Background: Rising health care expenditures and the adoption of bundled-care initiatives require efficient resource allocation for shoulder arthroplasty. To determine a reliable and accessible metric for implementing perioperative care pathways, we compared the accuracy of the Elixhauser Comorbidity Measure (ECM) and Charlson Comorbidity Index (CCI) for predicting adverse events and postoperative discharge destination after shoulder arthroplasty.
Materials and methods: The National Inpatient Sample was queried for patients who underwent total shoulder arthroplasty or reverse total shoulder arthroplasty between 2002 and 2014. Logistic regression models were constructed with basic demographic variables and either the ECM or the CCI to predict inpatient deaths, complications, extended length of stay, and discharge disposition. The predictive discrimination of each model was evaluated using the concordance statistic (C-statistic).
Results: We identified a total of 90,491 patients. The model incorporating both basic demographic variables and the complete set of ECM comorbidity variables provided the best predictive model, with a C-statistic of 0.867 for death, 0.752 for extended length of stay, and 0.81 for nonroutine discharge. The model's discrimination for postoperative complications was good, with C-statistics ranging from 0.641 to 0.879.
Conclusion: A predictive model using the ECM outperforms models using the CCI for anticipating resource utilization following shoulder arthroplasty. Our results may assist value-based reimbursement methods to promote quality of care and reduce health care expenditures.
Keywords: National Inpatient Sample; Shoulder arthroplasty; appropriate use; comorbidity indices; predictive analytics; resource utilization; risk stratification.
Copyright © 2018 Journal of Shoulder and Elbow Surgery Board of Trustees. Published by Elsevier Inc. All rights reserved.