Abstract
Time-based diets have gained popularity for their health benefits, but their effects on human longevity remain unclear, with most evidence from short-term human trials and animal studies. We determined the associations between eating window and mortality among U.S. adults. We conducted a prospective cohort study using NHANES 2003-2018 data linked to mortality records through December 2019. The analytic sample included 33,052 adults (aged 20 and above) with two complete 24-h dietary recalls collected at baseline. Eating window was defined as the time between first and last consumption of any food/beverage containing > 0 kcal within 24 h. We used survey-weighted Cox regression with Restricted Cubic Splines (RCS) to model nonlinear associations, treating eating window as both a continuous and categorical variable (< 8.0-≥ 15.0 h/day). Models were adjusted for sociodemographic, lifestyle, health, and dietary factors. Subgroup analyses were conducted by age, sex, and race/ethnicity. Over a median follow-up of 8 years, there were 4158 all-cause, 1277 cardiovascular, and 989 cancer deaths. RCS models showed a U-shaped association between eating window and mortality, with the lowest risk at ~11-12 h/day (p = 0.004). Shorter windows (≤ 8 h) were linked to ≥ 30% higher all-cause mortality, especially in older adults, and > 50% higher cardiovascular mortality in older adults, men, and Whites. Longer eating window categories (≥ 15 h/day) were associated with 25% higher all-cause mortality (95% CI: 1.01-1.55). Moderate eating windows (~11-12 h/day) are linked to the lowest mortality risk, with deviations associated with higher risk. Differences across demographic groups highlight the need for personalized guidance.