Medical Student Stanford University School of Medicine
Introduction: The five-item modified frailty index (mFI-5) and Charleson Comorbidity Index (CCI) have been previously used to predict healthcare outcomes such as complications and mortality in patients. However, as a result of the homogenous patient populations traditionally utilized to develop comorbidity indices, there is a potential for index underperformance in minority patient populations.
Methods: In this study, we attempted to evaluate and quantify this potential disparity in both mFI-5 and CCI performance when stratified by race. We collected retrospective insurance claim data on all patients that underwent spinal fusion procedures in the IBM Marketscan and Medicare Supplemental Databases. Index values were acquired 6 months prior to the first operation. Subjects without 6 months of continuous pre- and post-operative enrollment were excluded in this study. Patient populations with high and low proportions of non-Hispanic black people were identified through neighborhood distributions in the National Data Archive. Statistical analysis included Delongs tests for two roc curves based on race to evaluate the performance of the indices in the targeted patient population. The outcome of interest for all patients was prolonged hospital stay.
Results: In total, 185,039 patients were included in this study. When evaluating the CCI, the AUC for low and high proportions of black people were 0.5761 and 0.5575 respectively (p=0.0002). For the mFI-5, the AUC for low and high proportions of black people were 0.5657 and 0.5528 respectively (p=0.0114).
Conclusion : There is an apparent gap in the mFI-5 and CCI performance in different patient populations, particularly performing worse in communities with higher proportions of black patients. This gap can result in differences when it comes to patient management and treatment of minority patients, and further result in the propagation of healthcare disparities. These results support the continued initiative to modify new and existing indices to more accurately predict healthcare outcomes in an equitable manner.
