Paper
Outcomes for youth in residential care vs. intensive in-home therapy: a propensity score analysis
- issue: Issue 3 / 2008
- authors: Richard P. Barth, Johanna Greeson, Shenyang Guo, Sarah Hurley, Jocelyn Sisson
- keywords: Usa, youth, outcomes, residential care
- views: 3909
- downloaded: 1
- download pdf (64 Kb)
abstract
Background. The study compares outcomes for behaviorally troubled children receiving intensive in-home therapy (Iiht) derived from Multisystemic Therapy and those receiving residential care (RC). Propensity score matching identified matched pairs of youth (n=684) with equivalent propensity for Iiht. The vast majority of pre-treatment differences between the Residential care (RC) for children and youth is costly and restrictive. Concerns about its effectiveness are contributing to examination of alternative methods.
Intensive in-home treatments (Iiht) have been identified as a possible alternative to hospitalization and shown some success in reducing recidivism of juvenile offenders. Few direct comparisons have been made between IIHT and RC. Of those, the majority have been studies of Multisystemic Therapy (Mst) by the founders of this approach. Evaluations of Mst by other agencies who have been trained to provide Mst are very rare.
Research questions. What are the behavioural outcomes for youth who have received residential treatment and how do they compare to matched youth who have received intensive in-home services?
Method. This study uses propensity score matching to compare youth who had the propensity to be served in residential care but were recieving services at home. Propensity score matching involves the modeling of the likelihood that a youth would receive a treatment and the selection of matched youth who had similar demographic and risk factors and did not recieve the treatment. This is a way to approximate randomization.
Measures used in the propensity score matching procedure are presented first, followed by a description of the measure used in the analysis of youth's outcomes. Two sets of variables were created for the stepwise selection of conditioning variables. The first set contained all demographic variables, intake data from psychosocial histories, service-related variables, and poverty status of the parents: race (African American or Other), gender (male/female), age (in years: 0-11, 12-15, 16+), presenting problem of delinquency (yes/no), number of delinquency types (range from 0 to 3+), presenting problem of mental health issues (yes/no), number of mental health problems (range from 0 to 3+), presenting problem of maltreatment (yes/no), beyond parental control (yes/no), other criminal behavior (yes/no), commission of a sex offense (yes/no), siblings in out-of-home care (yes/no), past mental health treatment (yes/no), past inpatient treatment (yes/no), past outpatient treatment (yes/no), receipt of special education services (yes/no), parent's receipt of financial assistance (yes/no), and youth's prior legal charges (yes/no). Basically, the groups were balanced on these indicators.
Youth were identified if they had finished RC or Iiht and their families were contacted at one year to determine their current standing regarding school progress, involvement with the law, and whether they were living at home. The outcome variable was a three-level ordinal variable with the following categories (desirable, mixed, and undesirable). The categories were derived from the agency's one-year outcome variables that included living with family, educational progress, trouble with the law, and out-of-home placement. The desirable category consisted of those cases with indication for all of the following criteria: living with family, progress in school, no trouble with the law, and no re-placements. The mixed category consisted of those cases with indications for living with family, and at least one of the following criteria: no progress in school, trouble with the law, or a re-placement. The undesirable category consisted of those cases with indication of the child not living with family. Since the outcome was an ordinal variable, an ordinal logistic regression to predict the probability of falling into each of the three outcomes was conducted.
Logistic regression was conducted on outcome differences at one-year postdischarge. Two different methods for conducting propensity score matching were compared and sensitivity analysis indicated that the results were very similar.
Findings. In both analyses, the results for Iiht tended to be superior to those for RC. Iiht recipients had a greater tendency (.615) toward living with family, making progress in school, not experiencing trouble with the law, and placement stability than did RC youth (.558; p <.10), suggesting that Iiht is at least as effective for achieving positive outcomes. Thus, youth with similar presenting problems who were served intensively at home - rather than in residential care - tended to be more likely to realize the benefits intended by the program.
Recommendations. Given Iiht's reduced restrictiveness and cost, it should be developed more broadly and routinely used in place of residential care for youth who are routinely placed into residential care. Although there are some youth who were placed into residential care who had no youth in the counterpart group of those who were served at home, many youth with very serious problems were served at home and did as well or better than the youth in RC. Contrarily, many youth without such serious problems were served in RC but they did no better there than youth served at home. Given the clear finding that RC is not more effective than IIHT but lacks the advantages related to reduced cost and reduced restrictiveness, more IIHT needs to be developed. This will require a review of finance strategies that support RC, program development to train more IIHT staff, and a commitment to make intensive services more efficient and effective.
Key references
Burns, B. J., & Hoagwood, K. (2002). Community treatment for youth: Evidence-based interventions for severe emotional and behavior disorders. New York: Oxford University Press.
Henggeler, S. W., Rodick, J. D., Borduin, C. M., Hanson, C. L., Watson, S. M., & Urey, J. R. (1986). Multisystemic treatment of juvenile offenders: Effects on adolescent behavior and family interactions. Developmental Psychology, 22, 132-141.
Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41-55.
Contacts: Richard P. Barth, University of Maryland, School of Social Work, 525 West Baltimore Street, Baltimore, MD 21210, E-mail: rbarth@ssw.umaryland.edu, Phone 410 706 7794.