Finding per enrollee
Same either way
- An audited HCC is Confirmed, Confirmed Higher, Discrepant, Discrepant Lower, or granted an exception; the record decides [2]
Home / Newsletters / The HCC Operator
The HCC Operator · Issue 5 · Late November 2026
RADV PY 2022 kicks off in January: a 90-day record-readiness plan
Short answer: the CMS schedule puts PY 2023 in November, PY 2022 in January to February and PY 2025 in April to May [1]. The PY 2024 instructions show the sample sizes, records and attestations to expect [2]; extrapolation is designed in but not yet collected; and the 90-day plan starts with 30 charts from 2021 scored against the CMS checklist.
6 payment years · 14 monthsPY 2022 · 2021 recordsSamples 35 to 200 · 2 records per HCCFeb 5, 2027 · PY 2024 records due
Edition
Issue 5 · Nov 2026
Focus
RADV readiness
Horizon
90 days
Metric
30 charts
Effective date
Jan to Feb 2027
Length
5 min
Jan to Feb 2027
Planned initiation of PY 2022 RADV audits, per the CMS schedule [1]
2021
Data collection year whose records a PY 2022 audit will request; five years old on arrival [2]
Feb 5, 2027
PY 2024 record submission deadline, 11:59 pm EST [2]
550plans a year
MA plans CMS said it would audit, up from about 60 [3]
PY 2022 starts in January to February 2027
PY 2020, 2021 and 2024 already opened this year. Six payment years in 14 months; a group serving several plans sees every wave. Dates subject to change [1] [2].
- 1PY 2024 initiatedAug 28, 2026Records due February 5, 2027
- 2PY 2023 plannedNov 20262022 records
- 3PY 2022 plannedJan to Feb 20272021 records; notice lands while PY 2024 is being answered
- 4PY 2024 deadlineFeb 5, 2027Submission window closes
- 5PY 2025 plannedApr to May 20272024 records, first full-weight V28 year
What the PY 2024 instructions tell you about PY 2022
The August 28 methods document is the template; the window ran about 20 weeks. PY 2022 records are from 2021; CMS notes a ten-year retention requirement [2].
| Rule | PY 2024 instruction |
|---|---|
| Sample size | 200 for the ten largest contracts, then 100, 50 or 35 by sampling frame size |
| Records per enrollee | Maximum of twice the number of audited HCCs, any combination |
| Valid record | Legibly signed, face-to-face visit, acceptable provider, date of service in the data collection year |
| Attestation | CMS-generated form only, outpatient records only, first page of the PDF, signed by the treating practitioner |
| Invalid sources | Hospice, home health, lab only, superbill, non-face-to-face encounters |
| Coder review | Up to three independent reviews by certified coders per record |
Extrapolation is designed in, not yet collected
RADV scale, before and after the May 2025 strategy [3]
The 2023 extrapolation rule was vacated in 2025 and is under appeal; for PY 2024 CMS collects sampled-enrollee errors and reserves the extrapolated amount if legally permissible [2].
~60
Plans auditedbefore
~550
9x
Plans auditedper year
35
Records per planbefore
35 to 200
up to 6x
Records per plan
40
CMS codersbefore
~2,000
50x
CMS codersby Sep 1, 2025
BeforeAfterWorkforce
The four-phase response
Intake to discrepancies
- In our RADV audit response playbook; the dated schedule timeline is new this month
Two reads per chart
ASP update
- Our risk adjustment coding attaches evidence to every code, what a RADV reviewer looks for; engine on the HCC coding AI page
Takeaway
Pull 30 charts from 2021 dates of service for members who carried an HCC into 2022 and score each against the CMS record checklist. The failure count is your 90-day plan.
The 90 days, and the checklist
- 1Days 1 to 30own itName the record-request owner; one request template, one log
- 2Days 31 to 60fix itUnsigned electronic notes, missing credentials, history-of-cancer notes without current treatment, unlinked manifestations
- 3Days 61 to 90repeat itSecond 30-chart sample on 2024 dates of service for PY 2025; two-business-day turnaround target for plan requests
Right patient, right yearDate of service present and inside the data collection year.
Legible, valid provider typeCredential and signature on the record.
Diagnosis supports the audited categoryElectronic problem lists often never drop a diagnosis; a history of cancer without current treatment may not validate [4].
Log every requestRequest date, receipt date, intake status, owner.
Frequently asked questions
When do PY 2022 RADV audits start?
The CMS schedule lists January to February 2027 as the planned initiation, after PY 2023 in November 2026 and before PY 2025 in April to May 2027. CMS says the dates are subject to change and that it will update the schedule as needed.
Which records will a PY 2022 audit ask for?
Records from 2021 dates of service, because the data collection year is the year before the payment year. The record must document a face-to-face visit with an acceptable provider, be legibly signed with credentials, and carry a complete date of service.
Can a provider group be audited directly?
CMS audits the MA contract, not the provider. The plan requests records from the providers who documented the audited diagnoses, so a group's exposure is the speed and quality of its record retrieval and the documentation itself.
Is CMS extrapolating PY 2024 results?
CMS designed the PY 2024 audit to calculate extrapolated overpayments but has not decided whether to collect them. It will collect sampled-enrollee amounts and reserves extrapolation if legally permissible, with the 2023 rule under appeal.
- CMS: Medicare Advantage RADV Audit Schedule
- CMS: Payment Year 2024 MA Contract-Specific RADV Audit Methods and Instructions (August 28, 2026)
- CMS press release: CMS Rolls Out Aggressive Strategy to Enhance and Accelerate Medicare Advantage Audits (May 21, 2025)
- CMS: RADV Medical Record Checklist and Guidance
Checked October 25, 2026. Rules change; confirm against the source before relying on them. CMS has said the RADV audit schedule is subject to change.
Want the 30-chart sample run for you?
We pull the charts, score them against the CMS record checklist, and report the failure modes by provider with the retrieval log set up. Certified coders, two reads per chart.