Dietary intake and peripheral arterial disease incidence in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study.

Peripheral arterial disease (PAD) is a costly source of morbidity and mortality among older persons in the United States. Dietary intake plays a role in the development of atherosclerotic cardiovascular disease; however, few studies have examined the relation of food intake or dietary patterns with PAD. We examined the relation between habitual dietary intake at midlife and incident PAD over ∼20 y of follow-up. Among 14,082 participants enrolled in the ARIC (Atherosclerosis Risk in Communities) Study initially free of PAD, dietary intake was assessed at baseline in 1987-1989 by using a modified Harvard food-frequency questionnaire. Food groups were created, and principal components analysis was used to develop "healthy" and "Western" dietary patterns; both were categorized into quintiles or quartiles. Incident PAD was determined by an ankle-brachial index <0.9 assessed at 2 subsequent examinations and hospital discharge codes through 2012. Multivariate-adjusted Cox proportional hazards regression was used. During a mean follow-up of 19.9 y, 1569 participants developed incident PAD. In models adjusted for demographic characteristics, behaviors, and food groups, the HRs (95% CIs) for incident PAD increased across quintiles of meat consumption [quintile 1: reference, quintile 2: 1.38 (1.16, 1.65), quintile 3: 1.38 (1.16, 1.65), quintile 4: 1.45 (1.20, 1.74), quintile 5: 1.66 (1.36, 2.03); -trend <0.001]. Compared with those who drank no alcohol, those who had 1-6 drinks/wk had a lower risk of incident PAD [0.78 (0.68, 0.89)]. For coffee, ≥4 cups/d compared with none was inversely associated with incident PAD [quintile 5 compared with quintile 1: 0.84 (0.75, 1.00); -trend = 0.014]. There was no association between other food groups or patterns and incident PAD. In this prospective cohort study, greater meat consumption was associated with a higher risk, and moderate alcohol consumption was associated with a lower risk of incident PAD. Whether these associations are causal remains to be seen. This trial was registered at clinicaltrials.gov as NCT00005131.