Work continued on appropriate methods for selecting controls and on the reliability of exposure data from death certificates and interviews with next-of-kin in case-control studies of cancer. Problems in interpreting data from prevalent cohorts, like seropositive persons at risk for the acquired immunodeficiency syndrome, were studied. Methods for cancer risk projection for individuals and populations were applied to cohorts at high risk for breast cancer and to those exposed to radiation. New methods for the analysis of case-control studies with cluster sampling are in development, as well as methods to detect qualitative interactions in clinical trials and observational studies. Other projects include projection of long-term cancer rates, applications of polychotomous logistic regression, and development of computer software for epidemiologic analyses.
