Effect of Needle Size in EUS-Guided FNA Study with Rare Events Using Meta-Analysis

• Activity Number: 233
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Akshita Chawla*+ and Tapabrata Maiti and Vivek Pradhan
• Companies: Michigan State University and Michigan State University and Pfizer
• Keywords: EUS-FNA ; myocardial infarction ; needles ; pancreas ; rosiglitazone ; zero inflated Poisson random effects

Abstract

In clinical trials and other applications, meta-analysis is mainly conducted to summarize the effect size on primary or on key secondary endpoints. However, sometimes safety endpoints such as risk of having complications in pancreatic surgery or risk of myocardial infarction (MI) is the main point of interest in a drug study. As these endpoints are rare in nature, most studies report zero such incidences. As a workaround, either trials with both arms having zero events are deleted or 0.5 correction is applied. In randomized control trial (RCT) set-up, Cai, Parast and Ryan (Stat.Med. 2010,29 2078-2089) proposed methods based on Poisson random effect models to make inference on relative risks for two arms with rare event data. In this article, we give the general framework of how their assumption of having RCT can be relaxed and utilized for non-RCT studies to make inference on two treatments. We also propose two new approaches based on zero-inflated Poisson random effect models that are more appropriate with excessive zero counts data. We explicate our proposed models for two real world data by doing extensive simulations that corroborate the appropriateness of the proposed methods.