METHODS FOR POOLED ANALYSES OF EPIDEMIOLOGIC STUDIES

This paper presents a systematic methodology for the pooling and analy sis of previously conducted epidemiologic studies. It discusses the me thodologic issues to consider and the procedures to follow when conduc ting a pooled analysis of individual subject level data. I review rand om and fixed effects models to estimate pooled risks and the methods t o examine whether heterogeneity in study-specific effect estimates exi sts. Analytic strategies presented here can be used to assess sources of heterogeneity, particularly those originating from different study designs and methods. I describe methods for incorporating a qualitativ e assessment of the study design and data collection methods into the quantitative estimation of the overall effect. Finally, I make recomme ndations for future pooled analyses of epidemiologic studies.