EvidenceBased Medicine and Health Information Systems especially CDSSs:

1
(No Transcript)
2
Evidence-Based Medicine and Health Information
Systems (especially CDSSs)
Romeo and Juliet or Jekyll and Hyde?
3
Objectives

• To review the origins, aspirations, and evolution
  of evidence-based medicine (aka the groom)
• To review the strengths and weaknesses of
  computerized decision support systems (CDSS) (the
  bride)
• To consider how EBM and CDSSs are made for each
  other and yet might be star crossed or
  possibly transform each other into pure evil

4
(No Transcript)
5
(No Transcript)
6
Definition

• Evidence-based health care is...
• ...the explicit, judicious and conscientious use
  of current best evidence from health care
  research in making decisions about the health
  care of individuals or populations.

Research-enhanced health care
7
A source of confusion the term evidence is
commonly used for many phenomena of relevance to
clinical practice, not just health care research
evidence. For example, clinicians collect
evidence of patients circumstances by taking a
history, doing a physical examination, ordering
lab tests - and the patients wishes.
Evidence-based medicine is about evidence from
research.
8
Current times are unique for health care
practice...

• US100Billion/yr in health research
• Better applied research methods
• Information technology (eg HIS)
• World-wide demand for more for less from
  health services

9
EBM isan evolutionary attempt to harvest the
fruits of 50 years of public and private
investment in health care research.
10
a
b
c
? Knowledge Translation ?
Steps from evidence generation to clinical
application
Steps 1. generation of evidence from research
2. evidence summary and synthesis 3. forming
clinical policy 4. application of policy 5.
individual clinical decisions, including a)
patients circumstances, b) patients wishes, and
c) evidence from research
11
For evidence-based medicine to work, and for this
to matter

• all steps must occur in sequence
• and they must occur
• - faster and better
• - in more local settings
• than without EBM

12
The History of Thrombolysis
Grines et al, NEJM 1993 Immediate Percutaneous
Coronary Intervention (PCI) is more effective
than thrombolysis for acute MI
Antman, Lau, Kupelnick, Mosteller, and T
Chalmers, JAMA 1992
13
Step 1. Generating Research Evidence
14
Step 2. Synthesizing Research Evidence
15
How much synthesis do we need?

• ..at least 10 000 Cochrane reviews are needed to
  cover a substantial proportion of the studies
  relevant to health care that have already been
  identified
• Susan Mallett Mike Clarke
• ACP Journal Club. 2003 Jul-Aug139A11.

16
When will we have our 10,000 reviews?
Growth of Cochrane Reviews and Protocols
2003
protocols
2000 competed reviews as of mid-2004
1995
reviews

• between 2010 and 2015.
• MallettClarke, ACPJC 2003

17
Step 3. Developing Policy
18
Step 4. Applying evidence in practice
19
Information in the Internet Age

• Information in the Internet age may be
  constrained by a variant of Malthus law
• The amount of information is growing
  exponentially, but our attention is not
• The low cost of production of poor quality
  information results in high quality information
  being drowned out
• The cost of finding specific information rises as
  the amount of information increases

Coiera E. Information economics and the
internet. J Am Med Inform Assoc 20007215-21.
20
Are CDSSs clinically effective?
21
Effects of Computer-Based Clinical Decision
Support Systems on Health Care Provider
Performance and Patient Outcomes A Systematic
Review
Amit Garg, Neill Adhikari, Heather McDonald,
Patricia Rosas-Arellano, Justina Sam, Joseph
Beyene, PJ Devereaux, Brian Haynes
22

• Context Computer Decision Support Systems
• Software designed to directly aid in clinical
  decision making in which characteristics of
  individual patients are matched to a computerized
  knowledge base for the purpose of generating
  patient specific assessments or recommendations.

Rules / Algorithms

• INPUT
• Patient Characteristics
• Into Computer System
• Automated through EMR
• By extra research staff
• By existing health care staff
• By the patient
• By the practitioner

Computer

• OUTPUT
• Recommendations
• delivered to health
• care provider
• Directly by computer
• By pager
• By extra research staff
• By existing health care staff

• Outcomes
• Provider performance
• Patient outcomes

integrate into workflow
23
Examples of Clinical Decision Support Software
Alert Remind Critique Interpret Predict Diag
nose Assist
Highlight out of range K Remind about need for
hepatitis B Reject med order when allergy
present EKG Interpretation 10 year Framingham
Risk for Cardiac Disease Diagnose Suggest new
orders for active care
24
(No Transcript)
25

• Did CDSS improve practitioner performance?
• Update 103 studies
• counting positive results on 50 outcomes
  measured

Examined in 97 studies, 58 cited improvement
(60)
?
?
?
Some CDSSs provided more than 1 class of feedback
26

• Did CDSS improve patient outcome?
• Update 103 studies

Examined in 47 studies, 11 cited improvement
(23)
27
Preventative Care 40 studies
Improved Practitioner Performance - 71 -
Improved Patient Outcome - 27 -

• Most are REMINDERS
• chronic disease care hypertension, diabetes,
  cholesterol,
• 
  cardiovascular disease, asthma,
• renal
  disease, obesity
• cancer prevention cervical screening,
  mammography, stool OB
• infection prevention immunization
• periodic health exam
• Improvements in patient outcome
• intermediate outcomes blood pressure,
  cholesterol, HbA1C
• influenza vaccination reduced hospital
  admission

28
Active Care 53 studies
Improved Practitioner Performance - 57 -
Improved Patient Outcome - 21 -
Most are RECOMMENDATIONS. Range of problems,
for example - acute disease care (cardiac
disease) - chronic disease care (diabetes,
hypertension, lipids, depression) - appropriate
drug prescribing - incontinence in the elderly
- advanced directives - ventilator support -
infertility - health care utilization
(hospitalization, laboratory testing)
29
Improved Practitioner Performance - 64 -
Improved Patient Outcome - 19 -
Drug Dosing 28 studies
aminoglycosides wafarin
dosing - - -
- - (coumadin) - - - - -
- heparin

insulin
digoxin
-
- TPN
lidocaine
- theophylline
/ - - - -
- aminophylline multiple drugs
- - - -
Improved symptoms post tPA
HbA1C
Decreased length of hospital stay
30
Improved Practitioner Performance - 38 -
Improved Patient Outcome - 14 -
Diagnosis 13 studies
cardiac disease (EKG) - -
- - pediatric conditions
- hypertension -
- depression, psychiatry
- - - - - acute abdo
pain -
- general med conditions
-
admission to CCU
31
How to improve clinical effectiveness of CDSSs?
32

• Improve User acceptability of CDSS?
• The CDSS was integrated into the practitioners
  workflow (1-5)
• 1 strongly disagree
• 2 disagree
• 3 neutral
• 4 agree
• 5 strongly agree
• Practitioners using the CDSS were satisfied with
  the system
• 1 strongly disagree
• 2 disagree
• 3 neutral
• 4 agree
• 5 strongly agree

In 57 of studies, agree or strongly agree
In 10 of studies, agree or strongly agree
33

• Conclusions for CDSSs
• CDSSs can enhance clinical performance,
  particularly for preventive care, chronic disease
  management and drug dosing.
• Improvements in patient outcomes with CDSSs
  remain unsubstantiated, and require further
  study.
• The cost effectiveness of such systems requires
  further consideration.

34
Three deadly sins of decision support systems
according to Brian Haynes

• Telling me things that I already know
• Telling me things that I know are out of date
• Telling me things that I dont know that aint so

35
The McMaster PLUS project

• only a tiny proportion of all research is ready
  for application
• only a tiny fraction of the ready research is
  relevant to the practice of a given clinician
• only a tiny proportion of the relevant research
  for a given practitioner is interesting in the
  sense of being something important and new.

36
Evidence-Based Journals
Critical Appraisal Filters
3,000 articles/y meet critical appraisal and
content criteria (94 noise reduction)
50,000 articles/y from 120 journals
37
McMaster PLUS Project
20 articles/yr for clinicians (99.96 noise
reduction)
3,000 articles/y meet critical appraisal and
content criteria (98 noise reduction)
5-50 articles/y for authors of evidence-based
clinical topic reviews
http//hiru.mcmaster.ca/MORE
38
Step 4. Applying evidence in practice
39
Step 4. Applying evidence in practice
40
Step 5. Applying evidence in clinical decisions
41
WHO estimates US100B/yr for health-related
research

• not enough is for application research
• the balance is shifting slowly
• should there be a Nobel Prize for knowledge
  translation research?

42
(No Transcript)
43
Step 5. Applying evidence in health care
decisions
44
The brightest hope for getting all this right

• Current best evidence information systems that
  support evidence-based decision making by
• assisting policy makers at the local level
• coordinating information across the 4P levels
• helping practitioners to recommend effective
  treatments (CEQI)
• helping patients to follow effective treatments
• All technically possible if evidence and
  information systems were wed together in a
  harmonious state of matrimony.

45

• The low cost of production of poor quality
  information results in high quality information
  being drowned out
• The cost of finding specific information rises
  as the amount of information increases

46
(No Transcript)
47
(No Transcript)