Objective: We examined the Working Alliance Inventory (WAI) and Interpersonal Trust in Physician (ITP) in a forensic psychiatry hospital, where all patients were detained under mental health legislation for psychosis. We hypothesised that working alliance and trust are bilateral and can be measured.

Method: We adapted the WAI and ITP minimally so that patients rated both their treating psychiatrist and primary nurse. We also adapted them minimally so that clinicians could rate WAI and ITP with the patient. A total of 81 of 83 patients completed the assessments. The clinicians (seven consultant psychiatrists and 43 nurses) also completed a minimally altered version of the same questionnaires. All three (patient, nurse and psychiatrist) were blind to the ratings of the others.

Results: Cronbach's alpha was greater than 0.9 for both patient and clinician versions of the WAI and greater than 0.8 for the ITP. The WAI and ITP correlated with each other (Spearman r > 0.67 for patients, for psychiatrists and for nurses). Patients rated clinicians higher than clinicians rated patients. Ratings were higher in pre-discharge wards than in acute wards. Patients' ratings of WAI for their psychiatrist and nurse correlated r = 0.75, and patients rating of IPT for psychiatrist and nurse correlated 0.67. Psychiatrists correlated with nurses 0.38 for WAI, 0.53 for IPT. Psychiatrists and patients mutual ratings correlated r = 0.35 for WAI, 0.24 for IPT. Nurses and patients correlated r = 0.34 for WAI, 0.25 for IPT. All correlations were statistically significant. Mental state (PANSS) and global function (GAF) correlated with all ratings and confounded most patient-clinician correlations.

Conclusion: Working alliance and interpersonal trust can be measured reliably even in forensic settings. The extent to which they measure a mutual quality is unclear.