Safe IT systems?Safe Patients?

Professor Bryony Dean FranklinOctober 2012

CMSSQCMSSQCentre for Centre for MedicationMedication Safety & Service Safety & Service Quality Quality

Why are you still studying medication errors? There

won’t be any soon, once we have electronic prescribing…

Automation and IT in pharmacy...

Examples

• Electronic prescribing (+/- electronic medication administration records in hospital and care home)– with various levels of decision support

• Automated dispensing– Pharmacy based (“robots”)– Ward based (“vending machines”)– Aseptic compounding robots– Automated CD storage

• Barcode verification of medication and/or patients

• “Smart” IV pumps

A quiz

• Inpatient electronic prescribing with prescriber order entry – is it more prevalent in:

A. USA ?

B. UK ?

UK hospital electronic prescribing

• 101 (61%) of 165 hospital trusts responded in survey of English hospitals– 70 (70%) had at least one EP system in place– 56% of sites with EP had more than one system in

place. Four sites had more than 4 systems.– 63 different systems

• Nearly half of respondents had EP systems supporting in-patient prescribing (47.5%, n=48).

• Discharge prescribing in 65.3% (n=66) of sites. • Outpatients was the least catered for (5.9%, n=6).

Ahmed, Franklin and Barber, 2012

UK hospital electronic prescribing

• 101 (61%) of 165 hospital trusts responded in survey of English hospitals

– 70 (70%) had at least one EP system in place

– 56% of sites with EP had more than one system in place. Four sites had more than 4 systems.

– 63 different systems

• Nearly half of respondents had EP systems supporting in-patient prescribing (47.5%, n=48).

• Discharge prescribing in 65.3% (n=66) of sites.

• Outpatients was the least catered for (5.9%, n=6).

Ahmed, Franklin and Barber, 2012

US hospital CPOE

• ASHP national survey of pharmacy practice in hospital settings 2011

• Stratified random sample of 1401 hospitals• 40.1% response rate (n=562)• 34% of hospitals had computerised prescriber

order entry• 67% using electronic medication administration

records

US hospital CPOE

• ASHP national survey of pharmacy practice in hospital settings 2011

• Stratified random sample of 1401 hospitals• 40.1% response rate (n=562)

• 34% of hospitals had computerised prescriber order entry

• 67% using electronic medication administration records

A quiz

• Inpatient electronic prescribing with prescriber order entry – is it more prevalent in:

A. USA ?

B. UK ?

A quiz

• Inpatient electronic prescribing with prescriber order entry – is it more prevalent in:

A. USA ?

B. UK?

Automation of dispensing in hospitals

• Automated dispensing systems– Pharmacy based (“robots”)– Aseptic compounding robots – Ward based (“vending

machines”)• 6 (7%) of 91 UK respondents • (cf 89% in USA)

– Automated CD storage • 2 (2%) of 91 UK respondents

McLeod, Barber and Franklin, 2012

Aseptic compounding robot

Verifies bags using barcode Verifies vials using photo recognition

Ward-based automated storage

Verifies product on loading, using barcode

Automated CD storage

Are our IT systems safe?

Are our patients safe?

What’s the evidence?

International literature

• Studies of CPOE generally show benefits (17-81% reduction in errors)– But increasing realisation that new types of

error

Smart pumps

• Used in 68% US hospitals

• Drug “libraries” to permit checking of doses and infusion rates

• Require standardisation of concentrations etc

• Bypassing of the safety software is common

• Nuckols et al: Only 4% of preventable IV ADEs would be preventable with smart pumps

UK evaluations

• Electronic prescribing in hospitals– Most (but not all) evaluations show a modest reduction in

prescribing error

• Closed loop ward based automated dispensing system with barcode verification– More dramatic reduction in administration errors

• Dispensing robots– Reduction in “wrong content errors”

• Smart pumps?• Ward-based automated dispensing?

Why?

What is technology good at?

• Repetitive tasks, same every time• Follows the rules• Forcing functions

– Can’t proceed until you’ve completed all the fields

• More legible than handwriting• Reminders• Supporting formularies, protocols, standardisation

of treatment • Audit trail

But…

• Can be inflexible

• New error types– Selection errors from menus

– Menus often present very long lists of options which prescribers not familiar with

– Assumptions - “the computer must be right”

• New work processes may be required, which can themselves increase or decrease errors– Development of workarounds

• Alert overload

Unintended consequences

Selection errors

• Selection of penicillamine, instead of penicillin • Menu arranged alphabetically in hospital

system– Paracetamol soluble tablets– Paracetamol suspension– Paracetamol tablets

• Many patients prescribed paracetamol soluble tablets – At risk of hypernatraemia

Selection errors

• Selection of penicillamine, instead of penicillin • Menu arranged alphabetically in hospital

system– Paracetamol suspension– Paracetamol tablets– Paracetamol tablets soluble

Assumptions

• Human-computer interaction causes most deaths of all IT induced fatalities– Eg a UK hospital: ~1000 cancer patients under-

dosed with radiotherapy over 9 years. Decision support software incorporated in machine, staff did not know and applied a second, manual dose reduction calculation

– McKenzie ‘Knowing machines’ 1996

– Assumption that EP system would include allergy checking, when it didn’t...

Workarounds

Workarounds

• Increased patient identification from 17% of doses with manual system, to 81% with barcode system

• Why only 81%?• Staff sometimes found the

wristband hard to scan, and so stuck the barcode to the patient’s table…

Alert overload

• “If you have too many warnings from the computer then that makes you tend to override them, you become a bit more cavalier and that's a danger.” (Practice Study, PR6-GP3)

How do we maximise error reduction and minimise new errors?

1. Health warning

• Do not assume that benefits in other health systems / other countries will extrapolate to your own context

2. Systems aren’t “plug and play”

3. Local evaluation essential

When do we measure the effectiveness of the system?

When do we measure the effectiveness of the system?

With thanks to Nick Barber

Conclusions

• Huge potential patient safety benefits• Success in achieving these is dependent on

many other contextual and organisational factors• Local evaluation is essential

– Need some form of ongoing monitoring and refining of the system. And listening to users

• Need a good relationship with suppliers• Embedding systems into everyday practice is a

long-term project