How reliable are Medicare’s star ratings?

The nursing home star rating does what it claims. Sorted by overall stars, the facilities that score well genuinely have more nurses and lose fewer of them. Across 14,288 facilities reporting staffing, the gradient runs in one direction from one star to five, with no reversal.

That is worth saying plainly, because the rest of this page is about a limitation and it would be easy to read it as “the ratings are useless.” They are not. If you are comparing two nursing homes and one has five stars and the other has one, the five-star home almost certainly has more staff at the bedside.

The same staffing data, cut by who owns the building rather than by star rating, produces a gap most families are never shown.

Non-profit homes average roughly 43 minutes more nursing time per resident per day, about 19% more, and lose noticeably fewer staff each year. Turnover matters more than it sounds: the person who knows your mother usually ate breakfast at seven is the person who notices when she stops.

This is an average across thousands of facilities, not a verdict on any single one. Plenty of for-profit homes staff generously and some non-profits do not. Treat it as a question to ask on a tour, not a filter to apply before one.

Medicare gives home health agencies two separate ratings. The clinical star summarises measures drawn from clinical records. The patient experience star comes from a survey of patients themselves. Both appear on Care Compare. Most families read the first and assume it stands in for the second.

It does not, and the way it fails is not the way you would expect.

Patient experience climbs from one star to 4, and then falls back at five. Five-star agencies average 3.91 on patient experience, below the 4.06 averaged by 4-star agencies. The dip is small, and it is the direction that matters rather than the size: more clinical stars stop buying a better experience before you reach the top of the scale.

A single odd column usually means the measure is odd, not the world. So we checked the same thing against a different question: the raw share of patients who said they would recommend the agency, which is a percentage rather than a star and is reported by more agencies.

It has the same shape. Recommendation rates rise through the middle of the scale and come off the top, on a sample of 6,552 agencies. Two independent readings of what patients thought, agreeing with each other and disagreeing with the clinical star, is harder to explain as a quirk of how one rating gets built.

We are not claiming to know why. Agencies that excel on clinical measures may take on sicker patients, or may run the kind of tightly protocolised operation that scores well and feels brisk. The honest statement is the narrow one: the clinical star does not predict the patient’s account of the care, and at the top of the scale it points slightly the wrong way.

Averages hide the individual cases, and the individual cases are what you actually choose between. Of the 4,538 agencies carrying both ratings, 1,205 differ by two stars or more. That is 27% of them, better than one in four.

At the extremes: 310 agencies hold the top clinical rating while patients give them three stars or fewer, and 249 score two stars or fewer clinically while patients rate them four or better. Those are real agencies that a family would pick or reject on one number without ever seeing the other.

Read both home health ratings, and treat disagreement as information. An agency strong clinically and weak on experience is not disqualified. It tells you what to ask about: how often the same aide comes, whether anyone answers the phone at the weekend, how changes get explained.

For nursing homes, ask for the inputs. Nursing hours per resident per day and annual turnover are the two numbers behind the star, and a facility that will not give you them has told you something. Ownership is worth knowing before the tour, not to rule anywhere out, but so you know which questions to press.

Check the date on everything. An inspection from three years ago describes a building that may since have changed hands, administrator and half its staff.

Figures come from Centers for Medicare & Medicaid Services public reporting, the same source behind Medicare Care Compare. We reload it monthly. Everything above is computed at page build from our copy, so the numbers move when CMS republishes rather than when someone remembers to edit a sentence. The queries live in one file, lib/research/rating-divergence.ts.

Four limits worth stating:

• Patient experience is only reported for agencies with enough survey responses: 4,570 of 12,460 home health agencies. The comparison covers the subset carrying both ratings, not the whole field.
• These are averages across agencies, each counted once. They are not weighted by how many patients each one serves.
• Nothing here establishes cause. Ownership predicting staffing does not mean ownership determines care, and the five-star dip has several plausible explanations we cannot separate with this data.
• Assisted living is absent. It is licensed state by state with no national register, so no comparable national dataset exists. Our own coverage is deepest in six states.

How we handle the underlying ratings elsewhere on the site is set out in how we rank providers.

Are Medicare star ratings accurate?

They are accurate about what they measure, which is narrower than most families assume. For nursing homes, the overall star rating tracks staffing and turnover closely: in our analysis of 14,273 facilities reporting staffing, five-star homes averaged substantially more nursing hours per resident per day and much lower staff turnover than one-star homes, in a clean gradient. For home health agencies the picture is different. The clinical star rating and the separate patient-experience rating disagree by two or more stars at more than a quarter of the agencies carrying both.

Does a five-star home health agency mean patients are happier?

No, and mildly the opposite at the top end. Across agencies carrying both ratings, average patient experience rises from one star through four stars and then falls back at five stars. The same shape appears in the separate "would you recommend this agency" percentage, measured on a larger sample, so it is not an artefact of how the experience star is built. Five-star clinical agencies are excellent at the clinical measures Medicare scores. That is not the same as being the agency patients rate most highly.

What is the difference between the clinical star and the patient experience star?

The clinical star summarises process and outcome measures drawn from clinical records: whether care began in a timely way, whether patients improved at walking or bathing, whether they ended up back in hospital. The patient experience star comes from HHCAHPS, a survey of patients themselves, covering communication, how staff treated them, and whether they would recommend the agency. They are built from different inputs and answer different questions, which is why they can disagree about the same agency.

Are for-profit nursing homes worse than non-profit ones?

On staffing, the gap is consistent and large. Among the 14,273 facilities reporting staffing data, non-profit homes averaged roughly 0.7 more nursing hours per resident per day than for-profit homes, which is around 19% more nursing time, and had noticeably lower staff turnover. That is an average across thousands of facilities, not a verdict on any individual home: plenty of for-profit facilities staff well and some non-profits staff poorly. Use it to know what to ask about, not to rule a place out.

What should I look at instead of the star rating?

Look at the inputs rather than the summary. For nursing homes, ask for nursing hours per resident per day and annual staff turnover, and read the inspection findings rather than the number derived from them. For home health, read the clinical star and the patient experience rating separately and notice when they disagree, because that disagreement is information. In both cases the date matters: an inspection from three years ago describes a building under different management.

Where does this data come from?

Centers for Medicare & Medicaid Services public reporting, the same source behind Medicare Care Compare. We reload it monthly and the figures on this page are computed from our copy at page build rather than typed in by hand, so they move when CMS republishes. Nursing home and home health data are federal and cover every certified provider in the country. Assisted living is licensed state by state with no national register, so it is not part of this analysis.