The CMS National Plan and Provider Enumeration System — the NPI Registry — is a federal database of every licensed US healthcare provider, maintained by the Centers for Medicare and Medicaid Services, and accessible to anyone at no cost. It contains over 1.5 million individual provider records, each with a standardized specialty taxonomy code, a practice address, and in many cases a phone number and business fax. For a clinical research site trying to reach every physician who treats patients matching a trial's inclusion criteria, it is the most comprehensive starting point that exists.
Most sites are not using it. The ones that are have a systematic advantage in physician outreach that compounds over time.
What a Taxonomy Code Is and Why It Matters
Every licensed healthcare provider in the NPPES system is assigned one or more taxonomy codes — standardized identifiers that describe their specialty. An endocrinologist carries taxonomy code 207RE0101X. A cardiologist has 207RC0000X. A rheumatologist has 207RR0100X. An oncologist specializing in internal medicine has 207RI0200X.
For a clinical trial targeting a specific patient population, the relevant taxonomy codes identify every physician who could plausibly treat an eligible patient and refer them to your site. You do not need to know these physicians. You do not need to have a relationship with them. You need their contact information and a way to reach them, which the NPI Registry provides.
How to Query the Registry for Your Therapeutic Area
The NPPES API is available at npiregistry.cms.hhs.gov/api/ and accepts queries by taxonomy code, state, postal code, and provider name. A basic API call with a taxonomy code and state returns every licensed provider of that specialty registered in that state.
For geographic precision — finding physicians within a specific radius of the research site — the process requires geocoding. Each returned practice address is converted to latitude/longitude coordinates, then filtered to include only those within the configured mile radius. A site in Houston, Texas running a cardiology study querying for 207RC0000X within 20 miles typically returns 80 to 250 individual cardiologists, depending on market density. Each one treats patients who may meet the study's inclusion criteria.
What the Output Looks Like
A completed NPI query for a clinical trial produces a list of physicians with their full legal name, practice name, primary business address, and taxonomy code. Email addresses are typically not included in NPPES records and must be found separately — through professional email verification services that identify business email addresses from provider name and domain. Phone and fax numbers are frequently included and can be used for direct outreach.
The practical output of a well-executed NPI query is a contact list of 50 to 500 physicians — depending on therapeutic area, geographic density, and search radius — who treat patients eligible for the study. This is the starting point for a physician outreach campaign, not the endpoint.
Common Mistakes in NPI-Based Physician Identification
The most common error is querying too broadly — including every physician in a state rather than filtering by radius, which produces an unworkable list of thousands. The second most common error is using the wrong taxonomy code. A site running a diabetes trial that queries only for endocrinologists misses the much larger population of primary care physicians, internal medicine specialists, and family medicine providers who manage the majority of type 2 diabetes patients in the community setting.
For most therapeutic areas, the highest-volume referral source is not the specialist — it is the generalist who sees the largest number of patients with the relevant condition and who is most likely to have a patient meeting eligibility criteria at any given time. A well-constructed NPI query includes both specialist taxonomy codes and the primary care and internal medicine codes most likely to see the relevant patient population.
From NPI Data to a Working Outreach List
The NPI query is the identification step. The outreach list requires three additional steps: email address verification for each physician record, deduplication against physicians already known to the site, and prioritization by estimated patient volume based on practice size and specialty specificity. Sites running this process systematically — for every new study before activation — enter each enrollment period with a ready outreach list rather than building one after the study has opened and enrollment pressure is already present.