Calculating cancer latency using data from a nested case-control study of prostatic cancer

Assessing latency for neoplastic diseases is crucial for determining the causal effects of a complex mix of carcinogenic exposures. An initial assessment of cancer risks in a U.S. tire and rubber plant revealed a significant SMR of 140 for prostatic cancer. Using an industry-based, case-control death certificate study of prostatic malignancies, we found matched odds ratios of about 3 (p less than 0.025) for Batch Preparation, the work area with the greatest exposure to carbon black, solvents, and heavy metal oxides. To assess latency, we used the matched case-control series to calculate annual estimates of the odds ratio by determining the proportion of cases and controls employed for greater than 1 month in Batch Preparation during each year under study. This approach produced a plot with great fluctuations. To reduce variability in the resulting curve, a method was developed that measured the "etiologic fraction", which is its highest point represents an estimate of the peak of the latency distribution. For Batch Preparation the modal point was 29 years before death with the greatest risk occurring from employment in the mid-1940's. The latency method allows risk assessment for time and year of greatest exposure difference, thus suggesting appropriate prevention strategies. Applications of this method for other types of studies and exposures are discussed.