Erikka Loftfield

Ph.D., M.P.H.

NCI

Research Fellow

PLCO (Learn more about this study)

PLCO-322

Nov 8, 2017

Multivitamin Use, Mortality, and Cancer: Leveraging Repeat Measures to Clarify the Impact of Multivitamin Use on Mortality and Cancer Risk

In the United States, an estimated 31% of adults use multivitamins (MV), with prevalence of use even higher among women and older Americans. As people take multivitamins to enhance health and prevent disease, studies of the effects of MV use on total and cause-specific mortality are the ultimate measures of public health impact. However, despite widespread use, the scientific evidence on MV use and mortality is mixed and highly controversial, and several important questions remain unresolved. First, it is unclear how MV use changes over time among individuals, and how such changes may affect disease. One issue of concern is that patients with diagnosed disease may alter their MV intake, either increasing or decreasing their use because of perceived benefit or harm. Second, confounding by healthy lifestyle remains a major concern as MV users often differ from nonusers with respect to diet, smoking, physical activity and obesity. Finally, it is unclear how MV use is associated with cancer incidence among those who are cancer-free at baseline.

The proposed study will investigate the hypotheses that daily MV use is associated with lower risk of 1) total mortality, 2) cause-specific, and 3) cancer incidence by leveraging existing data from three, large, geographically diverse, U.S. cohorts (i.e., the NIH-AARP Diet and Health Study cohort; the Agricultural Health Study (AHS) cohort; and the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial cohort) with repeat assessments of MV use and extensive follow-up for mortality outcomes. With a combined sample size exceeding 670,000 adults and 107,000 deaths, the study will be substantially larger than most previous studies and will therefore be well-powered to test associations among population subgroups including those who are adherent or non-adherent to healthy lifestyle recommendations. Furthermore, the proposed study addresses key limitations of prior studies. First, each of the three cohorts included in the analysis obtained detailed baseline information on demographics and lifestyle factors that may confound the association between MV use and mortality. We can incorporate this data into multivariable models to obtain adjusted relative risk estimates, and because of the very large sample size, we can also stratify by important potential confounders and effect modifiers (e.g., smoking status) to thoroughly investigate the impact of these factors on risk estimates. Second, each of these cohorts obtained multiple assessments of current MV use as well as lifetime history of MV use for a large subset of participants. This longitudinal data will allow us to evaluate the impact of changes in MV use and other lifestyle factors on mortality. In addition, each of these cohorts has extended follow-up time (i.e., >15 years), which will allow us to assess potential reverse causality by comparing the association between MV use and mortality during different periods of follow-up.

Aim 1A: Prospectively evaluate the association of daily MV use, as compared to non-use, with total mortality and cause-specific mortality from the 10 leading causes of death in the United States in three large, geographically diverse cohorts.

Aim 1B: Investigate the impact of: 1) changes in MV use over time on risk of mortality using follow-up assessments, and 2) confounding by healthy lifestyle using detailed information on smoking, dietary intake, physical activity and obesity and leveraging the very large sample size to perform stratified analyses with sufficient statistical power to detect modest associations.

Aim 2A: Prospectively evaluate the association of daily MV use, as compared to non-use, with cancer incidence (for cancers with at least 250 cases across all cohorts combined).

Aim 2B: Investigate the impact of: 1) changes in MV use over time on risk of cancer using follow-up assessments, and 2) confounding by healthy lifestyle using detailed information on smoking, dietary intake, physical activity and obesity.