A Comparison of the Charlson and Elixhauser Comorbidity Measures to Predict Inpatient Mortality After Proximal Humerus Fracture

Abstract

Objectives: Proximal humerus fractures are very common in infirm elderly patients and are associated with appreciable inpatient mortality. We sought to compare the discriminative ability of the Charlson and Elixhauser comorbidity measures for predicting inpatient mortality after proximal humerus fractures.

Methods: Data from the Nationwide Inpatient Sample (2002-2011) were obtained. We constructed 2 main multivariable logistic regression models, with inpatient mortality as the dependent variable and 1 of the 2 comorbidity scores, as well as age and sex, as independent variables. A base model that contained only age and sex was also evaluated. The predictive performance of the Charlson and Elixhauser comorbidity measures was assessed and compared using the area under the receiver operating characteristic curve (AUC) derived from these regression models.

Results: Elixhauser comorbidity adjustment provided better discrimination of inpatient mortality [AUC = 0.840, 95% confidence interval (CI), 0.828-0.853] than the Charlson model (AUC = 0.786, 95% CI, 0.771-0.801) and the base model without comorbidity adjustment (AUC = 0.722, 95% CI, 0.705-0.740). In terms of relative improvement in predictive ability, the Elixhauser score performed 46% better than the Charlson score.

Conclusions: Given that inadequate comorbidity risk adjustment can unfairly penalize hospitals and surgeons that care for a disproportionate share of infirm and sick patients, wider adoption of the Elixhauser measure for mortality prediction after proximal humerus fracture-and perhaps other musculoskeletal injuries-merits to be considered.