INTRODUCTION: Although most are aware that overprescription contributes to the opioid use epidemic, surgeons continue to overprescribe postoperatively. The objective was to expand an opioid prediction calculator validated in a gynecologic oncology population to a diverse gynecologic population. METHODS: Patients aged 18 years or older who underwent major gynecologic surgery within an academic health system from November 1, 2023, to November 1, 2024, were eligible for inclusion. Patients completed preoperative and postoperative surveys. The original model used seven predictors: age, educational attainment, smoking history, anticipated pain medication use, anxiety regarding surgery, operative time, and pregabalin administration. Model adjustments were evaluated: anticipated instead of actual operating time, pregabalin removal. Outcome variable for model evaluation was number of pills used. Primary outcome was model performance. RESULTS: Data from 616 patients who underwent gynecologic surgery were included: 137 gynecology (22.2%), 154 gynecologic oncology (25.0%), 163 minimally invasive gynecologic surgery (26.4%), 163 urogynecology (26.4%). 226 patients (37%) used zero pills after discharge, and the median number of pills used was 2 (interquartile range, 0, 8). Ordinal concordance index (95% CI) of the original model in the new cohort was 0.70 (0.65–0.74) for predicting 2 or more pills, 0.72 (0.68–0.76) for 5 or more pills, and 0.74 (0.7–0.79) for 10 or more pills. Ordinal concordance index of the adjusted preoperative model was 0.66 (0.64–0.69) for predicting 2 or more pills, 0.66 (0.64–0.69) for 5 or more pills, and 0.66 (0.64–0.69) for 10 or more pills. CONCLUSIONS/IMPLICATIONS: The original model performance maintained accuracy and generalized to a population of patients following a greater diversity of gynecologic specialists and subspecialty surgeries. The modified preoperative model also demonstrated sufficient performance. Widespread implementation of this calculator may decrease unused opioids in the community.
Lim et al. (Thu,) studied this question.