Birth Injury Marketing
Open data

Take the files. There is no form.

Everything behind the intelligence layer, as CSV. Free to reuse with attribution, including commercially. No registration, no email, no rate limit, and no approval rights over what you write with it.

Counties
3,144counties.csv · 183 KB
Hospitals
4,658hospitals.csv · 875 KB
States
51states.csv · 4 KB
Licence
CC BY 4.0attribution, that is all
Four files. Generated from the same derived data the charts render from, so a figure in a chart and the same figure in a download cannot drift apart.
FileRowsWhat is in it
counties.csv3,144Births 2021–2024, population, lawyer establishments, births per office, delivering hospitals, hospitals with status not reported
states.csv51The same rolled up, plus counties with no lawyer office and counties with births but no confirmed delivering hospital
hospitals.csv4,658Obstetric service status per facility, ownership, type, Birthing Friendly designation, and the three withheld outcome measures
outcome-audit.csv13Published versus withheld for every CMS obstetric measure, with the footnote code and its verbatim definition
README.txtConventions, the finding below, known limits, and the citation
How a withheld value is represented, and why. Four ways to encode a value the source did not publish; three of them silently produce a wrong answer downstream. The source published no value Write 0sums low, silently Write blankaverages past it, silently Impute itinvents the measurement Write WITHHELDyour tool errors A plausible number nobody can tell is wrong An error you notice The error is the correct outcomewhich is why the column is a string, not a number
Figure 1 — the convention is deliberately inconvenient. Every other encoding lets a consumer compute an average over data that does not exist and get a number back. Making the column a string means the mistake surfaces at the moment it is made rather than in a published chart.
Completeness of each file. Share of each dataset that is published versus withheld at source. published withheld or not availablecounties.csv3,144 rows3,144states.csv51 rows51hospitals.csv — service status4,658 rows3,687971hospitals.csv — outcomecolumns4,426 reporting4,426
Figure 2 — what you are actually getting, before you download it. Two files complete, one mostly complete with 971 facilities hatched, and three columns that are withheld in their entirety. Published here rather than discovered after you have built something on it. Sources: Census Bureau, Vintage 2024 county population estimates (components of change), released June 2025; Census Bureau, County Business Patterns 2023, NAICS 541110 Offices of Lawyers, released May 2025; CMS Care Compare, Maternal Health · Hospital, reporting period 1 Jan – 31 Dec 2024.

One convention, and it matters

WITHHELD is a string, not a blank

A value the collecting agency did not publish is written as the literal string WITHHELD. It is never an empty cell and never a zero.

If you sum() a column containing it, your tool will error. That is deliberate. Those values are absent, not low, and quietly averaging past them produces a number that looks real and is not. An error is the correct outcome; a plausible average is the dangerous one.

Before you use the outcome columns in hospitals.csv

All three are WITHHELD for every row, because for the 2024 reporting period CMS published no hospital-level obstetric outcome scores at all — zero of 4,426. The columns are present so the absence is explicit rather than something you infer from a missing header. The full audit, and the write-up.

What this data cannot support. The limits, in the same place as the download, because a limit in a footnote is a limit designed not to be read.
Do not use it toBecause
Infer where claims are more likelyNothing here measures litigation. Lawyer establishment density measures where lawyers have offices.
Judge a hospitalObstetric outcome scores are withheld in their entirety. Birthing Friendly is a structural attestation, not a result.
Rank counties by "legal need"NAICS 541110 does not distinguish practice areas, and counts locations rather than firms or attorneys.
Read a county with no delivering hospital as having no obstetric careBirths are counted by mother’s residence; delivery happens wherever the hospital is, frequently in the next county.
Correlate clinical outcomes with legal activityDifferent quantities, different collectors, different purposes. We never join them and neither should you.

Want a different cut?

By year, by geography, by facility type, joined to something of your own — ask and we will produce it, free, usually within the week. We will also tell you where the data will not support the question, which is generally the more useful half of the reply. Ask for a cut →

A limit we will not cross. Nothing on this page links a clinical outcome rate to a legal claim. Adverse-event rates measure medicine. Litigation measures a legal process. They are different quantities, they are collected by different agencies for different reasons, and correlating them would be the fastest way to publish something false. Where we present clinical data and legal data together, it is to describe a market — never to imply that one causes, predicts, or justifies the other.

Sources

  1. U.S. Census Bureau — County Population Totals and Components of Change, Vintage 2024 — births by county, 2021–2024
  2. U.S. Census Bureau — County Business Patterns, 2023 — NAICS 541110, Offices of Lawyers; establishment counts by county
  3. CMS Care Compare — Maternal Health · Hospital — PC-02, PC-07a, PC-07b, SM-7; 2024 reporting period
  4. CMS Care Compare — Hospital General Information — ownership, facility type, Birthing Friendly designation
  5. CMS Care Compare — Footnote Crosswalk — definitions for suppression codes
  6. U.S. Census Bureau — Cartographic Boundary Files, 2023 (1:20,000,000) — county and state polygons; projected to Albers equal-area at build time

Cite this analysis

Free to reproduce with attribution, including commercially. Charts may be embedded as published.

Birth Injury Marketing. “Births, obstetric delivery capacity and legal services capacity by U.S. county.” 24 August 2026. https://birthinjurymarketing.com/data/

Journalists and researchers: the underlying data is already published as CSV at /data/, free and without conditions — no form. Need a different cut, by state, county, hospital or year? Ask, and we will also tell you what the data cannot support.

The layer these come from → · Every join, and the three datasets we could not get → · Found an error? →

Authorship & review

Author
Kevin Schwaner, Founder; Chief Marketing Officer. Credentials.
Last reviewed
Corrections
Found an error? Tell us and it goes in the dated corrections log, whether or not it flatters us.