Modifiable determinants of adolescents’ physical activity and sedentary behaviour in school, home and mixed settings: A DE-PASS systematic review and meta-analysis of self-reported data with a holistic analysis

$\textbf{Purpose}$
Despite physical activity (PA)'s health benefits, few adolescents meet recommended levels. This review examined associations between adolescents’ self-reported PA and sedentary behaviours (SB) and their modifiable determinants, assessing interventions across settings (school, home, mixed).

$\textbf{Methods}$
Five databases were searched from 2010 to 2023. Eligible studies included experimental studies assessing adolescents’ self-reported PA/SB and modifiable determinants at pre- and post-intervention. Within a social-ecological framework, Robust-Bayesian-Meta-Analyses (RoBMA) were conducted per setting, assessing heterogeneity and publication-bias. Single-study outcomes were narratively reported. A holistic classification of determinant domains was then applied to map measured determinants and targeted intervention components across individual-psychological, individual-behavioural, interpersonal, and environmental domains. Risk of bias and evidence certainty were also evaluated. PROSPERO-ID: CRD42021282874.

$\textbf{Results}$
Twelve RCTs and two CTs (school n = 10, home n = 2, and school-home n = 2 settings) were included. ... mehrThirteen studies (93%) incorporated theory-based approaches, and 13 studies (93%) addressed multiple determinant domains, although only four (29%) showed high-integration multi-domain designs. Of the 67 unique modifiable determinants identified, 94% were classified within individual/interpersonal categories. In particular, RoBMAs showed minimal or no pooled effects on PA/SB or determinants, except for weak evidence suggesting a small positive effect on planning in school setting, including action and coping planning, implementation intentions, or commitment to a plan of action, with substantial heterogeneity across studies. Risk of bias was due mainly to randomization and outcome measurement issues. The level of certainty from the evidence ranged from very low to low.

$\textbf{Discussion}$
Existing interventions predominantly target individual-level factors, with limited coordinated integration across socio-ecological domains. High-quality future trials should test whether well-designed, integrated multilevel interventions can more consistently modify diverse determinants of adolescents' PA/SB across setting and co-occur with psychosocial benefits.