Data pooling in occupational studies.

Abstract

Summarizing epidemiologic data from multiple studies can be accomplished either by a meta-analysis of published findings or by pooling data and performing new analyses of enlarged data sets. Meta-analysis is relatively easy and inexpensive to perform but usually is restricted to examination of overall risks. Data pooling offers a more direct approach for subcohort analyses and for estimation of dose-response relationships. The advantages of pooling data from multiple studies of workers with similar exposures pertain primarily to enlarged study bases and a greater potential for evaluating statistically stable dose-response relationships than can be accomplished by meta-analysis. Valid methods for data pooling that are developed for cohort mortality studies of relatively large worker populations ultimately can be applied to studies of other health outcomes using different study designs.