Member condition assignment is a critical component in Medicare Part C and D risk scores and revenue payments, and historical patterns can reveal useful insights into where the populations may be headed in the future.

Condition-based analytics as a foundational lens for understanding the Medicare population

In the Medicare Advantage (MA) and Part D programs, the Centers for Medicare and Medicaid Services (CMS) offers plan sponsors a fixed capitation payment to provide care for their enrolled population. Since Medicare-focused plan sponsors cannot underwrite individuals to align premiums with risk, as is common in some U.S. insurance markets, CMS varies its payments based on, among other things, a beneficiary’s diagnosed health conditions. The mechanism determining this specific payment variation is risk adjustment, which CMS implements through detailed and complex models—the CMS-HCC (hierarchical condition categories) risk model for Part C and the RxHCC risk model for Part D. The models aim to align plan sponsor capitation payments with the disease burden and anticipated costs of care of their enrolled populations. As such, it is critical for each plan sponsor to understand the actual prevalence of health conditions underlying its Medicare population, both to ensure it offers appropriately tailored care management programs and to minimize inaccurate and/or incomplete medical coding within its encounter data submissions.

This publication explores disease prevalence as expressly measured through conditions identified in the Medicare CMS-HCC and RxHCC risk adjustment models—hereafter referred to as “condition prevalence.” We specifically focus on the payment year (PY) 2024 (i.e., diagnosis submissions for services incurred in 2023) and the following versions of the CMS risk adjustment models:

• 2020 CMS-HCC (V24): Risk adjustment model used to develop Part C risk scores for PY 2020 through PY 2025¹
• 2024 CMS-HCC (V28): Risk adjustment model used to develop Part C risk scores for PY 2024 through present¹
• RxHCC (V08):² Risk adjustment model used to develop Part D risk scores for PY 2023 through present³

Our goal is to provide key insights into condition prevalence rates, empowering Medicare organizations to make informed planning, budgeting, and strategic decisions. 

Notable trends and insights in Medicare condition prevalence

Overview

Medicare risk adjustment has been in a state of flux, starting with the conversion from the V05 to the V08 model in Part D, continuing with the transition to the V28 model in Part C following the previous transition from RAPS to EDS submissions, and now moving back to Part D model changes with the inclusion of maximum fair price (MFP) drugs. All of these changes affected risk score or condition assignment in recent years and have increased the importance of keeping apprised of risk adjustment transitions and trends. To help Medicare organizations in this area, we provide a variety of summarized metrics related to HCC and RxHCC assignments, referred to collectively from this point as “HCCs” for simplicity.

To start, in the table in Figure 1, we present a few high-level metrics to survey the landscape before addressing more detailed results.

Some immediate takeaways are evident.

• The transition to the V28 model materially affected the number of identified conditions and, by extension, the associated risk scores.
• Even though both the Part C and D risk adjustment models utilize medical diagnoses and reasonably align with each other, the number of assigned conditions in Part D far outpaces Part C.

We include additional commentary on these patterns later in the paper.

While the above summary reflects the HCC level, we now analyze more detailed prevalence patterns by groupings of related conditions with a separate focus on Part C (both MA and MAPD plans) and Part D (MAPD plans only) results. We do include an interactive table at the bottom of this article’s webpage, allowing users to view information not presented, including additional summarizations for the Medicare Fee-for-Service (FFS) and PDP programs.

Condition prevalence with CMS-HCC Part C risk models

We begin the more detailed review of condition prevalence with the CMS-HCC risk models. Although the diagnosis codes and corresponding HCCs underlying the V24 and the V28 models do not completely overlap, we grouped the HCCs within each into clinically similar categories to allow for a more comparable set of metrics and help keep the information more manageable.

The graph in Figure 2 directly compares prevalence at our mapped condition category level, ranked highest to lowest on V24 results.

The calculated column on the right represents the prevalence change in V28 over V24. The prevalence rates for many condition categories under the V28 model are reasonably in line with V24, varying by 5% or less. The blood, vascular, and psychiatric categories are notable standouts, highlighting where CMS significantly altered risk adjustment eligible diagnosis codes, and the updated CMS-HCC model, therefore, identifies significantly fewer HCCs. These results are not unexpected, as CMS announced that V28 includes “the removal of several HCCs in order to reduce the impact on risk scores of MA coding variation from FFS.”⁴

While the aggregate results are informative, a closer look reveals several noteworthy underlying patterns. The table in Figure 3 shows the five most prevalent condition categories from the V28 model, split by population cohort.

There is quite a high degree of overlap in the most common conditions among cohorts. In fact, three condition categories overlap completely: cardiac, lung, and diabetes.

As might be expected, condition prevalence is generally higher among the institutional population, followed by full dual eligibles. In addition, the prevalence relationships among the populations vary significantly by condition, with, for example, diabetes and lung condition prevalences for institutional members being much closer to the other populations than, say, cardiac conditions.

Lastly, dementia—first included in risk adjustment with the introduction of the V24 model—emerges as the most common condition among institutional members, yet it does not appear in the top five conditions for the other two population groups.

From Figure 3, we were able to extract some useful high-level insights, which we believe underscores the importance of understanding that the results (and, therefore, the conclusions drawn) can vary significantly by the specific cohorts analyzed. This is particularly true when slicing the data by region, various population types (i.e., individual/group, stayers/leavers/joiners, age groups), individual HCCs, coverage type (FFS, MA), plan types (FFS and PDP, MAPD, MA only, etc.), and even specific plan sponsor entities. While this level of detail is available in the data, it is beyond the scope of this particular analysis.

Condition prevalence with RxHCC Part D risk models

Historically, Part D revenue has been much lower than Part C, which has generally meant less drug-specific risk score analysis and monitoring. This dynamic, however, has changed, and the interest level in risk scores and condition patterns has increased in the wake of the Part D risk adjusted revenue shifts promulgated in the Inflation Reduction Act of 2022 (IRA).⁵

In Figure 4, we again compare prevalence at our mapped condition category level, ranked highest to lowest. When mapping individual HCCs to condition categories, we maintained alignment with the Part C categories where clinically appropriate.

The most striking pattern is the elevated prevalence of metabolic and cardiovascular conditions. In fact, these two HCCs explain nearly all of the stark contrast in overall prevalence between the Part C and D markets shown in Figure 1.

• Unlike the CMS-HCC models, the RxHCC model considers diagnosis codes for disorders of lipoid metabolism (RxHCC 47) and hypertension (RxHCC 187)—both of which contribute significantly to high metabolic and cardiac prevalence in Part D.
• Were it not for these two categories, the remaining categories, in terms of maximum prevalence and the overall pattern from low to high, would look similar to Part C (although, the prevalence rates for the lowest-ranked categories are much lower than Part C).

Keeping with the Part C section format, we next rank the top five condition categories from the 2026 RxHCC model split by population cohort.

Based on these findings, several key takeaways emerge:

• The top condition categories in Part D are very similar to Part C, and the level of full category overlap across all cohorts is actually higher than Part C.
• We observe generally higher prevalence for institutional members, followed in Part D by low-income members—with the notable exception of metabolic conditions.
• There is not as much variability in the prevalence rate differences among population cohorts, with Part D non-low income being much closer to institutional than non-duals were to institutional in Part C.

Insightful benchmarking for improved healthcare decisions

As the Medicare risk adjustment landscape continues to evolve, understanding the nuanced patterns of condition prevalence within a plan sponsor’s enrolled population is more critical than ever. This analysis highlights how risk adjustment models and beneficiary characteristics—such as institutional status, dual eligibility, and income level—can materially influence condition assignment and, consequently, revenue outcomes. By benchmarking data against broader market trends, plan sponsors can identify areas of alignment and divergence, informing tailored interventions and strategic decisions around care management, coding accuracy, and financial planning.

Appendix A: Methodology

We developed the metrics in this paper from assigned beneficiary level conditions under the 2020 CMS-HCC (V24), 2024 CMS-HCC (V28), and RxHCC (V08) models, using nationwide 2023 Medicare diagnosis data and 2024 eligibility data in the CMS research 100% identifiable files (RIFs). We then summarized the results and created the comparisons presented herein.

We assigned HCCs using the following steps.

• Filtered claims consistent with the CMS Encounter Data System (EDS) logic
• Assigned member status using CMS eligibility files
• Ran each member through the respective CMS-HCC and RxHCC model software
• Mapped the resulting conditions to groupings based on a combination of guidance provided by CMS⁷ and Milliman’s clinical resources
• Excluded the following plan types: Program of All-Inclusive Care for the Elderly (PACE) and Part B Immunosuppressive Drug (Part B-ID)
• Excluded New to Medicare members (i.e., members with fewer than 12 months of Medicare coverage in the diagnosis period)
• For CMS-HCC results (Part C only), excluded the following members: hospice members, members with end-stage renal disease (ESRD), and members with Part A only or Part B only coverage

We then assigned plan and product to each contract/plan benefit package (PBP) using published CMS information. For our numerical assessments, we summarized HCC counts by select cohorts.

Appendix B: Caveats and limitations

Milliman developed certain models to produce the values included in this analysis. The intent of the models is to estimate market wide risk scores under certain CMS-HCC and RxHCC models. It may not be appropriate for any other purpose. We reviewed the models, including the inputs, calculations, and outputs. We believe they are consistent, reasonable, appropriate to the intended purpose, and compliant with generally accepted practice and relevant actuarial standards.

The models reflect data as inputs. We relied on the following information:

• 2023 diagnosis and 2024 eligibility information in the CMS RIF data
• The CMS-HCC and RxHCC model software for the applicable benefit years
• Publicly available data and information on the Part C and D risk adjustment programs from CMS
• Other publicly available information
• Our interpretation of federal guidance

We accepted this information without audit but reviewed it for general reasonableness. Our results and conclusions may not be appropriate wherever information is not accurate.

Actual results will differ from those developed in the paper for a variety of reasons, and the following limitations should be considered when analyzing our results.

• The datasets we used represent historical data with its own mix of population types, plan selections, utilization, and acuity. This data may not reflect any one Medicare population in a given state/market, or future claim patterns or cost levels in future periods (despite using the CMS risk adjustment model from future benefit years).
• Our condition assignment will not exactly match those calculated by CMS and used in plan payments.
• The datasets used do not have the same information available to CMS, including, but not limited to:
• Imperfect mapping of member institutional status
• Potentially incomplete retroactive data adjustments
• We excluded certain populations from the analysis.
• We assigned members as institutional regardless of their Medicaid eligibility status.
• Composite results in a given cohort reflect an average across many members, and any one member’s experience will likely deviate from the average of the cohort.
• At any time, CMS could update or refine risk adjustment rules, guidance, and/or regulations such that the results presented in this analysis no longer apply.

¹ CMS.gov. 2024 model software/ICD-10 mappings. Retrieved October 15, 2025, from https://www.cms.gov/medicare/health-plans/medicareadvtgspecratestats/risk-adjustors/2024-model-software/icd-10-mappings.

² The logic identifying V08 RxHCCs (but not assigning the risk scores) did not materially change across the applicable PYs. Therefore, we expect similar RxHCC patterns whether using the 2023, 2025, or 2026 RxHCC model, and we only include results based on the proposed 2026 software.

³ CMS.gov. Other model-related documents. Retrieved October 15, 2025, from https://www.cms.gov/medicare/health-plans/medicareadvtgspecratestats/risk-adjustors-items/riskothermodel-related.

⁴ CMS.gov (February 1, 2025). Advance notice of methodological changes for calendar year (CY) 2024 for Medicare Advantage (MA) capitation rates and Part C and Part D payment policies (p. 44). Retrieved November 7, 2025, from https://www.cms.gov/files/document/2024-advance-notice-pdf.pdf.

⁵ Gergan, R., Leciejewski, Z., Koenig, D., & Pierce, K. (January 2023). Medicare Part D risk and claim cost changes with the Inflation Reduction Act. Milliman. Retrieved November 7, 2025, from https://www.milliman.com/en/insight/medicare-part-d-risk-claim-cost-changes-inflation-reduction-act.

⁶ CMS.gov (February 1, 2025). Advance notice of methodological changes for calendar year (CY) 2024 for Medicare Advantage (MA) capitation rates and Part C and Part D payment policies (p. 50). Retrieved October 15, 2025, from https://www.cms.gov/files/document/2024-advance-notice-pdf.pdf.