Home › Guides › US healthcare data
US healthcare organization data: identifying facilities, drug sponsors and device makers
US healthcare is one of the most heavily documented sectors in public data. Every provider that bills insurance has an NPI; every approved drug names its sponsor; every device manufacturer registers with the FDA. This guide covers what each registry identifies, how they fit together, and where the personal-data line sits.
NPI: the identifier for every provider organisation
The National Provider Identifier is a ten-digit number required of any US healthcare provider that bills insurance. The registry behind it, NPPES, is fully public. The critical split:
| Type 1 | Type 2 | |
|---|---|---|
| Identifies | An individual clinician | An organisation |
| Examples | A physician, a nurse practitioner | Hospital, clinic, group practice, lab, pharmacy |
| Data nature | Personal data about a named person | Business record about a facility |
| B2B relevance | Limited, and higher risk | The right population |
Organisation records carry the legal name, practice address, often a phone number, and a taxonomy code classifying the specialty — the field that turns a list of facilities into a segmented target list. Selling imaging software means filtering to radiology taxonomies, not reading through every hospital in a state.
NPI registry organisation data →
FDA drug approvals: the sponsor view
FDA publishes approved drug products together with the sponsor — the company responsible for the application. Searching by sponsor gives you a company's approved portfolio, which answers a different question from the pipeline: not what a company is trying to do, but what it has already got through.
Pairing approvals with trial registrations is where this gets analytically interesting. Trials show intent; approvals show conversion. The ratio between them, per company, is a crude but real measure of execution — see the pharma pipeline guide for the trials side.
FDA drug approvals by sponsor →
Medical devices: establishment registrations
Device manufacturers must register their establishments with FDA and list the devices they produce. Both are published, giving you manufacturer identity, location and device categories.
For supply-chain research or medtech sales this is the equivalent of a manufacturing census: who makes what, and where. It is also the layer that reveals contract manufacturers — companies that appear as registered establishments without consumer-facing brands.
FDA device manufacturer data →
Putting the three together
| Question | Source |
|---|---|
| Which clinics of this specialty exist in this state? | NPI registry, Type 2, filtered by taxonomy |
| What has this pharma company brought to market? | FDA drug approvals by sponsor |
| Who manufactures this class of device? | FDA device establishment registrations |
| What is the company still developing? | Clinical trials by sponsor |
| Is this healthcare nonprofit financially healthy? | Form 990 financial filings |
| Who funds their research? | NIH and NSF award records |
The personal-data line in healthcare
Healthcare data carries a heavier compliance load than most B2B sectors, and the Type 1/Type 2 split is where the line falls most clearly. A Type 2 organisation record — a clinic's name, address and specialty — is a business record. A Type 1 record names an individual practitioner, and building marketing lists of named clinicians raises questions that the same activity against companies does not.
Worth stating plainly: none of these registries contain patient data. NPPES describes providers, FDA describes products and manufacturers, and trial registrations describe study designs. Patient-level information is not part of any of them.
Practical notes
- Taxonomy codes are the segmentation key. Without them the NPI registry is an undifferentiated list; with them it is a specialty-targeted database.
- Organisations move and merge. NPIs persist through address changes, so the registry is more stable than a scraped directory — but the address can lag a relocation.
- Sponsor names are legal entities. As with every federal dataset, match on the registered name rather than the brand.
Taxonomy codes: how the registry becomes a target list
The field that does the real work in NPPES is the taxonomy code — a hierarchical classification of what a provider does, assigned from a national standard rather than typed in freely. It has three levels: a broad grouping, a classification within it, and often a specialisation below that.
Without it, the registry is an undifferentiated list of several million providers. With it, you can construct precise populations: dialysis centres in three states, independent imaging facilities, skilled nursing rather than assisted living, clinical laboratories rather than physician offices that happen to run tests.
Two practical notes. First, an organisation can carry several taxonomies — a multi-specialty group legitimately does. Filter on "has this taxonomy" rather than "primary taxonomy equals", or you will silently drop the large multi-service facilities, which are usually the ones you most wanted. Second, taxonomy is self-selected at enrolment and updated irregularly, so it describes what the organisation registered as, not necessarily its current service mix.
NPIs are never reused, which is why the registry is trustworthy
An NPI is assigned once and is not recycled. When an organisation closes, its NPI is deactivated rather than reissued to someone else. That single design decision is why the registry works as a spine for healthcare data: a number you captured three years ago still refers to the same entity today, even if the name, address and ownership have all changed since.
It also means the registry carries its own history. Deactivated records remain, which lets you distinguish "this facility no longer operates" from "this facility was never in my data" — a distinction most scraped directories cannot make, because a closed business simply disappears from them.
Frequently asked questions
What is an NPI?
A ten-digit identifier required of US providers who bill insurance. Type 1 is an individual, Type 2 an organisation.
Which type should B2B teams use?
Type 2 — organisations. It is the right population commercially and keeps you on the business-record side of data protection.
Can I see a company's approved drugs?
Yes, FDA publishes approvals with the sponsor company, so a sponsor search returns the approved portfolio.
Does the registry include email?
No. Addresses and often phone numbers, plus the specialty taxonomy. Enrich separately for digital contact routes.