How We Get Our Data
Every listing comes from two public sources: the CMS National Downloadable File and the NPI Registry. We show the file date on every page, we haven't called the offices, and our spot check found contact details often differ between the two records. Call to confirm.
Sources and Dates
| CMS National Downloadable File (Doctors and Clinicians) | File dated October 30, 2026. Name, specialty, practice location, phone, group and Medicare fields. |
|---|---|
| NPI Registry (NPPES) monthly file | File dated November 2, 2026. Active status, dates, and a degree when CMS leaves it blank. |
| Next planned update | CMS plans its next file for December 17, 2026. We refresh about monthly, by hand for now. |
Links: CMS dataset, NPI Registry.
What Is Released So Far
4,820 doctors in 20 cities across 5 states, in 8 specialties. 63,648 doctors met our quality rule in those states and specialties, and we're releasing them in stages so each page can help someone. Cities were picked by a fixed, neutral order, not by doctor counts or guesses about demand.
Who Is Left Out
- Anyone not in Medicare's file. Doctors who don't bill Medicare are missing, and children's doctors will be thin.
- Clinicians who aren't MDs or DOs, for now.
- Anyone whose NPI has been deactivated.
- Doctors with no public practice location or no listed phone.
- Addresses that may be homes, and those CMS has hidden. For these we show the city only, with no street or phone. That includes any address marked as an apartment, lot or trailer, and any street where a doctor is the only person in the CMS file and nothing marks it as commercial. Our screen is a cautious guess: it hides some real small offices and can't catch every home.
- Everywhere outside the released cities.
- Any doctor who asked us to remove their page.
Our Spot Check, a Small Sample
On November 15, 2026 we compared 50 released doctors, chosen at random, 10 per state, with the live NPI Registry.
| Check | Matched, of 50 |
|---|---|
| Found in the registry and active | 50 |
| First name | 50 |
| Last name | 50 |
| Degree (MD or DO) | 49. In 1, the registry has no value. In 0, the sources differ, so we show no degree |
| City and state of any of our locations | 37 |
| Street of any of our locations | 32 |
| Phone at any of our locations | 23 |
50 of 50 were active, 32 of 50 street addresses and 23 of 50 phones matched the NPI Registry. These are two separate official records, a mismatch doesn't show which is right, and we haven't called the offices. Call to confirm.
The registry is often the older record: 20 of the phone misses and 14 of the street misses are on registry records last updated before 2024. That suggests staleness, but it proves nothing about any one record. Some CMS locations look like a hospital where a doctor also works, so a number can reach a main hospital line. We don't know how often.
Across all doctors who met our quality rule, 26,974 of 63,648 (42.4%) have a CMS phone that also appears in the registry's bulk file. For released doctors it's 2,120 of 4,820 (44.0%).
What "Call Group" Means
Most buttons say Call Office. A button says Call Group, or Call This Number if no group is listed, when there's evidence the number may be central: it is toll-free, its area code mostly appears in another state in the CMS file itself, or the group's own name says care is remote. On those pages we add: "This number may reach a central office, not a local one." When a doctor also has a local-looking number in the same city, we show that one first. We haven't called these numbers, so the label is a caution, not a finding, and an ordinary Call Office button can reach a main line too.
What the Data Can't Tell You
Neither source shows who is taking new patients, languages, private insurance, booking links, ratings or medical quality. We don't add these from other directories. Nothing here shows that a doctor is licensed, and we don't call anyone "verified". There are no paid placements, and lists run A to Z.
How We Handle Mistakes
The one-tap buttons on doctor pages and the corrections page reach a person, who checks against the source. A report alone never changes a page. If our processing caused the error, we fix it. If the source is wrong, we show how to fix it there. If data is ever more than 90 days old, we'll say so.