Recently, we took a first look at the Medicare Advantage 2024 Advance Notice, released by the Centers for Medicare and Medicaid Services (CMS). These proposed changes, which are slated for implementation in 2024, will have a significant impact on risk adjustment factor (RAF) scores for Medicare Advantage beneficiaries. Before we jump into an analysis of that impact, let’s review RAF scores.
What is a RAF score?
Risk Adjustment Factor (RAF) scores are part of the model used by CMS to estimate the associated cost of Medicare Advantage beneficiaries. The RAF score determines the amount paid by CMS to the health plan per beneficiary during the corresponding payment year. Medicare Advantage Organizations (MAOs) are paid at a higher rate for patients that have multiple conditions and conditions with greater levels of severity, as their anticipated costs of care will be higher.
How RAF scores are calculated
The RAF score is based on both the demographics and the disease risk scores for the beneficiary. The demographic score includes age, sex, residence (whether in the community, skilled nursing facility, or other institution), and disability status. The disease risk score is based on the reported diagnoses from patient encounters and their corresponding Hierarchical Condition Category (HCC) codes.
Generally, a higher RAF score indicates sicker patients and lower RAF scores indicate healthier patients. However, a low RAF score could also be an indicator of inaccurate coding due to lack of information on the patient record or a gap in care.
How RAF scores may be impacted by CMS-HCC Version 28
CMS made significant changes to the structure of the HCC model in Version 28 (V28), which will impact RAF scores for a large percentage of Medicare Advantage beneficiaries. These changes include:
- How the V28 HCC codes are named and numbered
- An expanded number of HCCs
- Changes to ICD-10-CM code to HCC mappings
- Changes to the HCC coefficient values
- Removal of 2,294 diagnosis codes that no longer map to a payment HCC
- Addition of 268 diagnosis codes that did not map to a payment CMS-HCC in V24
Reflecting fee-for-service updates
CMS used updated fee-for-service data years (including 2018 diagnoses and 2019 expenditures) to calibrate the V28 model. They stated the proposed model “results in more appropriate relative weights for the HCCs in the model because they reflect more recent utilization, coding and expenditure patterns in FFS Medicare.” CMS decided not to include HCCs (and diagnoses) in the V28 model if:
- The conditions did not accurately predict costs
- Coefficients were small
- The conditions they represent are uncommon
- Conditions that did not have “well-specified” diagnostic coding criteria
Fewer ICD-10-CM to HCC mappings
CMS has released a file titled “PY 2024 Proposed Clinical Revision Part C Model ICD-10-CM Mappings”, which provides detailed insight at the level of specific ICD-10-CM codes and HCC codes. It includes the proposed ICD-10-CM mappings to V28 HCC codes as well as current ICD-10-CM mappings to V24 HCC codes. Based on this file 2,294 ICD-10-CM codes that mapped to a payment HCC in V24 no longer map to a payment HCC in V28. Selected groups of related conditions impacted by the proposed changes are shown in our full analysis.
A relatively small number of diagnosis codes that did not previously map to a payment HCC will map to a payment HCC in V28 as proposed. Notable examples are available in tables included in our full analysis. Of the 268 “new” ICD-10-CM diagnosis codes that map to a CMS-HCC in V28, over 40% represent conditions not encountered in the majority of patients in the Medicare Advantage population.
Constraining will likely lower RAF scores
CMS acknowledged that the changes in the proposed rule could change beneficiary risk scores with or without a change in the patient’s health status. They stated the proposed model “results in more appropriate relative weights for the HCCs in the model because they reflect more recent utilization, coding and expenditure patterns in FFS Medicare.”
CMS used a process referred to as constraining, where related HCCs are given the same coefficients. A significant example of constraining in the V28 model affects the Diabetes diagnosis category. The contribution to the RAF score from diabetic disorders will not change regardless of whether the patient has uncomplicated diabetes or diabetes with complications. However, type 2 diabetes mellitus without complications (E11.9), for example, will receive a slightly higher coefficient in V28 than it currently does in V24 (i.e., from 0.105 to 0.166). Overall, this will result in a significant reduction in the RAF score for patients with acute or chronic complications from diabetes. The financial impact on MAOs and other stakeholders will depend on case mix.
The overall impact of the proposed changes on beneficiary RAF scores will depend on several factors, however RAF score in general will likely decline. CMS projects that the CY 2024 impact on MA risk scores of the proposed Part C CMS-HCC model is projected to be -3.12%. This projects to $11.0 billion in net savings to the Medicare Trust fund in 2024. Actual PMPM payment amounts are based on multiple addition factors.
Examples of the proposed changes on RAF scores
The following examples demonstrate the potential impact of the proposed changes on RAF scores (based on disease coefficients only) in 2023 vs. 2024.
In our first example, there is a significant negative impact on risk score based on disease coefficients in a Community, NonDual, Aged 73-year-old female beneficiary with multiple conditions. Only the disease coefficients for V24 and V28 are shown in the tables below.