Riehm · JAMA psychiatry 2019 · prospective cohort study · n=6595

Associations Between Time Spent Using Social Media and Internalizing and Externalizing Problems Among US Youth.

Level 3 - non-randomized controlled study

Prospective longitudinal cohort study

PubMed 31509167 · doi:10.1001/jamapsychiatry.2019.2325 · record verified 2026-08-26

What was done

This longitudinal cohort study analyzed 6,595 US adolescents (aged 12–15 years at Wave 1; 51.3% male) from Waves 1 to 3 (2013–2016) of the nationally representative Population Assessment of Tobacco and Health study. Exposure was self-reported daily time on social media at Wave 2 (none, ≤30 minutes, >30 minutes to ≤3 hours, >3 to ≤6 hours, >6 hours). Outcomes were past-year internalizing problems alone, externalizing problems alone, and comorbid internalizing and externalizing problems assessed at Wave 3 using the Global Appraisal of Individual Needs-Short Screener.

What was found

In adjusted analyses, daily social media use exceeding 3 hours compared to non-use was significantly associated with higher risk of internalizing problems alone (>3 to ≤6 hours: relative risk ratio [RRR], 1.60; 95% CI, 1.11–2.31; >6 hours: RRR, 1.78; 95% CI, 1.15–2.77) and comorbid internalizing and externalizing problems (>3 to ≤6 hours: RRR, 2.01; 95% CI, 1.51–2.66; >6 hours: RRR, 2.44; 95% CI, 1.73–3.43). Daily use of 3 hours or less was not significantly associated with elevated risk in adjusted models, and social media use was not significantly associated with externalizing problems alone after adjustment.

Why it matters

This study provides prospective, population-level evidence that heavy social media use (>3 hours per day) precedes elevated risk for internalizing psychiatric symptoms among adolescents, identifying a potential threshold for behavioral screening and intervention.

Limits

Both social media exposure and psychiatric symptoms were self-reported via screening questionnaires rather than objective screen-time tracking or formal clinical diagnostic interviews. Residual confounding cannot be fully excluded.