Developmental cascades in studies of adolescent and young adult substance use etiology: A systematic review.

Introduction: Frequently, developmental cascade models are used to examine causal linkages between early family risk and substance use etiology. When framed with longitudinal data, cascade models contribute to understanding developmental etiology by parsing stability from change in multiple domains of influence. This systematic review examines the research methods used in cascade studies of substance use etiology. Method: A systematic literature review involved four electronic literature databases (i.e., PsycINFO, MEDLINE, EMBASE, Web of Science). Specific terms referenced substance use etiology and developmental cascade effects. Inclusion requirements included cross-domain effects and repeated measures. Studies were eliminated based on including interventions or growth modeling that failed to differentiate time-specific effects. A risk assessment indicated adequate inter-rater reliability for the 18 studies included. Results: Conceptually, there was little evidence supporting hypothesized cascade effects that involved cross-domain risk mechanisms linking early parental socialization with later substance use. Methodologically, studies were characterized by modest sample sizes, lack of power, and relatively small effect sizes (ESavg. = 0.05 [SD = 0.046], range 0.003 - 0.19). Only half of the studies conducted formal statistical tests of indirect effects linking early socialization with later substance use. Conclusion: This review highlights there is very little evidence for developmental cascade effects involving early parental socialization and substance use etiology. Methodological and conceptual limitations may hamper detection of developmental cascade effects and further undermine our understanding of substance use etiology. Future studies may want to follow larger samples, over extended time frames and specify intermediate mechanism that contribute to vulnerability.