1. In this retrospective study, augmentation of the Centers for Medicare & Medicaid Services Hierarchical Condition Category cost prediction model with a validated claims-based frailty index significantly improved its accuracy at different levels of frailty.
2. While its estimates were more precise on average, the improved model was still not able to account for dramatic variation within each frailty category.
Evidence Rating Level: 2 (Good)
Study Rundown: In order to ensure that healthcare consumers are charged fairly for their costs of care, Medicare has adopted the Centers for Medicare & Medicaid Services Hierarchical Condition Category (CMS-HCC) risk-adjustment cost prediction model in a move toward value-based payment. While this model accounts for a substantial range of comorbidities, conditions such as frailty or reduced functional status can escape notice, shifting unexpected costs to providers. A recently-developed claims-based frailty index (CFI) has already been tested as a predictor of clinical outcomes, but its utility in forecasting costs is unknown. This study evaluated whether the addition of the CFI to the CMS-HCC enhanced model accuracy, finding that the CMS-HCC+CFI was able to provide a more comprehensive view of the factors affecting patients compared to the CMS-HCC alone. Some annualized Medicare costs (AMCs) that were unexplained under the standard model were projected by the CFI, and the hybrid model also better characterized patients with dual enrollment in Medicaid. These improvements come at no additional cost to clinicians as the index can be calculated directly from existing Medicare data, although in-person examinations may be more beneficial for correctly classifying patients. Of note, this model does not apply well to individual patients due to wide variation in cost within frailty categories.
Click here to read the study in Annals of Internal Medicine
Click here to read an accompanying editorial in Annals of Internal Medicine
In-Depth [retrospective cohort]: In this study, information about 16535 Medicare Current Beneficiary Survey (MCBS) participants was linked to their fee-for-service claims and administrative data from 2006 to 2013. Individuals were eligible for inclusion if they had at least 1 year of continuous enrollment in Medicare parts A and B to serve as a baseline and either remained enrolled for at least 1 additional year or died in the following year. Those receiving long-term care were excluded because frailty adjustment would not have revealed additional information beyond their institutionalization. Patients were stratified as robust, prefrail, mildly frail, or moderately to severely frail based on diagnoses (e.g. anemia or infection) and use of medical services or equipment (e.g. supplemental oxygen or wheelchairs). Further, 88% of patients were either robust or prefrail, and only 2.5% were moderately to severely frail. Multivariable regression was used to predict annualized Medicare costs (AMCs) for both the CMS-HCC and the CMS-HCC+CFI models, and data and calculated residuals were depicted in the form of kernel density plots. The CMS-HCC model overpredicted costs for robust patients while underpredicting costs for mildly and moderately to severely frail patients. Addition of the CFI, while not perfect, successfully reduced the magnitude of misprediction in all but the 4th cost quintile. For those who were dually enrolled in both Medicare and Medicaid, while the standard model underpredicted costs by an average of $632, the composite model had nearly perfect accuracy, underestimating costs by only $1.
Image: PD
©2020 2 Minute Medicine, Inc. All rights reserved. No works may be reproduced without expressed written consent from 2 Minute Medicine, Inc. Inquire about licensing here. No article should be construed as medical advice and is not intended as such by the authors or by 2 Minute Medicine, Inc.
