A Predictive Scoring Model for Perioperative Psychiatric Symptom Exacerbation in Schizophrenia Spectrum Disorders:...

Patients with schizophrenia spectrum disorders are vulnerable to clinically significant perioperative psychiatric deterioration, including delirium, which may require consultation-liaison psychiatry management and complicate medical and surgical care in general hospitals. Although postoperative delirium and nondelirious psychiatric worsening have been studied separately, no validated prediction tool has integrated these clinically relevant outcomes for perioperative risk assessment in patients with schizophrenia spectrum disorder.

To develop and internally validate a predictive scoring model for perioperative symptom exacerbation in patients with schizophrenia spectrum disorders.

We conducted a retrospective multicenter registry study of patients with schizophrenia spectrum disorder admitted to nonpsychiatric wards for surgical treatment at 3 general acute care hospitals in Japan between April 2017 and March 2022. Perioperative psychiatric symptom exacerbation was defined as clinically meaningful deterioration requiring specialist psychiatric management, operationalized as psychiatrist-initiated psychotropic medication changes lasting ≥4 days, psychiatric symptom-related physical restraints, or transfer to a psychiatric ward. Predictors retained in the final model were identified using multivariable logistic regression, and a simplified scoring system was internally validated using bootstrap resampling.

Among 200 patients, perioperative psychiatric symptom exacerbation occurred in 25 (12.5%). Emergency surgery, operative time ≥180 min, and admission from another hospital or facility were retained in the final multivariable model. The simplified score (0-3) showed acceptable discrimination (optimism-adjusted area under the curve 0.741). At a cutoff score ≥1, the model demonstrated high sensitivity (88.0%) and negative predictive value (95.5%), suggesting potential utility as a rule-out tool.

This preliminary scoring model showed high sensitivity and negative predictive value in internal validation and may provide a framework for future validation as a perioperative triage-support tool. External validation is required before routine clinical implementation.