Frontiers in Nutrition · Oct 31, 2026
Empirical dietary inflammatory pattern and cognitive-state transitions in older US adults: a multistate cohort study
By Zhipeng Liu, Min Lei, Huiwen You et al.
Background: Dietary inflammatory potential may be associated with cognitive aging, but evidence on cognitive-state transitions is limited. We examined the association of the empirical dietary inflammatory pattern (EDIP) with transitions from normal cognition to survey-classified cognitive impairment. Methods: We linked 2013 Health Care and Nutrition Study dietary data to 2014-2022 Health and Retirement Study cognitive classifications and mortality records for 4,997 adults aged 60 years or older without dementia at baseline. A piecewise continuous-time Markov model represented normal cognition, cognitive impairment, and death. The primary model assigned EDIP only to the normal-to-impairment transition; a key sensitivity analysis allowed EDIP associations with all four transitions. Results: Across 21,197 state observations, 1,615 observed normal-to-impairment transitions occurred. The primary hazard ratio (HR) per 1-SD higher EDIP was 1.074 (95% confidence interval [CI], 1.014-1.138; P = 0.014). With EDIP assigned to all transitions, the corresponding HR was 1.029 (95% CI, 0.963-1.099; P = 0.401). Under the primary model, four-year standardized cognitive-impairment probabilities among 4,068 participants with normal cognition at baseline were 17.0% and 18.9% at EDIP values of -1 and +1 SD, respectively, a difference of 1.87 percentage points (95% CI, 0.48-3.42). Conclusions: The modest positive association in the restricted primary model was not robust to allowing EDIP associations with all transitions. Single dietary assessment, possible reverse causation, and survey-based outcome classification further limit interpretation. These findings do not establish a causal dietary effect, an inflammatory mechanism, or a clinically actionable benefit.