Machine Learning Algorithms Predict Prolonged Opioid Use in Opioid-Naïve Primary Hip Arthroscopy Patients

Area of Science:

• Orthopaedic Surgery
• Machine Learning
• Pain Management

Background:

• Excessive opioid use post-orthopaedic surgery increases patient morbidity and healthcare costs.
• Predicting prolonged opioid use is crucial for effective pain management strategies.

Purpose of the Study:

• To develop and validate machine learning algorithms for predicting prolonged opioid use in opioid-naïve patients undergoing hip arthroscopy.

Main Methods:

• A registry of 775 opioid-naïve hip arthroscopy patients was queried.
• Five machine learning algorithms were trained and tested.
• Recursive feature elimination identified key predictive variables.

Main Results:

• 18.2% of patients required opioid refills post-surgery.
• The stochastic gradient boosting model showed the best predictive performance (c-statistic: 0.75).
• Key predictors included Harris hip score, age, BMI, preoperative pain, and worker's compensation status.

Conclusions:

• Machine learning models can accurately predict prolonged opioid use after hip arthroscopy.
• Further external validation is recommended before clinical implementation.
• An open-access web application for the algorithm is available.