CMS Hierarchical Condition Category 2014 Risk Scores Are Lower for Rural Medicare Beneficiaries Than for Urban Beneficiaries


The Centers for Medicare & Medicaid Services (CMS) uses Hierarchical Condition Categories (HCC) and demographic information to calculate beneficiary risk scores, which predict expected Medicare spending by beneficiaries. CMS-HCC risk scores may be underestimating expected healthcare utilization among rural beneficiaries compared to urban beneficiaries. Incorrect estimation of expected healthcare utilization can lead to important financial losses for providers.

The study investigates potential differences in rural and urban CMS-HCC risk scores by rurality, census region, and beneficiary race or ethnicity. Previous studies across a variety of health measures suggest that rural Medicare beneficiaries are sicker than urban beneficiaries. However, this study finds that average risk scores are lower for community‐dwelling and institutional setting beneficiaries in rural counties as compared to urban counties. In addition, the more rural an area, the lower the risk score.

Furthermore, in both rural and urban counties, average risk scores are generally higher for Black and Indigenous beneficiaries as compared to beneficiaries of other races, and average risk scores are lowest for beneficiaries residing in the West census region as compared to other census regions. Among community dwelling beneficiaries, risk scores are generally lower in rural counties when stratifying by census region and beneficiary race or ethnicity.

An analysis of average utilization by healthcare setting for community‐dwelling beneficiaries in metropolitan, micropolitan, and non‐core area counties in the year prior to the calculated risk score revealed some differences in utilization count data, primarily in hospital outpatient and office‐based settings. Thus, observed differences in risk scores in rural versus urban counties may be driven in part by differences in the intensity or types of healthcare interventions received. However, this study cannot rule out a role for coding practices and resources as a potential additional driver of the observed differences.

North Carolina Rural Health Research and Policy Analysis Center
Tyler Malone, Denise Kirk, Randy Randolph, Kristin Reiter