What is Medicare Advantage (MA) Risk Adjustment Data Validation (RADV)?

 The Centers for Medicare & Medicaid Services (CMS) pays MA plans a capitated amount per member per month on a risk-adjusted basis using the CMS Hierarchical Condition Category (HCC) risk adjustment model. Under the HCC model, CMS determines risk scores based on health status and demographic characteristics. MA enrollee HCCs are assigned based on diagnosis data collected from certain healthcare providers (e.g., hospitals, physicians, and outpatient facilities) and submitted by plans to CMS for payment.

Under federal rules, CMS has the authority to conduct annual audits to ensure MA payment integrity through a process called RADV.  Through these audits, CMS validates the accuracy of the diagnosis codes plans submitted for payment. Currently, for each health plan selected for RADV (5%–10% of MA contracts), CMS generates a random sample of enrollees and reviews medical records for the sample to validate their risk scores and the corresponding payments to the plan. CMS estimates payment error by taking the difference between the actual paid amount, based on plans’ submitted diagnoses, and the amount that would have been paid based on RADV-validated diagnoses. The process of applying the error rate of a sample of plan enrollees to the plan’s membership is known as extrapolation.

Overview of 2012 and 2018 Proposals to Modify the MA RADV Process

Over the last 10 years, policymakers proposed different approaches for when CMS can and should recoup overpayments made to MA plans. In December 2010, CMS proposed a methodology for selecting “a statistically valid sample of enrollees from each audited MA contract and extrapolating from the results of that sample audit to calculate a contract-level payment adjustment.” In February 2012, CMS released a “Notice of Final Payment Error Calculation Methodology for MA RADV Contract Level Audits” (the 2012 Notice).

The methodology described in the 2012 Notice included the sampling framework, extrapolation calculation, and the application of a technical adjustment to account for differences in documentation standards for RADV data and the Fee-For-Service (FFS) data used in the risk adjustment model calibration (known as the FFS Adjuster). The FFS Adjuster represents the level of diagnoses not supported by medical records in FFS but included in the calibration of the risk adjustment model and, by extension, plan payment. The FFS Adjuster sets a permissible level of payment error due to unsupported diagnosis codes and limits RADV audit recovery to payment errors above that level.

In November 2018, CMS published a draft rule (4185-P) that departed in part from the methodology outlined in the 2012 Notice. In particular, CMS proposed revising its methodology to exclude the FFS Adjuster in its payment recoupment calculations. To support its proposal, CMS cited an internal study finding “that errors in the FFS claims data do not have a systematic effect on the risk scores calculated by the CMS-HCC risk adjustment model estimation and, therefore, do not have any systematic effect on the payments made to MA organizations.”

Key Policy Changes in the Final 2023 MA RADV Rule

On January 30, CMS released the final MA RADV rule. CMS noted that it expects to recoup approximately $4.7 billion over 10 years through RADV based on the updated policies. Key changes include:

• Exclusion of the FFS Adjuster: CMS finalized its proposal not to include a FFS Adjuster in its final methodology for RADV recoveries, including extrapolation. CMS stated that a FFS Adjuster is not required to ensure “actuarial equivalence” between payments to MA plans and payments under the FFS program, as several stakeholders had argued.
• Extrapolation starting in Plan Year (PY) 2018: CMS revised its proposal to retroactively apply extrapolation to audits staring for PY 2011 and will instead apply extrapolation starting for PY 2018 audits. For PYs 2011–2017, CMS will only collect the non-extrapolated overpayments identified in RADV audits.
• Flexibility in sampling and extrapolation methodology: CMS notes the agency “will rely on any statistically valid method for sampling and extrapolation that is determined to be well-suited to a particular audit.” To determine which contracts to audit, CMS will focus on MA contracts and enrollees’ HCCs that have been flagged as risks for improper payment.
• Application of policies in the final rule to OIG audits: The Office of the Inspector General (OIG) also conducts RADV audits, though it uses a different methodology than CMS. The final rule notes that the provisions focused on extrapolation and sampling described above will also apply to OIG audits that occur outside of CMS’s RADV process.

The final rule addresses many comments raised by stakeholders in response to the CMS’s 2018 proposal. However, it also leaves several unanswered questions that could result in uncertainty for health plans seeking to prepare for future RADV audits, such as:

• What criteria will CMS use to determine which plans to audit? CMS has indicated that it will focus on contracts with higher levels of coding intensity or evidence of high error rates in prior audits. CMS has flexibility in how it will identify which contracts and HCCs are “high risk.” For example, CMS did not define “high coding intensity” or specify what evidence it will consider when identifying high error rates.
• What methodology will CMS apply to future audits? CMS conducted contract-level audits for 2011 to 2013 but changed to condition-specific audits for 2014 and 2015. The audit methodology will determine the extent of plan liability and is a critical piece of information needed to assess the possible impact of the finalized provisions. 
• What criteria will CMS use to determine when it will extrapolate results? By not outlining all the specific conditions under which CMS would choose not to extrapolate results, CMS could leave itself vulnerable to stakeholder concern that it is applying this authority inconsistently.
• What sampling methodology and approach to error estimation will CMS use? CMS has used either random sampling or stratified random sampling where variances can be calculated. Without more specific information on sampling and stratification, plans will not be able to replicate or validate CMS’s audit findings.
• Will CMS coordinate its audit activities with OIG’s? CMS was silent on whether it will coordinate audit activities with OIG to avoid duplication of effort. CMS also did not provide guidance on appeals procedures or reconciling differences in methodology between CMS and OIG.

CMS noted that, as it further develops its methodology and pursues RADV audits, it may—at its sole discretion—provide additional information on the sampling and extrapolation methodology through subregulatory guidance.

Potential Impact of the Final Rule

CMS’s policies in the final rule are consistent with the agency’s ongoing focus on program integrity. Because some items are left unresolved—notably the methodology CMS will employ for future RADV audits—the final rule creates uncertainty for MA plans. Plans may respond by adjusting their MA bids and revenue expectations, which may in turn affect premiums and benefits for MA enrollees.

The impact will vary substantially by individual MA plan based on factors such as plan size, frequency of selection for audits, error rates, and ability to fulfil audit documentation requests. The aspects of CMS’s approach and methodology that are currently undefined may present challenges for plans seeking to assess their risk of audit and future financial liabilities. As a result, most plans will likely build the risk of audit into their assumptions, which could drive up plan bids. Plans that have prepared for audit but are not selected may have a “windfall” that could impact their medical loss ratios. On the other hand, smaller plans or those with fewer resources may be unprepared to manage the uncertainties and costs associated with the RADV process. The extent of the changes may lead plans to assert legal claims related to the Administrative Procedure Act or CMS’s interpretation of whether the statute requires CMS to apply a FFS adjuster.

The final rule may also have an impact on providers. In particular, the burden on providers could increase if health plans increase oversight to enforce a zero-tolerance policy for errors in diagnostic coding in medical records and claims without similar expectations on the Medicare FFS side.

Avalere has deep expertise in RADV and MA plan payment methodologies. To learn more about the potential impact of the Final MA RADV Rule on your business, connect with us.