Background
Programs to support parents are the recommended strategy to reduce disruptive child behavior problems. Efforts have been made to demonstrate which program components (i.e., clusters of techniques taught) increase program effects, but these methods fail to account for the fact that components rarely operate in isolation. We examine how combinations of components cluster together to form program types and use network meta-analysis to estimate the relative effects of these program types.
Methods
We updated an existing systematic review of parenting programs for disruptive child behavior and identified 197 randomized trials. We modeled clusters of components in each trial arm and chose the best-fitting model. We subsequently took 20 draws from the probability distribution of the latent class for each arm, entered each draw into a network meta-analysis model and combined findings using Rubin’s rules. Combined estimates were bootstrapped to rank the clusters. We estimated main models and separate models for prevention and treatment settings.
Results
A five-class solution fit the data best: (1) behavior management; (2) behavior management with parental self-management; (3) behavior management with psychoeducation and relationship enhancement; (4) maximal component loading and (5) no/minimal component loading (i.e. control). In the main model and in treatment settings, all four program types were effective compared to no/minimal components. In prevention settings, only behavior management and behavior management with parental self-management were effective compared to no/minimal components. Probabilistic ranking showed that overall and in treatment settings, behavior management had the largest chance, and in prevention settings, behavior management with self-management had the largest chance, of being most effective compared to no/minimal components.
Conclusions
Programs with more focused content seem more likely to yield stronger effects, and different foci may be needed in treatment versus prevention settings. Next steps include identifying individual family differences in optimal program content.