A family intervention to reduce sexual risk behavior, substance use, and delinquency among newly homeless youth

Abstract

Purpose: We evaluate the efficacy of a short family intervention in reducing sexual risk behavior, drug use, and delinquent behaviors among homeless youth.

Methods: A randomized controlled trial of 151 families with a homeless adolescent aged 12 to 17 years. Between March 2006 and June 2009, adolescents were recruited from diverse sites in Southern California and were assessed at recruitment (baseline), and at 3, 6, and 12 months later. Families were randomly assigned to an intervention condition with five weekly home-based intervention sessions or a control condition (standard care). Main outcome measures reflect self-reported sexual risk behavior, substance use, and delinquent behaviors over the past 90 days.

Results: Sexual risk behavior (e.g., mean number of partners; p < .001), alcohol use (p = .003), hard drug use (p < .001), and delinquent behaviors (p = .001) decreased significantly more during 12 months in the intervention condition compared with the control condition. Marijuana use, however, significantly increased in the intervention condition compared with the control condition (p < .001).

Conclusions: An intervention to reengage families of homeless youth has significant benefits in reducing risk over 12 months.

Copyright © 2012 Society for Adolescent Health and Medicine. Published by Elsevier Inc. All rights reserved.

Figures

Figure 1

Estimated Regression Lines and Observed Values for Outcome Measures: Adolescents in the Intervention Condition (solid lines, black triangles) and Control Condition (broken lines, white diamonds)*† * Effect sizes are calculated from model estimated (ME) values as: ((ME control, 12-month – ME control, baseline) – (ME intervention,12-month – ME intervention, baseline)) / standard deviation for the baseline value in the control condition. † Significance testing is for the difference in slopes between the intervention and control conditions (time by treatment interaction). Analyses were conducted using SAS 9.1 software. Continuous measures were modeled with the PROC MIXED procedure. Dichotomous and Poisson-distributed count measures were modeled with the PROC GLIMMIX procedure. No significant intervention effect was found for whether the adolescent had been sexually active, had unprotected sex, number of times had sex, whether used alcohol or marijuana, whether used hard drugs.

Figure 1

Estimated Regression Lines and Observed Values for Outcome Measures: Adolescents in the Intervention Condition (solid lines, black triangles) and Control Condition (broken lines, white diamonds)*† * Effect sizes are calculated from model estimated (ME) values as: ((ME control, 12-month – ME control, baseline) – (ME intervention,12-month – ME intervention, baseline)) / standard deviation for the baseline value in the control condition. † Significance testing is for the difference in slopes between the intervention and control conditions (time by treatment interaction). Analyses were conducted using SAS 9.1 software. Continuous measures were modeled with the PROC MIXED procedure. Dichotomous and Poisson-distributed count measures were modeled with the PROC GLIMMIX procedure. No significant intervention effect was found for whether the adolescent had been sexually active, had unprotected sex, number of times had sex, whether used alcohol or marijuana, whether used hard drugs.