School-level variation in health outcomes in adolescence: analysis of three longitudinal studies in England.

School factors are associated with many health outcomes in adolescence. However, previous studies report inconsistent findings regarding the degree of school-level variation for health outcomes, particularly for risk behaviours. This study uses data from three large longitudinal studies in England to investigate school-level variation in a range of health indicators. Participants were drawn from the Longitudinal Study of Young People in England, the Me and My School Study and the Research with East London Adolescent Community Health Survey. Outcome variables included risk behaviours (smoking, alcohol/cannabis use, sexual behaviour), behavioural difficulties and victimisation, obesity and physical activity, mental and emotional health, and educational attainment. Multi-level models were used to calculate the proportion of variance in outcomes explained at school level, expressed as intraclass correlations (ICCs) adjusted for gender, ethnicity and socio-economic status of the participants. ICCs for health outcomes ranged from nearly nil to .28 and were almost uniformly lower than for attainment (.17-.23). Most adjusted ICCs were smaller than unadjusted values, suggesting that school-level variation partly reflects differences in pupil demographics. School-level variation was highest for risk behaviours. ICCs were largely comparable across datasets, as well as across years within datasets, suggesting that school-level variation in health remains fairly constant across adolescence. School-level variation in health outcomes remains significant after adjustment for individual demographic differences between schools, confirming likely effects for school environment. Variance is highest for risk behaviours, supporting the utility of school environment interventions for these outcomes.