RADV instructions show CMS is serious about attacking overpayments, but is the approach fair?
The Centers for Medicare and Medicaid Services (CMS) is true to its word. Late Friday the agency published methods and instructions for Risk Adjustment Data Validation (RADV) audits for Payment Year (PY) 2019. In addition, a companion Frequently Asked Questions (FAQ) document was published. The agency announced on May 21 it will audit all eligible Medicare Advantage (MA) plans for each payment year in all newly initiated audits and invest additional resources to expedite the completion of audits for payment years (PYs) 2018 through 2024.
Details of the announcement and some of my concerns are in this June 5 blog: https://www.healthcarelabyrinth.com/cms-to-audit-all-plans-on-risk-adjustment/ . On May 12, I also did a blog here ( https://www.healthcarelabyrinth.com/could-medicare-advantage-overpayments-be-considered-to-offset-deep-medicaid-cuts/ ) on the possibility that MA risk adjustment overpayments would be part of the budget reconciliation bill to increase savings and perhaps alleviate Medicaid reductions. But after some lobbying from MA interests, lawmakers appear to be backing away. But you never know what can happen on Capitol Hill. In the end, while caution is needed, I am a believer in some risk adjustment reforms.
What did CMS announce?
Here is my review of the key points from the methods and instructions document as well as the FAQs for PY 2019. The instructions are substantially similar to the PY 2018 one currently inflight with one major exception outlined below:
- CMS selected 45 MA contracts in the first batch of PY 2019 RADV audits. They include big plans Cambia (a Northwest Blue), Elevance Health, Humana, Molina, Scan, United, and Centene/WellCare. There will be additional rounds for PY 2019. The PY 2018 RADV audits had 60 contracts. It commenced in November and submissions were due in April. Plans await findings.
- Risk adjusted payments for PY 2019 were derived from diagnoses from face-to-face medical encounters with dates of service from January 1, 2018, through December 31, 2018.
- The PY 2019 audits begin immediately. Enrollee data lists (EDL), which has each member and the HCCs (risk adjustment groupers) being audited for each, were available as of 6/12/2025. Medical record submissions commence on 6/20/2025. Records must be submitted by 9/15/2025. CMS expects findings in early 2026.
- The audits will focus on risk adjusted payments that CMS believes may be at higher risk for overpayments. Specifically, the samples represent those ranked in the top decile of the improper payment prediction models (i.e., predicted to have the greatest reduction in their risk score as a result of a RADV audit). CMS says they then used a statistically valid random sample of 35 enrollees. For PY 2018, this same high-risk approach was used, but in addition CMS also reviewed enrollees who had all of their HCCs submitted derived from a linked or unlinked chart review.
- CMS describes in some detail the extrapolation process. It says risk adjustment discrepancies are aggregated for sampled enrollees to determine the average change in sampled enrollees’ risk scores as a result of the RADV audit. This is then used to estimate an overpayment amount for collection for the sampling frame using a statistically valid extrapolation technique.
- Plans must submit at least one valid medical record to support each audited CMS-HCC. For each sampled enrollee, the plan can submit a maximum number of medical records equal to two times the number of audited HCCs. If a sampled enrollee has three audited CMS-HCCs in the EDL, the maximum number of medical records a plan can submit for that enrollee is six. A plan may submit up to six medical records in any combination for the three audited HCCs. This policy appears more limiting than in the last. In effect, the average number of medical records that can be submitted for each HCC is two. For PY 2018 and prior, the count was five per HCC.
- CMS does a good job explaining the coversheet and the medical record requirements. One medical record could substantiate more than one audited HCC but must be clearly marked as so.
- Medical records lacking the necessary provider signature and/or credentials require a signed attestation to be considered valid and eligible for review.
- If a medical record file and coversheet combination is determined to be invalid, the plan may submit another one to replace it if the medical record submission window is still open. CMS encourages plans to submit coversheets and medical records as accurately and as early as possible. CMS will provide daily feedback and publish a feedback report.
- Medical record file and coversheet submissions determined to be valid will be sent to the diagnosis code abstraction process. Each submitted valid medical record may be subjected to up to three rounds of review by certified medical record coders.
- A plan experiencing an extraordinary circumstance may request a hardship exception.
Conclusion
Plans need to be worried about a few things. The 100% audit announcement and methods will likely embolden further lawsuits against the RADV rule.
The 100% RADV audit announcement came quickly and it is safe to say that many plans are ill-prepared to meet quick turnarounds for future rounds. Plans need to examine their end-to-end encounter processes for thoroughness and accuracy, look at potential wrong diagnoses being submitted, educate providers on RADV, and perhaps run mock audits.
Fewer records can now be submitted, which ups the ante on accuracy, timeliness, and completeness during the submission period.
Generally, the suspect sample audit approach and extrapolation should be a major concern. In the 2023 final RADV rule, CMS said it was “not adopting any particular statistical sampling methodology in this final rule. As previously noted, CMS will use statistically valid methods for sampling and extrapolation that we determine to be well-suited to a particular RADV audit.” We have seen the negative impacts of these discretionary approaches from CMS before. A case can surely be made that legitimate sampling methods and even the extrapolation approach could be reasonable. Government deserves to oversee its contractors via efficient means. But not publishing sampling and extrapolation methods is simply not transparent or equitable. This is ripe for legal challenge.
More importantly, plans should be most worried by the targeted RADV audit approach and how that ties to rigorous and ongoing 100% audits. Are plans at risk of greater extrapolated recoupment penalties (and greater negative publicity) by the targeted high-risk sample approach than if extrapolation occurred through comprehensive sample and auditing approaches? Does the government now conduct RADV audits in an Elon Musk, “DOGE-like” approach, storm-trooping for maximum recoupment to hit lower spending targets rather than a fair and transparent process rooted in real program integrity?
Could we see too much “off with their heads” here? CMS’ recent approach to Star regulation and appeals as well as its high-profile losses in court are instructive as plans prepare for RADV audits and their aftermath. If CMS is to go this route, it should be judicious, be transparent, and recognize the huge fallout that could occur on an already beleagured MA industry that is so crucial to Medicare’s future as well as the growing number of Medicare beneficiaries.
Additional materials:
CMS 5/21 Press Release: https://www.cms.gov/newsroom/press-releases/cms-rolls-out-aggressive-strategy-enhance-and-accelerate-medicare-advantage-audits
Methods and Instructions: Audit Methods and Guidance Contract-Specific Risk Adjustment Data Payment Year 2018
FAQs: https://www.cms.gov/files/document/py2019-radv-questions-and-answers.pdf
#medicareadvantage #riskadjustment #radv #overpayments #cms
JUNE 28 UPDATE:
CMS is working at lightning speed. About 2 weeks after announcing Round 1 of PY 2019 RADV audits, on June 25 the agency announced Round 2. Round 1 had 45 contracts. Round 2 has 310. That means that almost 50% of 732 contracts live in that year already have audit notices.
CMS reissued both its “method and instructions” document and FAQs on June 25 for Round 2. No material changes from the documents issued roughly two weeks ago. In essence the deadlines and timetable are about two weeks later for Round 2 compared with Round 1.
Updated FAQ link to see all 355 contracts now under audit: https://lnkd.in/eXmcYVtY
As well, see this CMS page for all the relevant documents: https://lnkd.in/e3Jr3CZ3
PY 2018 started in late 2024 is now concluded on the plan side, but the PY 2018 FAQs indicated that results would not be until mid-2026. Will these be accelerated as CMS says they will have results for PY 2019 in early 2026? Or will CMS take time to look at both years together? Like PY 2019, CMS used a high-risk suspect audit approach for PY 2018.
Some health plans are challenging the 2023 rule on a number of grounds, including on its retroactive recoupment, extrapolation, and lack of FFS adjuster.
The new 100% audit approach from CMS will likely mean more plans litigate.
While a case can be made that suspect auditing can be fair, it raises a real question about whether the industry will end up worse off with the suspect approach than a more comprehensive one.
— Marc S. Ryan
