Social Media and Mental Health Among Early Adolescents in Sweden: A Longitudinal Study With 2-Year Follow-Up (KUPOL Study)

Abstract

Purpose: The aim of this study is to assess the longitudinal associations between the frequency of social media use and symptoms of mental ill-health among Swedish adolescents.

Methods: Data came from KUPOL, a Swedish school-based longitudinal cohort accrued in 101 participating schools in 8 regions of Sweden. The study sample consisted of 3,501 adolescents in grade 8 (14-15 years, 51.5%, n = 1,765 girls) followed for 2 consecutive years. Daily social media use was measured as weighted average of self-reported use in weekdays and weekend days. Mental health was measured with the Strength and Difficulties Questionnaire (SDQ). A Random-Intercept Cross-Lagged Panel Model was applied to distinguish between-person from within-person associations between social media use and symptoms of mental ill-health.

Results: Median SDQ score at baseline was 9 (interquartile range [IQR] 6-14). Median social media use was 1.7 hours at baseline (interquartile range .6-3.0) and increased over the 3-year period. Adolescents with more social media use also reported higher SDQ scores, B (95% confidence interval [CI]) = 2.40 (2.03-2.77). On a within-person level, no cross-lagged associations were found between changes in social media use and subsequent changes in symptoms of mental ill-health after 1 year, B (95% CI) = .02 (-.12 to .16) or vice versa B (95% CI) = .00 (-.02 to .02). Weak cross-sectional associations were found between changes in social media use and concurrent changes in symptoms of mental ill-health, B (95% CI) = .24 (.00-.48).

Conclusions: Adolescents with higher use of social media report more symptoms of mental health problems, but there is no evidence for a longitudinal association between increased use and mental health problems. This suggests that social media may be rather an indicator than a risk factor for symptoms of mental ill-health.

Keywords: Adolescents; KUPOL; Mental health; Random-intercept cross-lagged paned model; Social media; Sweden.