Abstract
The authors conducted a controlled cohort study to determine the characteristics of patients irregularly discharged from inpatient substance-abuse treatment and to derive and validate an instrument to predict such early discharge. The sample was 220 subjects (110 consecutive patients who were discharged early and 110 matched control patients) admitted to a voluntary substance abuse treatmentunit of a general hospital. Structured chart audits were used to collect demographic and clinical data. Hierarchical logistic regression identified only three predictors of irregular discharge: predominant heroin use, failure to complete high school, and age younger than 30 years. These three items were used to derive and validate an instrument for predicting completion of treatment. Readily available information on age, education, and psychoactive substance history is useful in identifying the risk of early discharge from inpatient substance abuse treatment.