Frontiers in Nutrition · Nov 16, 2026

Food-Based Diet Quality and Mortality Across FIB-4 Risk Categories in Adults with Probable MASLD

By Tom Ryu, Baron Yang, Jaejun Lee et al.

Practical food questionnaires may inform nutritional assessment in metabolic dysfunction-associated steatotic liver disease (MASLD), but their prognostic interpretation across fibrosis risk requires clarification. This secondary UK Biobank analysis included 54,278 adults with probable MASLD defined using fatty liver index and available cardiometabolic criteria. An adapted seven-component food score was examined using adjusted Cox models, within-FIB-4 contrasts, continuous FIB-4 splines, and time-varying analyses. This UK Biobank analysis included 54,278 adults with probable MASLD defined using fatty liver index and available cardiometabolic criteria. An adapted seven-component food score was examined using adjusted Cox models, continuous FIB-4 splines, and time-varying analyses. During a median 15.4 years, 7,613 deaths occurred. The time-averaged all-cause mortality hazard ratio (HR) per score point was 0.96 (95% CI 0.95–0.98); interval-specific HRs were 0.93, 0.96, and 0.97 for 0–5, >5–10, and >10 years. Within the high-FIB-4 stratum (1,170 participants; 361 deaths), the HR for scores 4–7 versus 0–2 was 0.72 (95% CI 0.57–0.92). Overall categorical and continuous-spline interaction tests yielded P = 0.098 and P = 0.059, respectively, but proportional-hazards diagnostics indicated time-dependent interactions. These findings do not establish homogeneous associations across FIB-4 risk strata or histological fibrosis stages. A higher adapted food score was associated with modestly lower mortality, with variation over follow-up. Residual confounding by socioeconomic position, physical activity, ethnicity, energy intake, and health-seeking behavior, which were incompletely captured in the analysis export, prevents causal or individualized dietary conclusions.The time-averaged all-cause mortality hazard ratio (HR) per score point was 0.96 (95% CI 0.95–0.98), attenuating to 0.98 (95% CI 0.96–0.99) with extended adjustment. Interval-specific HRs were 0.93, 0.96, and 0.97 for 0–5, >5–10, and >10 years. Overall categorical and continuous-spline interaction tests yielded P = 0.098 and P = 0.059, respectively; subgroup analyses were exploratory. Approximate E-values for the primary estimate and its confidence limit nearest the null were 1.19 and 1.14 after hazard-to-risk-ratio conversion. A higher adapted food score was associated with modestly lower mortality, but modest unmeasured confounding could explain this association. The score may also mark broader health behaviours. Residual confounding, selection into the restricted cohort, and single baseline dietary measurement preclude stratum-specific dietary conclusions.

The paperJournal sitePopulations

AgingResearch.News