The SSVI, Sorted: What 6,673 Scores Show About Size, Ownership, and Noise

By now most hospice leaders have seen their number. The SSVI sits in a public CMS file next to each hospice's name, a single score between 0 and 16, and the first question every team asks is the same one: is ours bad? That question has no answer on its own. A score only means something against the field it came from, so we took the full field, all 6,673 hospices CMS scored for FY2025, and sorted it by the two things the score never mentions: how big the hospice is and who owns it.

We wrote in September about what the score is made of: half of it is one unadjusted spending total, parts of it flag hospices by construction, and the median is 6. This piece is the analysis underneath that argument, run on CMS's final rule data.

Sort by size and the score climbs on its own

There is an obstacle to running this analysis at all. CMS reports each hospice's size in only four coarse categories of hospice days per year, and the largest category is open-ended: everything above roughly 68 average daily census sits in one row, so a 70-census hospice and one at 1,100 are indistinguishable. That buries exactly the range where most mid-size and large hospices operate.

The file itself offers a way around the cap. CMS publishes each hospice's total non-hospice spending and the same spending per day of election, and dividing one by the other returns the hospice's actual days of hospice care for the year. That reconstruction reaches 6,528 of the 6,673 scored hospices, 98% of the field, and everywhere CMS's own coarse category can check it, the estimate lands inside the category. The 145 hospices it cannot reach reported zero or suppressed non-hospice spending, and their median total SSVI is 2.

Rebuild size that way, sort the field into census bands hospices would actually recognize, and the pattern is unmistakable.

| Average daily census | Hospices | Median spending score | Scoring 7–8 on spending | Median utilization score | Median total SSVI | |---|---|---|---|---|---| | Under 25 | 2,947 | 2 | 8% | 3 | 5 | | 25–50 | 1,311 | 5 | 18% | 2 | 7 | | 50–100 | 1,130 | 6 | 29% | 1 | 7 | | 100–300 | 947 | 7 | 66% | 1 | 8 | | 300 and up | 193 | 8 | 98% | 0 | 8 |

The spending half of the index is a raw dollar total sorted into eight bands, with the top band starting at $538,406 of non-hospice spending in a year. Nothing in it is per patient or per day, so more patients means more total dollars, and more total dollars means a higher band. The consequence shows up at the top of the table: among hospices above 300 census, 98% score a 7 or 8 on the spending half, and the median spending score is 8, the maximum the measure can give. Above roughly 100 census, that half of the SSVI stops behaving like a variation measure and starts behaving like a census disclosure.

The utilization half leans the other way, with a median of 3 in the smallest band falling to 0 in the largest, largely because several of its measures follow almost automatically from a small census. The most-tripped flag in the entire index, providing no continuous home care and no general inpatient care in a year, catches 45% of all hospices and is close to automatic for a small hospice that never had a patient needing either level of care.

The two halves pull against each other and they do not cancel. The median total climbs from 5 to 8 as census rises. A hospice in the 100 to 300 census range sitting on an 8 is not an outlier. It is the median hospice of its size, and a quarter of hospices that size sit at 9 or above.

The reconstruction carries two honest caveats. Dividing a rounded per-day figure into a total leaves a small error on any individual hospice's days, which is fine for sorting the field into bands and not a basis for precision claims about one hospice. And census here means Medicare hospice days divided by 365, so a hospice with meaningful non-Medicare census sits somewhat lower in the table than its true size.

The ownership gap the score cannot see

Ownership is where the analysis turned up the finding we did not expect.

CMS's own commentary in the final rule reports that for-profit hospices run roughly 167% higher non-hospice spending per patient day than nonprofits. That is CMS's number, from CMS's own per-day analysis, and you would expect it to show up in the spending score. It does not.

| Ownership | Hospices | Median spending score | Median utilization score | Median total SSVI | |---|---|---|---|---| | For-profit | 5,172 | 5 | 2 | 7 | | Nonprofit | 1,050 | 5 | 1 | 6 | | Government | 128 | 2 | 1 | 4 |

(The remaining 323 scored hospices are listed as unknown or other ownership in CMS's file and are left out of the table.)

The median spending score is identical, 5 and 5. The banding is built on total dollars, and nonprofit hospices are larger on average, so their lower per-day spending and their higher census wash out in the total. The entire one-point difference in median total SSVI between for-profit and nonprofit hospices comes from the utilization half.

We read that as a methodological finding, not a judgment on either group. The per-day gap CMS describes is real and worth understanding. The score CMS built simply cannot display it: a banding scheme on unadjusted totals erases the very variation CMS's own analysis surfaced. If the index exists to find outliers against their peers, that is a design limit worth knowing before anyone treats the spending half as a conduct measure.

Nearly one in five hospices swung three points in a year

One more property matters before reading any single score. Among the 6,319 hospices scored in both FY2024 and FY2025, with no change in methodology, 38% rose, 33% fell, and 18% moved three or more points on the sixteen-point scale. A one-year score is a snapshot of a noisy measure, which is why the two-year pair tells you more than either year alone.

How to read your own number

The analysis suggests reading your own score in three steps before reacting to it.

After those three steps, what remains is the part of the number a hospice actually controls: the coverage and relatedness determinations inside the spending dollars. Size explains the level, ownership mostly washes out, and noise explains much of the movement. The determinations are what is left, and they are the one input that is yours.

That is the part MerlinRx works on, surfacing the coverage decision where the care team makes it, so what gets billed outside the benefit reflects a deliberate call rather than a gap the hospice never saw.

If you want this read done on your own numbers, we offer an independent review of your medication coverage process: one conversation and a short write-up of what looks strong and what could be tighter, which you keep either way.

Schedule an independent review at merlinrx.info/schedule.