Vitamin and mineral supplements are among the most commonly used drugs in the United States, despite limited evidence on their benefits or risks. This paper describes the design, implementation, and participant characteristics of the VITamins And Lifestyle (VITAL) Study, a cohort study of the associations of supplement use with cancer risk. A total of 77,738 men and women in western Washington State, aged 50-76 years, entered the study in 2000-2002 by completing a detailed questionnaire on supplement use, diet, and other cancer risk factors, and 70% provided DNA through self-collected buccal cell specimens. Supplement users were targeted in recruitment: 66% used multivitamins, 46% used individual vitamin C, 47% used individual vitamin E, and 46% used calcium, typically for 5-8 of the past 10 years. Analyses to identify confounding factors, the main study limitation, showed that regular nonsteroidal anti-inflammatory drug use, intake of fruits and vegetables, and recreational physical activity were strongly associated with supplement use (p < 0.001). The authors describe a follow-up system in which cancers, deaths, and changes of residence are tracked efficiently, primarily through linkage to public databases. These methods may be useful to other researchers implementing a large cohort study or designing a passive follow-up system.

In the United States, dietary supplements contribute a large proportion of micronutrient intakes. Therefore, it is important to collect accurate information on supplement use for studies of micronutrients and disease risk. This report describes the test-retest reliability and validity of a detailed, self-administered mailed questionnaire on vitamin and mineral supplement use. Participants (n = 220) completed the questionnaire at baseline and 3 months later. During an in-person interview, participants provided spot urine and blood samples, and interviewers transcribed nutrient information from their supplement bottle labels. The questionnaire had very good test-retest reliability for mean supplement intake over the past 10 years, with intraclass correlations ranging from 0.69 for beta-carotene to 0.87 for vitamin E. Pearson's correlation coefficients comparing current supplemental intakes from the questionnaire and interviews/label transcriptions were high, ranging from 0.58 for beta-carotene to 0.82 for chromium; however, for some nutrients, median intakes from the questionnaire were slightly lower than from the interviews. Beta-carotene, vitamin C, and vitamin E (alpha-tocopherol) showed clear linear trends of increasing blood concentrations with higher self-reported supplemental intakes (Pearson's correlation coefficients adjusted for potential confounding factors and diet = 0.31, 0.29, and 0.69, respectively; all p < 0.0001). Creatinine-adjusted spot urinary calcium values were not associated with supplemental calcium intakes (Pearson's r = -0.07). This self-administered questionnaire demonstrated high reproducibility and validity for collecting detailed information on supplement use.