Empirical dietary inflammatory pattern and cognitive-state transitions in older US adults: a multistate cohort study

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). Conclusion 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.