HIV risk reduction among detained adolescents: a randomized, controlled trial

Abstract

Objectives: Criminally involved adolescents engage in high levels of alcohol-related risky sex. A theory-based sexual and alcohol risk-reduction intervention was designed, implemented, and evaluated in juvenile detention facilities.

Participants and methods: In a randomized, controlled trial, 484 detained adolescents received 1 of 3 group-based interventions: combined sexual and alcohol risk reduction (group psychosocial intervention [GPI] + group motivational enhancement therapy [GMET]); sexual risk reduction only (GPI); or HIV/sexually transmitted disease prevention information only (group information-only intervention [GINFO]). Follow-up data were obtained 3, 6, 9, and 12 months after the intervention. Behavioral outcomes were condom-use behavior, frequency of intercourse while drinking, and alcohol-related problems.

Results: Condom-use behavior measured as frequency of condom use during sex (ranging from never to always) decreased over time, although the GPI and GPI + GMET interventions mitigated this tendency at the 3-, 6-, and 9-month follow-up assessments. Although both active interventions were significantly more successful than the GINFO condition and the pattern of effects favored the GPI + GMET, there were no statistically significant differences between the GPI and GPI + GMET interventions.

Conclusions: Findings support the feasibility of integrating alcohol-specific sexual risk content into a theory-based sexual risk-reduction intervention and provide additional evidence that theory-based interventions are effective at reducing risky sex in this population. There was limited evidence of intervention effects on alcohol-use outcomes. Future research should focus on strengthening the GPI + GMET to most effectively target risky sexual behavior among at-risk adolescents.

Trial registration: ClinicalTrials.gov NCT00914719.

Figures

FIGURE 1

Diagram of participant flow through the waves of the study according to condition.

FIGURE 2

Observed and model estimated means for condom use according to condition. Note that the original model estimated demonstrated marginal fit to the data ( χ262[n=484]=62.92; P < .05; root-mean-square error of approximation [RMSEA]: 0.07 [95% confidence interval (CI): 0.04 – 0.10]; comparative fit index [CFI]: 0.87). Using an exploratory approach, we estimated improvements to model fit by allowing correlations among residual error variances. The final estimated model demonstrated good fit to the data ( χ322[n=484]=45.12, NS; RMSEA: 0.05 [95% CI: 0.00–0.09]; CFI: 0.94).

FIGURE 3

Observed and model estimated means for frequency of intercourse while drinking according to condition. Note that the original model estimated demonstrated marginal fit to the data ( χ362[n=484]=57.97; P < .05; RMSEA: 0.06 [95% CI: 0.03–0.09]; CFI: 0.75). Following the same approach of allowing correlations among residual error variances, the final estimated model demonstrated good fit to the data ( χ322[n=484]=37.13, NS; RMSEA: 0.03 [95% CI: 0.00–0.07]; CFI: 0.95).

FIGURE 4

RAPI means at baseline and at the 12-month follow-up according to condition.