Predictors of completion of an adolescent residential alcohol and other drug withdrawal program

Aims: This study extends the limited research on predictors of completion of adolescent residential drug and alcohol program. Design and Methods: A Queensland, Australia, five-bed state-wide voluntary adolescent drug and alcohol withdrawal service collected routine clinical pre-entry data over a period of 9 years (2000–2009) on 524 subjects. Potential predictors of program completion included socio-demographic, family functioning, past mental health history, mental health problems on admission, and patterns of drug use variables. The chi-square analysis identified significant predictors with subsequent multivariate logistic regression modeling. Given the loss of data, additional multiple imputation logistic regression analysis was utilized. Results: Initial predictors were aged 17 and over; a past history of psychiatric treatment or violence or assault; mental health problems identified on entry; absence of current alcohol use or ecstasy use. After regression and multiple imputation modeling, predictors were a history of psychiatric treatment and of violence/assault. Discussion and Conclusions: A history of co-morbidity rather than traditional markers of good outcomes such as sex, ethnic background, income support, and housing support predicted program retention. The study reinforces the important role of non-substance psychiatry co-morbidity in the treatment of substance use disorder