Assessing the Effectiveness of Interventions for Children Exposed to Violence: Findings from the National Evaluation of Safe Start Promising Approaches

Background:  OJJDP funded the Safe Start Promising Approaches initiative and its national evaluation to test promising programs for children exposed to violence. During Phase 2 of the initiative, 10 sites were funded to serve diverse groups of children of different ages, in different settings, and via different intervention models. The national evaluation examined the effectiveness of 10 evidence-informed, promising programs that attempted to reduce the harmful effects of children’s exposure to violence.

Methods: At seven sites, the design was a randomized controlled trial, with randomization occurring at the family level and eligible children recruited after families were referred to the program. At three sites, the design was a quasi-experimental design with the comparison group drawn from different programs, agencies, or shelters that served demographically comparable families but did not offer services specifically for children exposed to violence. For both groups, child outcomes (including behavior/conduct problems, social/emotional competence, depression, school behaviors/attitudes, family functioning, and violence exposure) and contextual information were assessed at baseline, 6, and 12 months. The measures used in this study were uniform across the national evaluation but prioritized within each site as to the relevance to the intervention under study. At most sites, study enrollment took place over 36 months.

Results: Across the 10 sites, over 2,800 families were enrolled in the study with about 1,900 of these families retained for the 6-month assessment, representing a 66 percent overall retention rate. For each site, we first looked at the power to detect the estimated effect size difference between the two groups being compared. After conducting power analysis, we had four sites powered to detect their estimated effect size. We then looked at differences within groups between baseline and the 6-month assessment for primary outcomes, with paired t-tests comparing individuals at 6 months to his or her own score at the baseline assessment. For five of the sites, there were differences from baseline to the 6-month follow-up for one or both of the groups, all in the expected direction. Finally, we examined intervention effects over time for primary outcomes using an intent-to-treat approach. There was no intervention effect associated with the primary outcomes for any of the sites in both unadjusted models and adjusted models controlled for demographic characteristics and violence exposure. 

Implications: The large number of families enrolled in the study demonstrated that it is feasible to recruit and engage families in a study of community-based behavioral health services across a range of settings. In general, families who participated in Safe Start improved over time. At two of the powered sites, case management and supportive group counseling provided to the control group appeared to be as effective as the intervention services. Overall, the effects of the tested programs were generally small in size. Sites found that is it challenging to form and maintain groups, lack of program staff buy-in can results in issues with engaging families in services and that substantial, culturally informed effort is needed to recruit and retain families exposed to violence for treatment and research.