Food variety score and all-cause mortality in stroke survivors: Minimal mediating effect of adiposity indices (A cross-sectional study with prospective mortality follow-up using NHANES 2005–2018)

Objective To investigate the association between Food Variety Score (FVS) and all-cause mortality among US stroke survivors, and to evaluate the mediating role of novel adiposity indices. Methods This population-based cross-sectional study included 1,186 eligible adult stroke survivors from NHANES 2005–2018 (7 cycles), with prospective all-cause mortality follow-up via National Death Index (NDI) linkage. Weighted Cox proportional hazards regression, restricted cubic splines, and counterfactual-based causal mediation analyses were applied with full covariate adjustment. Results Over a median follow-up of 69.0 months, 385 deaths occurred. Higher FVS was independently associated with lower all-cause mortality in a linear dose-dependent manner. Each 1-unit increase in FVS yielded a fully adjusted HR of 0.96 (95% CI: 0.93–0.98, P < 0.001); the highest versus lowest FVS quartile showed a 49% lower risk (HR = 0.51, 95% CI: 0.35–0.75). In exploratory mediation analyses, adiposity indices exhibited minimal overall mediating effects; only the visceral fat marker ABSI Z-score showed a weak but statistically significant indirect effect, which should be interpreted cautiously given the cross-sectional measurement of exposure and mediators; BRI and WHtR were non-significant. In exploratory subgroup analyses, age was the only variable showing a statistically significant interaction with FVS, though this finding requires cautious interpretation. Conclusions Greater dietary diversity is independently associated with lower all-cause mortality in US stroke survivors, with the association largely independent of adiposity pathways. These observational findings support the potential value of dietary diversity as a simple, feasible, and low-cost nutritional strategy for post-stroke secondary prevention, but causal inference is not warranted given the study design.