Predicting functional remission after antipsychotic discontinuation: a real-world study in schizophrenia

BACKGROUND: This real-world study aimed to characterize patients with schizophrenia who achieve sustained good functional outcomes after antipsychotic discontinuation and to develop the Functional Remission in Schizophrenia after Antipsychotic Discontinuation (FURSAD) predictive model. METHODS: We retrospectively identified individuals aged 18-65 years with schizophrenia (ICD-10) from the Shanghai Mental Health Center discharge database. Patients who discontinued antipsychotics for ≥1 year were classified as functional remission (FR) or functional non-remission (FNR) based on functioning assessments. Sociodemographic, clinical, and treatment-related data were extracted blindly from hospital records and structured interviews. RESULTS: Among 4,166 discharged patients screened, 180 met the inclusion criteria (FR: 116; FNR: 64). Six independent predictors were identified: total disease course, Clinical Global Impression-Severity (CGI-S) score, Positive and Negative Syndrome Scale (PANSS) emotional distress subscale score, use of first-generation antipsychotics, discontinuation due to treatment benefits, and discontinuation due to lack of insight. The logistic regression model showed strong predictive performance (AUC = 0.867, 95% CI 0.813-0.921), with 82.8% sensitivity and 81.9% specificity. Internal validation was performed via 10-fold cross-validation. CONCLUSION: Discontinuation motives and illness trajectory are relevant in predicting long-term functional outcomes. A limitation is that a substantial number of patients could not be recontacted or declined participation, which may introduce selection bias. The FURSAD nomogram may help clinicians estimate the probability of FR 4.5 years post-antipsychotic discontinuation in patients previously on antipsychotics for ≥3 years.