Accumulation and Time-Dependent Changes in Unfavorable Eating Behaviors and the Risk of Incident Metabolic Dysfunction-Associated and Alcohol-Related/Associated Steatotic Liver Disease: A Retrospective Cohort Study

Background/Objectives: Unfavorable eating behaviors may be associated with metabolic dysfunction-associated steatotic liver disease (MASLD) or metabolic dysfunction- and alcohol-associated liver disease (MetALD), but longitudinal evidence considering the accumulation and changes in these behaviors remains limited. We investigated the associations of skipping breakfast, eating dinner within 2 h before bedtime, and fast eating, individually and cumulatively, with incident MASLD/MetALD in Japanese adults undergoing repeated health examinations. Methods: This retrospective longitudinal cohort study included 29,889 adults without MASLD/MetALD at baseline who underwent health examinations between May 2015 and March 2026. The endpoint was defined as the development of MASLD or MetALD. An unfavorable eating behavior score ranging from 0 to 3 was calculated from the three behaviors. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Additional analyses incorporated time-dependent eating behaviors and covariates, landmark analyses, and subgroup analyses by sex and body mass index (BMI). Results: During 115,088 person-years of follow-up, 4829 participants reached the endpoint. In the mutually adjusted model, skipping breakfast, eating dinner within 2 h before bedtime, and fast eating were associated with incident combined MASLD/MetALD, with HRs of 1.19 (95% CI 1.11–1.27), 1.11 (1.05–1.18), and 1.06 (1.00–1.13), respectively. Compared with participants reporting no unfavorable eating behaviors, the fully adjusted HRs were 1.12 (1.04–1.20), 1.22 (1.13–1.33), and 1.44 (1.28–1.61) for those reporting 1, 2, and 3 behaviors, respectively. Each additional behavior was associated with an HR of 1.12 (1.08–1.15). The association of the cumulative score remained evident in time-dependent and landmark analyses. The cumulative-score association was stronger among participants with BMI < 23 kg/m2 than among those with BMI ≥ 23 kg/m2 (p for interaction < 0.001), whereas no statistically significant interaction by sex was observed. Conclusions: The accumulation of unfavorable eating behaviors was associated with a progressively higher incidence of MASLD/MetALD. Repeated assessment of simple eating behaviors during routine health examinations may help identify individuals at elevated risk, including those below conventional obesity thresholds.