See what a report looks like.

What is actually in a report

A single-facility report runs to twelve numbered sections, ordered the way someone who already knows the hospice would walk you through it. Headline first, then the thing most likely to explain the headline, and finally everything that would change how you read either.

00 · At a glance

One screen. The live-discharge rate, the Hospice Care Index score, the Medicare spending figure, the inspection count. Each row carries its comparison point and the section it comes from, so no figure here appears without its comparison or a route to the working behind it.

01 · Live-discharge rate

The measure the rest of the report exists to put in context: how often patients left this hospice alive instead of dying in its care. Shown against the national rate for the same CMS release, and against hospices of a similar discharge volume, because a small hospice and a large one are not really comparable on this.

02 · Medicare dollars at this facility

Total Medicare hospice payments and payment per beneficiary. This tells you how big the hospice is and nothing else. A large hospice bills more because it is large.

03 · Discharge-pattern components

The four CMS indicators that share the live-discharge denominator — early discharges, late discharges, and the two burdensome-transition measures. This is where a rate stops being one number and starts having a shape.

04 · Hospice Care Index

CMS's own ten-part composite, shown as CMS publishes it, with the national comparison labeled according to whether it is CMS's published figure or one we computed.

05 · Ownership & structure

Who owns the hospice, the legal entity Medicare actually enrolled (often nothing like the name on the front door), and any recorded change of control. A hospice changing hands is worth knowing about. On its own, it establishes nothing.

06 · Case mix & scale

Diagnosis and care-setting mix, and average daily census. Case mix is a possible explanation for a high live-discharge rate: dementia and debility trajectories outlast a six-month prognosis far more often than cancer does. This section is here so you can check that explanation, instead of assuming it either way.

07 · Family experience (CAHPS)

What bereaved families said. This is included with an important limit stated in the report itself: the CAHPS survey samples the families of patients who died, so live-discharged patients are outside the sampling frame entirely. Good scores here neither confirm nor refute anything in section 01.

08 · Inspection record

Every CMS Form 2567 survey and citation, by severity and date, against a national benchmark computed per survey. Count citations raw and a hospice inspected three times looks worse than one inspected once, purely for having been visited more.

09–10 · Over time

The same measures across consecutive CMS releases, so you can ask which way a hospice is moving and not just where it sits. One caveat, stated in the report itself: CMS's windows overlap, so consecutive releases share underlying data. Four releases are not four independent years.

11 · Methodology & provenance

Where every figure came from, which CMS file and release, and the limits section below.

Why each number says where it came from

Every figure in every report carries one of two labels, right next to the number:

• CMS-PUBLISHED — CMS calculated this and published it. We reproduced it and changed nothing.
• PRODUCT-COMPUTED — we calculated this from CMS's published data. It is arithmetic on public inputs, but the arithmetic is ours. If you disagree with the method, you have what you need to discount it.

That distinction is easy to let slide, and letting it slide would work in our favor: a figure we computed would borrow the authority of the agency's name at the top of the page. So we keep them apart. The moment a report is challenged, the first question anyone asks is which numbers came from the government and which came from us. You want to answer that out of the document in front of you, not from an email chain weeks later.

There is a third state, and it gets the most care of any of them: NOT ESTABLISHED. When CMS suppresses a figure, or a hospice is simply absent from a file, the report says so in those words. No zero gets printed. No gap is left sitting there for you to fill in with an assumption. A hospice with no published CAHPS score has not scored badly. It has not scored.

Why the facility counts differ between sections

Read two sections and you will notice they were computed over different numbers of hospices. That is deliberate, and the report says so where it happens.

Each comparison counts the hospices that actually had the figure being described. The live-discharge quartiles count hospices where CMS published both discharge denominators. The survey benchmark counts every hospice with a valid CCN and at least one survey in the inspection export, which is a different file on a different release schedule. Neither total is the true one, and neither corrects the other. A hospice missing from one is missing from that measure and no other. Every figure therefore names the set it was computed over, which is what lets two counts sit side by side without either of them being wrong.

The reproducibility manifest

At the foot of every report is a block you will probably never need. It records the exact CMS release date behind each file, a checksum for each one, the code version, the template version, and the moment the document was generated.

It is there for the day a report gets challenged, or two people discover they are holding copies that disagree. That block settles it. Anyone can point at a URL and call a figure sourced; the manifest pins the specific published data as of a specific date, which is what lets the same report be rebuilt and compared line by line. Hospice data is republished constantly. A figure that was right in May may not be the figure you get in November, and if neither of you knows which release you are holding, the disagreement has nowhere to go.

What the report will not do

Near the front of every report is a list, written as plain statements, of what the report does not establish. It is set out as a list rather than folded into the prose. It says that a high live-discharge rate is not evidence of misconduct, that we have not audited this hospice's clinical decisions, and that a rate cannot be broken down by reason because CMS does not publish that split per hospice.

It comes first on purpose. Stating the limits at the front means that every reader meets them before they see the figures.

Along the same lines: some things we decline to compute at all. We do not produce market concentration measures from service-area data, because the underlying file records where a hospice served at least one patient and carries no patient counts. It cannot support a market-share claim, so we do not make one look possible.

What a real report adds

The samples are complete in structure and fabricated in content. A real report is the same document with live CMS data for a named subject: the actual figures, the actual inspection history, the actual ownership record, and a manifest pinning the specific releases used. Reports come as a paginated PDF suitable for attaching to a filing.

Pricing covers the three report types and turnaround. The methodology sets out every definition in detail, including the ones we think are contestable. The notes apply the same method to public questions, with the data published alongside so anyone can recompute it.