Two of the nine long-stay measures in Five-Star don’t come from the MDS at all. They come from Medicare claims: the number of hospitalizations, and the number of outpatient emergency department visits, per 1,000 days your long-stay residents spent in the building. Each is worth up to 150 points, as much as any quality measure in the rating.
They share a denominator, a one-year window and a risk-adjustment method, so they are explained together here.
Who is counted
Only residents with traditional Medicare, Parts A and B. A resident is left out entirely if they aren’t a Medicare beneficiary, or if they were in a Medicare Advantage plan for any part of the stay. Medicaid-only residents and Medicare Advantage residents do not appear in either measure.
A resident becomes long-stay after 100 cumulative days in the building, counted across stays as long as there is no gap of 30 days in a row in the community or another institution. A resident is also excluded if there is no quarterly or comprehensive MDS assessment in the 120 days before the day they became long-stay, or if claims or MDS data the measure needs are missing.
The denominator: long-stay days
The measure counts days, not residents:
“A long-stay day is any day after a resident’s one-hundredth cumulative day in the nursing home or the beginning of the 12-month target period (whichever is later) and until the day of discharge, the day of death, or the end of the 12-month target period (whichever is earlier).”
Days the resident spends in hospice are taken out, and so are days they are out of the building for any reason, including in a hospital. Stays are built from the entry and discharge dates on your MDS records, so accurate entry and discharge tracking records matter here too.
What counts: hospitalizations
“The numerator for the measure is the number of admissions to an acute care or critical access hospital, for an inpatient stay or outpatient observation stay, occurring while the individual is a long-term nursing home resident.”
- Observation stays count, whatever the diagnosis. Sending a resident out for observation counts the same as an admission.
- Planned admissions don’t count. CMS’s Planned Readmissions Algorithm sorts inpatient admissions into planned and unplanned, using the principal diagnosis and procedure codes on the hospital claim.
- It counts events, not people. A resident sent out three times in the year adds three.
- Admissions while the resident is enrolled in hospice are not counted.
What counts: outpatient ED visits
“The numerator for the measure is the number of visits to an emergency department that did not result in an inpatient hospital stay or outpatient observation stay, occurring while the individual is a long-term nursing home resident.”
This is the treat-and-return visit: the resident goes to the ER and comes back the same way. Any diagnosis counts. An ER visit that turns into an observation stay or an admission is not counted here. It is counted once, in the hospitalization measure, unless the admission was a planned one. Visits while the resident is in hospice are not counted.
How the rate works. CMS’s own example: a building with 75 long-stay residents who spent 27,375 eligible days there in the year, with 28 unplanned hospitalizations and 7 observation stays. That is 35 events over 27.375 thousand days, a rate of 1.28 per 1,000 long-stay days, or about 3 residents sent to the hospital each month.
See your building’s hospitalization and ED visit rates, free.
Risk adjustment
Both measures are risk adjusted, which is why the MDS still matters. CMS predicts how many hospitalizations or ED visits each resident would be expected to have, from their age, sex, race and ethnicity, hospital stays in the year before, a comorbidity index built from hospital claims, and a set of MDS items. The building’s rate is its observed rate divided by its expected rate, times the national rate.
“The data for the risk adjustment model is derived from Medicare inpatient claims data prior to the day the resident became a long-stay resident (i.e., after 100 cumulative days in the facility) and from the most recent quarterly or comprehensive MDS assessment within 120 days prior to the day the resident became a long-stay resident.”
So one assessment per resident, the last quarterly or comprehensive one before day 101, sets their expected risk. The items used include end-stage prognosis (J1400), oxygen therapy, IV medications, hospice care, internal bleeding, shortness of breath, and diagnoses such as septicemia, stroke, Parkinson’s disease and respiratory failure. The lists differ slightly between the two measures. In July 2025 CMS added Section GG functional items to both models.
The practical point is accuracy, not volume. If that assessment leaves out a condition the resident really has, the building is compared as if the resident were healthier than they are.
How Five-Star scores them
Both are 150-point measures, split into ten bands on the national distribution, and each uses a full year of claims rather than four separate quarters. A building needs at least 20 stays in the year for the measure to be reported. Lower is better:
| Points | Hospitalizations per 1,000 days | ED visits per 1,000 days |
|---|---|---|
| 150 | 0 to 0.7179 | 0 to 0.4741 |
| 135 | 0.7180 to 0.9433 | 0.4742 to 0.6661 |
| 120 | 0.9434 to 1.1024 | 0.6662 to 0.8288 |
| 105 | 1.1025 to 1.2549 | 0.8289 to 0.9853 |
| 90 | 1.2550 to 1.4058 | 0.9854 to 1.1590 |
| 75 | 1.4059 to 1.5573 | 1.1591 to 1.3672 |
| 60 | 1.5574 to 1.7184 | 1.3673 to 1.6026 |
| 45 | 1.7185 to 1.9283 | 1.6027 to 1.9055 |
| 30 | 1.9284 to 2.2685 | 1.9056 to 2.4707 |
| 15 | 2.2686 and up | 2.4708 and up |
The cut points were last changed in July 2025, when the risk-adjustment models were updated, so that the spread of points across buildings stayed the same.
Because these come from claims, you can’t see them in your own MDS, and each rate is a full year of transfers that have already happened. SuperQM shows your published rates, and the points they earn, alongside your other long-stay measures.
