffs committed once again tomaking this change.

SummaryThe extensive analysis performed for

this project was the basis for facilitatingthe cooperation among the various clini-cians. The presentation of the data madedecisions about how to address the issuemuch easier, since little guess work wasneeded. The clinical staff was very com-fortable making data-driven decisions.Some of the changes seemed to be ele-mentary. Sometimes additional impetus isrequired to make difficult decisions,

which seemed to be the case in this situ-ation. The leadership knew they had poorservice delivery. The resistance to alteringschedules to staff the evening shift withpharmacists who had been working day-time hours for years was not trivial. Eachof the changes that were implementedcontributed to a significant improvementin the service delivery for this pharmacy.The staff committed to staffing theevening shift with more experiencedpharmacists.



FIGURE 5.

Before and After Process ImprovementsFIGURE 4.


Since the Institute of Medicine reportwas published in 1999,1 severalstrategies to reduce medication

errors have been advocated, including com-puterized prescriber order entry, bar codemedication administration, clinical pharma-cy services, automated drug dispensing cab-inets, and dispensing robotics. While thesestrategies can be effective in reducing theoverall rate of medication errors, they havelimited effectiveness in addressing intra-venous (IV) medication administrationerrors, which arguably pose the greatestrisk of harm. MEDMARXSM 2002 datashowed that the IV route of administrationfor medications often results in the mostserious medication error outcomes.2

IV infusions are administered to thesickest patients, often with a large numberof infusions being administered simultane-ously, with frequent dosage adjustments.Compared to oral and non-IV parenteralmedications, an IV infusion typically is not asingle administration event, but rather aseries of programming events under cir-

cumstances where the risk is greatest. In a seminal medication error study,

Bates et al found that 38% of the preventa-ble medication errors occur at the point ofadministration, and only 2% of these areintercepted.2 IV medications represent 61 %of the serious and life-threatening errors.3

Together, these findings further reinforcethe fact that IV administration is an areawhere errors have the greatest potential forpatient harm. Consequently, implementingmedication safety systems that can increaseinterception of the IV administration errorshas a high potential to reduce harm andprotect both patients and nurses—a strate-gy that is unique compared to other med-ication safety initiatives.

Intravenous Delivery DevicesThe increasing complexity of IV therapy

has led to the development of sophisticatedinfusion devices designed to deliver accu-rately a wide variety of therapies. The infu-sion devices, commonly referred to as"pumps," are the most widely used medical

devices in hospitals today. Approximately750,000 pumps are used to administer morethan one million IV doses per day in UnitedStates hospitals. Unlike medications, whichare prescribed and dispensed based on indi-vidual patient requirements, infusion devicestypically are not configured for individualpatient use by Biomedical Engineeringdepartments. Rather, they are configured atthe time of implementation to cover the fullspectrum of possible applications, from apre-term 600 g baby receiving a fraction of amilliliter per hour to an 80-kg trauma patientreceiving as much as a liter of fluid per hour.These general-purpose infusion devices aredesigned to be easy to use, require noauthorization to program, and have a10,000-fold rate and dose range that cansupport a wide range of infusion orders. Untilrecently, the infusion pumps had no capabil-ity to provide a "test of reasonableness" tothe programming of an IV medication orfluid. Consequently, infusion devices havebeen associated with some of the most seri-ous medication errors.

IV Medication Safety SystemA new generation of infusion devices

introduced in May 2001 comprises an inno-vative medication safety system that pro-vides an IV “safety net” for nurses at thebedside. Although shipped to hospitals as a"dumb" system with features similar toexisting legacy pumps, these new infusionsafety systems incorporate software thatcan be customized for each hospital's "bestpractices" for IV therapy. The best practicesare incorporated in the safety software tocreate multiple patient-care-area-specificlibraries that include medications, concen-trations, dosing units, and dose limits.Hospitals can now have the equivalent of 10infusion devices in one, with drug librariesand infusion rules designed for unique areasor patient types (referred to as "Profiles").After a clinician identifies the Profile for a


Key Points:

• The intravenous (IV) route of administration for medications often has the greatest potential for patient harm.

• An IV medication safety system incorporates customized software that applies hospital best practices for IV therapy, including medications,concentrations, dosing units, and dose limits; and maintains a log of "near misses" that can be used for analysis and process improvement.

• A review of IV medication safety system data sets from more than 65 hospitals revealed large variations in all aspects of IV therapy best practice rules, both intra- and inter-hospital.

• Standardizing concentrations, drug names, dosing units, dose limits,maximum infusion rates, weight limits and volume limits may help to improve patient safety.

• In improving IV medication safety, a safety system’s data collection capabilities may be as significant as their capability to intercept IV medication errors.

PROCEEDINGS

Variability in IV Therapy: A 65-hospital Analysisof IV Best Practices

Tim Vanderveen, MS, PharmD, Exec. Clinical Director, The ALARIS® Center for Medication Safety and Clinical Improvement, San Diego, CA

particular patient care area such as neona-tal intensive care unit (ICU), med/ surg, oradult ICU and selects a drug to be infused,the customized software applies hospitalbest practice rules to check device program-ming, and alerts the clinician if the pro-gramming exceeds the rules. This ”test ofreasonableness” thereby ensures a new levelof safety for IV therapy.

Standardized concentrations, non-editable drug dosing units, and minimumand maximum dosage limits are among thesafety elements of this new system. In addi-tion to reducing the opportunities for pro-gramming errors through incorporatingbest practice guidelines and providing alertswhen programming exceeds limits, an IVmedication safety system also maintains alog of the alerts that can be downloaded forfuture analysis and process improvement. InOctober 2002, Health Devices, published byECRI, evaluated all currently marketed gen-eral purpose infusion systems and conclud-ed that only pumps that had "dose errorreduction software" should be consideredfor purchase.4

Creating the BestPractice Rules

Customization of the software data-base for an IV medication safety system isaccomplished through review of existing IVpractices and creation of an extensive druglibrary. A master drug library is created that

includes the drug names and available con-centrations. Appropriate items from thisextensive library, which often numbers inthe hundreds of entries, are then copied toeach patient care area profile where a par-ticular drug/concentration will be used. Inaddition, minimum and maximum dosagelimits, including soft (can be overridden atclinician's discretion) and hard limits (can-not be overridden), are added at the sub-library level for each profile. For example,dopamine may have three concentrationentries in the master library (400 mg/250 mL;800 mg/250 mL; and 1,600 mg/250 mL). Ina specific Profile, dopamine may be avail-able in one, two, or all three of the concen-trations, or it may not be available, if thedrug is not used in that patient care area.All three combinations may be available inthe adult ICU, only one in the step downunit, and none in a med/surg unit.

For each entry, min/max dosing unitscan also be customized according to howthe drug is used. The maximum dose thatcan be programmed for dopamine beforean alert is provided may be 20 or 22mcg/kg/min in the adult ICU, whereas themaximum dose might be 5 mcg/kg/min inthe step-down unit. The ICU dopaminelimit may be designated as a "soft" limit,while the step-down limit may be a "hard"limit to reflect the different indications forthe same medication.

The process for developing andapproving the IV best practices varies

among hospitals, but is typically pharma-cy-driven, with final approval by thePharmacy and Therapeutics Committee orthe Medication Safety Committee. Beforeloading the best practices information intoan IV medication safety system, a line-by-line signoff is required. The most advancedsystems utilize a CD-ROM with multiplelevels of security, and the safety softwareand data set are transferred by the bio-medical engineers using a laptop computer.

Unexpected FindingsOne evaluation criterion noted in

ECRI's Health Devices for selecting smartinfusion technology is consultative supportfrom the vendor to guide hospitalsthrough the development of the best prac-tices drug library data set.4 ALARIS MedicalSystems initially approached this consulta-tive support through the development of apeer-reviewed "starter" drug library dataset. However, despite the collective wisdomand experience of a blue ribbon panel ofclinical experts representing multiple disci-plines, this data set was found to be ofminimal value. The reason for this becameobvious during a review of more than 65individual hospital data sets: there is largevariability in all aspects of the best practicerules (see Table). This variability is found indrug names, concentrations, dosing units,and dose limits, as well as in other per-formance limits such as maximum infusionrates, weight limits, volume limits, etc.

Further investigation focused on thevariability in drug names. A surprising find-ing in examining the 65 data sets was thatthe average drug entity had four differentnames. Some of this variability was to beexpected and included generic, trade and"local" names. Another source for variabili-ty was the creative use of the drug namesto include indication for use, as illustrated bythe various creative names used for tPA. Inthe 65 hospitals, tPA has been given 20different names (e.g., alteplase, tPA, andActivase®). In all but two entries, the indica-tion or another modifier (e.g., tPA–stroke,Activase®–MI, alteplase–PE) has been added.

Variability in IV Therapy: A 65-hospital Analysis of IV Best Practices


TABLE

Infusion Therapy: Variation in Practice from 65 hospital data sets:• Average of 64 drugs per hospital

— Average of 279 drug/concentrations per hospital• Multiple names for same drug

— Average of 4 names per drug • Inconsistent continuous dosage units for same drug

— 60% have more than one continuous dosage unit (Range = 1-8)— Average of 13 unique dosage units/hospital (Range = 3-19) Does not include bolus dosing–typically different from continuous dosing

• Multiple concentrations per drug— Average of 1.5 per drug, per hospital (Range = 1-9)— 7.5 per drug in all data sets (Range = 1-13)

• Minimal concentration standardization in peds/NICU— 67% of entries are “fill in the blank” concentrations


Expansion of the capacity of the druglibrary from the initial 40 entries to 100 andthen to as many as 1,000 per patient carearea has provided hospitals with the neces-sary flexibility to add the indication to thedrug names. The addition of the indicationor other modifier such as "weight based"(heparin) or "central line" (potassium) hasallowed the best practices to be made indi-cation-specific and to significantly tightenthe safe software dose limits to match theintended use.

A second unexpected finding was thelarge variability in drug concentrations. Onaverage there were 65 drug entities perhospital data set. (A drug entity is definedas the drug only, e.g., dopamine or dobuta-mine. If a hospital used both generic andtrade names for the same drug, this wascounted as a single drug entity.) Furtheranalysis determined that there were 279drug/concentration combinations for these65 drug entities. Many of the drug libraryentries had "fill in the blank" concentra-tions, including 67% of the NICU and pedi-atric profile entries. The large number ofdrug/concentration combinations conflictswith the Joint Commission on Accreditationof Healthcare Organizations (JCAHO)2003/2004 safety initiative for reducing thenumber of concentrations (discussed bySchafer in these Proceedings). Data setsfrom the 65 hospitals included 18 differenttPA concentrations and 17 fentanyl con-centrations. One hospital had 15 differentKCl concentrations listed in their data set;another had five epinephrine concentra-tions. While not all of these drug/concen-tration entries were available in the sameprofile, the large number of concentrationoptions was unexpected.

A third unexpected finding was thevariability in dosing units, e.g., mg/hr,mcg/kg/min, and units/hr. The range perdrug entity over the 65 hospitals was 1 to8, and 60% of drug entities were associat-ed with 2 or more dosing units. The averagenumber of unique dosing units per hospitaldata set was 13, with a range from 3 to 19.The smart infusion devices being used inthese hospitals had 42 possible dosing

units, including nanograms, micrograms,milligrams, grams, units, milliunits, and mil-liequivalents. As expected, the dosing unitsfor NICU and pediatrics were the moststandardized, since most medications inthese profiles are set up as weight-basedinfusions. Examples of the dosing unit vari-ability included amiodarone (6 differentdosing units), calcium gluconate (7), andmagnesium sulfate (8). The 44 differentamiodarone drug name entries had thefollowing dosing units: mcg/kg/hr;mcg/kg/min; mg/day; mg/hr; mg/kg/hr;mg/min. Another surprise was to find 64continuous delivery IV medications thathad 2 dosing units in the same patient carearea profile. It should be noted that thisanalysis did not include bolus dosing, PCAor epidural drug infusions.

Implications of Data SetVariability

In many cases, the introduction of anIV medication safety system resulted in thedevelopment of the first comprehensivedrug library intended to set forth a hospi-tal's best practice guidelines for IV infusiontherapy. Evaluating the best practices datasets from 65 hospitals has revealed vari-ability in IV therapy that may not havebeen previously recognized. Intra- andinter-hospital variations are a significantand consistent finding. Several of the clin-ical experts who participated in the peerreview process to define the "starter" dataset were surprised by this variability, whichexplains why a peer-reviewed data set wasnot effective, even as a starting point. Itshould also be pointed out that these 65hospital data sets were the result of exten-sive internal review by each hospital, andoften represent significant reduction invariability from pre-IV medication safetysystem use.

At this point it is not clear whatimpact this variability has on medicationsafety, but it may well be a critical factor.For example, using or not using a weight ina drug calculation in a 70-kg patient repre-sents a potential for a 70-fold under or

over-dose. The variability revealed in analy-sis of the drug library data sets from 65hospitals provides unexpected support tothe 2004 JCAHO National Patient SafetyGoal requiring organizations to standard-ize and limit the number of drug concen-trations available in the organization.JCAHO's requirement for limiting concen-trations, as well as principles of complexitytheory and human factors engineering(HFE), and findings from other industries(see Meisel in these Proceedings), suggestthat standardizing drug names, dosingunits, dose limits, maximum infusion rates,weight limits and volume limits may alsohelp to improve patient safety.

ConclusionAn IV medication safety system pro-

vides both a safety net for nurses that candetect and prevent IV drug administrationerrors, and a database to measure, monitor,manage, and improve infusion safety. Sincethis route of administration is the mostdangerous and there are infrequent andoften cursory double checks at the drugadministration step of medication use, anIV medication safety system would appearto have an excellent impact when consid-ered with other patient safety technologies.



2. Hicks, RW, DD Cousins, RL Williams. Summary of Information Submitted to MEDMARX S M in the Year 2002. The Quest forQuality. Rockville, MD: USP Center for the Advancement of Patient Safety; 2003.

3. Communication with DW Bates, October 2001.

4. Emergency Care Research Institute, Health Devices; 2002;31(10):353-87.

Activase® (Alteplase, recombinant) is manufacturedby Genentech.

Variability in IV Therapy: A 65-hospital Analysis of IV Best Practices

New JCAHO NationalPatient Safety Goals

Healthcare, public and private organ-izations continue to focus intently on thereduction of adverse events that patientsmay experience. The 2004 NationalPatient Safety Goals (NPSG) of theJoint Commission on Accreditation ofHealthcare Organizations (JCAHO) outlinerequirements that organizations mustcomply with or they will "receive a specialrequirement for improvement for thatgoal" (a Type I recommendation). One ofthese new goals is to "Improve the safetyof using high-alert medications"; specifi-cally, to "standardize and limit the numberof drug concentrations available in theorganization." The measurement systemsuggested for meeting this standard canbe found in the new MedicationManagement (MM) chapter of the JCAHOaccreditation manual in section 2.20. Thenew MM standards are effective January1, 2004.

New requirements included in thesepatient safety goals can conflict with wide-

spread current practices. Resolving suchconflicts will be challenging, yet criticallyimportant to improving patient safety.

Current Practice: TheRule of 6

The "standardization of drug concen-trations available within the organization"has particular relevance in the pediatricpatient population. Pediatric practitionershave utilized the Rule of 6 to calculatecontinuous medication infusions by vary-ing the concentration while keeping therate of infusion standard. The prescribercalculates the desired concentration ofdrug based on individualized factors,including patient weight, desired drugamount, and solution infusion rate:

6 x desired dose (mcg/kg/min) x patient wt = mg of drug to desired fluid rate (mL/hr) be added to

100 mL solution

Using the Rule of 6 to calculatehow much medication to use inpreparing an infusion, the practitionervaries the concentration of the drugwhile keeping the rate of infusion

standard, as shown in the followingexamples:

• 2-kg infant to receive dopamine 5 mcg/kg/min to run at 0.5 mL/hr

6 x 5 mcg/kg/min x 2 kg = 120 mg of dopamine 0.5 mL/hr in 100 mL of solution

• 20-kg toddler to receive dopamine 5 mcg/kg/min to run at 1 mL/hr

6 x 5 mcg/kg/min x 20 kg = 600 mg of dopamine

1 mL/hr in 100 mL of solution

Medications that are commonly pre-pared and administered using the Rule of6 include vasoactive medications (e.g.,dopamine, epinephrine, dobutamine, mil-rinone), antiarrhythmic agents (e.g., lido-caine, amiodarone), central nervous sys-tem medications (e.g., opioid agonists,benzodiazepines), and other medicationsrequiring continuous intravenous admin-istration.

JCAHO ClarificationDue to the scope and widespread use

of the Rule of 6 in pediatrics, clarificationof JCAHO's position on standardized dripswas requested (Table 1). The response pro-vided in October 2003 read:

"The Rule of 6 is not acceptable. Whentrying to determine how to give a medica-tion as an infusion ordered at a certainmilligram per hour or microgram per kilo-gram per hour, one can keep the rate ofinfusion standard (e.g., 1 milliliter/min)and vary the concentration (milligramsper milliliter) of the drug, or one can varythe rate of infusion while keeping the drugconcentrations standard. The NationalPatient Safety Goal requirement 3b(scored on MM 2.20, EP #8) states that thelatter (keeping the drug concentrationstandard and varying the rate of infusion)must be used.



Key Points:

• A 2004 JCAHO National Patient Safety Goal requires organizations to stan-dardize and limit the number of drug concentrations available in the organiza-tion; this conflicts with the widely used Rule of 6, which results in the use of varied drug concentrations.

• The JCAHO Standards Interpretation Group has stated that the Rule of 6 is not acceptable in meeting their standards.

• The majority of institutions surveyed in 2003 still used the Rule of 6; those that had changed to standardized concentrations felt the change was useful.

• Issues of concern include administration rates with standardized concentra-tions, half-life of drugs delivered at very slow infusion rates, and the learning curve in making the transition from the Rule of 6 in pediatric practice.

PROCEEDINGS

The JCAHO Focus on Limiting Concentrations of High-Alert Medications: Opportunities and Potential Impact

Christopher L. Shaffer, PharmD, BCPS, Pharmacy Clinical Manager,Department of Pharmaceutical and Nutrition Care, The Nebraska Medical Center, Omaha, NE


"The Rule of 6, commonly used inpediatrics, is just one of many methods ofquickly calculating a concentration of thedrug needed to achieve a given dose, whilekeeping the rate of infusion constant (1milliliter/min). This allows for an infinitenumber of drug concentrations and isdirectly opposed to the NPSG and 2004standard MM.2.20, EP #8 requirement ofhaving the number of concentrationsstandardized and limited.

In many hospitals, the pharmacydepartment has developed charts/tablesthat allow nurses to quickly determine theproper rate of infusion to achieve a desiredmg/kg/ hour using different standardizedconcentrations of the drug. In addition,'smart' infusion pumps can assist in thiscalculation, too."

Teresa Stewart, RN, MJAssociate Director, Standards Interpretation Group, JCAHO

Controversies andOpportunities

The enforcement of this new stan-dard is being scrutinized by the pediatriccommunity for several reasons (Table 2).The Rule of 6 has been a standard of prac-tice for approximately 25 years and wasoriginally intended to provide a simplifiedcalculation of doses and infusion rates.Since the smallest infusion rate that mostinfusion pumps could deliver was 0.5mL/hr, the Rule of 6 helped determine theindividualized concentration of medica-tion needed to deliver the prescribed dose.As a result, most pediatric healthcarepractitioners have been trained to utilizethe Rule of 6.

A survey conducted by Morganshowed that while most centers still usedthe Rule of 6, those that had changed tostandardized concentrations all felt thatthe change was useful (Table 3).

Citing the historic precedence of Ruleof 6, pediatric practitioners at centersusing Rule of 6 stated that a change inpractice could adversely affect patientsafety by introducing a new "standard" towhich professionals are not accustomed.

Centers using standard concentrationsacknowledged that education was ofparamount importance in the successfulconversion from the Rule of 6 but thatpediatric practitioners were satisfied afterthe conversion.

The Rule of 6 also is commonly dis-cussed in pediatric references. For exam-ple, Pediatric Advanced Life Support(PALS) describes the Rule of 6 as onemethod for calculating medication dripcalculations. The Harriet Lane Handbook,NeoFax, and Pediatric Dosage Handbookall describe the Rule of 6 as a useful cal-culation method for continuous medica-tion infusions. The JCAHO standard willrequire revision of these frequently usedreferences.

Another concern with standardizedconcentrations is variation in the rate offluid administration needed to providethe prescribed therapy. There is wide vari-ation in the weight of patients between a24-week gestational age neonate and an18-yr old adolescent. If too much fluid isused to administer the prescribed dose ofa medication, pediatric practitioners areconcerned that the use of standardizedconcentrations might result in fluid

The JCAHO Focus on Limiting Concentrations of High-Alert Medications

TABLE 3.

Survey of Existing Practices• Douglas Morgan MS, RPh Children's

Hospital of Iowa, University of Iowa Hospitals and Clinics

• Survey completed in July 2003 and again in October 2003.

• Results— 30 respondents: 15 MD, 5 RN, 7

RPh, 3 Unknown with 27 centers— 18 of 27 centers use Rule of 6— 2 of 18 were planning to change to

standard concentrations only— 9 of 9 centers that changed to

standard concentrations felt change was useful; use 2-4 concentrations

TABLE 2.

Rule of 6: Why a MajorIssue?

• Historical precedence (1981 Harriet Lane)— Formula to assist with unit conver -

sions along with limited pump technology of infusion rates

• Pediatric Advanced Life Support (PALS)— "Gold Standard" for educational

purposes• References

— Harriet Lane Handbook— NeoFax— Pediatric Dosing Handbook

• Standard of Practice?

TABLE 1. Standardized Concentrations

vs. Rule of 6• "If we are using the Rule of 6 for

drips for our pediatric patients, are we non-compliant with NPSG #3b?"

• JCAHO's response (October 30, 2003):— "The Rule of 6 is not acceptable.

When trying to determine how to give a medication for infusion ordered at a certain mg per hour or mcg per kg per hour, one can keep the rate of infusion standard and vary the concentration of the drug, or one can vary the rate of infusion while keeping the drug concentration standard. The NPSG #3b states that the latter (keeping the drug concentration standard and varying the rate of infusion) must be used.

— "The Rule of 6… allows for infinite number of drug concentrations and is directly opposed to the NPSG and 2004 standard MM2.20…"

imbalances. Institutions that have adopt-ed standardized concentrations have beenable to create 2 to 4 different concentra-tions for vasoactive medications that arebased on patient weight. Practitioners areconcerned that smaller patients mayreceive too much fluid when being givenvasoactive medications, particularly ifthey are also receiving total parentalnutrition. In this situation, medicationsmay need to be prepared in concentra-tions that result in extremely slow infu-sion flow rates that may be unachievablewith current infusion pumps.

There have been considerableadvances in infusion pump technologysince the inception of the Rule of 6.Current technology including smartpumps can be programmed to delivermedication at very slow rates.Practitioners must be aware that while apump can be programmed to deliver suchrates, there is a point beyond which med-ication delivered at these very slow ratescan have little pharmacological effect.

Consider the following example:With microbore tubing, a rate of 0.1 mL/hrwill deliver one drop of medication fluidto the patient end of the microbore tub-ing every 13 to 14 minutes. Since thehalf-life of dopamine is 2 to 4 minutes,this low rate of administration may pro-duce a diminished or variable response,

because the pump cannot delivermedication at a rate close to the rate thepatient eliminates it. One reason a patientmight appear to need a higher dose orhave a variable blood pressure response isthat the drug simply is not being deliveredreliably. This illustration demonstratesanother aspect that a pediatric practition-er must consider: does the concentration(whether standardized of individualized)and mechanism of administration pro-vide the optimal pharmacologic actionintended?

One criticism that JCAHO hasreceived in its publication of this standardis the lack of evidence to support a reduc-tion in medication errors. One aspectbeing overlooked is the number of med-ication errors attributed to the Rule of 6kept in the "file drawer." Although theRule of 6 is the standard of practice inpediatrics, the infinite number of individ-ualized concentrations that can be creat-ed suggests an exponential increase in thepossibility of medication errors even inthe same patient, who may receive vari-able concentrations of the same medica-tion. Unless there are collaborative datasuggesting that the Rule of 6 results indecreased errors as compared to stan-dardized concentrations, the evidence-based approach discussed by opponentsto JCAHO's position is invalid.

ConclusionThe NPSG and JCAHO's position in

standardizing medication concentrationsand eliminating the Rule of 6 have led toa great deal of controversy in the pedi-atric community. Although the Rule of 6 isa standard of practice in both the educa-tional and clinical setting, the infinitenumber of individualized concentrationsthat could be created with the Rule of 6appears to increase opportunity for med-ication errors compared to a limited num-ber of standardized concentrations.Pediatric practitioners should examineexisting resources (both technologic andeducational) in their application of thesenew standards until evidenced-basedmedicine demonstrates the superior safe-ty of one method over the other.Regardless, the debate over standardizedconcentrations versus the Rule of 6 canbe expected to continue for the foresee-able future.

The JCAHO Focus on Limiting Concentrations of High-Alert Medications



Medication safety is an importantproblem, as illustrated by theAdverse Drug Events Prevention

study.1 In that study, there were 6.5adverse drug events for every hundredadmissions, and about a third of thesewere preventable. Serious medicationerrors are the errors that either harmsomeone or have the potential to do so.While the largest proportion of seriousmedication errors occurs at the orderingstage, the second greatest proportionoccurs at the administration stage.

A major finding is that errors occur-ring late in a process are less likely to beintercepted. The medication use process isno exception: in one study, about half ofordering errors were caught before theyreached the patient vs. only 2% of admin-istration errors.2

Intravenous (IV) medications are vitalin the therapeutic management of hospi-talized patients, yet administering med-ications via this route is a vulnerable area(Table). Inpatients often receive several IVmedications concurrently, and these oftenare delivered with infusion pump sys-tems.3 Critically ill patients often receivepotent IV drugs that have narrow safetymargins and require careful titration.These patients, who are sicker thanpatients not requiring intensive care, may

be more vulnerable to adverse effects ofmedications. While IV medications areundoubtedly beneficial and can be lifesaving, errors in administering them havea high risk for severe adverse events andhave caused many fatalities.4

New infusion systems incorporatesignificant technologic improvements.3

One important safety advance has beenthe development of mechanisms that cannearly eliminate the risk of free-flow,which has caused many fatalities. Otherfeatures include enhanced functionality,convenience and portability. Additionalfeatures, however, can also add complexi-ty. To attempt to "engineer out" errors,some of the newest infusion systems havefeatures including drug/dose calculations,programmable volume and time calcula-tions, improved alarms and indicators,and most recently, inclusion of drug orpatient-specific decision support capabil-ities. All such systems should be designedusing human factors techniques, whichinclude both adherence to certain pre-cepts and actual testing of interfaces andthe devices themselves to assess what cango wrong.5

Medication Errors (MEs) associatedwith the use of IV systems have receivedattention most often as individual casereports, sometimes related to machine

malfunctions. An even bigger problem,which has resulted in a number of fatali-ties, has been the administration of over-ly high doses, often ten-fold more thanprescribed. Many of these events havebeen related to human calculation errorsand not to machine malfunctions.

There are few prospective dataregarding the incidence and nature ofserious MEs associated with IV infusionpump delivery systems.3,4 It is important tonote that while such data are important,the FDA does not generally require thembefore approving devices, and thus suchstudies are rarely performed.

Current StudyWe recently received support from

the Agency for Healthcare Research andQuality to establish a Center of Excellencein Patient Safety. One of the initial effortshas been to perform a prospective studyto assess the impact of an IV medicationsafety system that has both error-preven-tion and process-improvement data-col-lection capabilities. Our study had the fol-lowing goals: 1) to assess the incidenceand epidemiology of serious MEs associ-ated with IV infusion pump delivery sys-tems in critically ill patients; 2) to evalu-ate the impact of an IV medication safetysystem on the serious medication errorrate; and 3) to assess the impact onresource utilization of the device by com-paring the length of stay and costsbetween the intervention and controlgroups.

The study has only recently beencompleted, and we are still finalizing thedata collection. Nonetheless, some earlyqualitative findings are apparent. Early inthe study, it became clear how complexcritical drug infusion can be. Patientsincluded in this study were in the cardiac


Key Points:

• Administration errors may be particularly hazardous, because most are not intercepted with traditional approaches.

• Intravenous medication safety represents a particularly vulnerable area, especially in the ICU.

• An IV medication safety system has the potential to substantially improve safety, but studies will be needed to determine the extent towhich it will achieve that potential.

PROCEEDINGS

Intravenous Medication Safety ErrorsDavid W Bates, MD, MSc1,2,3; Jeffrey M. Rothschild, MD, MPh1,3; Carol Keohane, RN 1,3. Division of General Internal Medicine and Primary Care,

Brigham and Women's Hospital,1 Harvard School of Public Health,2 and Harvard Medical School, 3 Boston, MA

surgical intensive care unit (ICU), wheremultiple infusions were the rule.Handoffs occurred when patients camefrom the operating room to the ICU viathe recovery room or directly, and theirtreatments were more complex thananticipated. The study was to be per-formed as a randomized controlled trial,but this proved impractical, because itcould not be determined at the outset of anoperation to which cardiac surgery ICU thepatient would go at its conclusion.Instead, an "on-off" methodology was used.

It was also found that physicians andpharmacists often did not understandprogramming the systems and movingthem from place to place, but that nurseshad a better understanding of this. Whileissues around the initial programming ofthe systems were targeted initially, itquickly became apparent that this did notrepresent most of the programming.Issues related to reprogramming the sys-tems were found to be as or more important.

The strategy from the beginning wasto use the IV medication safety systemlogs to identify events, to determine whatalerts were being provided by the safetysoftware, and what the responses to thealerts were. Data were downloaded from

the systems regularly to obtain this infor-mation. The early version of the softwaredid not include a patient identifier, whichmade it difficult to associate the datawith a patient, although it was possible todo this in the study setting. It becameclear that for this information to be mostuseful for quality improvement, it shouldbe linked both with a patient identifierand a nurse identifier. These data couldthen be used in follow-up evaluation ofspecific cases for quality improvement.

These log data may be useful in avariety of ways. For example, it will beimportant to assess how often warningsare given, and whether some warningsshould be removed or changed withrespect to level of severity. In particular,for some medications "low end" warningsmay not be useful, as "taper to off" con-tinuous infusion orders are common andit is unclear what dose is too low beforemedication should be discontinued. It willalso be possible to ask questions such aswhether some clinicians are more likelythan others to override the alerts, thoughthis may raise concerns to many about"big brother." While nursing may raiselegitimate concerns regarding thisapproach, and malpractice protection inparticular is important, this will representan extremely valuable resource from thesafety perspective.

Another interesting preliminary find-ing related to identifying dose limits. Itwas found that there was wide variationwith respect to dose limits for IV medica-tions, but in order to use the safety sys-tem, hospital-wide consensus dose-limitstandards had to be reached. This issueclearly presents opportunities for error.

A number of issues remain. It is clearthat greater degree of agreement aboutmaximal doses for IV medications is need-ed. Some areas appear to be particularlychallenging: for example, oncology, inwhich a large number of different types ofhighly toxic infusions are used, and exper-imental protocols with atypical dosing

ranges exist. Developing alerts for bolus IVdoses is also difficult. It was also apparentthat software and technology to make itpossible to transfer orders from comput-erized ordering and pharmacy applica-tions to the IV medication safety systemmight make it possible to reduce errorrates even further. Forcing functions–while very powerful–must be used spar-ingly and judiciously.

ConclusionsIV drug safety represents a particu-

larly vulnerable area, especially in the ICU.Administration errors are especially haz-ardous compared to other stages of themedication process, because most are notintercepted with traditional approaches.IV medications are risky compared toother routes, because of the high toxicityof the medications involved and the rapidbioavailability when drugs are deliveredvia this route. An IV medication safetysystem has the potential to substantiallyimprove safety, but the extent to which itwill achieve that potential remains to bedetermined.

References1. Bates DW, Cullen D, Laird N, LA, et al.

Incidence of adverse drug events and potential adverse drug events: implicationsfor prevention. JAMA 1995; 274:29-34.

2. Leape LL, Bates DW, Cullen DJ, et al. Systemsanalysis of adverse drug events. JAMA1995; 274:35-43.

3. Schneider PJ. A review of the safety of intravenous drug delivery systems. HospitalPharmacy 1999; 34:1044-1056.

4. Bates DW, Cousins DD, Flynn E, et al. Consensus Development Conference Statement on the Safety of Intravenous Drug Delivery Systems: Balancing Safety and Cost. Hospital Pharmacy 2000; 35:150-155.

5. Gosbee J. Human factors engineering and patient safety. Quality & Safety in Health Care 2002; 11(4):352-354.

Intravenous Medication Safety Errors


TABLEIntravenous Drug Safety• Intravenous delivery a vulnerable

spot

— Hard to detect—true rate higher than realized

— Very high severity level

— Hard to intercept without changing systems

• Effective interfaces will be key because integration critical

• Expertise in “engineering out” errors pivotal

— Need to build approaches that help clinicians do what they intend, don’t rely on training

The recently published Institute ofMedicine (IOM) Report, "PatientSafety: Achieving a New Standard

for Care," details a plan for the collection,coding, and classification of patient safe-ty information.1 To achieve an acceptablestandard of patient safety, it is recom-mended that comprehensive patient safe-ty programs that include adverse eventand so-called "near miss" detection andanalysis be established.

Near Miss AnalysisThe importance of near miss analysis

is the subject of an entire chapter in theIOM report. Near misses are relatively fre-quent compared with actual errors andrepresent circumstances in which poten-tial adverse effects have been avoided.For both reasons, they represent richlearning opportunities.

As other industries including aviationhave realized, near miss analysis providesa method of understanding underlyingsystem weaknesses and root causes oferror, so that systems can be redesignedand simplified to prevent adverse eventsand improve safety. The new medicationmanagement standards of the JointCommission on the Accreditation ofHealthcare Organization (JCAHO) rein-force the importance of active surveil-lance and reporting systems to achieve abetter understanding of medicationerrors.

Analysis of near misses has theadvantage of fewer barriers than thereporting of actual adverse events. Nearmisses cause less concern about potentialissues of blame or malpractice litigation,compared with actual adverse or sentinelevents.

Andrew Chang, JD, MPH, project

director in JCAHO's Division of Researchrecently said, "Because of the abundanceof near misses, data from analyzing themmay provide the means to distinguishrandom fluctuations from actual trends,and therefore will be useful for statisticalmonitoring purposes and developing use-ful interventions to enhance patient safe-ty." He goes on to state that near missanalysis may provide a linkage betweenhighly visible yet rare actual failures andvery frequent but nearly invisible latentconditions. Such a linkage would help indeveloping a predictive model that couldbe used to recognize emerging errors andprevent them from occurring.2

IV Medication Safety SystemA new tool for active surveillance and

analysis has become available with theintroduction of an IV medication safetysystem designed to avert IV medicationerrors. An IV medication safety systemenables a hospital to define best infusionpractices, including dosing limits, and toincorporate rules based on these practicesinto safety software that can perform afinal "test of reasonableness" within theinfusion system at the point of care.

When an IV medication safety systemis programmed to deliver a drug outsideof best dosing practices, the safety soft-ware provides an alert when the "start"key is pressed. Infusion cannot begin untilthe alert is addressed. The alert gives a cli-nician an opportunity either to adjust thedosing parameters or to override the alertand proceed. Other conditions may alsolead to an alert. For example, if a channelon a multiple channel infusion is pro-grammed to deliver a drug already infus-ing on another channel, a "Same DrugInfusing" alert is provided. In this way,programming steps that otherwise might



Key Points:

• Detecting and analyzing "near misses" achieves an acceptable standard of patient safety as recommended by a recent Institute of Medicine report.

• Analysis of near misses may provide a link between actual failures that are highly visible but rare, and latent conditions that are very common but nearly invisible. This might help to develop a predictivemodel for recognizing emerging errors and preventing them from occurring.

• A new IV Medication Harm Index, which is being developed to help analyze intravenous (IV) medication near misses, estimates the potential harm averted through the use of an IV medication safety system.

• Analysis of 10-hospital aggregated data from an IV medication safetysystem shows that IV medication near misses occur at the following rates:

— Minimal harm potential = 2.2 per 1,000 patient days

— Moderate harm potential = 0.8 per 1,000 patient days

— Severe harm potential = 1.3 per 1,000 patient days

PROCEEDINGS

Potential for Harm in IV System Programming: A 10-Hospital Analysis

Rick Crass, PharmD, Senior Manager, Clinical Marketing, ALARIS Medical Systems, San Diego, CA

lead to a medication error are brought toa clinician's awareness and corrected asthey occur, thereby improving patientsafety and reducing the potential forharm.

More than just a smart pump, an IVmedication safety system includes con-tinuous quality improvement (CQI) logsthat record each programming alert, sothat aggregate data can be collectedwithin an institution to analyze the avert-ed errors and to gain a deeper under-standing about when and why they occur.All events associated with an alert arelogged for subsequent retrieval andanalysis. Some system programmingevents that do not lead to an alert butthat may also be instructive in learninghow medication errors occur are alsologged. For example, if a drug is selectedfrom the library to initiate an infusion andthen the selection cancelled, this event islogged. The safety system company oftenworks with hospitals employing the safe-ty software to create an aggregate data-base of alerts associated with the use ofthis technology.

Several of the other articles in theseProceedings discuss the individual experi-ence of hospitals employing an IV med-ication safety system and lessons learnedfrom analysis of CQI data. The remainder

of this article will focus on the findingsfrom the 10-hospital, aggregated froman IV medication safety system CQI data.

Aggregated Data on Near Misses

Figure 1 lists the top 10 most fre-quently reprogrammed IV medicationswhere the initial alert was "Dose AboveMaximum Limit." The list contains sever-al entries common to other lists of prob-lem-prone medications: drugs such asopiates, sedatives, heparin and insulin.3,4

Many types of near misses associatedwith programming errors have been seen,including programming the mL/hr infu-sion rate or volume to be infused as thedose, programming a "0" instead of a dec-imal point, factor-of-10 errors, and extradigits that are deleted when the infusionis reprogrammed following an alert.

Assessing Averted HarmA question frequently asked in hospi-

tals during the review of their CQI data is"Are all near misses created equal, or arethere some averted IV medication errorsmore likely to be associated with harmthan others?" To answer this question, aproject was begun in the first quarter of2003 to create a harm potential modelthat could be applied to near-miss events

recorded in the safety software CQI logs. Although there are tools that can be

used to define harm from adverse drugevents that actually occur (e.g., theNational Coordinating Council forMedication Error and Prevention [NCCMERP] tool), there has not been a tool toassess harm for medication errors that areprevented and therefore never actuallyoccur. Using an adaptation of a conceptput forth in an article from the medicalinformatics literature,5 a proposed harmindex model was developed that uses themagnitude of the dose at the time of a"Dose Above Maximum Limit" alert (e.g.,2.5 times, or 5 times upper limit) as thebasis for stratifying averted IV medicationerrors into three categories: minimalharm potential, moderate harm potential,and severe harm potential (Figure 2).

This model was applied to "DoseAbove Maximum Limits" alerts in the 10-hospital database. The results indicatedthat 51.3% of near misses had minimalharm potential, 19.3% had moderateharm potential, and 29.4% had severeharm potential. Based on the number ofpatient days during which alerts wereprovided in the 10 hospitals, the rate ofevents with minimal harm potential is 2.2per 1,000 patient days, 0.8/1,000 forevents with moderate harm potential, and1.3/1,000 for events with severe harmpotential. Although the minimal-harm-potential events typically would not beexpected to result in a need for addition-al medical intervention or costs, the mod-erate- and severe-harm-potential eventscould be expected to affect patient statusand costs if the error had not been avert-ed by the safety software. A previouslypublished white paper describes thismodel and its application to some nearmiss data from CQI data.6

IV Medication Harm IndexIn July 2003, an interprofessional

group of clinicians with experience with



FIGURE 2.

Categorizing Harm byMedication Type

FIGURE 1.

10 Most Frequently Repro-grammed IV Medications

(Dose Above Maximum with Reprogramming) n = 477

dopamineheparinfentanylmilrinonemidazolampropofolepinephrinemorphinevecuroniuminsulin

140

66

29

27

27

21

19

18

16

15

0 10 20 40 60 80 100 140

Note: All bars with indicate high risk medication.


patient safety and medication errors pro-grams was convened for a conference onIV medication harm. Representatives frommedicine, nursing, pharmacy, and nation-al groups with a stake in patient safety(e .g . , the Inst i tute for Heal thcareImprovement and the American HospitalAssociation) reviewed the harm modeldescribed above and proposed enhance-ments to it. The outcome of this meetingwas the proposed IV Medication HarmIndex that can be applied to evaluate IVmedication near miss data.

This new harm index uses informa-tion about the drug, its inherent risk, themagnitude of the averted dose error, thepatient's acuity as determined by usingthe patient care area location as a surro-gate marker, and the likelihood of detec-tion of an adverse drug event, if a med-ication error had reached a patient. Usingthe IV Medication Harm Index results in asingle number score for each near-missevent. The range of possible scores is 3.5for near misses with the least harmpotential to 14 for near misses with thegreatest harm potential.

The Harm Index calculation has beenperformed for the near miss data for"Dose Above Maximum Limits" for sevendrugs 10-hospita l data (F igure 3) .Additional work is underway to pilot theuse of the Harm Index.

Dr. Sullivan describes this conferenceand the resulting Harm Index in greaterdetail in her article in these Proceedings.

References1. Patient Safety: Achieving a New Standard

for Care, Washington, DC. National Academy Press; 2003.

2. Pew C, ed. Patient Safety Taxonomy: One Step to Reducing Errors. Oakbrook Terrace, IL:Joint Commission Perspectives on Patient Safety. 2002;2(8):3.

3. Cohen MR and CM Kilo. High-alert medica-tions: safeguarding against errors. In Cohen, MR, ed. Medication Errors . 1999; Washington DC, American Pharmaceutical Association; 1999.

4. Hicks RW, DD Cousins, R Williams. Summaryof information submitted to MEDMARXSM in the year 2002. The Quest for Quality.Rockville, MD: UPS Center for the Advancement of Patient Safety, 2003.

5. Anderson JG, Jay SJ, Anderson M, et al. Evaluating the capability of information technology to prevent adverse drug events:a computer simulation approach. J Am MedInform Assoc. 2002;9:479-490.

6. Crass R. Improving intravenous (IV) medica-tion safety at the point of care: retrospec-tive analysis of pooled data using an innovative IV Harm Assessment Index. ALARIS Medical Systems, Inc.


FIGURE 3.

Assessing Harm Potentialfor Near Misses

(Dose Above Maximum with Reprogramming) n = 477


Advances in IntravenousMedication Safety

Computerized intravenous (IV) infu-sion devices —so-called "IV medicationsafety systems"—incorporate institution-established dosing limits and otherparameters to provide a final check at thepoint of care to help prevent IV medica-tion errors. In addition, safety softwareautomatically records data on the "nearmisses" (programming errors) averted bythe safety system. Most importantly, thenew infusion technology provides clini-cians with tools to help prevent harm,which can now be part of the researchfocus and continuous quality improve-ment (CQI) efforts with regard to infusiontherapy.1

Error vs. HarmThe National Coordinating Council

for Medication Error and Prevention (NCCMERP) approved the following workingdefinition of medication error: "... any pre-ventable event that may cause or lead toinappropriate medication use or patientharm, while the medication is in the con-trol of the health care professional,patient, or consumer."2 The NCC MERP

definition of harm is "…death or tempo-rary or permanent impairment of bodyfunction/structure requiring intervention.Intervention may include monitoring thepatient's condition, change in therapy, oractive medical or surgical treatment."3

Since medication errors do not necessari-ly correlate to patient harm, a new assess-ment tool was needed to evaluate theseverity and potential harm of averted IVmedication errors. Harm was defined con-servatively, i.e., only in terms of whether aserious error would potentially be life-threatening.

Harm Assessment In July 2003 a consensus conference

was convened with nationally recognizedexperts in IV medication safety, includingphysicians, nurses, and pharmacists, todevelop such a tool. Conference partici-pants focused on four main areas: iden-tification of the content domain forpotential and preventable IV medicationerrors based on newly available real-timeclinical data; assignment of evidence-based degrees of severity to this clinicalmeasurement index; development ofvalidity and reliability as initialrespectable psychometric properties for

this tool; and finally, application of thenewly developed IV medication harmindex to a limited data set as a pilot test.

Methods As described below, standard meth-

ods in instrument development were usedin designing the IV Medication HarmIndex.

Identification of Content Domain.Under the direction of a facilitator, con-ference participants reached consensusregarding the content domain of poten-tial and preventable IV medication errors.Consensus was based on available dataacquired from an IV medication safetysystem, state-of-the-science evidence,and expert opinion. The conceptual defi-nition for IV Medication Harm Index wasthen determined to include specific drugrisk, degree of overdosing of specific drug,level of care as an indirect measure ofpatient acuity, and detectability ofadverse event based on specific drug. AnIV medication safety system readily pro-vides all data items needed for definingthe index.

Based on expert and evidence-baseddecisions regarding content domain andconceptual definition, the current versionof the IV Medication Harm Index consistsof three sub-scales (Table 1). As the risk ofharm or clinical severity of potential med-ication error consequences increases, thesub-scale scores and the total summatedscore for the measurement index increasein quantified values.

The Drug Risk/Overdosing sub-scaleis based on the combined consideration ofdrug risk and degree of overdosing, andhas a quantified severity score range of1.5 through 9. As reference guides forscoring, keys providing information ondrug risk and overdosing ranges, based on


Key Points:

• A conference of national experts in intravenous (IV) medication safety was convened to design an evidence-based tool to measure the clinical severity of prevented IV medication errors identified through previously unavailable data from an IV medication safety system.

• The newly developed IV Medication Harm Index consists of three sub-scales that estimate risk of harm and clinical severity of potential and preventable IV medication errors.

• Specific drugs having known high risk for harm through clinical experience, expert opinion, or available evidence corroborate high IV Medication Harm Index scores.

PROCEEDINGS

IV Medication Harm Index:Results Of A National Consensus Conference

Jacqueline Sullivan PhD, RN, CCRN, Director, Nursing Research, Quality and Outcomes, Hospital of the University of Pennsylvania, Philadelphia, PA

expert opinion and available evidence,accompany the instrument.

The Level of Care sub-scale is basedon the patient care unit where the specif-ic event was prevented and has a quanti-fied severity score range of 1 through 3.Since available evidence indicates thatpediatric and neonatal intensive careunits (ICUs) have the highest degree ofrisk associated with IV medication errors,these units have an assigned score of 3.Adult ICUs have relatively high degree ofrisk associated with IV medication errorsand have an assigned score of 2; interme-diate care units, a moderate degree of riskwith an assigned score of 1.2; and gener-al care units, a lower risk with an assignedscore of 1.

The Detectability of Adverse Eventsub-scale is based on the likelihood thatan error will be detected and has a quan-tified severity score range of 1 through 2.Specific drugs whose adverse events arelikely to be detected are scored as 1, whilethose whose adverse events are consid-ered less likely to be detected are scoredas 2, i.e., are associated with high risk.Drug-specific keys provide information ondetectability of adverse events and arebased on expert opinion and available evi-dence. These keys, which accompany the

instrument for use, are used as referenceguides for scoring.

Summated scores increase as thedegree of potential clinical severity, risk,and harm increase. The total summatedscore of the IV Medication Harm Index hasa potential range of 3.5 through 14.

Application of the IV MedicationHarm Index was demonstrated with aclinical case abstracted from de-identifiedreal-time clinical data. For CQI data thatdocument an attempted delivery ofheparin to an adult ICU patient at fourtimes over the maximum dose, the fol-lowing sub-scale score ranges areassigned. Based on reference keys, DrugRisk/Overdosing Range are both rated ashigh with a sub-scale score of 9. Level ofCare sub-scale, using the rating for anadult ICU, is assigned a score of 2. Sinceheparin is rated as a drug whosedetectability of adverse events is unlikely,based on previously described referencekeys, the Detectability of Adverse Eventsub-scale score is assigned a score of 2.Adding these three sub-scale scoresresults in a total summated score of 13for the IV Medication Harm Index for thispotential and prevented IV medicationerror (Table 2). Since clinical experienceand published literature confirm the highdegree of risk associated with IV heparinmedication errors, this clinical demon-stration preliminarily corroborates thelogic of this scale application.

Development of OperationalDef in i t i on . Both numerator anddenominator sub-scales are used in oper-ational definition for the IV MedicationHarm Index. The number of potential andpreventable IV medication errors compris-es the numerator, which may be furtherdescribed based on severity ranges andspecific drug. Based on expert opinion,available data, and current nationaltrends in quality data management, thedenominator selected for the index is1,000 patient days.

Establishment of Initial Psycho-metrics. Initial content validity for the IVMedication Harm Index was establishedusing the technique of quantification ofcontent validity, as described by Lynn.1 Themultidisciplinary experts at the confer-ence independently rated each sub-scaleand the tool in its entirety using a Likertscale ranging from 1 = not relevant to 4 =very relevant. The initial content validityscore established by the 20 experts usingthis technique is quite respectable with atotal mean quantified score of 3.451.

Inter-rater reliability for the IVMedication Harm Index was also quitepromising with a Pearson r correlationcoefficient of 0.9295. Inter-rater reliabili-ty was established by comparing agree-ment between consensus panel experts

IV Medication Harm Index: Results Of A National Consensus Conference


TABLE 2.

IV Medication Harm Index–Clinical Example

• Heparin administered to an adult ICU Patient at 4 times over the maxi-mum dose (High Overdosing Range) :– Drug Risk/Overdosing Range

(High/High) = 9– Level of Care (Adult ICU) = 2– Detectability (Unlikely) = 2– Summated Score = 13

©2002-2004 ALARIS Medical Systems, Inc. All rights reserved.

TABLE 1.

IV Medication Harm Index–Sub-scales

• Three Sub-scales based on:Drug Risk/Overdosing Range

(Score Range = 1.5-9)Level of Care

(Score Range = 1-3)Detectability of Adverse Event

(Score Range = 1-2)• Summarized Score Range:

3.5-14* Higher score = Greater Harm/Risk

©2002-2004 ALARISMedical Systems, Inc. All rights reserved.

Heparin Range: 7 - 13Mean: 12.25

Propofol Range: 10 - 13Mean: 11

Vasopressin Range: 5 - 9Mean: 7

TABLE 3.

IV Medication HarmIndex–Total Summated Score


during independent application of thetool to eight identical clinical cases from ade-identified data set.

Test-retest reliability for the IVMedication Harm Index was also quiteimpressive with a Pearson r correlationcoefficient of 0.9695. Test-retest reliabili-ty was demonstrated by having each indi-vidual panel expert apply the instrumentto the same eight clinical cases on twoseparate administrations spaced by atwo-week interval. Expert ratings werecompared between the first and secondadministration to determine the scale'sstability and consistency in evaluationover time.

Pilot TestingThe newly developed IV Medication

Harm Index was pilot tested during appli-cation to a limited data set. The measure-ment index was applied by a singleresearcher to software-acquired datadownloaded from 45 IV medication safetysystems currently in use in a medical ICUand a cardiothoracic surgical ICU of a ter-tiary-care academic health center. TheseIV systems had been used daily in bothICUs without interruption for four con-secutive months. Table 3 lists three of the

most frequently occurring potential andpreventable IV medication errors identi-fied through software data with applica-tion of their associated harm indexranges. Importantly, specific drugs knownthrough clinical experience, expert opin-ion, and available evidence to have a highrisk for harm have corroboratively high IVMedication Harm Index scores.

Future ApplicationsFurther refinement of this newly

developed instrument has the potential toimprove safety and quality in IV medica-tion administration. Plans include:

• Employment of 1,000 patient-daysdenominator

• Correlation of the IV MedicationHarm Index with a cost index

• Establishment of advanced psycho-metric properties

• Development of computerizedautomation of IV Medication Harm Index

ConclusionsAdvances in IV medication safety

system now can prevent potential dose-related IV medication errors and automat-ically gather real-time data describingthese near misses. A new assessment toolis required to accurately assess the sever-ity of harm that has been averted throughthe use of this technology. The IVMedication Harm Index has been shownin pilot testing to effectively measure theclinical severity of potential and prevent-able IV medication errors. Ongoing refine-ment and use of the measurement indexin analysis of real-time clinical data canincrease our knowledge and understand-ing of IV medication administration andthe associated risks (Table 4).

The availability of a standardized toolat the earliest stages of an IV medicationsafety system data collection and analysiscan make a significant contribution to IVmedication safety and quality improve-ment efforts. Unlike earlier error-preven-tion efforts that had to rely on retrospec-

tive data, IV medication safety system andits associated CQI data offer unprecedent-ed opportunities to learn from mistakesbefore they are made.

References1. Wilson K, Sullivan S. Preventing medication

errors with "smart" infusion technology. Am J Health-Syst Pharm. 2004;61:177-83.

2. http://www.nccmerp.org/ aboutmederrors.htm3. Cited in: USP, Summary of the 1999

Information Submited to MedMARxSM, Glossary:19.

4. Lynn M. Determination and quantification of content validity. Nurs Res . 1986;35(6):382-5.

TABLE 4.

Summary and ConclusionsIterative combination and refinement of

innovative safety technology withadvanced analytical strategies:

• Can effectively measure the clinical severity of potential and preventable IV medication errors

• Can increase our knowledge and understanding of IV medication administration and the associated risks

• Can make a significant contribution toIV medication safety and quality improvement efforts

• Offer unprecedented opportunities to learn from mistakes before they are made

©2002-2004 ALARISMedical Systems, Inc. All rights reserved.

IV Medication Harm Index: Results Of A National Consensus Conference


St. Joseph’s/Candler Health System(SJC) is a three-hospital systemcomprised of two acute-care, terti-

ary referral centers in Savannah and onerural hospital located in an outlying com-munity. The Savannah facilities include St.Joseph's Hospital, which has 304 bedsmade up of various adult medical andsurgical specialties, and the CandlerHospital, which has 340 beds and pro-vides adult and pediatric care. Candler isalso the primary maternity facility in thec i ty . Th is paper br ief ly descr ibesthe implementation of the Medley™Med ica t ion Safe ty Sys tem wi thGuardrails® Safety Software at these twohospitals and preliminary results from theanalysis of data derived from event logsthat accumulated automatically duringuse of the system.

During the last four years SJC hasbeen actively engaged in assessing newtechnology designed to improve medica-tion safety for its hospitalized patients.This assessment has included bedsidemedication verification and documenta-

tion using bar-coded labels for medica-tions, patients, and caregivers; computer-ized physician order entry (CPOE); and anIV medication safety system that incorpo-rates medication safety features withdose limits for drugs. Concurrently withtechnology assessment, SJC has imple-mented multiple strategies to engagephysicians, nursing staff and other care-givers to improve the recognition andreporting of medication errors, so that thedesign of the medication use process canbe improved in ways that reduce thepotential for error. Our activities andassessments validate the observation thatthe medication use process is extremelycomplex with many steps performed byhumans in conjunction with machines.More than 250 steps are involved in themedication administration process fromthe time of ordering of drugs by thephysician to their administration to thepatient by a nurse. Errors can occur inany step.

Prioritizing TechnologyInvestment

It is believed by the Institute for SafeMedication Practices that the combina-tion of CPOE, bar coding and safety infu-sion systems may finally provide a soliddefense against the most serious medica-tion errors.1 However, a dilemma for manyinstitutions is the cost of implementing allof these systems concurrently when fiscalconstraints limit the availability of fundsfor capital investment. SJC selected an IVmedication safety system for its initialcommitment of funds based on the beliefthat this could contribute the greatestbenefit to patient medication safety in theshortest amount of time. This conclusionwas reached in part as a result of statis-tics that indicate that 25% of all medica-tion doses administered in our hospitalsare given by the intravenous (IV) route(Table 1). IV medications have the greatestpotential for producing harm.2 Therefore,we concluded that an IV medicationsafety system, which incorporates dosinglimits and data logs for process improve-ment, would have the greatest impactfor risk reduction to patients treated inour institutions.

IV Medication SafetySystem Implementation

In October 2002 SJC implementedthe Medley™ System with the Guardrails®Software in all three of its hospitals. Thisimplementation was preceded by severalmonths of work performed by clinicalpharmacists, nurses and physicians at SJCto develop the infusion programming lim-its and other safety software parametersinherent to the system. Additionally, stafftraining using internet-based educationalmodules provided by the company was


Key Points:

• St. Joseph’s/Candler Health System selected an intravenous (IV) medicationsafety system for its initial investment of funds for hospital-wide implementation, having concluded that this technology, which incorporates dosing limits to avert intravenous medication errors, would have the greatest potential to reduce risk of harm to patients.

• Data logs in the IV medication safety system documented a 7.2% rateof pump reprogramming (i.e., "near misses"), which suggests that a significant potential for harm is being averted through the use of this technology.

• The safety software in this IV medication safety system is providing not only interdiction of untoward events but also information through its data storage and retrieval characteristics that is useful to continuous quality improvement of medication use at St. Joseph’s/ Candler Health System.

PROCEEDINGS

Using IV Medication Safety System Logs — A New Tool for Identifying Averted Harm

Ray R. Maddox, PharmD, Director, Clinical Pharmacy, Research and Pulmonary Medicine, St. Joseph’s/Candler Health System, Savannah, GA

conducted. All three hospitals were simul-taneously converted from the existing IVpumps to the new systems in a single 24-hour period without significant difficul-ties. The unusual success of the imple-mentation process is attributed to the col-laboration of multiple groups of people,including nurses, pharmacists, physicians,medical/surgical buyers, and administra-tors. This collaboration assured a highlevel of compliance with the use of thetechnology by nurses after the IV medicationsafety system was installed. The implemen-tation experience is outlined in Table 2.

ResultsIn the months since implementation,

"event" data have been accumulated inthe computer "brain" in each IV medica-tion safety system. An event is defined asany alert given to the nurse/caregiver thatrequires a reevaluation of their systemprogramming selection. All events are col-lected by the system until purged at des-ignated times by SJC. These data havebeen synthesized for analysis for the timeperiod October 2002 through June 2003,which represents nine months of our ini-tial experience with the safety software atSt. Joseph's and Candler Hospitals. Keyresults of these data are summarized inTable 3 and provide information from the545 systems in use in these two hospitals.The data indicate that most alerts (57%)

given to nurses were warnings of the pos-sibility of drug overdose.

Of the total 8,294 recorded events,7,317 (88%) were associated with theadministration of two drugs: propofol andoxytocin. In the ICU propofol is adminis-tered for ventilator-related sedation, andin the pregnant female oxytocin is admin-istered during the induction and postpar-tum delivery period. Propofol is a medica-tion frequently given as a bolus in addi-tion to a constant rate infusion to main-tain sedation in the ventilated patient.Our data include these bolus propofoldoses and provide insight into theamount of this medication per unit oftime administered to patients in the ICU.The data indicate that the bolus dosemode for system setting by nurses wasnot used appropriately during this timeperiod. Additionally, there may be need forfurther evaluation of clinical outcomesassociated with the current process anddosing methodologies of propofol forthese patients. The oxytocin data corre-spond to the current practice of highinfusion rates of the drug in the postpar-tum patient to expel afterbirth; these highinfusion rates exceeded those pro-grammed into the system for constantrate infusions during the induction period

and therefore, resulted in system alerts(events).

The data also indicate that 598events (7.2%) resulted in the nurse can-celing the administration process orresetting the system. These warningsinvolved multiple medications includingsome of those identified by the USPMEDMARXSM as being associated with thehighest liability for harm.2 As indicated inTable 4, 76 potential overdoses of heparinand six potential overdoses of insulinwere averted. In 30 instances for multipledrugs, doses greater than 10-fold themaximum drug-library limit werecancelled. We believe these 598 eventswere potentially serious medicationerrors that were prevented by theMedley™ System and that can be charac-terized as "near misses." Given that therewere 995,570 IV doses of drugs adminis-tered at SJC in our fiscal year 2003, a7.2% rate of system reprogramming sug-gests a significant potential for harm isbeing averted through use of an IV med-ication safety system.

In addition to averting harm, the sys-tem allowed us to discover that nursingstaff would be able to program the systemmore easily and deliver the correct dose ofmedications, if IV drug labels were refor-mulated to include the total volume andamount of drug. The technology alsoallowed us to redesign our heparin proto-

Using IV Medication Safety System Logs —A New Tool for Identifying Averted Harm


TABLE 1.SJC Medication

Statistics• 4,014,396 doses administered in

FY03 in 2 tertiary-care hospitals• 62% of doses were injectable• 40% of injectable doses were IV• 995,570 of all doses were IV

medications

TABLE 2.

SJC ImplementationExperience

• Drug library max-min infusion parameters development

• Extensive pre-implementation training

• System implemented in October 2002

• 600+ systems in 3 hospitals• 98%-100% compliance by nurses

with use of Guardrails® Data ("event") collection October 2002 –June 2003

• Data analysis from 525 systems at SJH and CH

TABLE 3.

Event Rates for 525 IVMedication Safety Systems

• 8,294 "events" resulted in warnings to nurses

• 4,746 (57%) dose greater than maximum

• 2,571 (31%) associated with propofol & oxytocin

• 598 (7.2%) reprogrammed or cancelled processes—presumed tobe averted medication errors

col. Our former heparin protocol requiredthe nurse to send an order form to thepharmacy to calculate a rate of infusionbased on milliliters per hour. With the IVmedication safety system, nurses can nowprogram the system by units per kilogramper hour. This one change eliminated atleast three steps in the medicationprocess, multiple calculations, and multi-ple opportunities for error.

Increased efficiency also resultedfrom another unexpected use for thesystem that involves our neonatal andpediatric patients. Previously, whenorders were written for infrequently useddrugs, pharmacists spent time research-ing and compiling information to deter-mine the correct concentration for apediatric patient. Now, they are able toquickly reference the dose-checking drugdatabase, knowing that the informationis adequately backed by the currentliterature.

ConclusionOur preliminary data suggest that in

the past, prior to the implementation ofan IV medication safety system, eventsmay have been occurring that wereobscured by "routine" clinical occurrencesin patients that may have been misinter-preted as part of the course of their dis-ease process versus iatrogenic in nature.The Medley™ System with Guardrails®Software is providing not only interdic-tion of untoward events but also infor-mation through its data storage andretrieval characteristics that is useful tocontinuous quality improvement (CQI) ofmedication use at St. Joseph’s /CandlerHealth System.

References1. Institute for Safe Medication Practices

Newsletter, February 7, 2002.2. USP MEDMARXSM, Analysis of Participating

Hospital Data, 2001.

Using IV Medication Safety System Logs—A New Tool for Identifying Averted Harm


TABLE 4.

Selected Data from SJC

Item SJH CH Total Comment

Reprogrammed 264 257 521 6.2% of events

Cancelled 32 45 77 1.0% of events

Overdoses Prevented

Heparin 34 42 76

Insulin 2 4 6

10-fold 13 10 23 multiple drugs

50-fold 2 5 7 heparin, dopamine,propofol

Context Children's Hospital and Health

Center, San Diego (CHSD) is an independ-ent, nonprofit healthcare organizationthat offers comprehensive pediatric med-ical care through secondary and tertiaryspecialty outpatient clinics and inpatientcare. Children are at high risk for medica-tion errors and adverse drug events(ADEs) for a number of interrelated rea-sons. First, drug absorption, transport,metabolism, and excretion vary by age.Second, the weights of children vary dra-matically. For these reasons, dosages mustbe carefully calculated with most medica-tions being prescribed on a weight basis.Mathematical errors, including misplaceddecimal points, can result in 10-folddosing errors. Third, most drugs come inonly a limited number of dosages, so cus-tom dosage preparation is generally nec-essary for children. This extra work alsocreates the possibility for errors. Fourth,children lack the physical reserves towithstand the adverse consequences oferrors that do occur. For these reasons,the ADE rate in pediatrics is three times

higher than in adults. Intravenous (IV)medications are involved in 54% ofpotential ADEs in pediatric inpatients.Prevention of IV medication errors needsto be a priority when considering invest-ments in patient safety technology(Figure 1).

Until recently, CHSD used four typesof infusion devices, each with their ownuser interface. Limited dosing informationwas available at the point of care, andpumps were permissive (i.e., any dosagecould be programmed; there were nodosages that the pumps could not deliverwithin pump's operating limits, e.g., 999mL/hr). A single caregiver set up a deviceand programmed it for use with a givenpatient. There was no double check. Tocompound these limitations, there was noeasy way to accurately discover medica-tion administration errors that wereoccurring, which ones were interceptedand which were not.

ChangeIn January 2002, in order to address

these limitations, CHSD decided 1) to

reduce the number of devices available(standardize) and 2) to invest in smartpumps. Smart pumps are programmablepumps that have comprehensive, user-determined drug libraries. These librariesare programmed with both so-called"hard" (impassable) and "soft" (passable)dosing limits for each drug. Alerts are sig-naled whenever these limits have beenexceeded. Smart pumps have continuousdisplays of the selected drug's name anddose, and of doses being infused outsideof soft limits (if any). Smart pumps alsohave a comprehensive log that recordsdosing limit alerts and subsequent actions(such as bypassing a soft-limit alert, orreprogramming after a hard-limit alert).

To achieve CHSD's objectives, a mul-tidisciplinary group of 15 senior nurses,three physicians (including anesthesiolo-gists and senior care area physicians), twopharmacists, one biomedical engineer,one materials management representa-tive, and CHSD safety personnel was con-vened to evaluate and select a pump forhospital-wide use. The inclusion of abreadth of nursing input was deliberate,both for nurses' knowledge as contentexperts and in the belief that nursing buy-in was essential to successful implemen-tation and use of the new devices. Thetimeline for selection was three months.

When the selection process had nar-rowed the field to two candidate pumps,these devices were made available to thegeneral nursing staff for their review andinput. This step provided additional use-ful information and served to engage thenursing staff at large in the decision-making process

After one pump was selected, agroup of five senior nurses, five physi-cians (including anesthesiologists andsenior care area physicians), one pharma-



Key Points:

• Intravenous (IV) medications are involved in a high percentage of potential adverse drug events, and prevention of IV medication errors needs to be a priority when considering investments in patientsafety technology.

• At Children’s Hospital and Health Center, enormous value has been derived from the use of smart pumps, which have intercepted significant IV medication errors and have identified the most common types of intercepted errors, the most common times and themost common locations at which they occurred.

• Successful conversion to smart pumps was the result of bedside nurses having ownership of the selection and implementation process and data detailing intercepted errors.

PROCEEDINGS

A Medical Center's Experience Using Smart InfusionPumps to Manage Medication Administration

Glenn Billman, MD, Medical Safety Officer, Children's Hospital and Health Center, San Diego, CA


cist, and the medical safety officer met todesign the drug library, including hardand soft dosing limits, for nearly 100drugs. Each of the five nurse/physicianpairs was responsible for providing input,review, and approving the profile for theircare area. The approval process wasfelt to be an important reflection ofthe personal accountability for thesecritical decisions.

The new pumps and their librarieswere evaluated briefly before proceedingto full deployment. An important discov-ery at this juncture led to the decision toembed the anesthesia drug library withineach of the care area profiles rather thanas a stand-alone directory. This changewas made to ensure that access to emer-gency medication dosing alerts wasunfettered. Once this change was made,the 450 new pumps were rotated intoservice in October 2002. The time intervalfrom selection of the smart pump manu-facturer and model to the go-live datewas two months.

MeasuresThe smart pump collects data that

can be used to evaluate patient safety

without labor- and time-intensive (andoften practicably infeasible) chart reviews.Pumps collect data on medicationdosages; the number of alert events (i.e.,when hard or soft limits are exceededduring programming); staff responses toalerts (i.e., canceling drug selection;reprogramming weight, dose, or durationinformation for the selected drug; over-riding the alert); and dates, times, andlocations. Data obtained from the pumpsare provided as an Excel database.

ResultsFrom CHSD's perspective, enormous

value has been derived from the use ofthe smart pumps. First, the pumps haveprovided an exciting window of investiga-tion into the medication administrationprocess. Before the introduction of thesmart pumps, performance monitoring ofthe medication infusion process reliedalmost entirely upon completing and sub-mitting occurrence reports. The informa-tion provided by the pumps is much morecomprehensive and detailed than theinformation from occurrence reports.Specifically, in six months after the pumpswere placed into action, over 4,000 alerts

were recorded by the devices. Second, the use of these pumps has

prevented many significant errors fromreaching the patient. While the majorityof the 4,000 Guardrails® Alerts wereinterpreted to reflect the aggressive phar-macologic management of patients with-in the critical care environment, about12% of the alerts led to the reprogram-ming of the infusion device (Figure 2). Theinstances when a pump was repro-grammed after an alert were interpretedas intercepted and prevented medicationerrors. The majority of errors that wererecorded exceeded the alert limits by afactor of less than 1.5. Some of the inter-cepted errors involved high-alert medica-tions at increased multiples (> 2.5 timesthe alert limit), suggesting that some ofthe errors had a significant potential tocause harm if the dose were actuallyadministered.

Of particular interest was the findingthat the magnitude of dosing errors thatresulted in reprogramming was notrandom; 10-fold errors were especiallyprominent. This was interpreted to reflecteither a misinterpreted physician order ora misplaced decimal point during infusionrate programming. Review of the datarevealed that 10-fold programming errorsmost commonly involved dopamine.Interventions resulting from this findingare underway.

At CHSD, the time-based data pro-vided by the smart pumps also revealed

A Medical Center's Experience Using Smart Infusion Pumps to Manage Medication Administration

FIGURE 1.Prioritizing Investment in Safety

FIGURE 2.Action Taken When

Programmed Dose ExceededGuardrails® Limit

that alerts occurred much more frequent-ly at 18:00 hours (Figure 3). A variety offactors were felt to have contributed tothis finding. Specifically, 18:00 hours isthe last hour of a 12-hour shift. It is alsothe peak trauma care period, and the timewhen most of the next day's electiveadmissions occur. Concurrently, 18:00hours has also been one of the pharma-cy's scheduled medication delivery times.In light of the data provided from thepumps, these competing conditions, tasksand responsibilities are now felt to havepredisposed to errors occurring at thistime of day. Discussions about methods toredistribute some of these tasks are underreview. Smart pump data will be used tomeasure the effectiveness of selectedinterventions.

Lessons learned Analysis of smart pump data identi-

fied the most common types of intercept-ed errors, the most common times andthe most common locations at which theyoccurred. These data have confirmed thatharm is not random and have helpedidentify opportunities for improvement.By focusing on errors that were caught(reprogrammed events), the pumps werecast in a non-threatening light—the kindof light most likely to sustain gains.

Modern medical management reliesheavily upon infusion therapy. Nursingstaff depends upon their ability to useinfusion devices. Nurses can therefore behighly resistant to changes involving theinfusion systems to which they are accus-tomed. In retrospect, two central factorswere felt to be most responsible for thesuccess of our conversion. First, thebedside nurses had ownership of theselection and implementation process.Second, the ability of the pumps to detailerrors that were intercepted and prevent-ed has dramatically underscored the sig-nificant personal value that the use of thepumps has provided to the nursing staff.

In the future, more real-time dataevaluation is planned. It is now recog-nized that data would be more useful ifpumps recorded the total number of infu-sions, so that error rates can be calculat-ed. Data regarding patient identity andstaff identity also will be helpful so that

additional information can be collected(via chart review) when alerts wereoverridden. The smart pump vendor iscurrently evaluating all three of theserequests.

While the work described here tookplace in a pediatric hospital, it speaks tothe importance of the ability for anysystem to respond to patient-specificconcerns, whether they are related to size,maturation, physiologic maturity, or vary-ing degrees of organ dysfunction. Theseare not concerns unique to pediatrics.They unite all organizations that care fordiverse patient populations. As such, thelessons learned at CHSD should haverelevance for all organizations whereinfusion pumps are used.

References1. Kaushal R, DW Bates, C Landrigan, et al.

Medication errors and adverse drug eventsin pediatric inpatients. JAMA . 2001;285:2114-20.

A Medical Center's Experience Using Smart Infusion Pumps to Manage Medication Administration


FIGURE 3.

Chronogram of Guardrails® Events

Does intravenous (IV) therapy needto be a higher priority patientsafety issue?

Participants agreed that IV therapy is ahigh-risk method for administeringmedications and needs to be a higherpriority as a patient safety issue.

• Most medication error studies have been done with drugs being adminis-tered through a non-intravenous route, so error rates with IV therapy have not been well documented.

• Early experiences with the continu-ous quality improvement (CQI) logs from smart pumps have shown that IV medication administration errors detected and prevented are actually quite common.

• There are more variations in drug concentrations and base solutions with IV therapy compared with otherroutes of administration, making standardization difficult.

• The doses of medication adminis-tered by the IV route often vary, change, and are determined by the weight of the patient, making errors more common.

• Because of these variations and the acuity of patients who receive their drug treatment by this route, it is harder to build standardization into the system of IV medication adminis-tration, compared to other routes of administration.

• Harm is more likely to result from errors in administering medications through the IV route than through other routes of administration.

"I think the reason it's not high on thepriority list is we don't have our handsaround the data, which are still beingascertained." Christopher Shaffer,PharmD, BCPS

"There may well be thousands ofpatients who die every year, and weattribute their deaths to the underlyingillness, when they are actually dying oferrors. Without IV medication safetysystems, we just don't know about it."David Bates, MD

"A 'near miss' may be a dose that wasprogrammed in units that were differ-ent from what was intended. A thou-sand times difference between amicrogram and milligram-those arehuge multipliers." Tim Vanderveen,PharmD, MS

"We've learned so much about IV ther-apy in the last couple of years based onthe smart pump CQI data; before wehad those data, we really did not knowthis enormous variability in best prac-tices, in concentrations, drug names,etc., exists." Rick Crass, PharmD

Are smart pumps are an effectiveway to improve and measure IV drugadministration safety?

It was agreed that IV medication safe-ty is likely to become a bigger issuebecause of improvements in ways toavert errors and detect near missesusing CQI logs from smart infusionpumps.

• Medication error detection systems currently in use are either not quan-titative (voluntary event reports), or too time consuming (chart review).

• Using information from these logs not only identifies the problems or

medications that are most common-ly associated with errors, but also theanalysis of this information can facilitate the identification of latent conditions that need to be corrected.

• Data logs can also be used to measure improvements in IV medica-tion safety.

"We're going to fix a lot of errors thathave never been reported, by usingthese pumps. It's a real-time surveil-lance tool. It's a real-time tool to dis-cover the mistakes that are going onout there as well as mitigate againstthem." Nancy Pratt, MSN

"We had spent three and a half monthsstandardizing concentrations of solu-tions, but we still had at least 82 criti-cal errors that could have resulted inharm, had the IV medication safetysystem not been in place." RayMaddox, PharmD

"When we've taken these 'near miss'data back out to the nurses, they per-ceive enormous value to the data, andthe data have been very compelling tothem, both to extend the change andto continue the process." GlennBillman, MD

"So, we've got to get this informationout to them (nurses) as soon as possi-ble, because, if you incentivize them,they you won't get the bypasses.They're not going to do the work-arounds if they understand what theimportance of it is to them and fortheir patients." Elizabeth Plant, DipClinPharm

"I have never had difficulty getting anurse involved in trying to understandor wanting to deal with the data athand." Norma Barr, RN, MN



PROCEEDINGS

Roundtable Discussion Summary

"What's not yet recognized in pharma-cy is the bigger piece of this IV medica-tion safety system, and that is its valuein terms of the CQI data as it relates toreduction of medication errors."Ray Maddox, PharmD

How important is improving the safe-ty culture in an organization as anantecedent to adopting new technolo-gy, such as smart pumps?

It was agreed that a culture of safety iscritical to making any change in anorganization, especially a change thatinvolves a new technology. This is truefor several reasons.

• First, a capital investment is required;this requires diverting money from revenue generating programs to those that improve patient safety.

• Second, the changes required to introduce new technology affects the work of staff, often increasing the amount of time required to do it.

• Third, the information generated by atechnology like smart pumps can only be used to make improvements if a culture of safety exists.

"A moment of insight is worth a life-time of experience." Oliver WendellHolmes (Note: Did NOT attend theconference!)

"The biggest driver of cultural changeis actually the senior people in theorganization." Dave Schlotterbeck, CEO

"One of the ways to get things toprogress more rapidly is when there isa broader understanding of what theissues are and where there's value.One of the things I believe is thathands-on caregivers don't actuallyhave the tools they need to do a goodjob. Given the tools, they will be waymore effective, and they actually wantto use the tools." Richard Kremsdorf, MD

"Compliance related to the use of thepumps is greater than 90% in ourorganization, and it is that waybecause noncompliance is addressedvery quickly and straightforwardlyby nursing leadership— not only atthe highest level, but also down tothe manager level in the clinicalareas. Another reason is that theselection process and purchaseprocess was driven clinically — theclinical decision was given to andmade by nurses at our institution."Ray Maddox, PharmD

"The front-line staff really are theimportant voices we need to hear."Kathy Rapala, RN, JD

"One way to find out about the cul-ture of safety is to ask 'Hey, what's itlike around here?' If people are nottalking about a culture of safety,then you don't have one." DaveSchlotterbeck, CEO

"Until we can make the rationale forusing new technologies such assmart pumps meaningful and com-pelling from a value perspective forthat front line practitioner, we'regoing to see people that don't under-stand, so they'll work around it."Kathy Rapala, RN, JD

"'Did the pump help you? Did it catchyou doing something wrong?' 'Well,yes.' As nurse executives and nurseleaders, we have to talk to nurses inthat kind of a manner, so that theyunderstand how important and howpowerful a tool like an IV medicationsafety system is." Victoria Rich, RN, PhD

What is the best organizationalinfrastructure to improve safety inorganizations and what strategiesare needed to improve medicationuse safety?

Many approaches to improving safetywith different infrastructures were dis-

cussed, each of which can be success-ful. It was concluded that:

• No one infrastructure or strategy canbe recommended for improving medication use safety– it is institu-tion-specific.

• Senior leadership and support are needed.

• Nurses, physicians, pharmacists, information technology, and finance all need to be involved.

"We have found in looking at med-ication safety that a system-widesafe medication practice committeehas proved successful. We have amultidisciplinary group at the tableincluding nursing, pharmacy, infor-mation technology, physicians. Thatcommittee is where we discussthings like technology; we're alsolooking at our CPOE system and theproblems that we are having withlack of interfaces between our phar-macy system and point of care sys-tems." Elaine Levy, RPh

"Any interdisciplinary group that'sactually figured out how all thesepieces are going to work and howwe're going to redesign work flow todeal with the technology piecescould be really invaluable toenhancing our ability to take care ofpatients." Rita Shane, PharmD,FASHP, FCSHP

"It takes various incentives, andsometimes mandates (publishedstudies, professional guidelines, andaccreditation standards) to help hos-pitals decide to spend money to buytechnology that improves patientsafety." Nat Sims, MD

"It only took five years, from 1985 to1990, to go from 5% of the hospitalsin this country who utilize oximetersand captnographs intraoperatively to




100% —and that's about as fast asanything can move." Ellison Pierce, MD

"The closer you can align your strategywith what the caregivers tell you, themore likely you are to be successful,because the resistance factor goesaway." Richard Kremsdorf, MD

"One way to herd cats is with tuna."Charles Denham, MD

"The fact of the matter is that everyhospital in this country is going to beinvesting in IV pumps sometime in thenext 10 years." Steve Meisel, PharmD

"The tipping point is that ECRI is nowrating any device that doesn't havedose error reduction software asnot a recommended device." TimVanderveen, PharmD, MS

"One of the things that did fuel the fireto get the pumps purchased was thefact that I could point to one of thosecategories of harmful events, andthey were pump failures. So it was aneasy sell from a safety perspective."Nancy Pratt, MSN

"Nurses want three things: they wantto be financially rewarded, they wantto be valued, and they want to beheard. So, the more we can involvethem in decision-making regardingsmart pumps, they'll make that deci-sion because it's intuitive to them."Joan Vitello, PhD, RN, FAAN, FAHA

"Feedback is the key. I took our pre-liminary findings on 'near misses'from the smart pump CQI data, and Ifed them right back, even as raw asthey were, to the front-line clinicians.When they have the tools they need,they respond." Jacqueline Sullivan,PhD, RN, CCRN

"Part of the reason I believe that thecompliance is greater than 90% at ourinstitution is that the selection process

and purchase process was driven clini-cally— the clinical decision was givenand made by nurses in our institution."Ray Maddox, PharmD

"I think most physicians, if you makethem aware of this situation —both thevariability and that 7% to 10% ofpump input requires reprogrammingthat may hurt the patient— I thinkthey'd be very amenable to change. Ithink they wouldn't have any objec-tions relating to that." Frank Overdyk,MSEE, MD

Is there a business case for investing inimprovements in medication use safe-ty? How important is an evidence baseto support the investment?

This generated the most discussion,with general agreement that:

•Innovators and early adopters are making patient safety investments and changes on intuition, rather than a strict business-case or evidence-base analysis.

• There is not currently enough information to make business-case or evidence-based decisions.

• This is needed for widespread adoption of new patient safety technology.

"The level of evidence here so far ispretty modest. I did the evidencereview for bar coding, and the last timethis was done, there was not evenenough evidence to make IV med-ication safety systems part of theevaluation." David Bates, MD

"There are a lot of technologies outthere that can make a difference,so three things that I think arenecessary to look at are the cost, theease of adoption, and how long it takesto make a difference." DaveSchlotterbeck, CEO

"Purchasing smart pumps was an easydecision. I had budgeted barcodemoney, and we spent it on the pumps.The bottom line is that the bar codemoney was safety money, and this wassafety technology. We were going tobuy new pumps, so I threw themoney in to getting the smart pumps,because that was progress and meantwe could reduce harm real-time."Nancy Pratt, MSN

"The question will be not whether (hos-pitals) will buy new IV pumps, butwhether they will buy IV pumps withthis technology built in or not." Steve Meisel, PharmD

"There is no question in my mind thatthis needs to be driven as a nursingdecision. You've got to have the phar-macy along and you've got to havematerials management along, but ifyou take this, you know, down anotherpath and then try and stuff it to thenurses, it's a non-starter. You let thenurses choose what they want; they'regoing to go with this pump because itmakes sense." Nancy Pratt, MSN

"As we examine how we wouldorganize ourselves to do that, we seeseven different silos, each of which areinvolved in only a piece of infusiontechnology: materials management,pharmacy, IS to a tiny extent, qualityand safety, biomedical engineering andpatient care services. Each of these hassmall amounts of capital under theircontrol which, if aggregated, could dothe whole project, but it's not centrallycoordinated." Nat Sims, MD

"It is really important to get thatinformation back out to those front-line people and help them understandwhy they should use the system anduse it appropriately and how by usingit appropriately, there is a businesscase for it." Kathy Rapala, RN, JD


"I now can clearly articulate the num-ber of incidents where I have beenable to intercept medication errorsfrom reaching the patient by usingsmart pump technology. That to meadds value to move this forward... As amember of administration, I am nowable to look at my staffing and beginto articulate areas of vulnerability.I can start shifting resources andunload some of those responsibilitiesthat are all happening at a particulartime of day. Now I have a very specif -ic indicator that allows me to trackthat ...There is enormous nursing sat-isfaction in being able to get to acommon platform; this is where theyperceive direct personal value tothem… Overall, in looking retrospec-tively I can say without question thatthis is one of the smartest things thatwe have done. Glenn Billman, MD

"I agree that it cannot be intuitive. Ithas to show a bottom line. I have beena COO. I know that. And I have beenwith CEO's that have always said tome, ‘You have to show me how muchyou can add to the bottom line if I buythese things.'" Victoria Rich, RN, PhD

"We saw that there was a businesscase (for the Medley™ System devices)because of cost avoidance. We wereable to appeal to intuition (that con -verting to safe pumps was the rightthing to do) on many levels, and allthe nurses that had used these pumpsrefused to go back (to our previouspumps).” Al Gould, RN, MSN, CCRN

"The average settlement cost for med-ication errors is a little over $143,000per settlement. If I had 82 criticalmedication errors that were prevent-ed, it's in excess of $11 million." RayMaddox, PharmD

"I do believe this will be part of anursing retention strategy... becausethose really helped to start to makeyou safe." Victoria Rich, RN, PhD

"I review cases for a legal firm, andthere's nothing for getting an admin-istrator's attention like a $2-3milliondollar settlement." Frank Overdyk,MSEE, MD

"Our expenses on liability aren't any-where close to anything we've seenpublished (but) it almost doesn'tmatter what the average cost is. Youcould sell this product based on onecase." Nancy Pratt, MSN

"On average, malpractice expenses aregoing up 40% per year, so anythingthat can stem the tide of that growthcan really speak to that issue." CharlesDenham, MD

"Anesthesia premiums have gonefrom $30,000 to $40,000 per year to$80,000 or so in 1983 dollars. So, Ithink you are wise in looking at themalpractice insurance side of all this."Ellison Pierce, MD

"I can tell you that we've taken a closelook at IV medication safety systems. Itjust hasn't been articulated in a waythat it needs to be articulated. If youreally measure a fully loaded, enter-prise-wide systems impact, I think it'sprobably got the best business caseout there." Charles Denham, MD

"I've played a lot with the model, andthere are four key inputs. The savingscome from the cost of events,though-put, malpractice, and thenumber of events. Malpractice endsup a distant third, and the biggestvariable is the number of adverse drugevents that you prevent. A lot of the

business case depends on what thatfigure is." David Bates, MD

"From what I heard today and in thepast, there is evidence– evidence fromnon-randomized trials and retrospec-tive data. It seems like it had a littlebit more of a harder or scientifictype of a connotation than intu-itive." Jacqueline Sullivan, PhD, RN,CCRN

"There will be a lot of competition forthese scarce capital resources, and Ithink the ROI evaluations are impor-tant. Smart pumps have a good ROI,and two other main benefits are thatthe ease of adoption and the time tomake a difference with smart pumpsare very different as compared to anyof the other approaches we men-tioned." David Bates, MD

"I don't think you make a businesscase separately from a clinical case;I think it's made together." KathyRapala, RN, JD



©2003-2004 ALARIS Medical Systems, Inc. All rights reserved.04/04 SSM# 1454B

SPEAKERS

May Adra, BS, PharmDDirector of Drug Information/Medication Safety CoordinatorDepartment of PharmacyTufts–New England Medical CenterBoston, MA

David Bates, MD, MScMedical Director of Clinical and Quality AnalysisBrigham and Women's HospitalBoston, MA

Glenn Billman, MDMedical Safety OfficerChildren's Hospital and Health CenterSan Diego, CA

Rick Crass, PharmDSenior Manager, Clinical MarketingALARIS Medical SystemsSan Diego, CA

Ray R. Maddox, PharmDDirector, Clinical Pharmacy Research and Pulmonary MedicineSt. Joseph's/Candler Health SystemSavannah, GA

Steven Meisel, PharmDDirector of Medication SafetyFairview Health ServicesMinneapolis, MI

Nancy Pratt, MSNSenior Vice President, Clinical EffectivenessSharp HealthCareSan Diego, CA

Peter Provonost, MD, PhDAssociate Professor, Anesthesiology and Critical Care MedicationThe Johns Hopkins Medical InstitutionsBaltimore, MD

Philip J. Schneider, MS, FASHP(Moderator)Director of Latiolais Leadership ProgramClinical Professor, The Ohio State UniversityColumbus, OH

Christopher L. Shaffer, PharmD, BCPSPharmacy Clinical ManagerDepartment of Pharmaceutical and Nutrition CareThe Nebraska Medical CenterOmaha, NE

Terri Simmonds, RNDirector, Critical Care and Patient Safety Institute for Healthcare Improvement Boston, MA

Jacqueline Sullivan, PhD, RN, CCRNDirector, Nursing, Quality and Outcomes Hospital of the University of PennsylvaniaPhiladelphia, PA

Tim Vanderveen, PharmD, MS Executive Clinical Director The ALARIS® Center for Medication Safety and Clinical ImprovementALARIS Medical SystemsSan Diego, CA

ATTENDEES

Kenneth E. Aaron, MDMedical Director, Performance ImprovementHoag HospitalNewport Beach, CA

Sue Alderson, MBADirector of PharmacyWilliam Osler Health CentreBrampton Memorial HospitalBrampton, Ontario

Mary AlexanderChief Executive OfficerInfusion Nurses SocietyNorwood, MA

Norma Barr, RN, MNDirector, Professional Practice and Systems SupportBJC HealthcareSt. Louis, MO

Mary Burkhardt, MS, RPh, FASHPProgram Manager, Veteran's Health Administration–VA National Center for Patient SafetyAnn Arbor, MI

Dennis Cada, PharmD, FASHP, FASCPEditor-in-Chief, Hospital PharmacyExecutive Editor, The FormularyFacts and ComparisonsLaguna Niguel, CA

Charles R. Denham, MDCEO, Health Care Concepts, Inc.Austin, TX

Cathy Denning, RN, MSNSenior Clinical Manager–SafetyNovationIrving, TX

Timothy Dresselhaus, MD, MPHAssociate Clinical Professor UCSD Department of Medicine AssociateChief, VA Medical ServiceUniversity of California, San DiegoLa Jolla, CA

Patti FisherClinical Risk Manager SpecialistLondon Health Sciences CentreLondon, Ontario

Al Gould, RN, MSN, CCRNClinical Nurse SpecialistNebraska Health SystemOmaha, NE

Richard Kremsdorf, MD5 Rights ConsultingSan Diego, CA

Elaine Levy, RPhSystem Director Pharmacy and Clinical NutritionSharp HealthCareSan Diego, CA

Mary Beth Navarra, RN, MBADirector, Automation PlanningMcKessonPittsburgh, PA

Joan Osborne, RN, BC, BS, BSN, MSN, ARNPDirector of Clinical Practice and ResearchBroward General Medical CenterFt. Lauderdale, FL

Frank J. Overdyk, MSEE, MDVice Chairman, Clinical OperationsAssociate Professor of AnesthesiologyMedical University of South CarolinaCharleston, SC

Stan Pestotnik, PharmD, MSCEO, TheraDoc Inc.Salt Lake City, UT

Ellison (Jeep) Pierce, MDBoston, MA

Elizabeth PlantChief PharmacistMPS(NZ)PGradDipClin Pharm (Distinction) ANZCPTaranaki District Health BoardNew Zealand

Kathryn G. Rapala, RN, JDDirector, Risk Management and Patient SafetyClarian Health PartnersIndianapolis, IN

Victoria Rich, RN, PhDChief Nursing OfficerHospital of the University of PennsylvaniaPhiladelphia, PA

Eric SacksInternet Services ManagerHealth Devices GroupECRIPlymouth Meeting, PA

Rita Shane, PharmD, FASHP, FCSHPDirector, Pharmacy ServicesCedars–Sinai Medical CenterLos Angeles, CA

Nat Sims, MDCardiac AnesthesiologistAnesthesia AssociatesMassachusetts General HospitalBoston, MA

William A. SpoonerSenior Vice President and CIOSharp HealthCareSan Diego, CA

John VanEeckhout, PharmDVice President, Clinical Services Child Health Corporation of AmericaShawnee Mission, KS

Joan Vitello, PhD, RN, FAAN, FAHAVice President, Patient Care ServicesChief Nursing OfficerSt. Anne’s HospitalFall River, MA

ALARIS MEDICAL SYSTEMS, INC.

Mark Bruns, Division Vice President, North American Sales

Joseph Condurso, Director, Marketing Applications

Rick Crass, PharmD, Senior Manager, Clinical Marketing

Sally Graver, Writer

Gamble Heffernan, Director, The ALARIS® Center forMedication Safety and Clinical Improvement

Claudia Russell, Division Vice President, Marketing

David Schlotterbeck, CEO

Geoff Siegel, Division Vice President, Product Development

Tim Vanderveen, PharmD, MS, Exec Clinical Director, The ALARIS® Center for Medication Safety and Clinical Improvement

ALARIS Medical Systems, Inc.Worldwide Headquarters, 10221 Wateridge Circle, San Diego, California 92121-2772

Fax: 1-858-458-7760Customer Service: 1-800-482-4822Website: www.alarismed.com/na

infusion safety: addressing harm with high-risk drug administration

Documents