CMS miscalculated a Five-Star quality measure. Here’s what it cost nursing homes.

CMS's risk adjustment for the long-stay walking measure (N035.06, “ability to walk independently worsened”) was wrong, and published rates ran about 2 percentage points too high, going back to at least 2024. We found it and reported it, and on October 6, 2026, CMS confirmed the bug and fixed it in iQIES. We estimate it cost about 640 facilities a quality measure star, about 200 an overall star, and Ohio facilities $15–20 million in Medicaid quality payments.

How we found it

SuperQM rebuilds every CMS quality measure from each facility's MDS, resident by resident, and reconciles it against CMS's iQIES reports. Our rule: match iQIES exactly, or explain every difference.

On the walking measure we couldn't explain it. Numerators, denominators and observed rates matched CMS exactly, building after building. The risk-adjusted rate didn't.

Risk adjustment is supposed to give each resident an expected chance of the outcome, from the risk factors in CMS's QM manual: prior function, cognition, vision, oxygen, age and sex (how the measure works). The adjusted rate then compares observed against expected:

logit(adjusted) = logit(observed) − logit(expected) + logit(national average)

That way, a building caring for higher-risk residents gets credit for it. CMS's expected rates barely varied between buildings, which is the pattern you'd see if the model used age and sex and dropped the clinical risk factors. We documented it and filed a ticket. CMS confirmed the bug on October 6.

What changed after the fix

• Our calculation matches iQIES to one decimal place in 45 of 49 building-quarters.
• Corrected rates were lower in 49 of 52 building-quarters we compared, by about 2.2 points on average.
• CMS's national average for the measure dropped. For 2025 Q4 it went from 15.1% to 12.2%.
• 2024 data shows the same pattern.

Here is one Ohio facility we work with, over the four quarters Ohio used for its July–December 2025 quality payment:

Same residents, different answer. The facility's Ohio QIP quality score goes from 31.75 to 33.25 against a 32.0 floor. That's the difference between missing the quality add-on and earning about $38 more per Medicaid day.

The national impact

We measured the correction directly at 20 buildings (66 before-and-after building-quarters), then applied it to every facility on Care Compare with 1,000 simulations each. Our Five-Star scoring reproduces CMS's published ratings for all 8,238 facilities with every measure published. Ohio and Texas were re-scored using each program's published rules.

Buildings with higher-risk residents were hurt most, because the faulty model gave them no credit for it.

What we can't say yet. These are estimates; each facility's exact number is in its own post-October 6 iQIES report. Star cut points and state floors were set on the erroneous data. CMS hasn't said whether it will republish Care Compare or notify states, and states haven't said whether they'll revisit payments.

What to do now

1. Get your breakdown. We'll tell you, for free, whether your buildings were affected. Request your breakdown below.
2. Check your iQIES report. Any MDS Facility-Level QM report run after October 6 reflects the fix. Compare it with Care Compare for the same quarters.
3. Ohio operators: talk to OHCA and your counsel before accepting HB 479 settlement payments, which cover the same periods and may waive other claims.

We weren't looking for a CMS bug. We were making sure our own numbers were right. Quality measures now drive stars, referrals and Medicaid dollars, and the people accountable for them should be able to see how every number is built. That's what SuperQM is for.

SuperLTC is not affiliated with CMS. Estimates are modelled from CMS's correction at the buildings we measured; exact figures come from each facility's post-October 6, 2026 iQIES report.