Mediators of Condom Use Intention in an Effective Sexual Health Promotion Intervention for American Indian Adolescents

Introduction: Many programs for sexually inexperienced youth rely primarily on sexual/reproductive health education to increase condom use intention (CUI), an established predictor of later condom use. However, programs may need to consider mediating variables predictive of CUI that go beyond education. This paper presents a mediation analysis testing whether previously reported baseline predictors of CUI among American Indian adolescents mediate the impact of the Respecting the Circle of Life (RCL) program on CUI.

Methods: The RCL evaluation was a cluster randomized controlled trial conducted with 267 youth ages 13-19 years old. RCL consists of 8 sessions, with specific emphasis on skills-based training in condom use, taught to age- and sex-specific peer groups (13-15 or 16-19). The RCL and control programs were implemented during an 8-day summer basketball camp. Youth completed a culturally adapted Youth Health Risk Behavior Inventory, collecting data on sociodemographics, risk/protective factors, behaviors, and the seven Protection Motivation Theoretical constructs undergirding the RCL program (self-efficacy, response efficacy, response cost, intrinsic reward, extrinsic reward, severity and vulnerability). Data were collected at baseline, immediately post-camp, and at 6 and 12-months post-camp. Prior analyses showed that CUI increased baseline to post-camp, and identified baseline predictors of CUI. Of the variables predictive of CUI, two statistically significantly improved baseline to post-camp. This analysis tests whether these variables (condom use self-efficacy and response efficacy) mediate the effect of the RCL intervention on CUI, using a multiple mediator model with standardized coefficients. MLR estimation was used, a logit link specified, and models were adjusted for cluster randomization. Indirect effects were estimated using the product of coefficients method, and asymmetric confidence intervals were calculated.

Results: Results showed that study group (RCL vs. Control) was statistically significantly associated with CUI, and both mediators. Both mediators were also significantly associated with CUI. The indirect paths from study group to CUI through each mediator were statistically significant, fully mediating the direct effect of RCL on CUI. Condom use self-efficacy had the greatest effect.

Conclusion: While research has consistently shown that increases in condom use self-efficacy and response efficacy are associated with improvements in CUI and actual condom use, our findings specifically delineate the role of these factors as mediators in the path between the RCL intervention and CUI. These results indicate that new and existing youth programs must provide skills training opportunities to maximize impacts on CUI and overall efficacy.