Renuka Chintapalli, BA, MA, MB BChir: No financial relationships to disclose
Introduction: Operative management of spinal metastatic disease is largely for symptom palliation rather than curative. The decision to operate revolves heavily around the expectation that postoperative survival will exceed the period required for postoperative recovery. Long-term postoperative opioid use is a clinically useful indicator of recovery, yet few studies have attempted to apply machine learning models to predict this outcome in spinal metastatic disease patients.
Methods: The IBM Watson Health MarketScan Database and Medicare Supplement were analyzed to identify adult patients who underwent surgery for extradural spinal metastatic disease between 2006 and 2023. Patients were required to have at least 6 months of continuous preadmission baseline data, and 6 months of continuous post-discharge follow-up. The primary outcome was prolonged opioid use, defined as filling a perioperative prescription followed by another between 90- and 180-days post-discharge. Five models (stochastic gradient boosting, support vector machine, neural network, random forest and penalized logistic regression) were trained on a 70% training sample and validated on the withheld 30%.
Results: A total of 732 patients were included, of which 341 (46.6%) had prolonged post-discharge opioid use. Most patients had disease in the thoracic region (57.2%), were operated on via a posterior approach (57.7%), had arthrodesis (55.7%) and were discharged home (75.7%). Prior opioid use (vs no prior opioid use, odds ratio [OR]:4.12, 95% CI:1.5-11.34), sacral localization (vs cervical, OR:3.87, 95% CI:1.2-12.41) and thoracic localization (vs cervical, OR:1.94, 95% CI:1.07, 3.53) were associated with greater odds of prolonged opioid use. The random forest algorithm had the best predictive performance in terms of discrimination (area under the curve [AUC]: 0.611), calibration (intercept: 0.18, slope: 0.613) and overall accuracy (Brier score: 0.24). The final algorithms were incorporated into an open access web application to provide predictions and patient-specific explanations of the results generated by the models.
Conclusion : We developed and validated parsimonious predictive models to estimate risk of prolonged opioid use after surgery for extradural spinal metastatic disease. Integrating these models into physician- and patient-facing interfaces may improve prognostication, enhance clinical decision-making, and ultimately optimize pain management to support more tailored postoperative care strategies.