• Addiction;
  • alliance;
  • comorbidity;
  • dropout;
  • drug use or abuse;
  • dual diagnosis;
  • engagement;
  • offenders;
  • resistance;
  • treatment


Aim  Treatment dropout is an important concern for professionals working in mental health. While this problem is common, the highest attrition rates have been observed in drug rehabilitation programmes. The present study focuses on the therapeutic alliance, a process variable that has been associated repeatedly with positive treatment outcome in the scientific literature. Respondent behaviour indicative of commitment or resistance to treatment was examined in combination with therapist prognoses.

Design  A total of 248 subjects, classified into three subpopulations (justice, n = 50; mental health, n = 53; comparison group, n = 145), participated in the study. Analyses aimed at predicting dropout were conducted using Cox proportional-hazards regressions. The moderating effect of sub-population was tested.

Measurements  Respondents completed a multi-dimensional measure of alliance [California Psychotherapeutic Alliance Scale (CALPAS-P)]. Therapists rated the behaviour of respondents in treatment and made prognoses about perseverance and improvement.

Findings  An increased risk of dropout was predicted when patients viewed themselves as less committed and perceived the therapist as less understanding and less involved. Therapist prognosis of perseverance was also predictive of dropout. The relationship between patient/therapist evaluations and dropout is affected differently across subpopulations by means of a moderation effect.

Conclusion  This paper demonstrates the capacity to predict dropout by measuring therapeutic alliance, therapist prognoses and therapist appraisal of patient behaviour. Moreover, the moderation effect of clinical subpopulation on treatment process variables and dropout is supported in the context of drug rehabilitation programmes.