Dietary Patterns Predict Mortality in a National Cohort: the National Health Interview Surveys, 1987 and 1992

We examined the association of mortality and dietary patterns using data from the National Health Interview Surveys of 1987 and 1992 (n = 10,084), aged >/=45 y at baseline (with 2287 deaths due to all causes over 5.9 median years of follow-up). The approximately 60-item FFQ administered at baseline was examined for mentions of foods and dietary behaviors recommended in current dietary guidance (fruits, vegetables, lean poultry and alternates, low-fat dairy, and whole grains), and the resulting patterns were expressed as follows: 1) a Recommended Foods and Behavior Score (RFBS), 2) factor scores from factor analysis, and 3) clusters from cluster analysis. The multivariate-adjusted relative risk (RR) of mortality for each of the 3 types of dietary patterns was examined using Cox proportional hazards regression analysis. In men, RR of all-cause mortality was 0.72 (95% CI: 0.56, 0.92, P for trend < 0.001) for RFBS, and 0.74 (95% CI: 0.57, 0.95, P for trend = 0.002) for the fruit-vegetable-whole grain factor score when comparing extreme quartiles. Membership in 1 of the 4 clusters also was associated with lower risk in men (RR = 0.82, 95% CI: 0.66, 1.01). For women, the RFBS was a modest inverse predictor of mortality after multivariate adjustment (RR = 0.80, 95% CI: 0.61, 1.04, P for trend = 0.04), but estimates for factor and cluster patterns were attenuated. The population-attributable fraction due to diet was 0.16 in men and 0.09 in women. Dietary patterns characterized by compliance with prevailing food-based dietary guidance were associated with a lower risk of all-cause mortality.