Intergenerational Transmission and Homotypic Continuity of Adolescent Depression: Tests of Moderation By Family Interventions and Youth Gender

Introduction: This study examined two developmental risk processes in youth, intergenerational transmission and homotypic continuity of depressive symptoms. We tested whether family-focused preventive interventions moderated these risk processes. We also examined whether risk processes differed by youth gender. Prior research findings documented the effects of the tested interventions on reduced levels of depressive symptoms. We thus expected that risk processes would be attenuated in the experimental conditions compared to the control condition of this trial. Evidence for such moderation would provide support for the value of family interventions beyond their direct effects on depression levels. Based on studies indicating that females tend to experience greater risk than males, it was expected that risk processes would be stronger for girls than boys.

Methods: Participants were families of 6^th graders in the rural Midwestern United States (97% White); 11 schools each were randomly assigned to a control condition (n = 208) or one of two family-focused intervention conditions (n = 459). Analyses used data collected from youth, their mothers, and their fathers at pretest and in the 12^th grade (n = 457 families; 69% retention), and from target youth at age 21 (n = 483; 72%). Measures of parent and adolescent depressive symptoms were obtained using the Symptom Checklist 90-Revised and Achenbach scales, respectively.

Results: Multiple group structural equation modeling was conducted using Mplus 7.11, first comparing family interventions vs. control and then comparing girls vs. boys. The multiple group family interventions model with cross-group constraints on all path coefficients fit the data well (RMSEA = .02, CFI = .93), and did not fit significantly worse than an unconstrained model (Χ² diff, p > .05). This suggested that risk processes were not moderated by family interventions In contrast, a constrained multiple group gender model fit significantly worse than an unconstrained model (Χ² diff, p < .05). Independent tests were conducted to settle on a final model (RMSEA = .03, CFI = .95), which provided some evidence for gender differences. For example, baseline maternal depressive symptoms was a significant (p < .05) predictor of late adolescent depressive symptoms for girls (β = .24), but not boys (β = -.03). Results offered some support for hypothesized gender moderation.

Conclusions: Results of the study highlight gender-specific risk processes, and have implications for the tailoring of preventive interventions to address risk factors unique to specific subgroups.