BRENNEN

Industries / Health / Healthcare / Hospice

Documentation

Recording each person's assessments, certification and care, and the payment rules, quality reports and inspections built on those records.

7

Milestones achieved, from 2006 to 2025.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 2006

    Sweden's register of deaths

    For the Swedish Register of Palliative Care, the doctor or nurse responsible for a person in their last week of life answers an online questionnaire about that care after the death. It became a National Quality Register in 2006. In 2010 it recorded more than 30,000 deaths in nursing homes, hospital wards, palliative units and private homes, 34% of all deaths in Sweden and 58% of cancer deaths.

    ebi.ac.uk

    Credit

    S. Lundström, B. Axelsson, P.-A. Heedman, G. Fransson and C. J. Fürst (Palliative Medicine, 2012)

  2. 2011

    Doctors must show reasons

    US rules require the certifying doctor to write, personally, a brief narrative of the clinical findings that support a life expectancy of six months or less. From 1 January 2011 a hospice doctor or nurse practitioner must also see the patient face to face within 30 days before the third benefit period. The OIG later found that in 14% of sampled 2012 inpatient stays the doctor had not met such requirements and seemed to have little part in the decision.

    LessonA signature rule did not by itself make sure a doctor was involved.

    law.cornell.eduoig.hhs.gov

    Credit

    CMS (42 CFR 418.22); HHS Office of Inspector General

  3. 2015

    Australia measures outcomes

    In Australia's Palliative Care Outcomes Collaboration (PCOC), clinicians rate each patient's symptoms and problems with standard scales during care, and services get benchmarked reports every six months. Across 30 services and 19,747 patients, outcomes improved over three years in every area measured except pain, and the gap between services narrowed.

    LessonThe authors call this the first national evidence that routine data collection and feedback improved palliative care outcomes at service level; pain did not improve.

    ebi.ac.ukuow.edu.au

    Credit

    D. C. Currow, K. Eagar and colleagues, PCOC (Supportive Care in Cancer, 2015)

  4. 2016

    Two rates for home care

    Medicare pays a hospice a set amount for each day a person is enrolled, whether or not anyone visits that day, at one of four levels of care. Routine home care makes up nearly 99% of hospice days. Because hospices give more care at the start and end of a stay, from January 2016 Medicare split that rate in two: in fiscal 2025, $225 a day for days 1 to 60 and $177 after that.

    medpac.gov

    Credit

    CMS; figures from MedPAC

  5. 2022

    A hospice boom audited

    The California State Auditor found that Los Angeles County's hospice agencies had grown by 1,500% since 2010, to 1,841 of the state's 2,836. In 2021 an agency there averaged 4.6 patients a day, against 56.3 in the rest of the state, and the auditor found signs of "a large-scale, targeted effort to defraud Medicare". The legislature had already stopped new hospice licences from 1 January 2022.

    LessonToo many licences for the number of people dying was itself a warning sign.

    information.auditor.ca.govpropublica.org

    Credit

    California State Auditor (report 2021-123, 29 March 2022); the moratorium followed a Los Angeles Times investigation, as ProPublica reports

  6. 2023

    Oversight in four states

    In July 2023 CMS put new hospices in Arizona, California, Nevada and Texas under a period of enhanced oversight, reviewing their claims before paying. It also made surprise visits to every Medicare hospice to check each was operating at its listed address, over 7,000 of them by mid-August 2023. By CMS's count, it had revoked the Medicare enrollment of 122 of 668 hospices reviewed by June 2025 and of over 200 by April 2026. It added Georgia and Ohio in December 2025, and on 13 May 2026 stopped all new hospice enrollment nationwide for six months.

    cms.goveverycrsreport.compublic-inspection.federalregister.gov

    Credit

    CMS; counts as reported by MedPAC and the Congressional Research Service

  7. 2025

    A new assessment replaces HIS

    From 2014 US hospices recorded the Hospice Item Set (HIS), standard data items, at admission and discharge. MedPAC found its composite measure had "topped out": nearly every hospice scored so high that it could no longer tell them apart. On 1 October 2025 CMS replaced it with HOPE (Hospice Outcomes and Patient Evaluation), which adds update visits during the first 30 days of care and more detail on symptoms.

    LessonA measure everyone passes stops measuring.

    cms.govcms.govmedpac.gov

    Credit

    CMS; MedPAC

Watch for

  1. A score for unusual hospices

    CMS's Service and Spending Variation Index (SSVI) scores every hospice on nine claims-based measures of how it uses care and how much Medicare spends outside hospice for its patients, and publishes the scores for fiscal years 2024 and 2025 by provider. CMS says a high score "can signal potentially inappropriate utilization or concerns with quality of care or compliance".

    Achieved whenan analysis outside CMS shows that high-scoring hospices are more often found, on audit or inspection, to have improper claims or poor care.

    cms.gov

    Claimed by

    CMS, July 2026 (FY 2027 final rule)

  2. AI scribes across a charity

    Sue Ryder, a UK charity, signed a five-year deal with Heidi to give more than 630 palliative, end-of-life and bereavement clinicians in its seven services an ambient AI scribe, software that listens to a consultation and drafts notes, letters and forms. It plans to assess the effect on documentation workload and on how consistent records are.

    Achieved whena published evaluation, ideally by an outside party, reports measured time on records before and after, and the error rate in drafted notes.

    itbrief.co.uk

    Claimed by

    Sue Ryder and Heidi, February 2026

Open

  1. 01

    Hospice inspection findings, counted

    A public dataset of every US hospice inspection finding that CMS has posted, counted by provider, state, year, deficiency tag and the kind of failure (for example, no plan of care, aide visits not supervised, pain not assessed), with a link back to each CMS document. It publishes counts and categories about hospices, not the text about patients.

    The brief
    Credit · unclaimed
  2. 02

    Test CMS's new outlier score

    An outside check of the SSVI. Join each hospice's published score with what was later found about it: inspection findings on QCOR, family survey (CAHPS) scores and, where public, the loss of its Medicare enrollment. Publish whether high scores predict problems.

    The brief
    Credit · unclaimed

Open 01 · the brief

Hospice inspection findings, counted

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public dataset of every US hospice inspection finding that CMS has posted, counted by provider, state, year, deficiency tag and the kind of failure (for example, no plan of care, aide visits not supervised, pain not assessed), with a link back to each CMS document. It publishes counts and categories about hospices, not the text about patients.

Why it is still open

Since October 2021, CMS's QCOR site has posted hospice statements of deficiencies, the forms on which inspectors write what they found, now through June 2026. They come as one PDF per finding with no bulk download, and the site filters only by provider, survey type and tag. The Association of Health Care Journalists runs a searchable archive of inspection reports for hospitals only; a 2021 student project on GitHub predates the hospice postings, and searches found no dataset or study built from them. The OIG's 2019 review, which found deficiencies at over 80% of surveyed hospices, analysed CMS data from 2012 to 2016 and read the full reports for only a sample of 50 serious deficiencies.

qcor.cms.govhealthjournalism.orgoig.hhs.gov

Built with what exists

a script or a browser agent such as Browser Use 1.0 to page through QCOR and save the PDFs (see the Note "Deleting 80% to start over"); pdftotext; a model that answers many fixed-choice questions about one document in a single request, such as Gemini Decision 1, to sort each finding into categories (see the Note "A thousand questions, one pass"); CMS's public hospice ownership file to group hospices by owner.

  1. First step

    Save one state's 2025 findings, sort 100 by hand into a first list of categories, and compare the model's sorting.

  2. Done looks like

    Every posted finding from October 2021 to June 2026 counted. On a sample of 300 findings sorted by two people, the automatic categories agree with them at least 90% of the time. Data and code are public and refreshed when QCOR adds a quarter.

Open 02 · the brief

Test CMS's new outlier score

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An outside check of the SSVI. Join each hospice's published score with what was later found about it: inspection findings on QCOR, family survey (CAHPS) scores and, where public, the loss of its Medicare enrollment. Publish whether high scores predict problems.

Why it is still open

CMS finalized the index in July 2026 and publishes hospice-level scores for fiscal years 2024 and 2025 as a spreadsheet. The California Association for Health Services at Home wrote that it has "fundamental validity problems that make it unsuitable as an enforcement tool" and asked CMS to validate it against compliance data first. Searches found no published validation.

cms.govcahsah.org

Built with what exists

CMS's SSVI spreadsheet; QCOR inspection findings; CAHPS Hospice scores from CMS's Provider Data Catalog; Python with pandas and DuckDB.

  1. First step

    Download the SSVI file and match the FY 2024 scores to QCOR findings from one large state.

  2. Done looks like

    Every hospice with an SSVI score matched to its later inspection and survey results, with method and code public, and a stated answer, with its margin of error, to whether hospices in the top tenth of scores had more serious findings than the rest.