Causal inference framework reveals Mediterranean diet superiority and inflammatory mediation pathways in mortality prevention : a comparative analysis of nine common dietary patterns

Abstract

Background/Objectives: While some dietary indices have been developed to assess diet quality and chronic disease risk, their comparative effectiveness within the same population remains unclear due to methodological limitations in observational studies. This study employs a causal inference framework to compare nine dietary indices for reducing all-cause and cardiovascular mortality, while investigating inflammatory pathways through multiple mediation analysis. Methods: Using dietary data from 33,881 adults (aged ≥ 20 years, median follow-up 92 months), we applied a causal directed acyclic graph to identify the minimum sufficient adjustment set and implemented generalized propensity score matching to address confounding. Robust Cox proportional hazards regression assessed associations between nine dietary indices—Dietary Inflammatory Index (DII), Composite Dietary Antioxidant Index (CDAI), Healthy Eating Index 2015/2020 (HEI-2015/2020), Alternate Healthy Eating Index (AHEI), Alternate Mediterranean Diet (aMED), Mediterranean Diet Index (MEDI), and Dietary Approaches to Stop Hypertension (DASH/DASHI)—and mortality outcomes. Multiple additive regression trees (MART) algorithm was used for multiple mediation analysis to examine inflammatory markers (PAR, SII, NPR, TyG, LMR, PLR, ELR, CRP) as mechanistic mediators. Results: Among 33,881 participants (mean age 47.07 years, 51.34% women), 4,230 deaths occurred, including 827 cardiovascular deaths. Under the causal inference framework, higher DII scores increased both all-cause (HR: 1.07; 95% CI: 1.02–1.12) and cardiovascular mortality risk (HR: 1.07; 95% CI: 1.04–1.10) by 7%. The aMED demonstrated the strongest protective association, reducing all-cause mortality by 12% (HR: 0.88; 95% CI: 0.80–0.97) and cardiovascular mortality by 11% (HR: 0.89; 95% CI: 0.80–0.98), followed by MEDI with similar magnitude effects. Other healthy dietary indices showed modest 1–3% risk reductions. Multiple mediation analysis revealed that inflammatory markers, particularly neutrophil-to-platelet ratio (NPR) and systemic immune-inflammation index (SII), significantly mediated diet–mortality associations across all indices, with C-reactive protein (CRP) serving as the most frequent mediator. Conclusions: Using causal inference methodology, the Mediterranean dietary pattern (aMED) shows the strongest causal association with reduced mortality risk, with inflammatory pathways serving as key mediating mechanisms. These findings provide robust evidence for prioritizing Mediterranean dietary patterns in public health interventions and clinical practice, while highlighting inflammation as a critical therapeutic target for dietary interventions aimed at reducing mortality risk.