1. RevMan for Registrars Paul Glue, Psychological Medicine What is EBM? What is EBM? Different approaches/tools Different approaches/tools Systematic reviews Introduction to Review Manager Practical example Practical example Relapse prevention drug trials in anxiety disorders Perform a RevMan analysis on a set of papers Perform a RevMan analysis on a set of papers Calculate NNT Calculate NNT

2. What is EBM? An approach to systematically apply the best available scientific evidence to assist with medical decision making. Also assesses the quality of evidence of the risks and benefits of treatments/interventions Cannot be applied to all medical decision- making (e.g. quality of life) Scope can be broad (e.g. EB Guidelines) or narrow (EB individual decision making)

3. Techniques Systematic reviews of published research studies to evaluate treatments/interventions. e.g.Cochrane Collaboration – organization defining methods, repository for completed systematic reviews. e.g.Cochrane Collaboration – organization defining methods, repository for completed systematic reviews. Requires authors to provide a detailed and repeatable plan of literature search and evidence evaluations. Requires authors to provide a detailed and repeatable plan of literature search and evidence evaluations. Explicit assessment of clinical trial quality Trial design considerations: e.g. clearly-defined eligibility criteria, validated endpoints, minimal missing data. Trial design considerations: e.g. clearly-defined eligibility criteria, validated endpoints, minimal missing data. Generalizability considerations: e.g. data from narrowly-defined patient populations may not be generalizable to clinical practice. Generalizability considerations: e.g. data from narrowly-defined patient populations may not be generalizable to clinical practice. Followup: Was there adequate time for defined outcomes to occur? Followup: Was there adequate time for defined outcomes to occur? Statistical power: Was the study size adequate to detect a difference between treatment arms? Statistical power: Was the study size adequate to detect a difference between treatment arms?

4. Techniques Systems to stratify evidence by quality e.g. US Preventive Services Task Force e.g. US Preventive Services Task Force Level I: Evidence obtained from at least one properly designed randomized controlled trial. Level I: Evidence obtained from at least one properly designed randomized controlled trial. Level II-1: … from well-designed controlled trials without randomization. Level II-1: … from well-designed controlled trials without randomization. Level II-2: … from well-designed cohort or case-control analytic studies, preferably from more than one centre or research group. Level II-2: … from well-designed cohort or case-control analytic studies, preferably from more than one centre or research group. Level II-3: … from multiple time series with or without the intervention. Level II-3: … from multiple time series with or without the intervention. Level III: Opinions of respected authorities, based on clinical experience, descriptive studies, or reports of expert committees. Level III: Opinions of respected authorities, based on clinical experience, descriptive studies, or reports of expert committees. Statistical measures Meta-analytic methods Meta-analytic methods ROC curves ROC curves NNT NNT

5. Randomized Controlled Trials Gold standard for EBM Randomized – subjects are allocated to treatment arms randomly (usually by a computer generated random code) Double blind – treatment allocation is hidden from Dr and patient Control arm – new treatment compared with (e.g.) active control, placebo (inactive) control

6. Questions to be studied today How effective are antidepressants at preventing relapse in anxiety disorders? All major anxiety disorders respond to antidepressants All major anxiety disorders respond to antidepressants Clinical guidances recommend that after response, treatment should continue for >6 months to prevent re-emergence of original disorder Clinical guidances recommend that after response, treatment should continue for >6 months to prevent re-emergence of original disorder However this has not been studied systematically However this has not been studied systematically Is the effect of antidepressants on relapse prevention the same for all anxiety disorders? Is the effect of antidepressants on relapse prevention the same for all anxiety disorders?

7. Antidepressant relapse prevention designs Open-label antidepressant 6-8 weeks Responders randomized to stay on AD/switch to placebo Nonresponders are discontinued 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.0 0 100 200 300 400 500 Days since randomization Estimated survival probability citalopram placebo -Robust results in depression trials -Example of enrichment of patient population

8. DSM-IV and anxiety

9. The activity today You will be allocated all the antidepressant relapse prevention trials (n=3-6) for a specific anxiety disorder You need to extract data (n relapse and n randomized) for each of the trials and enter this into RevMan Calculate Odds Ratios for the relapse prevention effect of antidepressants Use the relapse data to also calculate NNT Prepare a brief (<3 slide) presentation of your output to present to the whole group.

10. Collecting your data This exercise skips several of the key steps involved in a systematic review This exercise skips several of the key steps involved in a systematic review Literature search Literature search Selection of appropriate studies based on explicit criteria (i.e. protocol) Selection of appropriate studies based on explicit criteria (i.e. protocol) What data were collected What data were collected Analytical Methods Analytical Methods

11. Literature search terms When did you search… How did you search… Where did you search… How did you explore further…

12. Inclusion criteria

13. Data Collection Are you comparing like with like? Are endpoints the same?f

14. Analysis Methods

15. Sorting out data from your search Titles identified from electronic databases and screened for retrieval (n=5424) Full text articles assessed for eligibility (n=72) Excluded (n=5289) Excluded (n=63) Combination therapy (n=10) Placebo control only (n=22) Duplicate data (n=12) Different patient group (n=15) Review article (n=3) Unusable data (n=1) Publications screened (n=135) Studies included in meta-analysis (n=9)

16. Using RevMan Open RevMan Open up a review from file anxiety-relprev-4-10.rm5 anxiety-relprev-4-10.rm5

17. 2: Study titles have already been entered 3: Then click on Data and analyses 1: Click on Studies and References

18. Click on Add Comparison

19. … and click Next Panic disorder, GAD etc

20. Click on Add an outcome…. ….then click Finish

21. Choose Dichotomous (relapse vs non-relapse) …and click Next

22. Fill in labels Click next

23. Select M-H Click next Random effects model Odds Ratio

24. Edit new outcome Click finish

25. A blank table shell is created Click here to enter studies

26. Highlight multiple studies to enter with CTRL And click finish

27. Relapse on AD Enter data from each study Total on AD Total PBO Relapse on PBO

28. Stats Generate a graph for presentation Study weight Forest plot Individual and pooled ORs Can uncheck to assess effects of individual studies (sensitivity analysis)

29. Presentation Forest Plot generated by RevMan

30. Calculating NNT NNT = 1 (nonrelapsers on ADs/all AD)-(nonrelapsers on PBO/all PBO) (nonrelapsers on ADs/all AD)-(nonrelapsers on PBO/all PBO) Round NNT up to next whole number NNT 2-4 – strong effect for treatment NNTs >10 may still be useful for prophylaxis

31. Go!