Model Developed to Predict Postoperative Opioid Use After Gynecologic Surgery

Researchers developed a model to predict outpatient opioid use after gynecologic surgical procedures. They described their model and its application in a recent report in the journal JAMA Network Open.

“Approximately 6% to 8% of opioid-naive patients develop chronic use following surgery, and the probability increases with duration of opioid therapy,” the researchers explained in their report. Therefore, they sought to develop a model to aid clinicians in predicting opioid use following gynecologic procedures.

In development of this prognostic model, the researchers used a variety of statistical methods with data from a set of prospectively enrolled patients who were undergoing gynecologic surgical procedures at a single center between February 2018 and March 2019. Indications for surgical procedures were benign or malignant. Data related to these patients were used for the training cohort in model development (cohort 1), and data from a separate cohort that was enrolled during May 2019 through February 2020 were used as a testing cohort to validate the model (cohort 2).

The final model was refined based on data from both cohorts, and the researchers used this model to develop an online calculator. The number of opioid pills used in the postoperative period was the model’s response variable, with analyses based on thresholds of 5 or more pills, 10 or more pills, and 15 or more pills. The researchers used metrics of ordinal concordance and the Brier score to evaluate the model’s performance.

A total of 382 female adults who were undergoing either minimally invasive or open surgical procedures were included in development of the model. Cohort 1 included 216 patients and cohort 2 included 166.

Opioid medication had been prescribed to 98% of patients in cohort 1 and 92% in cohort 2 upon hospital discharge. After discharge, 38% of patients overall did not use any pills. The median number of pills used was 3 (IQR, 0-10), and the mean number of pills used was 7 (SD, 10).

Predictors included in the final model were age, educational attainment, smoking history, preoperative anticipated pain medication need, preoperative anxiety regarding surgery, total operating time, and preoperative administration of pregabalin.

The final model had an ordinal concordance value of 0.65 (95% CI, 0.62-0.68) in analyses involving predictions of 5 or more pills used, 10 or more pills used, and 15 or more pills used. Brier scores were 0.22, 0.18, and 0.14, respectively.

“The proposed patient-centered model allows for prediction of postoperative opioid use with easily obtained patient-level variables and can be used following a broad range of gynecological procedures,” the researchers concluded in their report.

Disclosures: Some authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of disclosures.

Reference

Rodriquez IV, McKeithan Cisa P, Monuszko K, et al. Development and validation of a model for opioid prescribing following gynecological surgery. JAMA Netw Open. 2022;5(7):e2222973. doi:10.1001/jamanetworkopen.2022.22973