DISCONTINUED To examine prospectively relations between diet and major cancers, especially those of the breast, large bowel, and prostate. Background: This study addresses two methodologic problems plaguing observational epidemiologic studies of diet and cancer: 1) dietary assessment error and 2) insufficient range of intake (dietary homogeneity). Methods: The studys unique two-phase design involves the oversampling of screenees within extreme categories of dietary intake (defined according to the joint distributions of fat, red meat, fiber, and fruit and vegetable intake) in order to ensure relative dietary heterogeneity of the cohort. The large cohort size reflects sample size calculations based in part on the expected measurement error. The final cohort (540,000, 320,000 men, 220,000 women)is drawn from the AARP membership aged 50-69. The primary data collection instrument is a food frequency questionnaire (FFQ) modified to capture a broad range of key nutrients and foods. The baseline questionnaire includes questions on cancer-related covariates, such as smoking, reproductive characteristics, physical activity, and family history. A calibration study was conducted that compares responses from the FFQ to those from 24-hour dietary recalls (n=2000, two recalls each). To obtain in the final cohort a sufficient number of study participants within the extreme intake categories, the baseline questionnaire was mailed to 3.5 million AARP members in California, Florida, Pennsylvania, New Jersey, Louisiana, North Carolina, and the Detroit and Atlanta metropolitan areas. A second questionnaire, containing questions on eating practices, past diet, medication usage, and more extensive questions on medical conditions, physical activity, and body size history, was mailed to the cohort a few months after its selection. The followup period is five years. Endpoint ascertainment will be carried out primarily through state and metropolitan area registries. Progress: Data from the baseline questionnaires (over 600,000) and calibration study are now being analyzed. Over 350,000 second questionnaires have been returned. A biologic specimen component for this study, most likely buccal cells collection, is being developed. - Human Subjects: Interview, Questionaires, or Surveys Only
