BRENNEN

Industries / Health / Healthcare / Dialysis

Quality data

Counting what happens to people on dialysis, clinic by clinic and country by country, and tying payment and public ratings to those counts.

7

Milestones achieved, from 1968 to 2024.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1968

    Japan counts every patient

    The Japanese Society for Dialysis Therapy began a yearly national survey in 1968, when it counted 251 dialysis patients, and has followed all patients as a cohort since 1983. Its survey for the end of 2023 went to 4,529 facilities: 98.7% returned the facility form and 94.6% the patient form, and it counted 343,508 people on dialysis.

    jsdt.or.jprrtjournal.biomedcentral.com

    Credit

    Japanese Society for Dialysis Therapy, after a 1966 pilot survey by Dr. Odaka

  2. 2003

    Same disease, different deaths

    The Dialysis Outcomes and Practice Patterns Study (DOPPS) followed representative samples of hemodialysis patients in seven countries from 1996. Among 16,720 patients, one-year death rates were 6.6% in Japan, 15.6% in Europe and 21.7% in the US; after allowing for age, sex, race and 25 other conditions, the US risk was 3.78 times Japan's.

    LessonDifferences between the patients explained only part of the gap between countries and clinics, which turned the study toward differences in practice.

    ebi.ac.ukebi.ac.uk

    Credit

    David A. Goodkin, Friedrich K. Port and colleagues (Journal of the American Society of Nephrology, 2003)

  3. 2010

    Clinic death rates released

    CMS's Dialysis Facility Compare website, online since 2001, showed each clinic's survival only as "better than expected", "worse than expected" or "as expected". After a two-year request under the Freedom of Information Act, CMS gave ProPublica clinic-by-clinic reports for 2002 to 2010 as its investigation of US dialysis was published in November 2010.

    LessonCMS had held the figures back on the ground that patients might misread them, by ProPublica's account, and released them only as the story neared publication.

    propublica.orgpmc.ncbi.nlm.nih.gov

    Credit

    Robin Fields, ProPublica

  4. 2011

    Pay tied to quality

    The ESRD Quality Incentive Program scores every Medicare dialysis clinic each year on a set of measures and cuts its Medicare payments by up to 2% if it falls short; each clinic must display its score certificate. CMS calls it the first program of its kind in Medicare, and it began scoring clinics in 2011.

    LessonReviewing 50 years of the kidney program in 2023, Erickson and colleagues wrote that substantial improvements in quality from dialysis pay-for-performance and related schemes "have not been observed".

    cms.govpmc.ncbi.nlm.nih.gov

    Credit

    US Congress and CMS

  5. 2015

    Stars for every clinic

    In January 2015 CMS added a one-to-five-star summary to Dialysis Facility Compare, built from nine of its 18 published measures and weighted by a statistical method called principal components analysis. The first version fixed how many clinics got each number of stars, like grading on a curve; critics said this hid improvement across the industry, and from October 2016 CMS let the spread change.

    LessonA rating can track where a clinic is: clinics in areas with more Black residents had lower odds of a high rating (adjusted odds ratio 0.55), by Abhijit Kshirsagar and colleagues.

    pmc.ncbi.nlm.nih.govebi.ac.uk

    Credit

    CMS, with the University of Michigan Kidney Epidemiology and Cost Center, which maintains the measures

  6. 2023

    Infections counted nationwide

    Dialysis clinics report bloodstream infections to the National Healthcare Safety Network (NHSN), run by the US Centers for Disease Control and Prevention (CDC). In 2020, 4,840 of the 7,097 reporting facilities logged 14,822 such infections, a third of them caused by Staphylococcus aureus. In seven areas with active surveillance these infections were 100 times more common in people on hemodialysis than in other adults, and in NHSN data they were six times more common with a central venous catheter (a tube in a large vein) than with a fistula (an artery joined to a vein in the arm).

    LessonThe CDC noted that clinic-reported NHSN counts run lower than its active surveillance finds.

    pmc.ncbi.nlm.nih.gov

    Credit

    Brian Rha, Isaac See and colleagues, CDC (MMWR, 2023)

  7. 2024

    Quality checks, country by country

    The International Society of Nephrology's Global Kidney Health Atlas surveyed 167 countries, home to 97.4% of the world's people, in 2022. In 98% of high-income countries more than three-quarters of dialysis centers regularly tested their water for bacteria, against 28% of low-income countries, and 81% of countries with hemodialysis could offer the standard schedule of three sessions a week.

    ebi.ac.uk

    Credit

    Udeme E. Ekrikpo, Aminu K. Bello and colleagues (Nephrology, 2024)

Watch for

  1. Hospital risk flags

    Renvio, which sells software for dialysis clinics, says that in an analysis of about 20,000 patients its AI model picked out a high-risk group in which nearly 87% were admitted to hospital within 60 days.

    Achieved whenthe model's accuracy, including how many patients it flags and how many admissions it misses, is published, and an outside evaluation finds fewer admissions in clinics that use it than in comparable clinics.

    renvio.com

    Claimed by

    Renvio, June 2026

Open

  1. 01

    Rebuild the star ratings openly

    Open-source code that recomputes CMS's dialysis clinic star ratings from the public Care Compare measure file and the published scoring tables, checks the result against the stars CMS posts for every clinic, and shows how stars would move under options now under review, such as rebaselining (resetting the cut-points to a new base year).

    The brief
    Credit · unclaimed
  2. 02

    Registry measures side by side

    A public crosswalk of ten common measures, such as the share of patients treated at home, the share starting dialysis with a catheter, survival and the transplant rate, across at least 15 national kidney registries, giving each registry's definition, denominator, year and latest value, and flagging where definitions differ.

    The brief
    Credit · unclaimed

Open 01 · the brief

Rebuild the star ratings openly

Credit · unclaimed

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

What to build

Open-source code that recomputes CMS's dialysis clinic star ratings from the public Care Compare measure file and the published scoring tables, checks the result against the stars CMS posts for every clinic, and shows how stars would move under options now under review, such as rebaselining (resetting the cut-points to a new base year).

Why it is still open

The method and scoring tables are public, the latest dated July 2026, and a technical expert panel is reviewing whether to rebaseline, but a GitHub search in November 2026 found only projects that chart or classify the published data, none that rebuilds the ratings.

dialysisdata.orgdialysisdata.org

Built with what exists

Python with pandas, or R; the dialysis facility files from CMS's Provider Data Catalog; the star rating technical notes and scoring tables on DialysisData.org.

  1. First step

    Download the latest dialysis facility file and recompute the domain scores for the clinics in one state.

  2. Done looks like

    The code reproduces CMS's star for at least 99% of rated clinics in two releases in a row, lists every mismatch with its cause, and publishes a comparison of stars under the current and one alternative baseline.

Open 02 · the brief

Registry measures side by side

Credit · unclaimed

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

What to build

A public crosswalk of ten common measures, such as the share of patients treated at home, the share starting dialysis with a catheter, survival and the transplant rate, across at least 15 national kidney registries, giving each registry's definition, denominator, year and latest value, and flagging where definitions differ.

Why it is still open

A 2022 review found 79 kidney failure registries in 77 countries whose data are analysed in "silos", and named poor interoperability among the barriers; the International Society of Nephrology's SharE-RR toolkit (2024) helps countries set up registries and maps them, but does not line up their measures.

ebi.ac.ukpmc.ncbi.nlm.nih.gov

Built with what exists

Registries' public annual reports (the US Renal Data System, the ERA Registry, ANZDATA, Japan's JSDT registry, the UK Renal Registry and others); pdftotext; a language model to draft the extraction, checked by hand; a CSV file with a DOI from Zenodo.

  1. First step

    Fill three measures for the US Renal Data System, the ERA Registry and the JSDT registry from their latest reports.

  2. Done looks like

    15 registries by 10 measures, every cell linked to a page and table number, and a second reader agreeing on at least 95% of a random fifth of the cells.