Every number on DocData is copied out of a file the US federal government publishes. Nothing is modelled, estimated, scored or inferred. This page says exactly which files, how far behind each one runs, what the agencies remove before publishing, and what none of it can tell you.
Datasets read
| Source | What it covers | Typical lag |
|---|---|---|
| NPPES NPI Registry | Identity, practice address, taxonomy (specialty), state licence numbers, enumeration date, active/deactivated status. | Live. Reflects whatever the clinician last filed with NPPES. |
| CMS Open Payments (Sunshine Act) | Every payment or transfer of value a drug or device manufacturer or GPO reported — general payments, research payments and ownership interests. | About six months. A calendar year is published the following June, after a review-and-dispute window. |
| CMS Medicare Physician & Other Practitioners | Fee-for-service Part B claims: submitted charges, allowed amounts, Medicare payments, service and beneficiary counts, patient case mix and HCC risk score. | About 18–24 months behind the current calendar year. |
| CMS Medicare Part D Prescribers | Part D prescription claims by prescriber and by prescriber-and-drug: claims, total drug cost, day supply, beneficiary counts, opioid and antibiotic measures. | About 18–24 months. |
| CMS Doctors & Clinicians (Care Compare) | What a patient sees on medicare.gov: medical school, graduation year, listed specialties, group affiliations and practice locations. | Refreshed by CMS on a rolling basis. |
| CMS Medicare enrollment files | Order-and-refer privileges, Medicare opt-out status, revalidation due dates. | Refreshed by CMS on a rolling basis. |
| HHS-OIG LEIE | The federal exclusion list — individuals and entities barred from federally funded healthcare programs. | OIG republishes monthly; DocData re-downloads weekly. |
The exact program years available at any moment are read from the live CMS catalogue and printed on each card and in the site footer, rather than hardcoded here.
What the agencies remove before publishing
These are not gaps in DocData. They are redactions CMS applies to the source files, and treating a redacted figure as a zero is the single most common way a dashboard built on this data becomes wrong.
- Small counts are withheld, not zeroed. CMS blanks any beneficiary count between 1 and 10 to protect patient privacy. DocData renders those as withheld and draws no bar. Withheld means somewhere from one to ten.
- The Part D brand/generic split disappears below a disclosure threshold. When either side falls below it, the split is not charted at all — a zero-width segment would read as "prescribes none of these".
- Suppressed rows drop out of totals. A prescriber's drug-level rows are published only above the threshold, so the drug table can sum to less than the yearly total on the same card.
- A deactivated NPI is published as an NPI and a date only. CMS asks redistributors of the NPPES file to withhold the name, address and specialty of a deactivated record, so DocData does.
Judgement calls in how the data is presented
- Manufacturer names are folded. Companies self-report their own name, so one firm arrives spelled several ways in a single clinician's records — "AbbVie Inc.", "AbbVie, Inc." and "ABBVIE INC." are one payer. Grouping on the raw string splits a company across rows and produces a visibly wrong ranking. Payers are folded on a normalised key and shown under their most common spelling.
- The product chart deliberately over-counts. One Open Payments record can name up to five products, and CMS does not divide the amount between them. Summing by product therefore exceeds the total. The card says so where it appears.
- Ownership records are usually $0. The reportable fact is the interest itself, not a cash transfer, so ownership gets its own table rather than vanishing into a zero total.
- Live Open Payments queries are capped. Where DocData reads the CMS API directly it takes at most 500 records per year per stream. Above that the total is labelled a floor rather than reported as if complete.
What this cannot tell you
- Medicare fee-for-service only. Every Medicare figure excludes Medicare Advantage, Medicaid and all commercial payers. A clinician with a mostly-MA panel will look far smaller here than their real practice volume.
- Nothing here is a quality measure. High billing, high prescribing volume or large industry payments are not, on their own, evidence of anything. Volume tracks panel size, specialty, setting and case mix before it tracks behaviour.
- Not a licensure or credentialing check. Licence status and disciplinary history live with state medical boards, and the federal exclusion list is a different list from state Medicaid exclusions and from SAM.gov debarment.
- Not medical advice, and not a basis for a hiring, contracting or treatment decision on its own.
- Same-name confusion is real. The exclusion card reports both an NPI-level match and a name-level match, and a name match is not the same person until you confirm it against the OIG record.
Corrections
Nothing on DocData is editable — these are federal publications, and DocData does not hold a copy anyone can amend. Each agency runs its own correction process, linked from the card that shows the data. Open Payments in particular has an annual review-and-dispute window, and disputes go to the reporting manufacturer rather than to CMS. Most identity errors — the stale address that follows a clinician across every commercial directory — start at NPPES or PECOS.
Privacy
Searches are not stored or logged. Requests to CMS are proxied because NPPES sends no CORS headers and the CMS endpoints need retries and caching, not because anything about the search is retained. "Use my location" sends the coordinate your browser reports to the US Census Bureau geocoder to turn it into a ZIP code, and nowhere else.
The one exception is the interest list. If you type an address into it, we store that address, the area you asked to be covered, and the record you were reading when you signed up — so that the report can be sent and can lead with something useful. Nothing else about your visit is attached to it, joining is entirely optional, and you can ask us to delete the row at any time. Not joining changes nothing about what the site does.
DocData is an independent project. It is not affiliated with, endorsed by, or operated by CMS, HHS, or any government agency.