Potential Problems with Tian, Cai, Pfeffer, Piankov, Cremieux, and Wei's Exact Stratified Analysis of Rosiglitazone Cardiovascular Deaths

• Activity Number: 239
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Fraser Smith*+
• Companies: FDA/CDER
• Keywords: meta-analysis ; Rosiglitazone ; Tian ; proc StatXact ; CMH test ; fixed effects

Abstract

Lu Tian et al. proposed using an exact fixed effects inference procedure for meta-analysis of independent 2 by 2 tables and found no association between Rosiglitazone use and CVD death (p=0.83). Tian et al. noted that Nissan and Wolski's analyses that found an association between Rosiglitazone and the risk of myocardial infarction and CVD death excluded studies with zero events and that large-sample approximations may be inaccurate and lead to invalid conclusions when the individual study sample sizes are small, the number of studies is not large, or the event rates are low. However results from other analyses using CMH tests and proc StatXact before and after combining studies with zero events and random effects models also appeared comparable to those obtained by Nissan and Wolski. Unfortunately Tian et al.'s method may have been underpowered for analyzing Rosiglitazone CVD deaths.