Within-person associations of reallocating recreational screen time to moderate-to-vigorous physical activity with depressive symptoms among university students with disabilities: an intensive longitudinal study.

Objective: College students with disabilities are at high risk of depression, and their daily life is dominated by excessive recreational screen time. Previous studies have relied on self-reported data and focused primarily on mean associations, potentially obscuring within-person heterogeneity. This study examined whether modeled reallocation of recreational screen time to moderate-to-vigorous physical activity was associated with daily PHQ-8 scores and whether the estimated within-person association varied across conditional quantiles of the daily PHQ-8 distribution. Methods: 1,248 college students with disabilities were followed up for 12 months, combined with ecological momentary assessment. Recreational screen time and moderate-to-vigorous physical activity were objectively measured by mobile phone sensing and accelerometer, respectively. Fixed-effect isotemporal substitution model was used to analyze the average association, and fixed-effect quantile regression was used to examine the heterogeneity of the estimated association across conditional quantiles of daily PHQ-8 scores. This was an observational study; no exercise intervention was imposed. Results: Participants had an average of 317 min of daily recreational screen time and only 28 min of moderate-to-vigorous physical activity. A modeled 60-min reallocation of recreational screen time to moderate-to-vigorous physical activity was associated with a 1.65-point lower daily PHQ-8 score on average. The estimated association was weak and non-significant at the 10th conditional quantile and larger at the 90th conditional quantile. Exploratory subgroup estimates were numerically largest among students with mental disorders, high chronic pain, or high assistive technology dependence. Exploratory participant-level path analysis identified indirect statistical associations involving rumination and sense of accomplishment; because temporal ordering was unavailable, these estimates were not interpreted as mediated proportions or mechanisms. Conclusion: The estimated within-person association between modeled reallocation of recreational screen time to moderate-to-vigorous physical activity and daily PHQ-8 scores varied across the conditional outcome distribution, with larger negative coefficients at higher conditional quantiles. These observational findings are hypothesis-generating and should not be interpreted as causal effects, clinical benefits, or a basis for subgroup targeting or resource allocation.