NPI Registry Scraper: bulk NPI lookup and NPPES provider search

This NPI Registry scraper runs an NPI lookup for a list of numbers, or searches the official NPPES NPI Registry by name, specialty and location, and returns one clean row per US healthcare provider. Each row carries the NPI number, provider name, specialties, licenses, practice and mailing addresses, phone numbers and the source's registration and update dates, for individual practitioners and for organizations.

Use it to enrich an existing provider list, prepare a directory import, build a lead list of organizations in a specialty or location, check provider details during onboarding, or attach official registry identifiers to CRM records. Every delivered row includes npi as text and provider_id with the same value. These stable identifiers let you join provider records to other US healthcare directories when those directories publish an NPI.

NPI Registry input and search options

NPI lookup by number

For an existing list, supply npi_numbers as an array of strings. For example, NPI 1003000126 belongs to ARDALAN ENKESHAFI, M.D. in the source response used for this actor's examples. This is a real registry record, and its details can change when the provider updates NPPES.

{
  "npi_numbers": ["1003000126"],
  "max_items": 100
}

NPPES provider search by name, specialty and location

For discovery, use first_name, last_name, organization_name, taxonomy_description, city, state, postal_code and country_code. Set provider_type to individual, organization or any. The specialty filter expects a description such as Dentist or Cardiovascular Disease. It does not accept a taxonomy code such as 207RC0000X. The source may match any of a provider's taxonomies, so the primary specialty on a returned row can differ from the searched specialty.

A name can use one trailing wildcard after at least two characters, such as mayo* for organization names. use_first_name_alias includes aliases in first-name searches. A US postcode accepts five digits, nine digits or a prefix followed by *, such as 752*. A state alone does not define a valid search: combine it with a name, specialty, city or ZIP code.

{
  "taxonomy_description": "Cardiovascular Disease",
  "city": "boston",
  "state": "MA",
  "address_purpose": "PRIMARY",
  "max_items": 100
}

The default address_purpose is any, which lets the source match a mailing address as well as a practice address. Choose PRIMARY when the returned primary practice address must match your location filters. SECONDARY searches secondary practice locations. LOCATION covers primary and secondary practices, while MAILING searches mailing addresses. Explicit NPI lookups are independent of these search filters. When both a list and search fields are supplied, the actor looks up the list first, then searches, removing duplicate NPIs across both phases.

Output fields

Each dataset row represents one provider. The principal fields are npi, name, provider_type, specialty, taxonomy_code, license_number, license_state, address, street, city, state, postal_code, country_code, phone, fax, enumeration_date, last_updated, status and url.

Individuals also include published name parts, credentials and sex. Organizations include organization_name, subpart information, the parent organization when supplied, and an authorized_official object. taxonomies preserves every published taxonomy and its primary flag. specialties lists unique descriptions with the primary description first when available. Retired taxonomy codes can have a missing description; the actor retains the code and returns a null specialty without guessing a label.

Some active providers have no primary taxonomy flagged. For those records, primary_taxonomy, taxonomy_code, specialty, license_number and license_state are null, while all taxonomies and specialty descriptions remain available. A missing mailing address produces mailing_address: null; a missing primary practice address leaves the practice address fields null. Missing secondary practice locations produce an empty list. Records without a usable NPI, name or provider type are skipped with a warning, while the remaining providers are delivered.

mailing_address is separate from the primary practice address. practice_locations contains secondary practices. identifiers, other_names and endpoints preserve additional public IDs, aliases and electronic endpoint details. Missing text is null, including the source's -- name markers. Phone numbers remain in their published form. US ZIP+4 values receive a hyphen, and postal_code_raw retains the original value. The actor does not infer an email address from a phone number or endpoint.

Output example

The row below is an excerpt of a real record returned for NPI 1003000126 (lookup on 5 October 2026); nested taxonomy, identifier and endpoint lists are shortened here and complete in the dataset.

{
  "npi": "1003000126",
  "name": "ARDALAN ENKESHAFI, M.D.",
  "provider_type": "individual",
  "credential": "M.D.",
  "specialty": "Hospitalist",
  "taxonomy_code": "208M00000X",
  "license_number": "MD600003480",
  "license_state": "DC",
  "specialties": ["Hospitalist", "Internal Medicine"],
  "address": "6410 ROCKLEDGE DR STE 304, BETHESDA, MD 20817-1841",
  "city": "BETHESDA",
  "state": "MD",
  "postal_code": "20817-1841",
  "phone": "443-602-6207",
  "enumeration_date": "2007-08-31",
  "last_updated": "2025-05-28",
  "status": "A",
  "url": "https://npiregistry.cms.hhs.gov/provider-view/1003000126"
}

Datasets export to JSON, CSV, Excel or XML from the Apify console or API, so the same rows can go straight into a spreadsheet, a CRM import or a data warehouse.

Use cases

• Lead generation for healthcare sales. Search a specialty in a city or ZIP prefix, such as cardiologists in Boston or dentists in 752*, and export names, practice addresses and phone numbers for outreach or territory planning.
• Provider data enrichment. Start from a CRM or claims export that already holds NPI numbers and run a bulk NPI lookup to add specialties, license numbers, addresses and the latest update date.
• Credentialing and onboarding checks. Pull the registry's published taxonomies, license fields, status and dates for each provider as source data for an existing review process.
• Directory building and joins. Use npi as the join key between this registry and other US provider directories, review sites or internal systems that publish the same identifier.
• Change monitoring. Re-run the same NPI list on a schedule and compare last_updated, addresses or phone numbers to see which providers changed their NPPES record.

How it compares

Option	Bulk NPI list	Search filters	Structured export	Over 1,200 matches
NPI Registry website	One provider at a time	Yes	No, manual copying	No
NPPES API directly	One request per NPI or page	Yes	Raw JSON, nested	You write paging and splitting
NPPES full download file	Whole country only	No	Very large CSV, needs processing	Yes, but no filters
This NPI Registry scraper	Yes, up to 50,000 rows per job	Yes	Flat rows plus full nested data	ZIP splitting with deduplication

The registry website is fine for a single NPI lookup. The full NPPES download suits teams that want the entire national file and can process several gigabytes. This actor sits in between: targeted searches and list lookups with clean, deduplicated rows, ready to export.

Volume, continuation and coverage

max_items limits the total number of unique delivered providers, with a default of 100 and an upper limit of 50,000. It applies across the entire job, including the npi_numbers list. Increase max_items to the size of your list to process all its valid, unique NPIs. If the limit leaves NPI numbers unprocessed, collection stops with partial output and a warning stating how many numbers were not looked up. Larger collections continue automatically between bounded requests until that limit is reached. The run summary includes warnings and indicates any incomplete coverage; API callers can resume using cursor while keeping the other input values unchanged.

NPPES exposes at most 1,200 records for one set of search criteria. A full window produces an explicit warning. Enable split_large_searches to search narrower US ZIP prefixes and deduplicate their results by NPI. This can retrieve more providers, although a full exact ZIP code still requires narrower name or specialty filters. ZIP splitting cannot guarantee worldwide coverage; searches without a US location restriction warn that foreign or missing postcodes may be omitted. Providers sharing a mailing or practice address may occur in several searches, but duplicate NPIs are delivered only once.

Pricing for NPI Registry lookups

The actor charges one provider event for each delivered provider row. Empty results, invalid NPI values and missing providers do not produce provider charges. Duplicate NPIs do not produce extra rows or extra provider charges. Consult the actor's Pricing tab for the current amount per event and your account's usage limits.

Users without a paid Apify plan receive at most ten providers per run. Larger requested limits are capped, and ZIP splitting is disabled for these runs. Paying users can request the documented volume. The output summary reports whether the free cap applied, along with delivery counts and warnings.

FAQ

Does an NPI prove that a provider is licensed or credentialed?

An NPI is an identifier. The actor returns the registry's published taxonomy and license fields, status and dates. It does not independently verify a license, professional qualification, eligibility or credentialing decision. Use the returned license numbers and dates as source data for the verification process you already use.

Why does a Boston search return a provider with a different primary practice city?

The default address mode can match the provider's mailing address or another practice location. Select address_purpose: "PRIMARY" to restrict a location search to the primary practice address. The other address objects remain available so you can inspect how the source record is structured.

What happens to an invalid or unknown NPI?

Invalid checksums and malformed NPI values are skipped with a warning when other usable input remains. A wholly invalid list is rejected. An unknown but valid NPI adds a missing-provider warning; it never creates a placeholder provider row. A valid search with no matching records returns an empty result with no provider event charges.

Can I retrieve every US provider with one broad search?

Broad searches encounter the source's search window. ZIP splitting expands coverage, and full ZIP searches require further filtering. Check the summary warnings before treating a collection as complete. NPPES does not expose a total match count: its result count describes a single returned page.

Can I look up an NPI number by provider name?

Yes. Leave npi_numbers empty and search with first_name and last_name for individuals or organization_name for organizations, optionally combined with state, city or a specialty. Each matching provider comes back with its NPI number.

Does the NPI Registry include email addresses?

No. NPPES does not publish email addresses, and the actor does not guess one from a phone number or endpoint. Electronic endpoints, where a provider published them, are returned unchanged in endpoints.

Can I use this as an NPI Registry API?

Yes. Start the actor through the Apify API with the same JSON input shown above and read the dataset when the run finishes. Larger collections continue automatically within a run; if the summary reports incomplete coverage, API callers can resume with cursor while keeping the other input values unchanged.

Where does the information come from?

The source is the official CMS NPPES NPI Registry API, version 2.1. Each row links to the official provider page and includes fetched_at. Deactivated or unpublished records may be absent from this API. The source can update between requests, so repeated exports may reflect changed provider information.