SummaryAn analysis of CMS’s Hierarchical Condition Category (HCC) model shows that fully dual-eligible beneficiaries have the highest risk scores.
Medicare provides health coverage for seniors, those who are disabled, and those with end stage renal disease (ESRD). These individuals receive coverage either from Original Medicare—also known as Medicare Fee for Service (FFS)—or from private Medicare Advantage (MA) plans. Currently 24 million individuals, or about 35% of all Medicare beneficiaries, are enrolled in MA plans.
The Centers for Medicare and Medicaid Services (CMS) make risk-adjusted payments to MA plans for the beneficiaries they enroll so that the plans are compensated for the healthcare needs of their respective beneficiary population. The CMS’s HCC risk-adjustment model includes utilization, diagnoses, and costs for the FFS population to predict costs for beneficiaries in MA. The average beneficiary is given a risk score of 1.0, and the score is adjusted depending on the clinical and demographic characteristics of the beneficiary.
Because beneficiary costs vary significantly across the Medicare population, risk-adjusted payments reflect the relative health statuses of these different groups. As mandated by the 21st Century Cures Act, the CMS estimates risk scores separately for 6 segments of Medicare beneficiaries based on the reason for entitlement (i.e., aged or disabled) and Medicaid/dual eligibility (i.e., non-dual, partial dual, and full dual). In addition, the CMS estimates risk scores for new enrollees, those with ESRD, and institutional beneficiaries using separate models.
Variations of Risk Scores Are Most Prominent Between Aged and Disabled Full-Dual Enrollees
To understand how costs vary across different groups of Medicare enrollees, Avalere calculated the average risk score for each segment of community beneficiaries in Medicare FFS using the CMS HCC model v23. To calculate these risk scores, Avalere used diagnoses from 2018 claims for a cohort of individuals enrolled in Medicare FFS in 2019. Community beneficiaries are those enrollees who had 12 months of enrollment in 2018.
Note: These risk scores are not normalized
Dual-eligible beneficiaries (both full and partials) have higher average risk scores than non-duals. In addition, aged beneficiaries have higher average risk scores than the disabled across all dual statuses. For non-dual and partial dual beneficiaries, the risk scores for the aged are only slightly higher than those for the disabled, indicating that these populations are scored similarly and have comparable expected costs based on health status. However, aged full duals have much higher risk scores than disabled full duals. The risk scores used in this analysis were the raw risk scores—that is, these were not adjusted by the normalization factor that the CMS applies when using the risk scores in payment.
Changes in Risk Scores Can Impact Plan Payment
The CMS has broad statutory authority to develop and recalibrate the risk-adjustment models. Since the model’s introduction, the CMS has used this authority—plus additional direction from Congress—to adjust the model several times following the passage of the 21st Century Cures Act. Since risk adjustment is integral to the MA plan payment process, even small adjustments in risk scores can result in significant impacts to plan payment.
To receive Avalere updates, connect with us.
Funding for this research was provided by Humana. Avalere retained full editorial control.
Avalere ran version 23 of the CMS HCC risk-adjustment model on FFS members enrolled in 2019 and with diagnoses in 2018. The risk-adjustment software produced risk scores for each of the 6 community segments. Avalere attributed the community risk scores to enrollees based on:
- 12 months of Medicare Part A and B enrollment in 2018
- Age as of February 2019
- Dual status as of July 2019
Avalere used Medicaid dual status as of July 2019 to determine each member’s community segment because enrollment data for the full 2019 calendar year were not available at the time of analysis. Finally, Avalere estimated the average risk score for each of the 6 community models by taking the mean of the attributed risk scores across enrollees in each segment.
produces measurable results. Let's work together.