Methods have been developed and applied that incorporate treatment effects and changes in the surveillance definition into backcalculation models for projecting AIDS incidence. A method of estimating the probability of perinatal transmission from partially censored follow-up data has been devised. Work on the effects of measurement errors in exposure measurements has been published showing that unexpected biases can result from combining exposure categories. New methods were developed for analyzing case-control data with a validation sample that relates error-prone measurements to """"""""gold standard"""""""" measurements. Further work is underway to find improved methods of analysis and to understand the implications of an imperfect """"""""gold standard."""""""" Work has been published on the reliability of next-of-kin data and death certificate data on occupation. Methods are in development to analyze clinical trial data for categorical responses measured several times. A comprehensive series of three papers on the theory and practice of control selection for case-control studies is in press. Methods of inference, including variance calculations, have been developed for estimates of absolute risk derived from population-based case-control data. A model for predicting breast cancer risk based on data from the Breast Cancer Detection and Demonstration Project is being validated against internal and external data sources.
