The therapeutic alliance in early psychosis

Abstract

Research with nonpsychotic patients has indicated that the therapeutic alliance makes an important contribution to outcome of psychological treatments. Although recent years have seen a resurgence of enthusiasm for psychotherapeutic interventions for psychotic patients, particularly following the publication of large‐scale RCTs of cognitive‐behaviour therapy (CBT) for schizophrenia patients, there have been few attempts to assess the contribution that the patient–therapist relationship makes to these interventions, which have been usually described in terms of particular techniques. However, in a study of psychodynamic therapy for chronic schizophrenia patients, Frank and Gunderson (1990) reported that patients were slow to form an adequate alliance, although the quality of the alliance was a predictor of long‐term outcome.In this paper, we will review evidence from previous studies, and present data on the therapeutic alliance and other process variables taken during a large randomized controlled trial of CBT for first and second episode schizophrenia patients (the SoCRATES study). A total of 315 patients entered the study, and were randomized to treatment as usual, supportive counselling or CBT, with a 5‐week therapy envelope commencing within 2 weeks of admission to hospital. Baseline and follow‐up assessments were conducted by psychiatrists blind to group assignment, and 75% of the patients were retained in the study at 18 month follow‐up. We have reported the main outcome data elsewhere; statistically significant but modest benefits were found for both CBT and counselling. Therapeutic alliance measures were completed by both patients and therapists in thetwo psychological treatment groups at the third and 10th session. All measures had acceptable reliability but therapist assessments of the alliance correlated poorly with patients' perceptions. Both patient and therapist alliance ratings at session 3 predicted duration of treatment. Duration of treatment, in turn, predicted therapy response. The findings indicate the importance of investigating process variables as mediators of response to treatments for early psychosis.