A decision analytic Markov model was used to combine published data and assess costs and outcomes associated with the two interventions. The time horizon of the model was 25 years. A public sector and a societal perspective were adopted in the economic analysis.
Extensive literature searches were performed to obtain clinical and effectiveness data using National Health Services Economic Evaluation Database, Cochrane Database of Systematic Reviews, PsycINFO and MEDLINE databases and Google Scholar. Google searches were used to identify grey literature and unpublished studies. The main estimate of effectiveness was the impact of parenting interventions on childhood disorders that persist into adulthood. The effectiveness of the intervention was derived using data from studies included in a systematic review of randomised controlled trials (see Other Publications of Related Interest).
Monetary benefit and utility valuations:
Measure of benefit:
The probability that a child with conduct disorder at age five continued to show conduct problems at later ages.
Public sector costs included in the analysis were provision of the intervention, National Health Service, Social Services Department, Department for Education and the Criminal Justice Service. From a wider societal perspective the costs included voluntary sector, lost output due to crime, victim costs due to crime and other costs of crime (such as anticipation of crime, property damage and victim services). Costs of the parenting programme were derived from a commissioning toolkit that included staff, overheads, materials, catering, childcare and training and supervision. Costs to the public sector were derived from previously published studies. Costs of crime were derived from national reports, published studies and governmental data. Costs were presented in 2008/2008 UK pounds (£) and discounted using an annual rate of 3.5%.
Analysis of uncertainty:
A series of scenario analyses were performed. A best case showed the maximum potential cost saving from improved outcomes due to the intervention given low drop-out and recidivism and high rates of intervention effectiveness. The worst case tested whether there would be any cost savings given worst assumptions in drop-out and recidivism rates and low effectiveness of the interventions. Results were presented in a table.