ED FROM Charach A, Yeung E, Climans T, et al. Childhood attention-defi cit/hyperactivity disorder and future substance use disorders: comparative meta-analyses. J Am Acad Child Adolesc Psychiatry 2011;50:9–21. Correspondence to: Alice Charach, Department of Psychiatry, Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada; email@example.com Sources of funding: Ontario Mental Health Foundation, Ontario Ministry of Health and Long Term Care, McMaster University Evidence-Based Practice Center, sponsored by the US Agency for Healthcare Research and Quality, Hospital for Sick Children Department of Psychiatry Endowment Fund and the Canadian Institutes of Health Research. OM M EN TA RY The clinical and public health signifi cance of adolescent and adult substance use disorders (SUDs) suggests that identifying their early risk factors constitutes an important scientifi c priority. Crucially, early risk factors that precede the onset of SUD represent logical targets for intervention. Although childhood attention-defi cit/hyperactivity disorder (ADHD) is often correlated with SUD, the meta-analysis by Charach and colleagues provides persuasive evidence that ADHD prospectively predicts later SUD. Given that temporal ordering is necessary to differentiate correlates from risk factors,1 ADHD is a risk factor for SUD. Furthermore, because prospective longitudinal studies typically consist of fewer participants than cross-sectional studies, meta-analysis rigorously describes fi ndings from smaller studies. Clinically, this study suggests that individuals with previous ADHD must be rigorously assessed for SUD. Given that ADHD and SUD are particularly resistant to treatment, timely assessment and intervention is suggested. Prevention programmes targeting ADHD may interrupt the development of SUD or make SUD more amenable to intervention. Finally, clinicians should separately consider the positive and negative reinforcement properties of SUD in people with ADHD, given that ADHD is associated with differences in sensitivity to reward and punishment. There are several important limitations of this study, however. First, conduct problems (CPs) were conceptualised as mediators of childhood ADHD and SUD, although CP may moderate or even confound the association (ie, ADHD and SUD are spuriously related because of CP).2 Careful evaluation of CP in studies of ADHD and SUD, including formal tests of mediation, is needed to specify its precise role. Second, following evidence of signifi cant heterogeneity for SUD, tests of potential moderators, including age and sex, may have identifi ed study factors which systematically contributed to variability, but this was not examined in the study.