Every field carries the date it was last confirmed.
Most provider databases are sold on a single number and a promise. This page is the other thing: what a record actually holds, how current each part of it is, how a work email earns the right to be shown to you, and what we do not hold at all. Every count on this page is read from the same API a customer calls.
Seven kinds of fact about one person, reconciled onto one record.
Each group is refreshed on its own schedule, and each field carries the date it was last confirmed - so a record is never uniformly “current” or “stale”, and you can see which parts of it to lean on.
| Group | What it holds | Refreshed |
|---|---|---|
| Identity | Name and credentials as filed — the whole name, and the parts it was filed in: prefix, first, middle, last and suffix for a person, its own name for an organisation. The ten-digit identifier, whether the record is a person or an organisation, whether it is active, and the date the identifier was first issued. | Weekly |
| Specialty | Primary and secondary taxonomies, the classification and specialisation each one sits under, and the state each was filed against. Asking for a classification matches every specialty beneath it. | Weekly |
| Practice location | Practice and mailing address, city, state and ZIP, practice phone and fax, and the additional locations an organisation files under the same identifier. | Weekly |
| Affiliations | The employer or group a provider bills under, billing reassignments, and the organisation identifier where the employer has one of its own. | Monthly |
| Licences | Licence number and state as filed against each taxonomy on the record. | Monthly |
| Standing | Federal programme exclusions with the dates they began, and opt-out and revocation flags - so you can see, before you spend anything, whether a provider can be billed for at all. | Monthly |
| Work email | A work email where one has been confirmed, carrying the tier it was confirmed at and the date. Shown to signed-in customers, one provider at a time, never on a public page. | Continuous |
These are groups, not fields. Every individual field - what it means, what it is used for, how complete it is, what it costs, and its name in the API, in an export column and in the append tool - is on the field reference.
A provider's home address and personal mobile are a separate, licensed product: available under contract only, matched to the named provider, and never included in an export or a list. They are revealed one provider at a time, counted, and logged.
What a provider bills, and what they prescribe.
The rest of this page is who a provider is. This part is what they do: the procedures they billed Medicare for, code by code, and the drugs they prescribed under Part D, drug by drug. You can filter on either without knowing a code — search “total knee replacement” or “Eliquis” — and read the detail for a provider you care about. The limits are below, in full, because they decide whether this is any use to you.
| What it holds | Year | Clinicians with a published row |
|---|---|---|
| Procedures billedEvery procedure code a provider billed, at each place of service, with the volume and the published average charge, allowed amount and payment. | 2024 | 1,144,245 of 7,492,373 15% of active individual providers |
| PrescribingEvery drug a prescriber wrote under Part D, by brand and by generic name, with claims, beneficiaries and drug cost as published. | 2024 | 1,135,392 of 7,492,373 15% of active individual providers |
Both counts were measured on 2026-10-05. They are counts of providers with at least one published row, not estimates, and they are read from the platform's own measurement rather than typed onto this page. Counting distinct providers across ten million procedure lines is not something a page load can do while you wait, which is why they carry a date instead of claiming to be live.
Read this before you price it in
- This is published Medicare data for 2024 — fee-for-service only. It is not all-payor claims: a provider's commercial, Medicaid and Medicare Advantage volume is not in it.
- It covers about one in seven clinicians. 1,144,245 of 7,492,373 active individual providers have a published procedure line and 1,135,392 have a published prescribing row, measured 2026-10-05. The other six in seven are blank on these columns, and some of your own list will be too — though the share is far higher in the specialties that bill Medicare Part B for a living, and it is published per specialty rather than left as a national average.
- A provider with nothing here reads "No published Medicare volume", and that is one phrase because it has to be: low-volume providers are removed from the published file before we ever receive it, so it covers a provider with no Medicare volume, one below the publication threshold, and one who does not bill Medicare at all. Nothing in the published data separates them, and we will not pretend to tell them apart.
- One year is held and no history yet, so nothing here supports a trend: we cannot tell you a procedure count is rising or that somebody recently started performing one.
- Every figure is an aggregate published about the provider. There is no patient-level data here, in any form.
Through the API, one published procedure line and one published prescribing line each count as one row against your plan's row allowance — the same allowance an export spends — so thirty procedure lines for one surgeon are thirty rows. Asking about a provider whose Medicare volume was never published returns nothing and costs no rows, and the procedure and drug pickers cost no rows at all. Reading the detail on screen in the app is included, like search, and filtering a search on a procedure or a drug costs nothing beyond the providers it returns. In a file it is priced differently, because a file carries the summary and not the lines: the claims columns widen a row and never add one, so an export or an append costs exactly the same rows with them as without. How it compares with an all-payor claims vendor is set out on why NPI Mark; the individual fields, and what each one is for, are on the field reference.
The national share is not your specialty's share.
About one in seven clinicians has a published procedure line, and that average carries every profession that never bills Medicare Part B at all — dentists, most therapists, practices that see no Medicare patients. Measured per specialty, over the same population and in the same pass, the picture is different enough to matter. These are the largest specialties we publish a list for, highest share first. Each one links to its own list page, where the same figure appears beside the contact coverage for that segment.
| Specialty | Clinicians | With a procedure line | With prescribing |
|---|---|---|---|
| Emergency medicine physicians | 76,255 | 62%47,605 clinicians | 52%39,632 clinicians |
| Internal medicine physicians | 309,091 | 60%186,299 clinicians | 64%196,934 clinicians |
| Nurse Anesthetist, Certified Registered | 76,235 | 50%38,331 clinicians | under 1%142 clinicians |
| Family medicine physicians | 166,934 | 49%82,549 clinicians | 66%110,716 clinicians |
| Physician assistants | 223,117 | 45%99,538 clinicians | 50%112,480 clinicians |
| Nurse practitioners | 524,962 | 35%184,256 clinicians | 43%226,577 clinicians |
| Psychiatry & Neurology | 97,937 | 35%34,042 clinicians | 49%47,586 clinicians |
| Chiropractors | 104,387 | 30%31,166 clinicians | under 1%16 clinicians |
| Physical therapists | 289,848 | 26%75,727 clinicians | under 1%9 clinicians |
| Occupational therapists | 159,711 | 7%10,899 clinicians | under 1%2 clinicians |
| Psychologists | 156,662 | 6%9,272 clinicians | under 1%171 clinicians |
| Social workers | 461,944 | 3%13,324 clinicians | under 1%9 clinicians |
| Pediatricians | 112,220 | 1%1,581 clinicians | 3%3,725 clinicians |
| Student in an Organized Health Care Education/Training Program | 274,265 | 1%3,526 clinicians | 9%24,239 clinicians |
| Speech-language pathologists | 203,476 | 1%2,380 clinicians | under 1%4 clinicians |
| Counselors | 648,077 | under 1%3,063 clinicians | under 1%72 clinicians |
| Registered nurses | 249,039 | under 1%1,082 clinicians | under 1%1,127 clinicians |
| Dentists | 274,769 | under 1%1,121 clinicians | 42%116,320 clinicians |
| Marriage & Family Therapist | 102,168 | under 1%354 clinicians | under 1%9 clinicians |
| Behavior Analyst | 110,160 | under 1%19 clinicians | under 1%9 clinicians |
| Case Manager/Care Coordinator | 162,767 | under 1%13 clinicians | under 1%9 clinicians |
| Pharmacists | 328,748 | under 1%11 clinicians | 8%25,956 clinicians |
| Behavior Technician | 595,326 | under 1%12 clinicians | under 1%8 clinicians |
| Technician | 111,664 | under 1%2 clinicians | under 1%3 clinicians |
Counted over active individual providers filed under each specialty - the same population as the national figure above, so the two can be read together and the specialties sum back to it. That is a different base from the record counts on a list page, which are after removals and include the organisations filed under the same specialty. Every published specialty carries its own figure on its own list page, and all of them are in fields.json.
A high share here means the published Medicare data covers that specialty well. It says nothing about how complete our contact data is for it — different datasets, measured over different populations — and the email, phone and location fill rates are counted and published separately on the same page.
Everything in the boundary above applies to every row here: published Medicare fee-for-service data for 2024, low-volume rows removed before publication so a blank cannot be told from no volume, one year and no history, and provider-level aggregates only.
The places they practise, and who runs them.
Certified facilities carry everything a provider record does, plus what only an organisation has: a certification, a size, an owner, a chain and a leadership team.
| Group | What it holds | Refreshed |
|---|---|---|
| Family | Which kind of certified organisation it is - hospital, nursing home, hospice, home health agency, dialysis centre, or another certified type - and the category beneath that. | Monthly |
| Certification | The certification number where the organisation holds one, the date it was certified, and whether it is still in the programme. One that has left is dropped from search and from its own page rather than left standing. | Monthly |
| Size and ownership | Certified beds, ownership type, the chain it reports under, whether it offers emergency services, and its published quality ratings. | Monthly |
| Health system | Whether it belongs to a system or chain large enough to publish, which one, and what else that system operates. | Monthly |
| Leadership | How many named leadership contacts the organisation has, by role - chief executives, finance, nursing, medical directors, administrators, compliance, board members and owners. Counts on public pages; the names themselves for signed-in customers. | Monthly |
| Linked providers | The clinicians who bill under the organisation or practise at the facility, counted by specialty, and the other locations filed under the same identifier. | Weekly |
48,350 certified facilities on file, 2,702,752 certified beds between them. 24 systems and chains have a page of their own, which takes 5 or more active facilities.
A tier with its reasoning attached, not a green tick.
A work email is the one field on a record that cannot simply be read off a filing, so it is the one field we score. Each address is measured against signals we can each name - whether its domain belongs to a real employer or is a consumer mailbox, whether a verifier found it deliverable, whether the address matches the provider's own name, whether it is a shared desk address rather than a person, how many other providers carry the identical address, and whether the domain resolves at all.
| Tier | What it means for you |
|---|---|
| Trusted | Safe to send. This is the only tier the append tool will use unless you widen it. |
| Probable | Good enough to tell you where somebody works, not good enough to mail without care. |
| Doubtful | Kept and shown to you, and never acted on. |
| Rejected | Not usable, and said so rather than quietly dropped. |
Every score keeps its own arithmetic
A tier stores the signals that produced it and what each one contributed, so a badge is explainable rather than asserted. When you ask why an address is only probable, there is an answer, and it is the same answer every time.
Unmeasured is never treated as fine
An address that has not been scored yet fails every trust filter and is not used to derive anything. The alternative - defaulting an unknown to acceptable - is how a database quietly fills with addresses nobody ever checked.
Across the published specialty lists - 9,469,127 providers - 4,892,855 carry a work email, 3,084,198 of them in the trusted tier, and 9,467,101 carry a practice phone. The same counts are published per specialty and per state on the list pages, so you can check coverage for your own segment before you buy anything.
Suppression applies to every search, page, count and file.
Everywhere, not just the search box
A request reaches the public pages, the signed-in search, the counts the directory publishes, every export and every appended file. A suppression that held in one place and not another would be no suppression at all.
Withheld, and never just missing
On a page a withheld value says it was withheld, because a quiet blank reads as “nothing on file” and sends somebody looking for it somewhere else. In a file the value is left out and counted in the job result, so the gap is a number you can check rather than one you have to notice.
It keeps applying
Records are rebuilt from filings every week, so a removal has to survive the rebuild. The mark is held against the identifier rather than the row, which is why the record is never simply deleted: a deleted record is one the next refresh recreates.
Healthcare professionals can ask for their own details to be removed at remove my information. Customers are held to the same line in the terms.
The part of a data sheet nobody else writes down.
A buyer can verify what we claim to hold by looking. What we do not hold has to be said out loud, because absence is the one thing a search cannot demonstrate.
No patient data, in any form
Not identified, not de-identified, not aggregated, not a single line with a person behind it. The platform knows who a provider is and where they practise. It knows nothing about anybody they have treated, and it never will.
No all-payor claims
What we hold is published Medicare fee-for-service volume for 2024, covering about one in seven clinicians nationally, and more of the specialties that bill it most. A vendor licensing all-payor claims sees commercial, Medicaid and Medicare Advantage volume across nearly every clinician, and sees it sooner. That is a real difference and we are not going to argue it away — we set out where each one wins.
No clinical records
No diagnoses, no notes, no outcomes attributable to an individual patient. Where we carry a facility's quality rating it is the rating as published, about the facility, and we neither recompute it nor infer anything beneath it.
No consumer marketing file mixed in
A provider record is a provider record. Nothing from a consumer audience is folded into it, blended into a count, or shipped inside a provider export. The licensed home address and mobile are a contract of their own, revealed per provider and never listed.
No scores we cannot show you the working for
No propensity number, no likelihood-to-buy, no invented completeness grade. Where we do score something - a work email - the arithmetic travels with it. Changes we report are dated facts instead.
What buyers ask before they trust a file.
Where does the data come from?
The public record: what providers and the organisations they work for file about themselves in federal filings, plus the published certification and quality records for facilities. We assemble and verify it; we do not invent any part of it, and every field carries the date it was last confirmed so you can see how current it is before you rely on it.
What does the confirmed date actually mean?
The date that field was last confirmed against a refresh - not the date the page was built, and not the date the record was created. A field nobody has confirmed recently says so, which is the point: you can decide what is fresh enough for what you are doing rather than trusting a single figure for the whole database.
How complete is the contact coverage?
Practice phone is on the large majority of records because it is filed with the public record. Work email is smaller by a wide margin and always will be, because it has to be confirmed rather than filed. We publish the counts per specialty and per state on the list pages so you can check coverage for your own segment before you buy anything.
Do you have claims data?
We have published Medicare data - what a provider billed, procedure by procedure, and what they prescribed under Part D - for 2024, one data year. We do not have all-payor claims. The difference matters twice: ours is Medicare fee-for-service only, so commercial, Medicaid and Medicare Advantage volume is not in it, and it covers about one in seven clinicians nationally rather than nearly all of them - far more in the specialties that bill Medicare Part B most, which is published per specialty on this page - because low-volume providers are removed from the published file before we receive it. If what you need is every payor for every clinician, a claims licensee has it and we do not. If what you need is to find the clinicians who demonstrably do a procedure or write a drug, and reach them, that is what this is for.
Why is a provider blank on the procedure columns?
Because nothing was published for them, and we cannot tell you which of three reasons applies. Low-volume rows are removed from the published file before we ever see it, so a blank covers a provider with no Medicare volume, one whose volume was below the publication threshold, and one who does not bill Medicare at all. Nothing in the published data separates them, so we show one phrase for all three rather than a badge that implies we know.
Is there a field-by-field data dictionary?
Yes, at /fields. The tables on this page are groups; the reference has one entry per field with what it means in plain English, what it is actually used for, how complete it is where we measure that, what it costs on your plan, and the name it travels under in the API, in an export column and in the append tool. The same list is machine-readable at /fields.json.
Is there any patient data in here?
No. None, in any form: not identified, not de-identified, not aggregated. The platform holds who a provider is, what they practise, where they practise, who they work for, and - from published Medicare data - what they billed and prescribed. That last part is an aggregate published about the provider: a count of procedures and of prescriptions, attached to the provider, with nobody behind it. There is no record of any individual patient, no visit, no diagnosis and no prescription traceable to a person.
What happens when a provider asks to be removed?
The request applies to every search, every page, every count and every file, and it keeps applying - a later refresh does not bring the values back. On a page a suppressed field says it was withheld rather than reading as blank; in a file the value is left out and counted in the job result, so the gap is a number rather than an absence somebody mistakes for nothing on file. The record is never deleted outright, because deleting it is what would let the next refresh recreate it.
Can I check a claim on this page?
Every count here is read live from the same API a customer calls, and the same numbers appear on the directory, the list pages and the facility pages. Search is open without an account, so the fastest way to judge the data is to look up a provider you already know and see whether we have them right.
Judge it on a provider you already know.
Search is open without an account. Look somebody up, check what we have against what you know, and decide from that rather than from a page.