Digital Screen Time and Mental Health Difficulties Among Preschool Children in Western China: Cross-Sectional Study

Abstract Background With rising digital media use among young children, concerns have grown regarding its impact on early socioemotional development. However, evidence regarding the dose-response relationship between digital screen time and children’s mental health, as well as the potential indirect statistical association with family factors, remains limited. Objective The study aimed to explore the nonlinear association between screen time and mental health outcomes in preschool children, identify a potential threshold, and examine the indirect statistical association of parental discipline in this relationship. Methods This cross-sectional study included 21,366 parent-child dyads from Western China. Participants were recruited through stratified cluster sampling from 189 public kindergartens across 13 districts between February 28 and March 5, 2025. Caregivers of children aged 3 to 6 years completed questionnaires administered by trained kindergarten teachers, reporting children’s average daily screen time (weekdays/weekends) over the preceding month. Children’s mental health was assessed using the Strengths and Difficulties Questionnaire. Generalized additive models and piecewise logistic regression identified nonlinear patterns and thresholds. Mediation analysis quantified the indirect statistical association of parental discipline, and moderation analyses examined interaction effects. Sensitivity analyses included threshold dichotomization, exclusion of extreme values, E-value calculation, multiple imputation for missing data, and reverse-direction mediation. Results A significant nonlinear association was found, with a saturating dose-response pattern. A 2-piecewise logistic regression reparameterization yielded a policy-interpretable summary of approximately 70 minutes per day (95% CI 66‐75) for elevated clinical risk. Below this summary estimate, each additional hour of screen time was associated with 65% greater odds of total difficulties (odds ratio 1.65, 95% CI 1.46‐1.85); above the summary estimate, the association was greatly attenuated (odds ratio 1.07, 95% CI 1.01‐1.13). However, the smooth generalized additive models fit marginally better (ΔAkaike information criterion=−6.78), indicating that the association is better characterized as a graded, saturating pattern rather than a sharply defined break point. Weekday screen time showed a stronger dose-response pattern than weekend screen time. Subgroup analyses suggested tentative moderation by the number of children and household registration ( hukou , the birth-based urban vs rural classification that determines access to locally funded services such as education and health care), although these effects were not significant after false discovery rate correction. Parental discipline accounted for 20.5% of the association with total difficulties (95% CI 15.8%‐26.2%), with similar patterns for externalizing (21.2%, 95% CI 15.7%‐28.4%) and internalizing problems (23.6%, 95% CI 17.2%‐33.5%). Multiple sensitivity analyses confirmed robustness of the findings. Conclusions Daily digital screen time shows a nonlinear association with preschool children’s mental health outcomes. Reducing weekday screen exposure and optimizing parental discipline represent promising intervention targets, based on cross-sectionally derived correlational mediation patterns. These findings offer locally relevant, evidence-based benchmarks for screen-time guidance and highlight modifiable parenting factors for early childhood digital health interventions.