Using chi-Squared Automatic Interaction Detection (CHAID) modelling to identify groups of methadone treatment clients experiencing significantly poorer treatment outcomes

Abstract

In times of scarce resources it is important for services to make evidence based decisions when identifying clients with poor outcomes. chi-Squared Automatic Interaction Detection (CHAID) modelling was used to identify characteristics of clients experiencing statistically significant poor outcomes. A national, longitudinal study recruited and interviewed, using the Maudsley Addiction Profile (MAP), 215 clients starting methadone treatment and 78% were interviewed one year later. Four CHAID analyses were conducted to model the interactions between the primary outcome variable, used heroin in the last 90days prior to one year interview and variables on drug use, treatment history, social functioning and demographics. Results revealed that regardless of these other variables, males over 22years of age consistently demonstrated significantly poorer outcomes than all other clients. CHAID models can be easily applied by service providers to provide ongoing evidence on clients exhibiting poor outcomes and requiring priority within services.