Estimating health risks in studies of the health effects of asbestos.

Epidemiologic studies of respiratory cancer hazards associated with asbestos exposure in working populations should provide some estimate not only of whether an excess in cancer exists but of the magnitude of this excess. While most studies seem to be in agreement as to the existence of an excess, there is considerable disagreement as to its magnitude. For 11 studies, estimates of relative risk for respiratory cancer range from about 1.2 to 9.2. Four features of these 11 studies can be identified that account for some of the variation in results: some studies include workers whose exposure to asbestos was too recent to be likely to produce cancer by the end of the follow-up period, and results are diluted by the presence of these workers; in most studies, the wrong population was used for estimating expected numbers of deaths; in some studies, death certificates for the study population were corrected based on a review of other medical records, but then comparisons were made with expected deaths derived from uncorrected death certificates; finally, in most studies, level and duration of exposure is unknown, and this is clearly a factor that determines the magnitude of the respiratory cancer risk. For 6 of the 11 studies it was possible from published material to correct results roughly for the first 3 of these factors and much of the variation disappeared. Problems in the 11 studies reviewed here should be considered in planning and reporting on epidemiologic investigations in the future.