Transcript Evidence Based Primary Care

Practice of Evidence-Based
Medicine
9th April 2001
Dr Martin Dawes
University of Oxford
1
Who am I?
•
•
•
•
•
•
•
General Practitioner for 17 years
Worked in a town practice 5 partners
Started research in 1985
Investigating hypertension
Director Centre Evidence-Based Medicine
Wife is a physiotherapist
2 children aged 19 & 17
Contents
•
•
•
•
•
•
Why EBM?
What is EBM?
Clinical problem exercise
Knowledge
CAT’s
15 seconds to FAQ’s
3
Increasing Medical Knowledge
• 27 Kg of Guidelines
• New papers per day
– 3,000
• Medline New articles
– 1,000
• RCT’s
– 46
4
Increasing Knowledge
Number of articles on Hypertension cited in
Medline by Year
8000
6000
4000
Articles
2000
0
1966
1976
1986
1996
5
How Much Do We Read in Clinical
Practice?
•
•
•
•
•
1/2 hour or less/week:
1 hour:
1 1/2 hours:
2 hours:
3 hours or more:
3%
46%*
23%
20%
8%
6
Where do we get knowledge?
• Print Source:
–
–
–
–
–
General/speciality textbooks
Pharmaceutical textbooks
Journals
Drug Company Information
Self Made Compendia
• Human Sources
•
Reported
%
62
Observed
%
27
25
14
18
1
4
3
9
7
1
7
33
53
Gorman P et al. Bull Med Libr Assoc 1994; 82(2): 140-6
7
Educational Strategy
•
•
•
•
•
•
Talks by experts not GP’s
Subjects are chosen by others
Usually relate to secondary care
Often about topics we feel comfortable with
Usually large lectures
Very little follow up
8
How many do we see
• 100 - 200 consultations a week
• 5,000 -10,000 per year
• It is difficult to be sure that one is doing the
right thing for all these cases.
• One relies on experience
• Use 2 million pieces of information - stored
in your memory
9
Aren’t we doing OK anyway?
• in 1994 17% of family Practitioners thought
that aspirin had no effect on the survival of
patients having an MI (evidence 1988)
• Variation in prescribing, diagnostic tests,
supervision of chronic conditions
10
Why Evidence Based Health Care?
•
•
•
•
•
•
•
Too many patients
Too many problems
Too many journals
Information overload
No time to read
Read what I am familiar with
Avoid difficult issues
11
What is Evidence-Based
Medicine?
•
•
•
•
•
•
See a patient
Ask a question
Seek the best evidence for that question
Appraise that evidence
Apply the evidence
Monitor the change
12
Structured Question - Why Bother?
• What do you think about ACE inhibitors
and heart disease
• You know about EBHC - can you look it
up?
• ACE & Heart Disease 2,755 articles on
Medline through PubMed
• Need structure to my question
13
Foreground
Questions
Background
Questions
Experience with Condition
14
To be satisfactorily answered the question
must be focused and well articulated for all
4 parts of its "anatomy"
• the patient or problem being addressed;
• the intervention or exposure being
considered;
• the comparison intervention or exposure,
when relevant;
• the clinical outcomes of interest.
15
How can you recognize and formulate
clinical questions as they occur?
• pay careful attention to the questions that
spontaneously occur to you.
• listen for the question behind the question
• what can I use for a sprain
• might become
• is a topical NSAID like aspirin more effective than
paracetamol at enabling resumption of sport at 1
week?
16
What if too many questions arise?
• patients may have several active problems
– possible questions about diagnosis, prognosis, therapy
for each problem
– your questions may be too numerous to even ask, let
alone answer.
• What is the most important issue for this patient
now?
• Which question, when answered, will help me
most?
• then selecting from the many the few questions
that are most important to answer right away.
17
Three Part Question
1.In adults with heart disease
2. does the addition of an ACE inhibitor to
their present treatment
3. reduce mortality
• intervention - looking for addition
• endpoints - mortality (could be morbidity)
18
Exercise 5 Mins
• Clinical case
• Important or frequent question
• Structure
– Problem
44 yr old M
Chest Pain
80 yr old OA
knee
– Intervention
– Comparison
– Outcome
ECG
Exclude
Top NSAID
Oral Nsaid
Less
MI
Pain
19
What is the best evidence?
• Friend once told me
• Group of old
Professors
• Case series
• Systematic Review
• RCT
• Case control trial
• Case
– http://cebm.jr2.ox.ac.uk
/docs/levels.html
20
Search for Evidence
•
•
•
•
•
•
•
Correct Database(s)
High Sensitivity
Wildcards * ?
MESH
Related Articles
Then specify (limit) ie Systematic Reviews
Language, human, etc
21
Plan Search
Patient
Problem
Intervention Outcome
Adult*
Angina
ACE
Survive
Diabetic
Heart
Disease
Angiotensin
Not Dead
Middle
Aged
MI
Enalapril
Mortality
22
Appraise Evidence
• Easy if it has been done before
– Journal of Evidence-Based Medicine
(demonstrate)
– American College of Physicians journal
– Cochrane database
• Have an appraisal course
• Use a journal club
23
Appraise using JAMA Guide
Evidence-Based Medicine Working Group
• Are the results valid?
• What are the results?
• Will the results help me in patient care?
• http://medicine.ucsf.edu/resources/guideline
s/users.html
24
Requirements of New Treatment
• The drug must have specific defined mode of action.
• It must be tested on a well defined disease.
• The time of action must be observed.
• The effect of the drug must be seen to occur constantly in
many cases.
• The experimentation must be done with the human body,
for testing a drug on a lion or a horse might not prove
anything about its effect on man.
• Ibn Sina [981-1037]
25
Types of health care professional
Evidence
generator
Evidence Users
Evidence
Finders
Evidence Ignorer
26
Current Information
•
•
•
•
•
Experiential
Colleagues
Patients
Guidelines
Books
• Research
Source
Fast
Variable
Fast
Lost
Fast
Access
Fast
Fast
Fast
Buried
Lengthy
Quality
Dubious
Unclear
hmmm
Good
Variable
Slow
Lengthy
Variable
27
What I want
•
•
•
•
•
•
not to have to wade through papers
not to have to go on line with Medline
to have an index of the strength of evidence
to have a one word answer
to see questions that already have answers
15 seconds to medical knowledge
Critically Appraised Topic
•
•
•
•
•
•
•
Declarative title
Question
Name of paper
Search terms
Design
Setting
Patients
•
•
•
•
•
•
Intervention
Outcome Measures
Results
Table
Conclusion
Commentary
29
A suspicious breast lump in a young woman is not necessarily cancer.
Suspicion of a breast lump in women under 50 is not highly predictive of breast cancer (12.3%) and reassurance can
be given. In older women breast lumps are more likley to be malignant (41%).
Citation/s: Seltzer M, The significance of Breast Complaints as correlated with age and breast cancer. Am Surg 1992 58(7)
413-7
Three-part Clinical Question: In women presenting with breast lumps what is the likilehood that the lump is a cancer?
Search Terms: MESH Breast-neoplasms/diagnosis, predictive value of tests, lump. 2 articles selected
The Study Patients: Women ( 93% white) presenting at the author's personal practice between 1987 and 1990 in New
Hersey, USA.
Prognostic Factor: Breast Lump clinically needing biopsy and patients accepting referral (n=399)
The Outcome: Breast Cancer
There was a well-defined sample at a uniform (early) stage of illness. Follow-up was long enough; follow-up was complete.
Can't tell if there were blind, objective outcome criteria. Adjustment was made for other prognostic factors. There was no
validation in an independent test-set of patients.
Prognostic Factor
Outcome
Result
Measure
Confidence Interval
Independent?
Breast Lump (age
Cancer
12.3%
Biopsy
8.55-15.9
no
Breast Lump (age>=50) requiring biopsy
Cancer
41.2%
Biopsy
31.3-51.7
no
Comments: This is an American study that is only one doctor's practice. However it is similar to findings from
another study in this cat bank.
Appraised by: Dr M DawesCEBM; 22 August 2000
30
Email: [email protected]
Update By: When new data published - August 2001
Global Medical Knowledge
Database (Alex)
• Database (Index) of questions
–
–
–
–
With answers linked (indexed)
Yes/ No/Maybe
Strength of Evidence
Set out as in EBM (CAT - Critically Appraised
Topic)
• Objective, Design, Setting, Patients, Intervention,
Main Results, Conclusion, Commentary
– Date/ Who Appraised
31
32
Aims for Alex
• To appraise articles about problems that are
common or serious
• To translate these into answered questions
• To have a database of questions sorted by
topic, drug, key word etc
• To make appraised evidence available
within 15 seconds
• On your desktop or Psion
33
Software to generate knowledge
34
How will it work?
•
•
•
•
•
If 1% of the health professionals
From around the world
Wrote two CAT’s per year
What a database we would have!!
Needs peer review, updating, free access,
and paper versions.
• This is the future……..
35