Risk Adjustment

Topic Overview

In collaboration with West Health, the Duke-Margolis Institute for Health Policy has launched a project to modernize the Medicare risk adjustment system. Specifically, and as further described in a recent technical white paper, "Modernizing Medicare Risk Adjustment and Performance Measurement," Duke-Margolis aims to use standardized clinical and pharmacy data already available in the electronic health record and currently utilized to support screening and management of health risk factors, as an alternative to using diagnoses codes for risk adjustment. This would allow Medicare payments to reflect the cost of early diagnosis and efficient management of common population health risk factors that are not currently incorporated in Medicare’s risk adjustment system, which focuses on “downstream” complications and procedures that are reliably captured in claims and encounter data. Guided by stakeholders committed to generating evidence on a rapid-cycle basis, the project seeks to enhance payment accuracy, reduce administrative burden, and promote prevention-oriented care for Medicare beneficiaries.

Background

With long-term bipartisan support, Medicare is shifting from a fee-for-service (FFS) “volume-based” approach to a “paying for health” framework, where health plans and providers are increasingly paid at the person-level with accountability for improving quality and reducing costs. Medicare’s Hierarchical Condition Category (HCC) model, used for risk-adjusting payments, redistributes billions of dollars to accountable plans and providers, such as Medicare Advantage (MA) and the Medicare Shared Savings Program (MSSP), influencing care for many beneficiaries. However, the HCC system, largely unchanged from its design over 20 years ago, relies on FFS claims data despite significant advances in electronic health records (EHRs) and data-sharing technologies like HL7 Bulk FHIR. This reliance creates inaccuracies, incentivizes intensive coding practices, and imposes administrative burdens as providers transition from FFS, contributing to potential overpayments and a “hamster wheel” of documentation not directly tied to patient care.

The Center for Medicare and Medicaid Services (CMS) has attempted to address this with the now fully phased-in “v28” HCC model, reducing certain diagnoses to curb overpayments. But v28 in turn created a new problem: many high prevalence conditions that are underdiagnosed and undertreated in the Medicare population now lead to little or no direct impact on MA and MSSP payments. These include conditions such as asymptomatic chronic kidney disease (CKD), asymptomatic hypertension, obesity, and cardiometabolic syndrome without complications. Leveraging existing EHR data—e.g., lab values like Estimated Glomerular Filtration Rate (eGFR) for CKD or Body Mass Index (BMI) for obesity—to predict health care costs and associated care delivery offers a path to modernize risk adjustment.