We propose to conduct a prospective cohort study to evaluate a series of hypotheses relating nutritional factors to the incidence of coronary heart disease (CHD) and cancer, particularly large bowel cancer. Specific hypotheses include that increased risk of CHD is related to diets high in saturated fat, cholesterol, animal protein, and hydrogenerated vegetable oil, and low in polyunsaturated fat, fiber from specific sources, vitamins A, C, and E, calcium, selenium, and chromium. For large bowel cancer we will address hypotheses that higher risk is associated with high intake of total fat, saturated fat, polyunsaturated fat, cholesterol, and with low intake of specific sources of fiber, vitamins A, C, and E, calcium, and specific vegetables such as cabbage, cauliflower, and broccoli. To accomplish these aims we plan to enroll approximately 57,000 adult males between the ages of 40 to 75 years who are dentists, veterinarians, osteopaths, podiatrists, optometrists, and hospital pharmacists. Participants will complete a mailed general medical and health questionnaire at baseline in addition to an intensively validated semiquantitative food frequency questionnaire (SFFQ). At one a year, toenail specimens will be collected and catalogued for future nested case-control analyses of CHD risk in relation to levels of calcium, selenium and chromium. Follow-up questionnaires to update exposure informatin and ascertain non-fatal endpoints will be mailed at two-year intervals; all reported cases of non-fatal MI and cancer will be documented with hospital records and/or pathology reports. Fatal events will be ascertained with the National Death Index and documented using all available records. In order to standardize SFFQ nutrient scores against measurements of absolute intake, two 1-week diet records will be obtained from a random sample of 150 Boston-area participants. During four years of follow-up, we anticipate 1,214 cases of non-fatal MI, 996 of fatal CHD, and 274 of large bowel cancer. This will provide sufficient power to detect relative risks of 1.4, 1.4, and 1.8 (assuming = 0.05 and Beta = 0.80) for these three endpoints, comparing extreme quintiles of dietary intake levels derived from the SFFQ and slightly less power for analyses relating to elemental analyses of nails. However, an objective will be to provide quantitative dose-response estimates rather than simple tests of hypotheses.
Showing the most recent 10 out of 440 publications
