The register · live count
What the public has sent, counted in the open
Every row is a count of what was sent through this tool, with its N and its dates. None has been delivered to an agency yet.
How this is counted
- A thumb on a figure
- Bound to the published rowof the agency that published it
- Can be handed to that agency
- Answers
- One sender can send more than one answer, so a count of answers can be larger than the count of senders printed beside it.
- Sender
- A sender is one network on one day, counted once. It is never a verified person: a household on one connection is one sender, so this count runs low, never high.
…
survey answers
…
written interviews
…
thumbs on federal figures
…
reports of uncounted care
Loading…
How this page reads at scale: an illustration at 200 answers, not real data →
What these counts could change: the policy brief, by agency and fee schedule →
Site use (a separate counter)
Counts of actions, never of who took them. One number per day for each thing below, with no cookie, no address and nothing anyone typed. Our own tests and known bots are left out.
Reading the counts…
Which cost weighed most, ranked by those who carried it
loadingThinking about the whole time you were looking for a diagnosis, rank these from heaviest to lightest.
Corrections on published federal figures
loading
Each ledger line carries a thumb, bound to the exact published row its price came from.
- A thumb on a ledger linefits, or does not fit
- Bound to one published rowthe exact row the price came from
- Grouped by who published itthe body first, then the file the row is in
How this is counted
Enough thumbs on one row is something the body that published it can act on.
- Grouped by
- The body that published the row, then the file it is in.
- Read by place
- 52 of the 274 rows also carry a figure for each of the 109 Medicare localities, so a correction can be read against the figure for the place the person actually lives.
- Sent on
- Nothing here mails anyone automatically: we publish no agency inbox we have not verified. Every correction copies out as a citation block (the row, the file, the figure and the count) that a person at the publishing body can act on directly.
What that resists, and what it does not: how the count is kept honest →
Who published the figures each section is about
274 published figures: CMS 257 · AHRQ MEPS 12 · AHRQ HCUP 2 · BLS 2 · GSA 1
Each count on this page is published with its N, its denominator, its dates and where the answers came from. Nothing identifies anyone.
Corrections
- Counts
- A thumb on a published federal figure, bound to one row of one published file.
- Sources
- Of the 274 rows this tool prices, 257 were published by CMS, 12 come from AHRQ's MEPS survey, 2 from AHRQ's HCUP, 2 from the Bureau of Labor Statistics and 1 from the General Services Administration. Two of the MEPS rows are a peer-reviewed analysis of that survey rather than an AHRQ publication, and every screen that prints them says so.
- Grouped
- Under the body that published it.
Care that produced zero rows
- Counts
- An event no dataset recorded.
- About
- What MEPS and HCUP (AHRQ) and Medicare claims (CMS) do not record at all.
- Corrects
- Nothing: no published figure exists to correct. This is the count of the absence.
The burden ranking
- Counts
- Which cost weighed most.
- Why
- It is the stated basis for weighing kinds of burden, the question NIH health economists put to every cost-of-illness estimate.
- Published as
- A distribution with its N, with no weighting scheme of ours underneath.
Care that produced zero rows in federal cost data
loading
The count of care that leaves no row, across 0 reports.
- Recorded in MEPS, HCUP and CMS claimsCare that was delivered and billed.
- Leaves no rowA visit that never happened, a test that was refused, months lost waiting.
Written interviews
loading
- A written interviewHeld privately, never served.
- What changes because of it is published hereAs “someone said this, so we changed that”.
- Quotes
- Only where the writer chose to be quoted.
Loading…
Take the data
De-identified files an agency, a researcher or a committee staffer can use without asking us. No key, no sign-in.
Burden rankings
One row per answer: the five burdens in the order they were ranked, with the date and the channel.
Corrections, ready to route
Every thumb joined to the federal file behind the figure.
- Each row
- The file’s name, URL and SHA-256, the line the figure was read from, the audit verdict, the counts and the fit rate.
- The JSON
- A DCAT distribution with its own column contract, the price-table version, the audit date, and the CY2024 charge an uninsured person is billed against where CMS publishes one.
Uncounted care
Care that produced no row in any federal cost file, counted by category with its denominator.
Data dictionary
Every field in every export above: what it means, its type, and how it is counted.
Everything on this page
The whole register as one document, the same one this page reads.
How every figure is made
The file, the row and the arithmetic behind each of the 274 published figures.
Method: every figure on this page is a count of responses exactly as entered.
- No weighting scheme of ours
- No imputation
- No extrapolation to a population
How this is counted
- Sample
- Self-selected and recruited through the channels listed. It describes the answers received and nothing wider.
- Every number
- Published with a raw N, its denominator and a date, so it can be read as what it is.
- Withheld when
- A cell holds fewer than 11 responses in a named place. It is withheld and counted as withheld.
- Why 11
- That threshold is ours, not an obligation we are under. 11 is the floor CMS puts on researchers using its own beneficiary files: under the CMS cell-size suppression policy no cell holding 1 to 10 beneficiaries may be reported at all.
- Routing
- The corrections file carries the federal file, its SHA-256 and the line each figure was read from, so a correction can be routed to the office that published the number.