Seniors have been selecting Medicare Advantage plans over traditional Medicare for quite some time now. There are a number of reasons for this: additional benefits, networks of high-performing physicians, reduced out-of-pocket expenses for services and limits on total out-of-pocket expenses incurred in a given year, to name a few. These benefits are particularly appealing to the Medicare-eligible beneficiaries most adversely impacted by socioeconomic factors. For that reason, Medicare Advantage plans have been working to advance health equity, improve social determinants of health, and expand value-based care arrangements with their networks of healthcare providers. In 2021, the Centers for Medicare and Medicaid Services (CMS) published a study showing the impact Medicare Advantage plans had on racial, ethnic, sex, and rural-urban inequities from 2009 to 2018.1
Medicare Advantage plans are able to support these populations due to the structure of the program. The plans receive a risk-adjusted per capita capitation payment and then assume the financial risk for medical care of their plans' members. They are also incentivized to meet certain quality standards through the CMS Stars rating program. In many cases, these plans will contract with providers in a similar way — sharing in the financial risk, working with plans to accurately capture the risk profiles of their attributed members, and incentivizing high quality care through bonus programs.
The Medicare Advantage program has been a leader in advancing value-based care and controlling the growth of medical costs for their members due to the alignment of strategic objectives across the key stakeholders. Members want to have their medical costs controlled, physicians want to have happy and healthy patients, health plans want to keep their members healthy to continue receiving their capitation payments, and CMS wants to protect the Medicare Trust Fund. There have been attempts to varying degrees of success to replicate this model across other health care lines of business, but none have been as successful as the Medicare Advantage program.
For these reasons, there has been expansive growth of Medicare Advantage membership, number of plans, and value-based provider organizations. Unfortunately, not all of that growth has been with the best intentions. A few bad actors have utilized this program to commit fraud by adding baseless diagnoses to medical records to increase the perceived risk of a population and drive larger capitation payments from Medicare. CMS has been actively revising the program and its auditing tool, the Risk Adjustment Data Validation Audit (RADV), to combat this fraud while simultaneously pushing towards their stated goals of driving accountability, advancing health equity, supporting innovation, addressing affordability, and partnering to achieve systemic transformation.
In the Advance Notice Call Letter, CMS has proposed to make material changes to their risk adjustment model, the CMS-HCC Model. Unfortunately, these changes are contradictory to CMS's stated goals and would reverse years of advancements in health equity, innovation, and affordability. Starting with Section G (page 43) of the advance notice, CMS proposes a reclassification of the hierarchical condition categories (HCCs) used to determine member risk scores. The basis for this reclassification is an inadequate assessment4 CMS conducted to determine variances in coding of certain conditions for beneficiaries covered by traditional Medicare and members of Medicare Advantage plans.
The Core Problem: Correlation Mistaken for Causation
I say that this assessment is inadequate because the findings demonstrate a correlation of certain conditions being documented within the Medicare Advantage populations compared to traditional Medicare, but fail to identify the causation of that correlation. The danger of making such a material change to this program without fully understanding the causation can lead to unintended consequences — which is what I believe will happen if the proposed changes go into effect.
There are a number of factors that need to be accounted for in order to understand the true causation of this assessment. For instance, are the populations these programs serve identical? Are they assessing the coding variation across the same set of providers? Are the practice patterns or care models the same? The answer to those questions is inevitably no.
As stated above, Medicare Advantage populations tend to have higher proportions of members that are socioeconomically challenged. With a higher prevalence of socioeconomic challenges comes a higher prevalence of disease and chronic illnesses. These members are often in managed care plans where the care is coordinated and members that are not actively seeking care are engaged by the plan and care teams to ensure that the appropriate preventive and health maintenance care services are performed.
Traditional Medicare is not structured as a managed care product and therefore, beneficiaries are less likely to be outreached for preventive measures and less likely to be identified as having a chronic illness in its early stages. CMS and CMMI have attempted to integrate a level of managed care into the traditional Medicare program through various programs and demonstrations to develop Accountable Care Organizations (ACOs). The most recent, the ACO REACH program, evolved out of the Medicare Direct Contracting program that had a stated goal to "broaden participation in CMS Innovation Center models by allowing model participation by organizations new to Medicare FFS, such as physician-managed organizations that currently operate exclusively in the MA program."
Differing Provider Networks Skew the Assessment
Medicare Advantage plans manage the care their members receive through networks of contracted providers. These plans spend an enormous amount of time and resources balancing the need for access to care and maintaining the utmost quality and efficient providers within their networks. Traditional Medicare allows beneficiaries to see any provider that accepts Medicare. This allows for more access to care under traditional Medicare but also skews the stated CMS assessment by having a larger portion of the traditional Medicare data coming from physicians that are less efficient and effective than those in the Medicare Advantage networks.
The physicians that are excluded from Medicare Advantage networks are less likely to provide complete and accurate profiles of patient risk on their claim submissions or medical records. These physicians are also less likely to provide the same level of holistic care and focus on maintaining health and wellbeing. The fee-for-service (FFS) reimbursement model does not provide the right incentives for that level of care.
Medicare Advantage plans often contract with providers through value-based arrangements that provide the incentives for providers to maintain the health of their attributed populations — which includes the early detection and prevention of chronic illnesses. These providers are rewarded for keeping patients healthy as opposed to the volume of visits incurred. In order to effectively succeed in that mission, these providers must collect and maintain data on their patients, provide preventive services, and work to keep patients' conditions from exacerbating to the point that they need high-cost medical interventions.
When taking into account these differences across the MA and traditional Medicare programs, the findings of CMS's assessment would indicate a lack of complete diagnosis data within the traditional Medicare program instead of an excess of data within the MA populations. CMS has misinterpreted the causation of the findings from their assessment and relied on Principle 10 of the model development to develop their proposed changes. Principle 10 states:
Discretionary diagnostic categories should be excluded from payment models. Diagnoses that are particularly subject to intentional or unintentional discretionary coding variation or inappropriate coding by health plans/providers, or that are not clinically or empirically credible as cost predictors, should not increase cost predictions. Excluding these diagnoses reduces the sensitivity of the model to coding variation and coding proliferation.
CMS incorrectly assumes that the categories and diagnoses which they intend to remove from the HCC model are subject to discretionary coding variation, inappropriately coded, or not a credible cost predictor. The variation in coding can be explained by the sample biases discussed above. Any inappropriate coding of these conditions should be addressed through the RADV audit and oversight responsibilities of the plans.
Furthermore, this directly contradicts Principle 5 of the HCC Model, which states:
The diagnostic classification should encourage specific coding. Vague diagnostic codes should be grouped with less severe and lower-paying diagnostic categories to provide incentives for more specific diagnostic coding.
The proposed changes would reward providers that are less specific and complete in their documentation and harm the providers that are actively treating, managing, and documenting the health status of their members.
The Trust Fund Argument Doesn't Hold Up
If the proposed changes to the HCC model were enacted, CMS estimates that the Medicare Trust Fund will save $11 billion in 2024. This statement presents another inadequate analysis. The estimation of savings to the Medicare Trust Fund does not take into account the impact that these changes will have on providers' ability to finance their value-based care operations and plans' abilities to continue offerings that are affordable to those impacted by health inequities.
When considering the impact of those adverse consequences, the savings from changing risk adjustment will lead to critically higher medical costs, decreased quality and affordability for members, and will further accelerate the depletion of the Medicare Trust Fund — the exact opposite of the stated intent.
A Better Path Forward
Plans, providers, and Medicare recipients should take the opportunity to comment on this materially adverse proposal. Plans will also need to start developing their strategy in the event that CMS continues with the proposed changes — including evaluating network configurations and contract terms.
There are other changes listed in the advance notice that have a lesser effect on the program as a whole. Some of these proposed changes are beneficial: the expansion of the Part D Low Income Subsidy from 135% to 150% of the federal poverty line, and further integrating SDoH metrics into the Star rating methodology, are both positive developments.
CMS would be wise to take a holistic look at their proposals and the intended outcomes of each. Rather than completely recategorizing the CMS-HCC model, I would suggest expanding upon the existing version with the addition of categories related to Social Determinants of Health (SDoH) ICD-10 codes. The ingestion of that data is an important quality aspect as reflected in the Stars methodology — and the impact and variation of different SDoH factors should also be reflected in the risk adjustment methodology. That would serve to better predict future costs and create a greater incentive for providers and plans to capture that data and take steps to address the barriers their members and patients face in receiving equitable health outcomes.
