Risk of Obesity in Patients Treated with Clozapine or Olanzapine: A Simple, Clinically Useful Prediction Tool

Clozapine and olanzapine are associated with clinically relevant weight gain and metabolic risk.

To compare the risk of obesity (BMI ⩾ 30) between clozapine and olanzapine and to develop a parsimonious prediction model with performance and clinical-utility evaluation.

We fit a multivariable logistic regression on complete cases using index drug, daily dose, prolactin (PRL), cortisol, free thyroxine (FT4), and HDL. Discrimination (AUC), probability accuracy (Brier score), calibration (intercept/slope and deciles), and decision-curve analysis (DCA) were reported.

In the classification cohort, obesity counts were equal for clozapine (9/19) and olanzapine (9/19). The model achieved strong internal discrimination (AUC 0.869) with good calibration (slope 1.00; intercept -0.00) and Brier 0.123. DCA indicated net benefit at practical thresholds (∼5-15%), supporting low-harm preventive actions.

Drug-related risk can be individualized with a lean model using routine variables. External validation and unit harmonization-especially for HDL-are recommended before broad deployment.