By Dr. Harry Saag, former National Medical Director, Walgreens
Background – The History of Medicare Risk Adjustment
For a quick refresher, risk adjustment factors are a critical part in administering Medicare Advantage (MA) plans as well as a foundational component of many value-based models. Specifically, MA plans receive capitated revenue per patient determined by the severity of illness of each patient, which in theory, should correlate to that patient’s projected health care spend.
Since 2000, the risk adjustment model has been based on the Hierarchical Condition Categories (HCC) model whereby specific diagnosis codes map to an HCC which then correlates to a dollar amount for that condition. There are also payment adjustments for demographic factors, Medicaid status, and regional coefficients, but for the purposes of today’s article we will focus exclusively on the HCC model when discussing impact on reimbursement.
Increased Scrutiny on Risk Adjustment Methodology
Version 24 of the HCC system has been in place for several years but has come under scrutiny based on federal government OIG reports showing unusual patterns of certain diagnosis codes that do not correlate to future medical spend[1]. The report found that 12 health conditions accounted for billions of premium dollars:
Let’s look at a common diagnosis such as Diabetes to get more granular on how these changes impact an MA plan’s bottom line. In v24 methodology, a diabetic will have a different risk adjustment factor depending on if they are coded as:
- A diabetic without complications
- A diabetic with acute complications
- A diabetic with chronic complications
Recall that this determination is made based on which ICD code is selected by the clinician. It’s not uncommon for a very busy clinician to select the first diabetes diagnosis code that pops up on their electronic health record, as opposed to sifting through the multitude of diabetes codes and selecting the code that risk adjusts to the highest level. In this updated version called v28, CMS is consolidating all 3 of these risk codes to one code which will take pressure off doctors to select the ‘most accurate’ code. This concept is called ‘constraining’ codes in the risk adjustment world, and v28 also constrains congestive heart failure codes to simplify reimbursement for this disease state as well.
The below chart[2] demonstrates the constraining concept where all diabetes codes will constrain to a single HCC score. You will notice the initial three codes in the chart (labeled OLD) differ in HCC risk score from 0.105 to 0.302 depending on whether the patient has diabetes with or without complications. However, in the updated version of codes (labeled NEW) all three constrain to the same HCC score of 0.166. The end result of this constraining is that those clinicians who were previously selecting ‘diabetes without complications’ will see an increase in their risk adjustment factor while those who typically selected’ diabetes with complications’ will see a decrease in their risk adjustment directly impacting their capitation premium or revenue.
From a clinical perspective, these changes make sense. Almost any senior who has diabetes will meet some form of criteria for showing complications. It is the exception that a senior with diabetes would have zero signs of complications. Further, the way the specific ICD code is selected for each individual patient is likely based on the reimbursement model for that specific clinician. A clinician in a value-based arrangement is going to be aware of the difference in reimbursement between selecting ‘no complications’ versus ‘chronic complications,’ however a clinician who is reimbursed a pure relative value unit (RVU) model is more likely to select the first diabetes code that pops up on the screen since the ICD code selected does not directly impact the patient’s actual care. Constraining these codes should, in theory, simplify complexity and re-focus clinician brain space on the treatment of the patient’s diabetes as opposed to also trying to remember the highest reimbursable ICD code.
Debate will continue to rage around these changes with CMS predicting the v28 model will lead to roughly a 2% reduction in MA risk scores, which translates to an estimated $7.6 billion dollars in risk premium[3].
The X Factor – The Incoming Administration and Potential Changes
With a new incoming administration, anything could happen regarding potential revisions to v28 or the timeline to implementation. Currently, v28 is being phased in over a 3-year period with it being fully live in 2026. But many MA plans and risk-bearing ACO’s may push the incoming administration to revisit this timeline or suggest wholesale changes to the v28 model.
2025 will prove to be an insightful year as the transition to v28 continues with the only thing remaining certain is that more change is on the horizon.
[1] https://oig.hhs.gov/oei/reports/OEI-03-17-00474.pdf
[2] https://www.aafp.org/pubs/fpm/issues/2023/1100/hcc-update.pdf
[3] https://www.mwe.com/insights/cms-finalizes-risk-adjustment-model-in-2024-rate-announcement-for-medicare-advantage-and-part-d/?utm_source=chatgpt.com
About the Author
Dr. Harry Saag is an internal medicine physician and former National Medical Director at Walgreens. He has worked in the value-based care space for the past decade both at Walgreens and NYU Langone Health where he served as Medical Director for their Clinically Integrated Network. He received his MD from the University of Alabama at Birmingham and holds a Masters Degree in Healthcare Management from Harvard T.H.Chan School of Public Health. His academic work has been published in JAMA, JAMA Internal Medicine, and the Journal of General Internal Medicine, among others. He lives in Florida with his wife and 3 dogs
