Bridging Disciplinary Boundaries (January 11 - 14, 2007)

Purpose: Most research on substance use and victimization has been unable to look concurrently at the effects of physical child abuse (CA), intimate partner violence (IPV), and sexual assault on substance use despite the high rates of comorbidity between these forms of victimization. Further, previous research has suggested that substance use may be a response to psychological distress (PD) in the form of self-medication. In this study, : 1) we examine the relative importance of CA, IPV, and adolescent sexual assault in predicting later psychological distress and substance use; and 2) we test the self-medication hypothesis by evaluating the role of psychological distress as a potential mediator through which victimization increases substance use among adolescent mothers. Method: This study uses longitudinal data from three time points (each 6 months apart) of a racially diverse sample of 234 adolescent mothers recruited from community settings. In order to address the issue of temporal sequencing of events, we take our measures of victimization from age 18, our measures of psychological distress from 18.5, and substance use from 19 years. Our outcome measure for substance use (SU) is a latent construct composed of a measure of frequency of alcohol use in the past month and a measure of illicit drug use. Lifetime IPV and CA are latent constructs measured retrospectively using the Conflict Tactics Scale. Sexual assault is a measured variable assessed as the mean number of “yes” responses to four questions regarding forced sex during the past year. Psychological distress (PD) is a latent construct measured using the depression, anxiety, interpersonal sensitivity, and hostility subscales from the Brief Symptom Inventory. We tested our hypotheses using structural equation modeling. Results: In our initial analysis, we used confirmatory factor analysis to estimate the measurement model. All factor loadings were significant at p<.01, showing that the indicators composing each latent construct are suitable for these analyses. The standardized factor loadings for each indicator range from .95 (minor parental violence) to .50 (alcohol use). Overall, the measurement model fit the data well (CFI = .97; TLI = .95, RMSEA = .057). We next estimated the structural model which indicated that IPV and sexual assault predicted PD, but contrary to our hypothesis, that PD did not mediate the relationship with substance use. Instead, both IPV and sexual assault had unique direct effects on later substance use. Parental violence was not significantly associated with either PD or substance use. Overall, the model fit the data well explaining 19% of the variance in substance use. Implications: Our findings suggest that IPV and sexual assault are unique predictors of substance use in the year following the report of violence. Contrary to the self-medication hypothesis, this relationship is direct and not mediated through the effect of victimization on increasing psychological distress. When examined in context with IPV and sexual assault, parental violence was not significantly associated with either depressive symptoms or alcohol and illicit drug use. Alternative explanations for the mechanism through which victimization increases alcohol and drug use are needed.