Updated Calculator Improves Calibration of Opioid Use Predictions After Gynecologic Surgery

A prognostic study in JAMA Network Open evaluated a postoperative opioid-use calculator originally developed in gynecologic oncology and updated it for preoperative prescribing. Adults undergoing gynecologic surgery at a quaternary health system were assessed from November 2023 to November 2024, with follow-up up to six weeks; minor procedures were excluded. Of 835 patients who underwent surgery, 616 had complete opioid-use outcome data. The primary outcome was the number of opioid pills used.

Median use was 2 pills (IQR, 0-7), and 226 patients (36.7%) used none. The original model had C statistics of 0.70, 0.72, and 0.74 for predicting use of at least 2, 5, and 10 pills, respectively. Calibration indicated overprediction at the higher thresholds. The updated model incorporated anticipated pain medication use, age, anticipated operating time, education, smoking, and anxiety. Its optimism-corrected C statistics were 0.71, 0.73, and 0.73, with calibration intercepts near zero and slopes ranging from 0.93 to 1.15.

These findings suggest improved agreement between predicted and observed use, rather than improved discrimination. Individualized estimates may support opioid stewardship, but the updated model underwent only internal validation and requires testing at other institutions. Incomplete opioid-use outcomes and the single-health-system setting also warrant caution. The study evaluated prediction performance, not whether calculator-guided prescribing reduces excess prescriptions or improves pain control.