How Medicare calculates the ratings
Families see stars and scores and reasonably assume they are simple. They are not. Here is how CMS actually builds them, in plain English, taken directly from CMS’s own technical documents. Where CMS publishes an exact number, I give it; where it does not, I say so rather than guess.
Nursing home Five-Star ratings
CMS gives each nursing home three “domain” ratings, health inspections, staffing and quality measures, and then combines them into an overall rating of 1 to 5 stars. All of what follows is from CMS’s Technical Users’ Guide (September 2026).
1. Health inspection rating
- What goes in: the two most recent standard health inspections, plus complaint investigations and focused infection-control inspections from the last 36 months. (Before July 2025, CMS used the three most recent standard inspections.)
- Points for each citation: every deficiency gets points based on how serious it is (severity) and how many residents it affected (scope). More serious, more widespread problems get more points. Extra points apply to deficiencies that CMS classifies as “substandard quality of care.”
| Severity | Isolated | Pattern | Widespread |
|---|---|---|---|
| Immediate jeopardy to resident health or safety | J: 50 (75) | K: 100 (125) | L: 150 (175) |
| Actual harm that is not immediate jeopardy | G: 20 | H: 35 (40) | I: 45 (50) |
| No actual harm, potential for more than minimal harm | D: 4 | E: 8 | F: 16 (20) |
| No actual harm, potential for minimal harm | A: 0 | B: 0 | C: 0 |
- If an immediate-jeopardy citation is marked “past non-compliance,” it receives the points of a G-level citation (20).
- Revisits: no points for the first revisit; if a home needs a second, third or fourth revisit to show it fixed problems, CMS adds 50%, 70% or 85% of the inspection score for that cycle.
- Weighting by time: the most recent standard inspection cycle counts 3/4 and the previous one 1/4. Complaint and infection-control citations from the last 12 months count 3/4; those from 13 to 36 months ago count 1/4.
- Compared within the state: stars are set by comparing homes in the same state. The top 10% (fewest weighted points) get 5 stars, the next 70% are split evenly into 4, 3 and 2 stars (about 23.33% each), and the bottom 20% get 1 star. CMS recalibrates the cut-offs every month. This means a 3-star home in Ohio and a 3-star home in Indiana are each being compared with their own state, not with each other.
2. Staffing rating
The staffing rating comes from payroll records (CMS’s Payroll-Based Journal, PBJ) that homes submit every quarter, due 45 days after the quarter ends. Staffing levels are case-mix adjusted, meaning CMS adjusts for how much care the home’s residents need. PBJ does not count hospice staff or private-duty aides a family pays for.
| Measure | Max points | How points are assigned |
|---|---|---|
| Case-mix adjusted total nurse staffing hours per resident per day (RN + LPN + aide) | 100 | National deciles, 10 to 100 |
| Case-mix adjusted RN hours per resident per day | 100 | National deciles, 10 to 100 |
| Case-mix adjusted total nurse staffing hours on weekends | 50 | National deciles, 5 to 50 |
| Total nursing staff turnover | 50 | National deciles, lower turnover earns more |
| RN turnover | 50 | National deciles, lower turnover earns more |
| Administrator turnover | 30 | 0 departures: 30 · 1 departure: 25 · 2 or more: 10 |
Points are added up (maximum 380; rescaled if a measure is missing) and converted to stars: 1★ under 155, 2★ 155–204, 3★ 205–254, 4★ 255–319, 5★ 320–380 (Table 3). Turnover uses six quarters of data.
Automatic 1-star staffing ratings apply when a home does not submit staffing data, reports four or more days in a quarter with no RN hours while residents were present, or fails a CMS staffing data audit.
3. Quality measure (QM) rating
CMS uses 15 measures: 9 for long-stay residents and 6 for short-stay (rehab) residents. Most come from the resident assessments (MDS) the home completes; some come from Medicare claims. MDS measures use the four most recent quarters. Some measures are risk-adjusted for differences in residents.
- Ability to walk independently worsened
- Need for help with daily activities increased
- Pressure ulcers
- Catheter left in the bladder
- Urinary tract infection
- Falls with major injury
- Antipsychotic medication
- Hospitalizations per 1,000 resident days (claims)
- Outpatient emergency visits per 1,000 resident days (claims)
- At or above expected ability to care for themselves and move around at discharge
- New or worsened pressure ulcers/injuries
- Newly received an antipsychotic medication
- Re-hospitalized after admission (claims)
- Outpatient emergency visit (claims)
- Successful return to home and community (claims)
Each measure earns points by where the home falls in the national distribution. Some measures are worth up to 150 points and others up to 100, weighted by CMS for room to improve and clinical importance. Long-stay and short-stay totals are scaled to count equally. As of July 2026 (Table 5), an overall QM score of 1,494–2,300 earns 5 stars and 299–927 earns 1 star. CMS says it plans to raise these thresholds every six months by half of the increase in scores, so the bar rises over time.
4. How the overall rating is put together
- Start with the health inspection rating.
- Add one star if the staffing rating is 5 stars; subtract one star if it is 1 star.
- Add one star if the quality measure rating is 5 stars; subtract one star if it is 1 star.
The result cannot go above 5 or below 1. If the health inspection rating is 1 star, the overall rating can be raised by no more than one star. CMS does not assign fixed percentage weights; inspections are the foundation, and staffing and quality can move the overall rating by up to two stars. A home with no health inspection rating gets no overall rating, and homes in the Special Focus Facility program are not given star ratings at all.
The abuse icon and other icons
CMS shows an abuse icon when a home was cited for abuse (federal tags F600, F602 or F603) where residents were harmed (severity G or higher) on the most recent standard inspection or a complaint or infection-control inspection in the past 12 months, or when the home has repeat abuse citations at level D or higher in both the recent and the previous period. A home with the abuse icon has its health inspection rating capped at 2 stars, so its overall rating can be no higher than 4 stars. The icon is removed at the first monthly refresh after the home no longer meets the criteria.
Since September 2026, CMS also shows a trophy icon for “high-performing” homes, described in the guide as homes with fewer citations, higher staffing and other quality indicators (CMS memo QSO-26-14-NH).
Where this comes from
- CMS, “Design for Care Compare Nursing Home Five-Star Quality Rating System: Technical Users’ Guide”, September 2026.
- CMS, Five-Star Quality Rating System page, page last modified October 27, 2026.
Hospice measures
Hospices do not get a single overall star rating. CMS publishes a family caregiver survey star rating, plus several quality measures through the Hospice Quality Reporting Program (HQRP), which uses three data sources: the HOPE patient assessment (which replaced the older HIS starting October 1, 2025), Medicare claims, and the CAHPS® Hospice Survey (CMS HQRP Current Measures, last modified Nov 3, 2026).
Family caregiver survey star rating (CAHPS® Hospice Survey)
After a patient dies in hospice care, a family member or friend is mailed (or called) a survey about the care. Per the CAHPS Hospice Technical Notes for the August and November 2026 star ratings:
- Time period: the August and November 2026 ratings cover patients who died October 1, 2023 – September 30, 2025 (eight quarters). Star ratings are updated every six months.
- Minimum: a hospice needs at least 75 completed surveys in that period to receive a star rating.
- Top-box scores: each measure is the share of caregivers who chose the most positive answer (for example, “always” or “definitely yes”), adjusted for differences in patients (case mix) and in how the survey was given (mode).
- Measures used (during the transition to the revised survey, August 2026 to November 2027): communication with family, getting timely help, treating the patient with respect, emotional and spiritual support, help for pain and symptoms, and the hospice rating and willingness to recommend (these last two count one-half each). Getting hospice care training is left out during the transition.
- From scores to stars: CMS sets cut-offs using a statistical clustering method on recent data, gives a star rating for each measure, averages them, and rounds: under 1.5 → 1 star; 1.5 to under 2.5 → 2; 2.5 to under 3.5 → 3; 3.5 to under 4.5 → 4; 4.5 and up → 5.
- Survey change: a revised version of the survey began with patients who died in the second quarter of 2025, so newer and older results are not perfectly comparable.
Hospice Care Index (HCI)
The HCI is built entirely from Medicare claims. It looks at ten indicators of how care was delivered between admission and discharge. A hospice earns one point for each indicator where it meets CMS’s criterion, for a score from 0 to 10; each indicator counts equally. Hospices with fewer than 20 discharges over the two years of data get no score. The ten indicators are:
- Continuous home care or general inpatient care provided
- Gaps in skilled nursing visits
- Early live discharges
- Late live discharges
- Burdensome transitions type 1 (live discharge, hospital stay, hospice readmission)
- Burdensome transitions type 2 (live discharge, hospital stay, death in hospital)
- Medicare spending per patient
- Skilled nursing minutes per routine home care day
- Skilled nursing minutes on weekends
- Visits near death
CMS’s summary page does not list the exact criterion for each indicator; those are in CMS’s HQRP Quality Measure User’s Manual, which I have not reproduced here. On each hospice page, I show the hospice’s value for each indicator next to the state and national values.
Hospice Visits in the Last Days of Life (HVLDL)
From claims: the share of patients who got in-person visits from a registered nurse or medical social worker on at least two of the last three days of life. Patients are left out if they did not die in hospice care, were in hospice fewer than three days, or got continuous home care, respite or general inpatient care in their last three days.
Other reported measures
The Comprehensive Assessment at Admission measure (from the patient assessment) is the share of patients for whom the hospice completed all seven applicable care processes at admission: asking about beliefs and values (if desired), treatment preferences, pain screening, pain assessment, shortness-of-breath screening, shortness-of-breath treatment, and a bowel regimen for patients on opioids. Most hospices score near 100%, so it does little to separate one from another.
Where this comes from
- CAHPS® Hospice Survey, “Technical Notes for CAHPS Hospice Survey Star Ratings, August and November 2026”, 2026.
- CMS, Hospice Quality Reporting Program “Current Measures” page, page last modified November 3, 2026.
Data lag and update schedule
- Nursing homes: CMS refreshes Care Compare monthly; health inspection results are added as they reach CMS’s database, staffing ratings are calculated and posted quarterly, and quality measure ratings are updated quarterly, typically in January, April, July and October, per the Technical Users’ Guide (September 2026). Staffing data are reported a quarter at a time and are due 45 days after the quarter ends.
- Hospice: data are refreshed quarterly, but each release looks back over long periods. The family survey stars on this site cover deaths up to two years before they were published, and the Hospice Care Index uses two years of claims ending well before the release date.
- On this site: every rating and chart shows its measurement period and the CMS release date. See Sources and methods.
What the ratings don’t capture
CMS itself cautions that no rating system can address everything that matters, such as specialty care like rehabilitation or dementia care, or how easy it will be for family to visit, and that a home very close to family may be a better choice than a higher-rated one far away. CMS advises using the ratings together with other information, including a visit and your state long-term care ombudsman (CMS Five-Star page). From the bedside, here is what numbers can’t show you: how quickly the on-call nurse calls you back at 2 a.m.; whether the same nurse and aide come each week; how staff speak to a frightened person; whether a hospice will come to a small rural county quickly; and how a home feels at dinner time. Use the ratings to make a short list, then call and visit. My guide to reading the ratings has more.
Where this comes from
- CMS, “Design for Care Compare Nursing Home Five-Star Quality Rating System: Technical Users’ Guide”, September 2026.
- CMS, Five-Star Quality Rating System page, page last modified October 27, 2026.
- CAHPS® Hospice Survey, “Technical Notes for CAHPS Hospice Survey Star Ratings, August and November 2026”, 2026.
- CMS, Hospice Quality Reporting Program “Current Measures” page, page last modified November 3, 2026.