New Study Finds Significant Flaws In How CMS Audits Medicare Advantage Plans

posted by AHIP

on July 25, 2018

Overview of RADV Audits

Currently, CMS utilizes what is known as “Risk Adjustment Data Validation”, or RADV, audits to review the documentation of diagnoses submitted by plans to CMS. CMS adjusts payments to MA plans based on the health status of their enrollees, which is measured through the calculation of a ‘risk score’.

The ‘risk score’ is calculated based on the demographic characteristics (e.g., age, gender) and diagnoses (i.e., diseases) of the plan’s members, and are designed to compensate health plans for their members’ expected health care costs.

In February of 2012, CMS published its RADV audit methodology, which includes selecting a sample of enrollees for a group of MA plans and reviewing medical records to determine if the diagnoses submitted by the plans were supported by the medical records for each enrollee in the sample. Based on the audit results, CMS calculates a new risk score for any enrollee with a diagnosis that was not adequately documented in a medical record by excluding the condition. CMS then calculates a “payment error”, or the difference between the original monthly payment to the plan for each enrollee and the corrected monthly payment for each enrollee.

The CMS methodology would then extrapolate the payment error from the sample of enrollees to the entire MA plan population to determine a recovery amount. Finally, CMS would adjust the payment recovery amount for coding errors in the traditional Medicare population – also known as the ‘fee-for-service (FFS) adjuster’.

At this point, CMS has yet to close out the ongoing 2011-2013 audits because it has not yet finalized the FFS adjuster component of its methodology.

Key Findings from New Study on RADV Audits

The current CMS approach to RADV audits of MA plans has several significant and impactful methodological issues, according to a recent study by the Wakely Consulting Group.

In its study, Wakely reviewed the CMS approach to conduct RADV audits outlined in its February 2012 RADV audit methodology document and applied for 2011 to 2013 RADV audits. The study identified several significant areas of concern including:

  1. CMS’ extrapolation approach is subject to a high degree of randomness and could result in inequitable treatment of similar contracts, because contracts with similar average error rates can have very different payment penalties.
  2. The methodology is sensitive to which beneficiaries/conditions are included in the sample, because certain diseases can have a disproportionate impact on the payment error. As one example, based on a simulation of the RADV process, a single unsupported cancer diagnosis could increase the payment penalty by 16.7%.
  3. The methodology could drive bias against higher enrollment contracts and contracts with low absolute risk scores. The sampling approach makes proportionally higher penalties more likely for larger enrollment contracts compared to smaller contracts.
  4. Because the methodology does not take into account diagnosis-specific error rates, which are acknowledged by CMS to vary, penalties from one sample may be higher due to the “luck of the draw” for which diagnoses are selected (i.e., two different samples from the same plan could yield different payment penalties due to randomly selected diagnoses having a higher incidence of coding errors in the industry.)
  5. The methodology can result in a payment penalty higher than the actual payment error, because CMS is calculating a range around the average observed payment error.

AHIP Policy Recommendations

AHIP believes that due to the methodological issues identified by Wakely, CMS should close out the 2011-2013 RADV audits without extrapolating the payment errors identified as these flaws cannot be solved (since CMS already conducted the payment audits).

To download the full study, click the button below.