<![CDATA[Purpose: Adolescents are increasingly immersed in social media environments that encourage curated self-presentation and social comparison. Global prevalence estimates from the World Health Organization indicate that 5.5% of adolescents aged 15–19 meet diagnostic criteria for anxiety disorders. While prior research emphasized screen time duration, emerging evidence suggests that how young people engage— primarily through passive, emotionally reactive, or appearance-focused behaviors—may be more predictive of internalising symptoms such as anxiety and depression. This systematic review aimed to synthesize observational evidence on associations between adolescent social media use and internalising symptoms, with particular attention to engagement patterns, psychosocial mediators, and contextual moderators. Methods: A comprehensive search across PubMed, Scopus, EBSCOhost, and AI-assisted platforms was conducted between April 14 and 25, 2025. Inclusion criteria were: observational design, adolescent population (10–19 years), validated measures of anxiety/depression, and exposure to social media use. Ten studies published between 2018 and 2025 met eligibility criteria (N=10). Data were synthesized narratively, with quality assessed using the JBI checklist. Results: Across studies, problematic use and passive scrolling were more strongly associated with anxiety and depression than total time spent online. Sleep disruption and appearance-based comparison consistently emerged as mediators, while gender, emotional reactivity, and socioeconomic background moderated vulnerability. Girls and gender- diverse adolescents reported higher psychological reactivity. Protective factors included physical activity and family support. Most studies were cross-sectional; only one referenced neurobiological pathways. Conclusion: Digital mental health risks in adolescents are driven less by screen exposure time and more by emotionally charged engagement styles. Interventions should prioritize resilience-building and digital literacy, while future research must incorporate longitudinal and biopsychosocial frameworks to capture the complexity of these associations better.]]>
