A strategy for merging and analyzing work history data in industry-wide occupational epidemiological studies

Study populations examined in epidemiologic investigations of occupational disease risks often are assembled by the pooling of employee data from several workplaces that share common exposure factors. The primary objectives of this approach are to enhance the representativeness of the overall study population and to obtain sufficient employee sample sizes in exposure subgroups of interest. Among the many epidemiologic aspects that must be considered carefully in such industry-wide studies is the inter- and intra-company or plant comparability of employee work history data. Currently, the literature is void of articles that specifically address the fundamental data reduction and statistical analysis issues related to merging work history data from several distinct cohorts. This paper describes the basic methodologic problems associated with the pooling of work history data and proposes a joint job title/job exposure-based uniform coding scheme (UCS) that facilitates the aggregation of data from similar or diverse cohorts in a variety of epidemiologic study settings. The utility of the UCS as both an efficient data coding structure and a flexible basis for statistical analysis is described within the context of a popular computer software program for analyzing occupational cohort data. The fundamental features of the UCS are illustrated using data from a recent industry-wide study of copper and zinc smelter workers.