Mediation in Prevention Trials Targeting Adolescent Internalizing: A Content Review and Analysis of Recent Trials

Introduction:  There is growing emphasis on evaluating mechanisms through which interventions have preventive impact as a means of identifying key targets of next-generation prevention programs. Although prevention scientists have long valued studying mechanisms, it is not clear how often this has been implemented in randomized trials designed to reduce risk for emotional disorders. In this project, we sought to assess the current state of this field through systematic review and analysis of recent prevention trials designed to reduce risk in adolescents.

Method: As part of a larger project, we collaborated with prevention trials researchers to bring together data from 19 longitudinal trials targeting adolescents. These included trials directly targeting the prevention of depression, targeting adolescents who were experiencing major life transitions including parental divorce or death of a parent, or targeting risk mechanisms for externalizing disorders that were also associated with internalizing. We used information from this dataset and from all publications based on individual trials to identify and catalogue all putative mediators. We then content-analyzed this list, and assessed the current status of evidence concerning each mediator.

Results:  We identified a total of 120 putative mediators of preventive intervention effects on adolescent internalizing. These comprised three partially overlapping content sets involving individual adolescent cognition and behavior, family processes, and interpersonal functioning in other spheres. There was very little overlap of specific measures across trials; what overlap existed was due primarily to common use of measures by the same team of investigators across different trials of the same intervention program. Many measures, particularly in the area of family process, appeared to cover similar aspects of family functioning and often used similar or identical construct labels. However, content analysis of these measures found very little overlap in specific item content. Of these 120 putative mediators, reports were available on tests of mediation for 60. These tests included analyses concerning the impact of intervention on the mediator (42 significant, 18 not significant), and full tests of mediation (11 significant, 10 not significant).

Discussion: Although prevention science has long emphasized the importance of studying mediation, practice appears to have lagged in the area of depression prevention. We suggest that three initiatives will be key to moving the field forward: (1) development of common measures with strong evidence of validity; (2) use of these measures in all trials; (3) a standard protocol for analyzing and reporting full tests of mediation across all putative mediators in a trial.