The objectives of this project are to develop, adapt, expand and evaluate methodological procedures useful in epidemiologic studies of cancer. Methods of design of case-control studies were given particular emphasis this year, attention focusing on procedures for selecting an unbiased sample from cohort data, selecting controls using random digit dialing, and developing formulae for sample size when exposure is continuous. Prospective and retrospective designs were described within a unified framework based on the proportional hazards model. Adaptations were made to include multiple disease types and recurrent disease, and to estimate population attributable risk with multiple risk factors in case-control studies. Methods for collecting and evaluating dietary data were evaluated. Other work focused on evaluating surrogate responses and assessing quality of response for next-of-kin of dead controls in case- control interview studies. Investigation of general epidemiologic techniques for studying environmental cancer continued.
