Case Study: Evaluating the risk benefit of oral anti-diabetes agents given the recent FDA guidance in T2 diabetes

Question 1

Background

Answer 1

• To quantify background risk of CVD in a variety of subpopulations
• Appraise RR of CVD of alternative treatments
• Infer risk in conjunction with benefit in absolute and relative terms

Question 2

Rosiglitazone Meta analysis

Answer 2

• 42 of 116 studies selected
• Fixed effects model
• Limitations (no original source data, no time to event analysis, no dose-response data, events in ineligible trials ignored)

Question 3

Meta-analysis

Answer 3

• is the statistical procedure for combining data from multiple studies that address a set of related research hypotheses.
• when effect size is common, meta analysis can identify the common effect
• when effect size is not common meta analysis can identify the reason for the variation

Question 4

Fixed effects model

Answer 4

stipulates that studies under analysis constitute the entire population of units (only within study variation considered to influence the uncertainty of results)

Question 5

Random effects model

Answer 5

stipulates that both within study sampling error (variance) and between study variation are included in the assessment of uncertainty (gives wider (CI than fixed effects models)

Question 6

Two faces of meta analyses

Answer 6

• Aggregation (combine study results)–>focus on chance (power)
• Explanation (compare study results)–> emphasis on validity and generalizability

Question 7

RECORD Trial

Answer 7

Randomized, multi center open label active control safety study
-primary endpoint of time to first hospitalization for CVD event or death from CVD cause