Skip to content
FonteumThe Graph

By use case

Exclusion & monitoring (self-serve)Free roster screen — no accountExclusion & sanctions screeningCredentialing & provider-data enrichmentAudit evidence & defensible programsProvider data for AI / RAGM&A & network diligence

By buyer

Compliance & riskJournalists & newsroomsDevelopers & AI teams

By industry

HealthcareProviders & facilitiesSanctionsOFAC / EU / UK / UN / OIG / SAMFederal contractingSAM · USASpending · FAPIIS

The capability layer

APIREST + bulk accessFor AI agentsModel Context Protocol serverFHIR R4 APIBulk exportAudit packSigned report artifacts; source facts remain separately citedReconciliationSource-vs-source diffsAI answers with sourcesHow records link upHistoryRetained versions for named sources

The differentiator

Coverage & sourcesThe catalogFreshnessMethodologyCare CompareFacility qualityAI answers, comparedBrowse all datasets →
Research

The dev on-ramp

DocsAPI referenceConnect your AI agent (MCP)One-paste installFHIR sandboxLive API surfaceQuickstartStatusChangelogSDKs & integrations
Pricing
Sign inFree roster screen →Get a signed certificate →

Solutions

Exclusion & monitoring (self-serve)Exclusion & sanctions screeningCredentialing & provider-data enrichmentAudit evidence & defensible programsProvider data for AI / RAGM&A & network diligenceCompliance & riskJournalists & newsroomsDevelopers & AI teamsHealthcareSanctionsFederal contracting

Platform

APIFor AI agentsFHIR R4 APIBulk exportAudit packReconciliationAI answers with sourcesHow records link upHistory

Data

Coverage & sourcesFreshnessMethodologyCare CompareAI answers, comparedBrowse all datasets →
Research

Developers

DocsAPI referenceConnect your AI agent (MCP)FHIR sandboxQuickstartStatusChangelogSDKs & integrations
Pricing
Sign inFree roster screen →Get a signed certificate →
Research › Provider Directory Accuracy Index › Methodology

Fonteum Research · Methodology Disclosure

Provider Directory Accuracy Index

Version: pdai/v1

Overview

The Provider Directory Accuracy Index (PDAI) measures the degree to which three federal provider data sources agree on four key fields for each individual NPI. A high agreement score indicates that the same provider's information is consistent across federal sources — a necessary (though not sufficient) condition for directory accuracy. A low score flags systematic data quality problems that can lead to ghost networks and No Surprises Act compliance risk.

Fonteum does not independently verify provider information. This is a cross-source agreement score — not a ground-truth accuracy measurement.

Data sources

CMS NPPES — National Plan and Provider Enumeration System

Self-reported provider data. The authoritative NPI registry. Providers are responsible for keeping their NPPES record current; update frequency varies by practice. Source tier: Tier-2 (federal public records, bulk download).

CMS Care Compare — Provider data

CMS-curated provider information published via the Care Compare portal. Reflects CMS enrollment records and curation workflows. Source tier: Tier-2 (federal public records).

CMS PECOS — Provider Enrollment, Chain, and Ownership System

Medicare enrollment data. Represents the billing and enrollment record that determines Medicare payment eligibility. Source tier: Tier-2 (federal public records).

Scored fields

FieldDefinitionNormalization
practice_addressPrimary practice city + stateLowercase; city normalized; state as 2-letter USPS code
primary_specialtyPrimary taxonomy code (NUCC 10-digit)Exact code match; no synonym expansion in v1
org_affiliationOrganizational CCN or parent NPI6-char CCN or 10-digit organizational NPI, uppercase
telecomPrimary phone number10-digit digits only, no formatting

Scoring algorithm (pdai/v1)

  1. NPI intersection: Include only NPIs present in ≥2 of the three sources. NPIs in only one source are excluded from scoring (no comparison possible).
  2. Pairwise field comparison: For each NPI, compare all source pairs (NPPES × Care Compare, NPPES × PECOS, Care Compare × PECOS when all three are present). Each pair contributes one comparison per field.
  3. Field agreement rate: Count comparisons where both sources have a non-null value and the values match exactly. Agreement rate = matches / comparisons.
  4. Coverage filter: Fields where fewer than 50% of matched NPIs have a non-null value in ≥2 sources are excluded from the composite score (insufficient coverage to be representative).
  5. Composite score: Simple average of field agreement rates for fields that pass the coverage filter.
  6. Insufficient sample: Scopes with fewer than 100 matched NPIs return no score and are flagged insufficient_sample = true.

Aggregation scopes

  • National: All matched NPIs across all states.
  • State: NPIs filtered by practice_address state component. A provider with a multi-state practice is included in each state where they have a matched record.
  • Specialty: NPIs filtered by primary_specialty NUCC taxonomy code.

Exclusions

  • NPIs present in only one source.
  • Organizational NPIs (Type-2 NPIs) — scored separately in a future methodology version; excluded from v1 to avoid mixing individual and group practices.
  • Scopes with fewer than 100 matched NPIs (insufficient sample).
  • Fields with null values in both sources (no comparison possible).

Version history

pdai/v1

Initial release. Four scored fields: practice_address, primary_specialty, org_affiliation, telecom. Simple average composite with 50% coverage filter. National + state + specialty scopes. Insufficient sample threshold: 100 NPIs.

Limitations

  • Agreement is a proxy for accuracy — fields that are wrong the same way in all sources score as accurate.
  • Address granularity is city + state only in v1; ZIP-level and street-level matching is deferred.
  • Specialty codes are matched exactly; taxonomy synonym expansion is not performed. Source-specific code mapping variations may inflate disagreement counts.
  • Snapshot timing differs across sources — NPPES, Care Compare, and PECOS have different update cadences. Field-level staleness is not currently accounted for.
  • Fonteum does not independently verify, inspect, or certify any provider. This index describes data consistency, not provider quality.
← Back to Provider Directory Accuracy Index
See also
  • HH Quality Analysis
  • NPPES Anatomy
  • Healthcare Access
  • Care Quality
  • Real ACT Compliance

What’s on file, by the numbers

Platform snapshot · 2026-07-15

13.4Mproviders & companiesProviders, organizations, owners, and facilities on file
26.2Msource-linked factsSource-linked field facts in the dated platform snapshot
—sources liveCrosswalk-resolved sources with a proved content transition in the preceding 45 days
111sources integratedActive registry rows; integration does not establish a load
13state Medicaid jurisdictionsDistinct states represented in the state-exclusions serving table

Built on the authoritative federal record

The primary sources, named on every page.

These are the federal agencies whose public datasets Fonteum ingests and attributes — the issuing authorities, not customers or partners. Every figure on the site links back to one of them.

  • CMS
  • HHS-OIG
  • HRSA
  • FDA
  • NLM
  • NUCC
  • Census
  • BLS
  • BEA

See the full source registry, with license and refresh cadence for each →

Reproducible by design

Published figures name their source and date.

Source and date

Published research identifies its government file and observation date. Source-file SHA-256 coverage is disclosed separately; facts do not currently link deterministically to signatures.

Reproducible SQL

Each study ships the exact query behind its figures, run against the same dated copy of the federal file we used. Re-run it yourself.

Daily observations

The platform records table row counts daily. Those observations detect local drift; they do not imply that an upstream publisher released or Fonteum ingested new data that day.

Named medical review

Reviewed by Jennifer Montecillo, MD, medical reviewer. Non-practicing medical reviewer.

Read the full provenance and attestation methodology →

Fonteum
Platform
Platform overviewAPIFor AI agents (MCP)FHIR R4 APIBulk exportAudit packReconciliationAI answers with sourcesHow records link upHistory
Solutions
All solutionsExclusion & sanctions screeningCredentialing & enrichmentAudit evidenceProvider data for AI / RAGUse casesM&A & network diligenceCompliance & riskJournalists & newsroomsDevelopers & AI teams
Data & sources
Coverage & sourcesBrowse all datasetsState Medicaid exclusionsFreshnessMethodologyCare CompareNursing homesRecall recordsSanctions screeningOIG LEIE listOwnershipStaffingDeficienciesSpecial Focus Facilities
Federal contracting
OverviewAwards during active exclusionFederal debarment scorecardProcurement questionsContractor lookup8(a) certification guide
Developers
Developer hubDocsAPI referenceQuickstartStatusChangelogSDKs & integrationsWebhooks
Research & guides
Research hubLearnHow it worksGuidesHealthcare provider dataExclusion & sanctions screeningProvider credentialing dataHealthcare data for AIHospital margin gapProvider access gapsGlossaryComparisonsSourced data vs plain LLM answersCitationsManifestoWhy Fonteum
Company
AboutPressCustomersPricingContactEditorial policyCorrections
Trust & legal
TrustTrust markQualitySecurityPrivacy policyTerms of serviceAPI termsMedical disclaimer

Reviewed by Jennifer Montecillo, MD, medical reviewer. Non-practicing medical reviewer.

© 2026 Fonteum LLC. All rights reserved.

·hello@fonteum.com

A public-records graph that exposes source and observation metadata where supplied.

Fonteum's provenance ledger contained 26.2M source-linked facts on July 12, 2026. All but 14 carried a source-file SHA-256; 0 linked deterministically to a signature. Inspect a supplied snapshot id at fonteum.com/verify · source-mark coverage and limitations.
Request access→