Meta-analysis and occupational epidemiology.

Abstract

Meta-analysis is increasingly appearing in the epidemiologic literature. Although originally performed in the context of experimental or randomized controlled study designs and with regard to problems that are amenable to these designs, the method has also been applied to studies in observational epidemiologic settings. This movement has generated considerable debate about the validity of meta-analysis in observational studies, which can be said to be rather shaky. Meta-analysis is also appearing in occupational epidemiology, and a recent application to studies of industrial cohorts exposed to organic solvents has illustrated many of the arguments against meta-analysis of observational study. The propensity of meta-analysis for disarming critical approaches to individual study methods such as the standardized morbidity or mortality ratio (SMR) as effect measure, and to application of commonly used biostatistical techniques is illustrated. A cautionary note is sounded.