challenges and opportunities for patient safety with clinical … · 2019. 11. 18. · professor...
TRANSCRIPT
Challenges and opportunities for patient
safety with clinical information systems
Professor Johanna Westbrook PhD, FACMI, FACHI, FTSE,
Centre for Health Systems and Safety Research
Australian Institute of Health Innovation
Macquarie University, Australia
Patient Safety and Cost Pressure, Zurich, 15th November 2019
Australian Institute of Health Innovation
3
AIHI
Australian Institute of Health Innovation
Centre for Healthcare Resilience and Implementation Science – Prof J Braithwaite
Centre for Health Informatics – Prof E Coiera
Centre for Health Systems and Safety Research – Prof J Westbrook
Centre for Health Systems & Safety Research, AIHI
Applied research targeting safety
Centre for Health Systems & Safety Research
Focus on applied research to improve patient safety and investigate
the role of information technology in supporting improvements
➢ Medication Safety
➢ Electronic Decision Support and Human Factors
➢ Diagnostic Informatics
➢ Work Innovation and Communication
➢ Aged and Community Care Services
➢ Data Analytics for patient safety
Meth
od
s Develo
pm
en
t
Kahoot.it
Patient safety economics
• Patient safety is a critical policy issue and
active engagement with stakeholders is
essential.
• The cost to patients, healthcare
systems and societies is considerable.
• Greater investment in prevention is
justified and solid foundations for patient
safety need to be in place.
9
Disability Adjusted Life Years (DALYs) associated with adverse
events causing patient harm in OECD countries 2015
~65 DALYS/100,000 population
No. of years lost due to ill health/disability resulting from adverse events
OECD “best buys”
• VTE prevention strategies
• Infection control interventions (protocols to improve safety of central line catheters, urinary catheters & managing ventilated patients)
• Procedural checklists
• Medication management and reconciliation (across settings)
• Pressure injury prevention protocols
• Patient hydration and nutrition standards.
[Slawomirski, Auraaen and Klazinga 2017]
Agency for Healthcare Research and Quality (AHRQ)
• Between 2010 and 2015 - 21% decline in adverse events for Medicare patients treated in hospital
• Estimated to have avoided 125,000 deaths; saved US$28 Billion in heath costs as result of safety improvements.
• Greatest benefits accrued from reductions in adverse drug events – 42%
Benefits of Safety Improvement Activities
▪ 10% of care delivered to patients causes harm
▪ 30% of health care is low value or wasteful
▪ 60% of care delivered is consistent with consensus or evidence-based guidelines
Technology is providing enormous new possibilities to address these challenges
Challenges for Health Systems Internationally
60:30:10 Problem (Braithwaite et al)
13
To design & implement safe and effective
systems and technology
it is fundamental to understand clinical work
+
how safe care is delivered in practice
Clinical work conceptualised as linear, governed by
rules, regulations, policies and procedures
Work-as-imagined (WAI) – Hollnagel et al 2017
16
But healthcare really looks like this … Work-as-done (WAD)
17
Health Care is a complex adaptive system
❖Agents are autonomous often pursuing different agendas
❖Behaviour is emergent
❖Agents work in networks. They share some common rules for behaving and work together without a central source of direction.
❖Dynamic and use experimentation. Trial things and then adapt behaviours.
18
A complex adaptive system in action
Aims for ‘as few
things as possible to
go wrong’.
Error reduction is
through a ‘find & fix’
approach to
problems
or
‘whack-a-mole’!
Traditional approach to patient safety
Approaches to Patient Safety
21
The amazing thing about health care isn’t that it produces adverse events
in 10% of all cases, but that it produces safe care in 90% of cases.
Complementary Approaches to Patient Safety
WHO Goal:Reduce severe, avoidable harm related to medications by >50% over 5 years
globallyCHSSR | AIHI | MACQUARIE UNIVERSITY
• Global costs associated with medication errors – US$41 billion annually
• Medication errors – single most preventable cause of patient harm
• Information technology shows promise in making a significant impact on medication safety
23
Medication Safety - Change from paper to electronic – impact on safety
Sample: 3200 patient admissions; >17,000 prescribing errors
Prescribing errors declined by >50% (p<0.0001)
44% (p=0.0002) reduction in serious prescribing error rate25/100 admissions 14/100 admissions(95%CI 21-29) (95%CI 10-18)
No significant change on the control wards (p=0.4)
Do electronic medication management systems (eMM) reduce errors?
➢ eMM – resulted in a reduction of ~ US$50 per admission
➢ Entire hospital with 39,000 annual admissions = releasing
US$ 1.9M each year
26
New Errors facilitated by the eMM!
❖ Reviewed 1164 post eMMS prescribing errors
❖ 493 (42.4%) facilitated by the eMM, but low risk of patient harm 11/493
❖ Most frequent type Incorrect selection from drop-down menus = 43%
82% no harm to patients
13% low harm
4% moderate
1% severe
4 contributed to patient death
Benefits and potential risks
Embedding a model of linear workflow
“..Systems often appear to be imbued with a formal stepwise notion of health care work.” (Ash et al 2004)
Inflexibility in systems can led to errors
E.G. A drug ordered 3 times per day was discontinued, but one dose had been given.
The system would not allow the nurse to chart the one dose because the system considered it an incomplete execution of a task.
30
Island Health CEO apologizes in wake
of report on IHealth system
A report into the $174-million electronic health record system in Nanaimo [Canada] validates safety risks raised by doctors, and suggests Island Health should have spent more time tailoring the software to the needs of front-line workers before introducing it.
Cindy E. Harnett / Times Colonist November 17, 2016
Continued discord between IT systems & clinical work processes need more adaptive systems
31OFFICE I FACULTY I DEPARTMENT
Increased mortality following CPOE
Significant impacts of the system on workflow, including that:
❖ interventions such as medications could not be initiated as quickly as with the paper-based system
❖staff were taken away from clinical areas in order to attend to processes associated with the system
❖Communication was heavily reliant on the system rather than oral communication, reducing opportunities for ensuring messages had been received, and quick changes following consultation.
Technology constrained rather than facilitated work
Han et al 2006, Pediatrics
Medication safety in paediatrics
5 year NHMRC Partnership Grant ($1.1m)
Sydney Children’s Hospital Network
NSW Kids & Families
eHealth NSW Ministry for Health
Is eMM the Solution?
Research - Policy and
management driven &
funded
33
Measuring changes in prescribing error rates pre and post eMM
• Stepped Wedge Cluster Randomised Controlled Trial (n=3549 patients; 26,369 medication orders)
• Wards randomised for IT implementation over 10 weeks; collected data at baseline and each week
• Examined clinical prescribing errors (e.g wrong dose, wrong drug) for all patients & procedural errors (e.g incomplete order) for a random 68.3% (n= 2424 patients) sample.
• Longest followup period for trial was 10 weeks post eMM implementation
• Additional 1 year followup 12 months post eMM (1046 patients)
• Total of 35,003 orders reviewed for 4595 patients
Stepped-wedge cluster RCT study design
34
Randomised wards; measured medication error rates at each step
Clinical Prescribing Errors – Changes in error rates pre, post-eMM and 12 months post eMM
CHSSR | AIHI | MACQUARIE UNIVERSITY – Not for distribution 35
Paper (n=1865 patients)
Maximum time 10 weeks post eMM
(n=1684 patients)
One year post eMM
(n= 1046 patients)
No. of Errors / Orders
Error rate per
100 orders
95% CINo. of Errors
/ Orders
Error rate per 100
orders
95% CINo. of Errors
/ Orders
Error rate per
100 orders
95% CI
1690 / 11322 14.9 14.2, 15.7 2652 / 15047 17.6 17.0, 18.3 995 / 8634 11.5 10.8, 12.3
A significant 22.8% reduction in clinical prescribing error rate - pre eMM compared to 12 months
post eMM (14.9/100 orders to 11.5/100)
Following eMM implementation there was an initial significant increase in clinical errors (14.9 to 17.6/100 orders)
Clinical Prescribing Error Rates by Ward pre, post-eMM and 12 months post eMM
CHSSR | AIHI | MACQUARIE UNIVERSITY – Not for distribution 36
0
5
10
15
20
25
30
Paper eMM eMM-1YrClin
ical
pre
scri
bin
g er
rors
per
10
0 o
rder
s
Comm Trav Surgical Orthopaedic
Edgar Stephens Wade Turner
Clubbe Middleton Hunter Baillie
Results of 122 interviews with
staff (drs & nurses) at different times
post eMMimplementation
Issues Identified
CHSSR | AIHI | MACQUARIE UNIVERSITY 38
Workarounds - Nurse with drug trolley
Guarding Against Selective Attention
“The Invisible Gorilla Strikes Again” Drew et al 2013
40
“The Invisible Gorilla Strikes Again”
24 radiologists were asked to review ~ 200 CT
scans of five patients for typical lung cancer
screening.
Drew et al 2013 Psychological Science
11/16/2019
43
Results
➢ 20 out of 24 radiologists missed the gorilla
➢ 25 non-trained reviewers all missed the gorilla
“It’s important to be willing to look for more than one
thing, to set yourself up for success.”
Evidence that targeted decision support can be highly effective
Impact of decision support on repeat laboratory test ordering rates
46
Repeat testing for 5073 children under 1 year in ICUs significantly (p<0.0001) declinedfollowing the introduction of electronic test ordering
Li et al. 2014 What is the effect of e-pathology ordering on test re-ordering for paediatric patients? Studies in Health Technology and Informatics, 204, IOS press, 74-79.
48
But…….
A large body of work demonstrating that doctors override alerts (i.e. click past alerts
without following recommendations), up to 95% of alerts
Alert fatigue - mental state resulting from excessive numbers of
alerts being triggered
Leads to:
• User frustration and annoyance
• Prescribers overwhelmed by alerts
• Learn to ignore all alerts
49
When and why decision support may be effective?
What impact does eMMS decision support have during ward rounds?
▪58.5 hrs,14 teams,
96 orders
▪48% of medication orders
triggered alerts
▪17% read
▪No orders changed
Junior doctors’ response to computerised alerts at night 16:30-22:30
• Observational study - 65 hours
• 78% of alerts were read
• 5% resulted in a change in prescribing
Context Matters
“How many alerts can you fire at users before they become ineffective? “
An alert regarding - Alert Fatigue!
52
53
How do hospitals decide what alerts to
implement?
❖Survey of 26 Australian hospitals (Page N, et al, JPPR, 2018)
❖All had Allergy and Drug-Drug interaction checking
❖69% also had dose-range checking
Configuration decisions fueled by a perception that alerts change
prescriber behaviour and improve patient outcomes – But very few local
evaluations
54
❖ Drug-condition – 5/6 studies showed positive effects
❖ Drug-drug interaction – 2/6 studies
❖ Corollary order alerts – 1/6 studies
❖ No one studied the combined effects of multiple alert types
An evidence-based approach to decision support selection
33% of patients, 67% of ICU patients experience potential DDI
Few studies report harm
Bucsa et al, 2013 - Only ~2% were harmed by DDIs (e.g. Bradycardia; abdominal pain)
15,000 alerts???
Does the size of the problem warrant the solution?
55
Drug-drug Interaction (DDI) Alerts
What is the size and nature of the problem to be addressed?
56
❖Understand the nature of the problems to be addressed
❖Recognise IT systems need to be integrated into complex adaptive systems
❖Measure effects and implementation processes using rigorous methods – monitor for both expected and unexpected changes
❖Importance of collaborations between researchers, implementers, clinicians, consumers, policy makers working together to reap the benefits of clinical IT systems
Opportunities & challenges for patient safety with
Clinical information systems