Decision support and systematic reviews

1
Decision support and systematic reviews

• Ilkka Kunnamo, MD
• The Finnish Medical Society Duodecim
• ilkka.kunnamo_at_duodecim.fi
• Dublin 23.10.2006

2
What is computerized decision support?

• Combining medical knowledge with individual
  patient data with a computer to provide
  patient-specific guidance

3
Original publication
Systematic review (Cochrane review)
Guideline (EBM Guidelines)
Local workshop
Automatic reminder by computer
Local guidelines
Small group
Clinical practice
4
What is the evidence that decision support works?
5
Clinical decision support systems for neonatal
careK Tan, PRF Dear, SJ NewelCochrane Database
of Systematic Reviews 2005, Issue2. Art. No.
CD004211. 24 January 2005

• Two studies were found for computer aided
  prescribing and one study for computer aided
  physiological monitoring.
• 1 out of 3 studies showed some evidence of
  effectiveness

6
 Interventions to improve outpatient referrals
from primary care to secondary careGrimshaw JM,
Winkens RAG, Shirran L, Cunningham C,MayhewA,
Thomas R, Fraser CCochrane Database of
Systematic Reviews 2005, Issue 3. Art. No.
CD005471. 13 May 2005

• Active local educational interventions involving
  secondary care specialists and structured
  referral sheets are the only interventions shown
  to impact on referral rates based on current
  evidence.

7
Improving health professionals management and
the organisation of care for overweight and obese
peopleHarvey EL, Glenny A-M, Kirk SFL,
Summerbell CDCochrane Database of Systematic
Reviews 2001, Issue 2. Art. No. CD000984. 13
January 2001

• Two studies on reminders showed 13.5 and 29
  increase in interventions, but no patient
  outcomes were reported.

8
Personalised risk communication for informed
decision making about taking screening
testsEdwards AGK, Evans R, Dundon J, Haigh S,
Hood K, Elwyn GJCochrane Database of Systematic
Reviews18 October 2006

• Rersonalised risk communication (written, spoken
  or visually presented) increases uptake of
  screening (odds ratio (OR) 1.31, CI 0.98 to 1.77.
  
• Presenting numerical calculations of risk, the OR
  for test uptake was 0.82, CI 0.65 to 1.03).
• Categorised into high, medium or low risk, the OR
  was 1.42 (95 CI 1.07 to 1.89).
• Simply listing personal risk factors the OR was
  1.42 (95 CI 0.95 to 2.12).

9
On-screen computer reminders effects on
professional practice and health care outcomes
(Protocol)Gordon RB, Grimshaw JM, Eccles M, Rowe
RE, Wyatt JCCochrane Database of Systematic
Reviews 1998, Issue 2. Art. No. CD001096
10
Systematic review on the effectiveness of
computerized clinical decision support systems

• 100 controlled trials
• In 64 of the trials performance of clinicians
  improved by 50 or more
• Suggestions for diagnosis 40
• Preventive interventions (reminders) 76
• Suggestions for medication doses 66

Garg AX et al. Effects of computerized clinical
decision support systems on practitioner
performance and patient outcomes. JAMA
20052931223-1238
11
Factors and elements predicting the success of a
clinical DS system

• Automatic provision of reminders as part of
  clinician workflow
• Providing clear recommendations as opposed to
  providing only assessment about the situation /
  patients condition
• Providing decision support at the time and
  location of decision making
• Using a computer to generate decision support
• Of systems possessing all 4 features, 30 out of
  32 (94) improved the quality of patient care

Kawamoto K ym. Improving clinical practice using
clinical decision support systems a systematic
review of trials to identify features critical
to success. BMJ 2005330765-768
12
(No Transcript)
13
(No Transcript)
14
Actions taken in Finland

• A comprehensive set of guidelines made available
  in a national health portal
• Every Finnish physician read averagely 1.5
  guidelines from the Internet each day in 2005

15
Use of guideline databases in the Finnish health
portal
Documents opened
16
Actions taken in Finland

• A comprehensive set of guidelines made available
  in a national health portal
• Every Finnish physician reads on average 1.5
  guidelines from the Internet each day in 2005
• All patient records in a central archive by 2010
• Access by patients permission for professionals
• Direct access for patients themselves
• Essential data coded in XML (HL7 CDA R2 format)
• Diagnoses
• Drugs
• Test results
• Procedures
• Treatment plans

17
Actions taken in Finland

• A comprehensive set of guidelines made available
  in a national health portal
• Every Finnish physician reads on the average 1.5
  guidelines from the Internet each day in 2005
• All patient records in a central archive by 2010
• Access by patients permission for professionals
• Direct access for patients themselves
• Essential data coded in XML (HL7 CDA R2 format)
• A national decision support database and web
  service based on best evidence
• EBM Guidelines and its evidence summaries linked
  to Cochrane reviews
• National guidelines

18
What does a decision support system do?

• Displays guidelines at the right moment
• Displays reminders to clinicians (prevents
  forgettings)
• Warns about drug interactions, drug allergies,
  adverse effects
• Calculates patient risks (UKPDS, SCORE)
• Lists possible diagnoses based on patients data
  and checks criteria for diagnoses
• Suggests additional examinations, validates
  indications for treatment
• Pre-fills screens or forms automatically

19
Decision support database

• Descriptions of the functionality of the decision
  support
• Evidence summaries (from EBM Guidelines)
• Scripts written in programming language

20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
Smith John
020248-139Y
55.1 y
Type 2 diabetes High BP Rheumatoid
arthritis Osteoporosis
12.10.2002
HbA1c
9.4
7.5
Medication
25.03.2003
New
10/2003
Insulin H Protaphan 32 yks. ilt.
Metformin 500 mg 2 x
2 Aspirin 100 mg 1 x
1 Simvastatin 20 mg 1 x 1 Enalapril
20 mg 1 x 1 Amlodipin 5 mg
1 x 1
i
R
12.10.2001
fB-Gluc
9.8
4-6
08/2004
i
R
i
R
05/2003
01.12.2001
LDL-chol
2.9
3.0
i
R
R
i
12.10.2002
BP
136/88
140/90
R
i
Search Dg
21.01.2003 Karstula I. Kunnamo 01.11.2002
KSKS J. Saltevo
20.8.2004/I Kunnamo Swelling in right knee for 2
weeks. Considerable hydrops. No other joints
active. 25 ml of fluid was aspirated, and steroid
injected. BP in home measurements 145
158/92-98. Not successful in reducing weight.
Willing to start antihypertensive medication.
Drug orders Enalapril (Renitec) 20 mg x 1
(new) Methyl prednisolone (Solomet) 40 mg i.a.
Smoking
Yes
Certificate for drug remuneration 20.04.2003
Reminders
Data entry
CV risk
14
Reminders and guidelines are shown automatically
within the electronic patient record
Determine S-K and S-Creatinine (enalapril started
20.10.2002)
Treatment of type 2 diabetes Lifestyle
counseling in type 2 diabetes Insulin treatment
of type 2 diabetes Smoking cessation Care
pathway/type 2 diabetes, Central Finland
Search
Diagnosis and treatment of rheumatoid
arthritis Care pathway for type 2 diabetes in
C-Finland Osteoporosis (patient information
leaflet)
The colour indicates national treatment target
(guideline)
Guidelines
Lab Diagnoses Care plan Forms
28
Virtual health check
Results of virtual health check (performed by
computer as a batch run)
Alanen
Aku
Kemppi
Kimmo
Lieko
Vappu
All decision support scripts are executed (over
the night), and resulting reminders are listed
for all patients.
Vinkeä
Tuomo
29
First name the problem

• J01.1 Acute maxillary sinusitis
• Recommendation
• Antibiotics are indicated if the symptoms have
  lasted over 7 days

30
and get a ready-made presctiption

• Amoxicillin Generics 750 mg
• D. No XV S
• One tablet twice daily for sinusitis

31
Interactive forms

• Structured referral forms
• Physicians certificates
• Forms to assist history-taking
• Questionnaires to be filled by patients
• AUDIT questionnaire for alcohol use
• The forms will extract part of the data directly
  from the EPR

32
Computer-generated letter

• Dear Mrs. Smith,
• The laboratory tests taken on June 1st were as
  follows Haemoglobin was 134, which is excellent.
  The kidney test (creatinine 86) and the liver
  test (ALT 32) were normal. You can continue your
  medication. The next tests are due in October.
  The laboratory referral is attached.
• Faithfully yours,
• Dr. Jones

33
Decision support test site

• The test site allows simulation of all types of
  decision support on virtual patients

34
EBMeSD Study Group

• Pekka Rissanen, Tampere University
• Minna Kaila, Finohta, Tampere University Hospital
• Ilkka Kunnamo, Duodecim, Karstula Health Centre
• Tiina Kortteisto, Tampere University
• Marjukka Mäkelä, Finohta
• Jukkapekka Jousimaa, Duodecim
• Jorma Komulainen, Duodecim
• Taina Mäntyranta, Centre for Pharmacotherapy
  Development Rohto

35
Background

• In Finland, electronic guidelines are frequently
  used via one health portal that also disseminates
  the Cochrane Library
• All Cochrane reviews relevant to clinicians are
  summarized and linked to the electronic
  guidelines
• However, evidence is not always searched and used
  when it is available
• Electronic decision support system (EDSS) may
  bridge the gap

36
Objective

• To study the effectiveness of an EDSS system in
  the management of multiple acute and chronic
  health problems both in primary and secondary
  health care
• effects on speed and success of implementation of
  recommended care
• effects on patient outcomes
• effects on costs

37
PICO

• Participants
• volunteering physicians and patients both in
  primary and secondary health care
• Intervention
• evidence-based EDSS system that is created by
  guidelines groups AND linked to the computerized
  patient records
• timely reminders, tailored to individual patient
  issues
• Control
• No EDSS, automatic registration of practices
• Outcome
• speed and success of implementation of
  recommended care by guidelines
• patient outcomes

38
On-screen computer reminders effects on
professional practice and health care outcomes
(Protocol)Gordon RB, Grimshaw JM, Eccles M, Rowe
RE, Wyatt JC

• Primary hypotheses
• On-screen computer based reminders are more
  effective compared with no intervention.
• On-screen computer based reminders are more
  effective compared with other interventions.
• On-screen computer based reminders combined with
  other interventions are more effective compared
  with no intervention.
• On-screen computer based reminders combined with
  other interventions are more effective compared
  with on-screen computer based reminders alone
  (effect of educational co-interventions).

39

• Secondary hypotheses - Type of reminder
• Reminders that include some individual
  patient-specific information are more effective
  than generic reminders.
• Reminders that indicate that a response should be
  recorded or given are more effective than
  reminders that do not require a response.
  Interactive forms.
• Development of the reminder
• Reminders developed with the involvement of
  target clinicians are more effective than
  reminders developed without their participation.
  Local scripts.
• Delivery of the reminder
• Reminders delivered within a consultation are
  more effective than reminders delivered outwith a
  consultation (virtual health check).
• Reminders delivered by or with the knowledge of
  the target clinician's colleagues or patients
  (criteria for procedures or specialist
  consultations) are more effective than reminders
  delivered by other means.

40

• Content of the reminder
• Reminders that offer explicit advice are more
  effective than reminders that offer implicit
  advice or no advice at all.
• Reminders that include an explanation of their
  content or advice are more effective than
  reminders that do not include this.
• Reminders that are explicitly from or justified
  by reference to an influential source are more
  effective than reminders from another source. An
  influential source is a person or body likely to
  be perceived as credible by the target clinician.
• Behaviour targeted by the reminder
• Reminders will vary in effectiveness
  systematically according to the targeted
  activity. Various types of decision support
  studied.

41
Feasibility study focus groups

• Focus group interviews 7 groups, 39 physicians
  interviewed
• Both urban and rural physicians of different ages
  around Finland
• Interviews were done between October 2005 and
  January 2006
• Two moderators (MK and HV), interviews were
  audiotaped, transcribed, coded and interpreted

42
Focus group findings

• Physicians expect tailored and flexible solutions
  for their information needs
• They wish to choose the scope and depth of EDSS
• Concerns on time-management issues
• EDSS may serve better for purposes of therapy,
  less for diagnosis
• Generally, attitudes were positive and physicians
  easily made up suggestions for EDSS topics
• Experience of malfunctioning information
  technology systems may be a barrier

43
RCT outline
Health centre 1
Health centre n
Hospital 1
Hospital n
Implementation of the EDSS that utilizes
individual patient data
Selection of health topics and outcomes
Intervention(EDSS provided)
Control(No EDSS or different set of EDSS)
Primary outcome changes in health
Secondary outcomes changes in process of care
One-year follow up
44
Current status

• Basic funding for the study guaranteed
• Pilot testing of the EDSS is taking place
• Topics for EDSS are prepared and selected
• Baseline measurements in 2007
• practice characteristics
• organizational issues
• work load
• physician characteristics
• IT use and resources
• attitudes towards guidelines
• RCT starts in 2008

45
Pending questions

• Whom to randomize
• physicians or clusters of health care units?
• What topics to include
• variety of acute and chronic diseases
• potential of health benefit
• Ethical issues
• informed consent
• trial monitoring
• How to keep track of potential confounders
• changes in health services, co-interventions

46
Conclusions

• Evidence-based guidelines and national standards
  makes studying the EDSS feasible in Finland
• Physicians are used to using graded evidence,
  also Cochrane reviews, in guidelines
• Attitudes are generally supportive towards active
  implementation of evidence via EDSS
• Planning the RCT and the cost-effectiveness study
  needs a lot of thought and piloting
• All constructive criticism welcomed

47
For more information www.kaypahoito.fi/decisionsu
pport/decisionsupport.htm