Title: How do CME speakers use research results to support therapeutic recommendations A quantitative and q
1How do CME speakers use research results to
support therapeutic recommendations?A
quantitative and qualitative study
SACME Rancho Mirage, CA April, 2009
- Michael Allen, Tanya Hill, Richard
Handfield-Jones, Mike Fleming, Doug Sinclair, Tom
Elmslie
2Research team
- Dalhousie CME
- Michael Allen
- Tanya Hill
- Michael Fleming
- Doug Sinclair
- University of Ottawa CME
- Richard Handfield-Jones
- Foundation for Medical Practice Education
- Tom Elmslie
3Research Team
Im Doug Sinclair from Dalhousie CME. I can find
time to do anything.
Im Richard Handfield-Jones, the PI from
University of Ottawa CME. Im in the army now and
AWOL.
Im Mike Fleming from Dalhousie CME and Im not
sure how I got involved in this.
Im Tanya Hill the research associate from
Dalhousie CME. I dont care who says theyre the
PI, Im running this show.
Im Mike Allen, the PI from Dalhousie CME
Im Tom Elmslie. Im an ideas man and helped with
experimental design.
4Funding
- Royal College of Physicians and Surgeons
- of Canada
- Nova Scotia Department of Health through
- Drug Evaluation Alliance of Nova Scotia
5Outline
- Background
- Objectives
- Methods
- Results
- Key messages
6Think about this RCT
- Outcomes non-fatal myocardial infarction and
death from coronary heart disease - Subjects No history of coronary heart disease
(primary prevention) - Duration 3.3 years
- N 5100 patients in the control and 5100 patients
in the drug group - The patient characteristics are
- 80 male
- mean age 63 yrs.
- mean blood pressure 164/95
7What is your interpretation of the following
results? How likely might you be to prescribe or
take the drug based on each one?
- The drug led to a 36 decrease in the incidence
of non-fatal MI and CHD death (relative risk
reduction). - The drug decreased the rate of non-fatal MI and
CHD death from 3.0 to 1.9, an absolute risk
reduction of 1.1. - You would have to treat 94 patients for 3.3 years
to prevent one non-fatal MI or a death from CHD
(number needed to treat). - The 95 confidence intervals around the previous
result (ie, treat 94 patients for 3.3 years to
avoid one non-fatal MI or CHD death) are 60 and
215. - At the end of 3.3 years, 97.0 of patients who
don't take the drug will remain free of a cardiac
event and 98.1 of patients who take the drug
will remain free of a cardiac event (inverse
absolute RR).
8Patients not having MI or dying Placebo
9Patients not having MI or dying Treatment
10- Forrow L, Taylor WC, Arnold RM. Absolutely
relative How research results are summarized can
affect treatment decisions. Am J Med 1992
92121-4. - Naylor CD, Chen E, Strauss B. Measured
enthusiasm Does the method of reporting trial
results alter perceptions of therapeutic
effectiveness? Ann Intern Med 1992 117916-21. - Nikolajevic-Sarunac J, et al. Effects of
information framing on the intentions of family
physicians to prescribe long-term hormone
replacement therapy. J Gen Intern Med 1999
14(10)591-598 - Nexoe J, Kristiansen IS, Gyrd-Hansen D, Nielsen
JB. Influence of number needed to treat, costs
and outcome on preferences for a preventive drug.
Fam Pract 2005 22(1)126-131.
11Objectives
- To determine
- How much emphasis speakers place on presenting
data in absolute vs relative terms - How familiar FPs and speakers are with
statistical terms (especially absolute and
relative terms) - The attitudes and preferences of FPs about having
research data presented - How well speakers link therapeutic
recommendations to research data
12Methods
- 2 annual 3-day Family Medicine review conferences
- Halifax (Dalhousie University CME)
- 230 registrants
- 42 presentations (lecture and concurrent
sessions) - Ottawa (University of Ottawa CME)
- 209 registrants
- 38 presentations (lecture and concurrent
sessions) - 3 data collection methods for each conference
13Methods Data Collection
- Questionnaire
- Registrants and Speakers
- Evaluate understanding of statistical terms
- Focus groups
- 1.5 hour with FPs from each site
- Determine preferences about types of statistics
used - Analysis of presentations
- PowerPoint files
- Video recording of presentations
14Methods Data Analysis
- Questionnaire
- Descriptive statistics
- Focus groups
- Recorded and transcribed
- Content analysis
- Analysis of presentations
- PowerPoint files
- How research data presented slide counts
- Videorecording
- Linking of research data to recommendations
Likert scales
Two researchers
15Results Questionnaire
- Registrants (N121 29 response rate)
- 97 physicians 81 certificant of CFPC
- 51 male 49 female
- Avg. yrs practice 18.0
- 27 completed EBM workshop in last 10 years
- Speakers (N20 33 response rate)
- 19 specialists, 1 family physician
- 70 male 30 female
- Avg. yrs practice 18.0
- 30 completed EBM workshop in last 10 yrs
16Results Questionnaire
17Results Questionnaire (Registrants N121)
18Results Questionnaire (Speakers N20)
19Results Questionnaire
"How much emphasis should CME speakers place on
presenting research results when making
therapeutic recommendations?"
Percent responses
20Results Focus group
- 2 focus groups
- 9 males 5 females
- Most preferred statistics
- Numbers needed to treat
- Inverse ARR
- Confidence Intervals
21Results Focus group
- 2 focus groups
- 9 males 5 females
- Thats a pretty impressive statistic (RRR)
- Youd never see that in a drug trial (Inverse
ARR) - I responded to that in a different way it also
reminded me of the responsibility we have to
share this information with our patients (Happy
faces table) - Youre getting less and less enthusiastic about
prescribing - Most preferred statistics
- Numbers needed to treat
- Inverse ARR
- Confidence intervals
22Results Focus group
- Trust in speakers and CME providers
- Trust that specialists can critically evaluate
research - Skepticism about specialist ability to interpret
data - Attend CME because specialists simplify data and
make recommendations supported by research - Trust CME offices to ensure programs present data
accurately and completely - Disclose conflict of interest
23Recommendations from focus group
- Give a short stats refresher at the start of a
conference - Provide a 1-page handout on stats in the
registration package - Use consistent approach use the same format for
graphs and stats - Provide more research details when presenting
newer areas of medicine, pivotal trials, or
during drug sponsored company events
24Preliminary results PowerPoints (N26)
- General terms
- Graphs
- Risk (event rate)
- Prevalence
- Absolute or relative not specified
- Absolute terms
- Absolute risk reduction or increase
- Number needed to treat or harm
- Relative terms
- Relative risk
- Relative RR
- Odds ratio
- Hazard ratio
95 confidence intervals
25Preliminary results PowerPoints (N26)
Number of slides with . . .
26Preliminary results Powerpoints (N26)
Number of slides with . . .
27Preliminary results PowerPoints (N26)
Number of slides with . . .
28Preliminary results Presentations (N26)
Number of slides with . . .
29Preliminary results Videotapes (n19)
30Conclusions
- Most FPs have low to moderate level of
statistical comprehension - Speakers have higher level of statistical
comprehension than FPs - FPs want recommendations to be supported with
research - FPs want speakers to critically evaluate the
research - Most speakers present little data in relative or
absolute terms or with 95 CIs - Focus group recommendations for CME providers
- Provide short review on statistics for FPs
- Develop consistent format for presentation of
research data
31Next Steps
- Provide CME for FPs to help them understand
research data - Develop format for presenting data that FPs find
helpful and understandable - Provide faculty development for speakers
- Add question to evaluation form re presentation
of research data
32Questions?