Methods of analysis for case-control studies with cluster sampling of controls are being adapted for application to a study of non-melanoma skin cancer. New methods for estimating absolute risk from population- based case-control studies have been developed and compared, and the method used for estimating individualized risks of breast cancer has been extended to permit valid variance estimates. A computer program has been written to permit physicians to estimate individualized risk of breast cancer easily. Recently completed work shows that attributable risk can be estimated even if some unexposed subjects are misclassified as exposed. A series of studies have elucidated the role of energy adjustments in dietary studies and highlighted difficulties of interpretation. Several projects were completed on the design and analysis of randomized trials, and new methods of analysis, including covariate adjustment and handling of missing data, were developed for randomized community intervention trials. The method of backcalculation has been generalized to estimate age-specific HIV infection rates. New methods were developed to obtain smooth estimates of the AIDS incubation distribution and infection rates in cohorts of hemophiliacs. New methods of age-specific cancer prevalence estimation were developed to determine whether cancer rates are excessive among patients in AIDS registries.
