Self-Reported Eating Speed Is Dose-Dependently and Bidirectionally Associated with Long-Term Underweight- and Obesity-Related Weight-History Groups

Background: Body mass index (BMI) changes across the life course, yet most studies rely on single time-point measures and focus primarily on obesity. Whether self-reported eating speed is associated with long-term weight-history groups, including persistent underweight, remains unclear. Methods: This retrospective cohort study used Japanese health checkup data from 564,198 participants. According to BMI history in their twenties, participants were classified into normal-weight, underweight-history, and obesity-history groups. The underweight-history and obesity-history groups were further stratified according to the proportion of the total observation period spent in the respective weight state (0–25%, 25–50%, 50–75%, and 75–100%). Repeated BMI measurements were analyzed using linear mixed-effects models to characterize longitudinal differences across weight-history groups. Associations of self-reported eating speed with these groups were examined using sex-stratified multivariable logistic regression models. Results: BMI differed clearly across weight-history groups in both sexes. Self-reported eating speed showed a dose-dependent and bidirectional association. In women, the odds of fast eating increased progressively with obesity duration, reaching an adjusted odds ratio (OR) of 1.47 (95% CI 1.40–1.53) in the 75–100% obesity group; in men, the corresponding OR was 2.40 (95% CI 2.35–2.45). Conversely, the odds of slow eating increased with underweight duration, reaching ORs of 1.89 (95% CI 1.82–1.96) in women and 2.45 (95% CI 2.35–2.56) in men in the 75–100% underweight group. In additional analyses, self-reported eating speed also showed persistence over follow-up: the mean proportion of follow-up responses matching the baseline self-reported eating-speed category was 72.33% (95% CI 71.25–73.42) in women and 79.90% (95% CI 79.55–80.25) in men in the 75–100% obesity group for fast eating, and 72.84% (95% CI 72.00–73.68) in women and 68.24% (95% CI 67.08–69.39) in men in the 75–100% underweight group for slow eating. Conclusions: Self-reported eating speed showed a consistent bidirectional association with long-term weight-history groups in both sexes and remained relatively persistent over follow-up. These findings suggest that self-reported eating speed may be a behavioral correlate of long-term weight-history groups.