Comparison of methods for determining occupational exposure in a case-control interview study of chronic lymphocytic leukemia.

A comparison was made between two indirect methods for linking reported lifetime occupations and industries of employment with specific exposures and with directly reported exposure information, using data from a population-based case-control interview study of chronic lymphocytic leukemia (CLL). Exposures previously suspected of being associated with CLL were examined using a job-exposure matrix developed by Hoar et al and a linkage between observed occupational exposures and specific occupations, by industry, based on data collected in the National Occupational Hazard Survey (NOHS). In general, concordance on exposure for the two occupation-exposure linkage methods was fairly poor, although it was better for some of the exposures studied (butadiene and asbestos, for which most kappas were between 0.40 and 0.60) than for others (carbon tetrachloride and benzene with kappas ranging from 0.01 to 0.12). Higher proportions of cases and controls directly reported exposure to benzene and asbestos than was determined using the two more indirect methods, neither of which showed consistently greater agreement with direct reporting.