CMS does not sample your membership. It samples a frame.
CMS begins by building a sampling frame at the enrollee level, using eight specific criteria from Section 4.1. Both the audit sample and any extrapolation are tied to this same group.
Criterion 1 is the filter this page is built on, because it is the one that can be computed from published data. To meet it, a member has to be enrolled in the audited contract for the full thirteen months, from January 2023 through January 2024. If someone joined partway through 2023, or left, or died during that time, they don't make the cut. Neither does a member moved to a different contract number by a consolidation during that window. That member never left and never died, but wasn't in one contract for all thirteen months.
This sets a ceiling on the audited population for each contract in the PY2024 audit, based on what is in the published data. Criteria 1 and 7 below are filled in for your contract from published data. The others are examples. Swap in your actual figures and the rest of the calculation changes accordingly.
Verify criterion 1 against your own enrollment before you rely on it. Mine is built from the CMS monthly files and the Star Ratings disenrollment measure, which CMS publishes for 474 of the 590. Yours is built from the members themselves. If the two disagree, yours is right and I would like to know by how much.
590contracts in the PY2024 audit
431where disenrollment binds criterion 1
43where net enrollment change binds
0.79median criterion 1 ceiling
The three figures after the contract count are computed on the 474 contracts
for which CMS publishes the Star Ratings disenrollment measure. It publishes no value for the
other 116. For 46 of them the 2025 measure data file reads "Not enough data available". The
other 70 are not in the file. Those 116 are on this page and are marked on their own card, and
they are out of the three figures above. Each of the 116 takes the median published
disenrollment rate in its size band. The bands split all 590 contracts into five groups of 118
by January 2023 enrollment, and two of them share a median of 0.16. There are only four
imputed rates in the set, so a marked contract shares its figure with every other marked
contract in its size band.
474. Kirsha LLC's count, not a CMS figure. Of the 590 PY2024 audited contracts,
474 have a percentage in 2025 Star Ratings C26, "Members Choosing to Leave the Plan". Of the
other 116, 46 read "Not enough data available" and 70 are not in the file.
Source: CMS, 2025 Star Ratings Data Tables (ZIP), file 2025 Star Ratings Data
Table, Measure Data (Dec 2 2024), C26, data time frame 01/01/2023 to 12/31/2023.
431, 43 and 0.79. Kirsha LLC's computation, not a CMS figure. For each of the
474 PY2024 audited contracts with a published 2025 Star Ratings "Members Choosing to Leave the
Plan" rate (C26), Kirsha takes the smaller of two limits on continuous enrollment from January
2023 through January 2024. The first is the smaller of the two January enrollment counts, over
January 2023. The second is one less the C26 rate, less 0.0395. The second is the smaller for
431 contracts and the first for 43. The median across the 474 is 0.7905, shown as 0.79.
Sources: CMS, 2025 Star Ratings Data Table, Measure Data (Dec 2 2024), C26, data
time frame 01/01/2023 to 12/31/2023. CMS, Monthly Enrollment by Plan, January 2023 and January
2024. CMS Program Statistics, tables MDCR ENROLL AB 38 and MDCR ENROLL AB 15, CY2023.