Total ultra-processed food intake and selected food-group exposures in relation to non-alcoholic fatty liver disease: a systematic review and meta-analysis.

BACKGROUND: Ultra-processed food (UPF) consumption has been linked to cardiometabolic diseases, but the association with non-alcoholic fatty liver disease (NAFLD) remains unclear. Moreover, whether all UPFs confer similar risk is unknown. OBJECTIVE: To systematically review and meta-analyze the association between total UPF intake (strictly defined by NOVA classification) and selected UPF-related food-group exposures [red/processed meat and sugar-sweetened beverages (SSBs)] and NAFLD risk in adults. METHODS: We searched PubMed, Web of Science, and Cochrane Library up to March 31, 2026, for observational studies reporting adjusted effect estimates for total UPF intake or selected food-group exposures and NAFLD. Study quality was assessed using the Newcastle-Ottawa Scale (NOS) or AHRQ criteria. Random-effects models pooled odds ratios (ORs) comparing highest vs. lowest intake categories. Subgroup and sensitivity analyses explored heterogeneity. RESULTS: Twenty-four studies were included in the systematic review. Of these, nine studies (5 prospective cohorts, 4 cross-sectional) comprising 250,349 participants were included for total UPF intake. Higher total UPF consumption was associated with a 26% increased NAFLD risk (pooled OR = 1.26; 95% CI: 1.16-1.37; I2 = 53%; P < 0.001). Red/processed meat consumption (6 studies) showed a stronger association (pooled OR = 1.70; 95% CI: 1.29-2.24; I2 = 76%; P < 0.001). SSB consumption (3 studies) was associated with a 50% increased risk (pooled OR = 1.50; 95% CI: 1.23-1.84; I2 = 0%; P < 0.001). Sensitivity analyses confirmed robustness. CONCLUSION: Total UPF intake is associated with a moderate increase in NAFLD risk. Red/processed meat and sugar-sweetened beverages show stronger associations, suggesting that dietary recommendations may prioritize these high-risk food groups, although blanket avoidance of all UPFs is not directly supported by the current data.