Social Interaction, Life Satisfaction, and Follow-Up Sleep Quality in Community-Dwelling Adults: A Three-Year Longitudinal Cohort Study in Japan

Background/Objectives: Social interaction may be associated with sleep health, but longitudinal evidence among working-age adults is limited, and ceiling-concentrated scores may complicate linear modeling. We examined baseline social interaction, three-year follow-up sleep quality, and the exploratory role of life satisfaction. Methods: Data came from 487 adults aged 20–58 years surveyed in 2020 and 2023. Social interaction was assessed using the 18-item Index of Social Interaction (ISI). Normal sleep required at least 6 h of sleep and adequate restfulness. Models were adjusted for baseline sleep, demographics, lifestyle, health, depression, and anxiety. The ISI functional form was assessed using natural cubic splines and a piecewise model with a knot at 14. Results: In the initial linear summary model, a higher ISI score was associated with normal follow-up sleep (OR = 1.24, 95% CI: 1.09–1.42). The spline model fit better (likelihood-ratio χ2 = 19.69, df = 2, p < 0.001). Below 14, each 1-point increase was associated with higher odds of normal sleep (OR = 2.81, 95% CI: 1.64–4.82); no additional association was observed at 14 or higher (OR = 0.99, 95% CI: 0.83–1.19). The ISI × life satisfaction interaction was sensitive to parameterization. Spline-based analysis showed an average indirect association through life satisfaction (estimate = 0.0077, 95% bootstrap CI: 0.0019–0.0153), while direct and total associations were not significant. ISI was associated with restfulness but not duration. Conclusions: The ISI–sleep association was nonlinear, concentrated in lower scores, and plateaued near the upper end. Life-satisfaction findings were exploratory; concurrent measurement precluded causal mediation inference.