Leading predictors and their associations with combination opioid pain therapy in older adults with cancer: Application of machine learning approaches

Abstract

Combined use of opioids and other pharmacological therapies used for pain management, such as non-steroidal anti-inflammatory drugs (NSAIDs), benzodiazepines, gabapentinoids, and/or skeletal muscle relaxants (SMRs), in older adult cancer survivors can increase the risk of mortality. The objective of this study was to identify the leading predictors of opioid combination therapy with other therapies that may be inappropriate in older adults with cancer using interpretable machine learning approaches. A retrospective cohort design of older (> 66 years at diagnosis) cancer survivors (N = 2,682) diagnosed with primary and incident cancer in 2014. The Surveillance, Epidemiology, and End Results (SEER) cancer registry linked with Medicare claims database was used. Recursive feature elimination with random forest was used to extract the optimal number of features out of 119 for predictive modeling. The eXtreme Gradient Boosting (XGBoost), SHapley Additive exPlanations (SHAP), and global feature importance were used to identify the leading predictors and their associations with opioid combination therapy. Overall, 37.4% of older adults with cancer used opioid combination therapy. We included 34 features in the final predictive model.The predictive model had a good high area under the curve (AUC, 0. 758) and high recall (0.821) with the test dataset. We included 34 features in the final predictive model. Baseline opioids, NSAIDs, benzodiazepines, gabapentinoids, chemotherapy, surgery, and female sex generally positively predicted opioid combination therapy. We observed relationships of zip code percentage residents and Native American residents living below poverty with opioid combination therapy. Patient-level baseline medication use, biological factors, cancer treatment, and zip code-level poverty were leading predictors of opioid combination therapy. Our study findings contribute to the knowledge for targeted interventions to reduce the risk of opioid combination therapy.