Among 2,455 adults aged 60 and older in NHANES 2011–2014, each one standard-deviation increase in the Dietary Inflammatory Index (DII) — a 33-parameter nutrient-based score — was associated with a 0.091-point lower global cognitive Z-score (95% CI: −0.143 to −0.040; P=0.005). By contrast, the Food Inflammation Score for Individuals (FISI), which aggregates inflammatory potential at the whole-food level, showed no significant association (β=−0.019; P=0.341). Critically, neither score meaningfully improved discrimination for clinically low cognitive performance, with AUROC changes under 0.001.
This preprint, not yet peer-reviewed, surfaces a methodologically important tension in dietary inflammation research: two widely used scoring systems, ostensibly measuring the same construct, yield discordant results in the same dataset. The DII's advantage likely stems from its sensitivity to specific pro-inflammatory micronutrients and fatty acid ratios that whole-food aggregation can obscure. For practitioners and researchers, this is a cautionary finding — assuming DII and FISI are interchangeable could lead to conflicting conclusions across studies and misguided dietary guidance for cognitive health in aging populations.
Limitations are substantial: the cross-sectional design precludes causal inference, two 24-hour dietary recalls carry measurement error, and domain-level associations did not survive false-discovery-rate correction. Still, the study's large NHANES sample and rigorous survey weighting lend credibility. This is an incremental but clarifying contribution — less about diet-cognition mechanisms and more about measurement precision in nutritional epidemiology.
