Prediction of compliance with outpatient referral in patients with schizophrenia and psychoactive substance use disorders

Background: Most patients with concurrent schizophrenia and psychoactive substance use disorders may be adequately treated as outpatients. However, many do not comply with outpatient referrals and are therefore at heightened risk for rehospitalization. Method: Drawing on standardized interview data collected during an index hospitalization, we developed a logistic regression model to predict compliance with outpatient treatment. The model was tested on a confirmatory sample, and its sensitivity and specificity were further evaluated in a cross-validation study of 1000 random samples. Results: In a reference sample, the logistic function distinguished compliant from noncompliant patients in 37 (76%) of 49 cases. In a confirmatory sample, compliance status was predicted for 11 (78%) of 14 patients with a sensitivity of 1.00 and a specificity of 0.67. Women and patients with negative syndrome schizophrenia were compliant with outpatient referral, whereas those with mixed syndromes were most likely to be noncompliant. Cross-validation supports the stability of the model. Conclusion: While most persons with schizophrenia and concurrent substance abuse comply with integrated outpatient treatment, most who cannot may be predicted in advance.