The American Cancer Society Cancer Prevention Study II Nutrition Cohort

Prospective cohort (Cancer Prevention Study II Nutrition Cohort) established in 1992–1993 as a subgroup of the CPS-II; baseline questionnaire mailed to 516,671 CPS-II men and women aged 50–74 across 21 states; ~90% follow-up; completed questionnaires from 55,105 CPS-II men and 50,937 CPS-II women; a second mail-out expanded to 86,406 men and 97,788 women; final Nutrition Cohort includes 162,408 cancer-free participants at baseline (77,048 men; 85,360 women); predominantly White (97%), married (88%), with 38% college graduates; ages at baseline largely 50–69; data collected include demographics, medical and family history, reproductive history (women), use of medicines and vitamins, occupational exposures, brief dietary and physical activity assessments, and hormone use (women); follow-up every 2 years to update exposures and identify incident cancer and deaths; cancer incidence and mortality verified via medical records, registry linkages, and death certificates; blood samples from 39,200; buccal cell DNA from 50,000; FFQ dietary assessment validated against 24-hour recalls; questionnaires processed by optical scanning; aims to study cancer occurrence, mortality, and survival in relation to obesity/weight change, physical activity at life stages, vitamin supplement use, exogenous hormone use, other medications (aspirin/NSAIDs), and cancer screening modalities;

A large, long-term prospective resource created to study cancer risk and survival in relation to obesity/weight change, physical activity at various life stages, vitamin supplement use, exogenous hormone use, and medications, plus cancer screening modalities. It includes over 160,000 cancer-free adults (77,048 men; 85,360 women) from 21 states, with detailed dietary and lifestyle data, repeated exposure assessments, and biological samples (blood from 39,200; buccal DNA from 50,000). This design enables robust internal analyses and nested case–control studies of cancer outcomes, and its generalizability lies mainly in internal comparisons rather than extrapolation to the general population; it also supports future gene–environment interaction analyses as DNA data are used. The cohort is a selected sample intended to optimize data quality and follow-up rather than to perfectly represent the general population.

Not a random, population-based sample; selected cohort designed to optimize data quality and follow-up, which may limit generalizability to the broader population. Dietary assessment via FFQ tended to underestimate total energy and many nutrients in 1992–1993, with validity varying by nutrient (low for some micronutrients like zinc and vitamin E); repeated exposure assessments rely on self-reported data and measurement error may persist; some baseline data had missing values.