Pre- and in-treatment predictors of retention in methadone treatment using survival analysis

To determine the effects of pre- and in-treatment variables on patient retention in methadone treatment.

Retrospective longitudinal study of an admission cohort sample for up to 3 years of treatment or until discharge.

Six methadone maintenance programs operating 15 clinics in New York City.

A sample of 1206 admissions to these clinics during 1989-90.

Data were abstracted from patients' medical charts. Time in treatment for up to 3 years was the dependent variable, analyzed as a survival function. Pretreatment variables were: gender, race/ethnicity, age at admission, employment, education, marital status, living arrangements, child care responsibility, criminal justice status, life-time arrests, referral source, age at first heroin use, polydrug use, route of drug administration, mental health status and methadone treatment history. In-treatment variables were: patient problems during treatment, clinic response to patient problems, patient treatment strengths, methadone dosage, heroin use and cocaine use.

The estimated median treatment duration was 23 months. Cox proportional hazards modeling determined that two pretreatment variables (older age and no criminal justice involvement) were associated with longer retention in treatment. Among in-treatment variables, constructive clinic responses to patient problems, higher methadone dosage, more patient treatment strengths and less heroin and cocaine use during treatment were associated with longer retention.

Events during treatment are crucial for patient retention in methadone treatment. Only two of 16 pretreatment variables, compared with five of six during-treatment variables, had significant effects on length of stay in a multivariate model.