Table 2.

Primary analytical methods, continuity corrections, assumptions, and outcomes

Method, measure, effect type, and data sources	Single zero event trials	Zero total event trials	Continuity correction	Assumptions to satisfy or difficulties to consider	Outcomes
Peto, odds ratio, fixed effect
IPD only, IPD+RECORD, IPD+summary, IPD+summary+RECORD	Included	Excluded	None	Event rates <1%, balanced groups (treatment arms), small/moderate treatment effects	Analyses with IPD only: composite outcome, heart failure, myocardial infarction, cardiovascular related deaths, non-cardiovascular related deaths Analyses including summary data: myocardial infarction, cardiovascular related deaths
Mantel-Haenszel or Dersimonian and Laird (inverse variance), odds ratio or relative risk, fixed or random effects
IPD only, IPD+RECORD, IPD+summary, IPD+summary+RECORD	Included	Excluded or Included	Constant continuity correction of 0.5 or treatment arm continuity correction	Sample must be “large” overall (crude totals across all studies needs to be ≥5),44 might perform comparably or better than Peto’s method at even rates of 5-10%44	Analyses with IPD only: composite outcome, heart failure, myocardial infarction, cardiovascular related deaths, non-cardiovascular related deaths Analyses including summary data: myocardial infarction, cardiovascular related deaths

IPD=individual patient level data; RECORD=Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes study.