Prinz, R.J., Sanders, M.R., Shapiro, C.J., Whitaker, D.J., & Lutzker, J.R. (2009). Population–based prevention of child maltreatment: The U.S. Triple P System Population Trial. Prevention Science, 10(1), 1-12.
Child maltreatment is a significant public health problem; the U.S. Triple P — Positive Parenting Program — a comprehensive population-level system of parenting and family support — has been designed to deal with the problem. The program consists of five levels, involving increasing amounts of skill training and support to go with child behavior problems of increasing difficulty, and involves a broad array of service providers from different settings. A complete set of backup resources has been developed — manuals and courses for providers/practitioners and coordinated resource materials for parents. The authors cite extensive earlier evidence for the effectiveness of Triple P, based on a number of smaller-scale controlled outcome studies of program components; this current study was planned to test the extent to which the system can reduce the prevalence of child maltreatment at a population level.
The evaluation design involved stratified random assignment of 18 counties in a southeastern state to 1) Triple P, and 2) services as usual conditions, controlling for county population, poverty rate and child abuse rate. In the Triple P counties a total of 649 service providers were trained, via 69 professional training courses. Providers reported delivering Triple P to a large number of families; estimates from follow-up telephone interviews varied from 8,883 to 13,560. Program outcome indicators were three population statistics reported for each county by independent data-collection systems; these were rates per 1000 children for:
-Substantiated cases reported by child protective services.
-Out-of-home placements recorded through the foster care system
-Hospitalizations and emergency room visits due to maltreatment, recorded by medical staff.
To confirm sample equivalence, values for the Triple P and control counties were retrospectively compared for the 5-year period before the study; no significant differences were found. Pre-post difference scores for the two groups were then compared for the 24-month intervention period (t-tests, with county as the unit of analysis). For all three indicators, the Triple P counties were significantly better (p < .01 - .03); effect sizes were in the large to very large range (Cohen’s d = 1.09 – 1.22). For substantiated maltreatment, post-intervention rates/1000 children were 11.74 (Triple P) vs. 15.06 (Control). The authors argued that these results showed that a public health model like this program could achieve a preventive impact on child maltreatment at a population level.
The authors believed that this population trial was the first study of its kind to randomize communities to condition, implement evidence-based parenting interventions as a prevention strategy, and then demonstrate positive impact on population indicators of child maltreatment. A particular strength of this study was the fact that the three outcome measures were as objective as possible and largely independent of program staff; for example, hospitalization and emergency room reports of child injuries were made by medical staff, not involved in Triple P training. However, these are relatively gross measures that might underestimate the prevalence of harmful parenting practices. It is important to remember that effect sizes found at a population level like this are relevant for policy formulation, but should not be confused with effect sizes found for individual families or children in more traditional clinical trials.