Prediction Model Can Forecast Post-Operative Outpatient Opioid Use

A team of researchers has developed a prediction model capable of forecasting post-operative outpatient opioid use following gynecological surgery, which is currently being implemented in the clinical setting.

Opioid misuse and addiction are major public health challenges, resulting in a widespread opioid crisis that has been difficult to contain. According to the Centers for Disease Control and Prevention (CDC), prescription opioids have had a significant role in driving the crisis since the 1990s, when the number of opioids prescribed to patients for pain management began to increase. Similarly, the number of overdoses and deaths from prescription opioids has also climbed during the same period.

Further, over 263,000 people in the US have died from overdoses involving prescription opioids from 1999 to 2020, and overdose deaths involving prescription opioids increased almost five-fold during those years, CDC data shows. However, despite these increases in the number of opioids prescribed and the adverse outcomes associated with them, the pain levels that US patients report has not significantly changed or increased.

Changes in opioid prescribing practices have sought to address these issues, but over-prescribing opioids following surgery is still a problem, the study authors state. Some researchers have turned to data analytics and statistical models to develop more effective, individualized approaches to pain management and opioid prescribing.

In this study, the researchers developed statistical models for predicting outpatient opioid use following gynecological surgery. The models were created using medical data from two cohorts of patients undergoing gynecologic oncology surgery: a training cohort from Feb. 1, 2018, to March 1, 2019, and a testing cohort for internal model validation from May 2019 to February 2020. Patients were eligible if they were 18 or older, spoke English, and were scheduled for open, laparoscopic, or robotic-assisted abdominal surgery.

Additional data for model development and prediction variable selection were also collected, including patient-completed preoperative surveys and weekly postoperative assessments for up to six weeks following gynecological surgery by the clinician. The preoperative survey collected information related to baseline patient characteristics, the Pain Catastrophizing Scale, preoperative opioid use, preoperative anxiety, and expectations for postoperative pain.

Overall, age, educational attainment, smoking history, anticipated pain medication use, anxiety regarding surgery, operative time, and preoperative pregabalin administration were found to be significant predictors for opioid use following hospital discharge. Since both patient cohorts in the study were relatively small, the researchers discovered that a model combining both cohorts achieved the highest performance. The average number of opioid pills prescribed to patients over the course of the study was seven, but 38 percent of participants used zero opioid pills following discharge.

These findings indicate the potential clinical utility of a model capable of providing individualized estimates of outpatient opioid use, the study authors state. The most successful model from this study is currently being implemented in clinical settings, and the researchers plan to continue this research in other surgical populations in the future.