Separate the facts before connecting the context.

Plain-language method

1. Identify the doctor

NPI is the preferred stable identifier. Names, location, credentials, and specialty help a user distinguish people with similar names, but a name-only fuzzy match is not enough to publish a financial relationship.

2. Read the reported payments

CMS Open Payments records are summarized by program year, reporting company, payment category, and products explicitly named on individual payment records. General Payments are not combined with Research Payments or Ownership/Investment Interest into one headline total.

3. Keep two product relationships separate

Payment-linked product: a specific Open Payments record explicitly named a drug, biologic, device, or medical supply. If the same publishable brand drug also appears in this doctor’s public Medicare Part D data, the site can show that separate overlap.

Same-company prescribing overlap: the exact Open Payments reporting entity paid the doctor; the same entity reported a branded drug or biological on at least 10 General Payment records involving at least 2 distinct recipient NPIs in that program year; and the doctor has the same non-generic brand in public Part D data. This relationship does not assert ownership, manufacturing, affiliate status, or that the individual payment was for the drug.

4. Preserve the limits

Medicare Part D does not represent a doctor’s entire practice. Missing public rows can reflect dataset scope, year differences, or CMS publication thresholds. A positive overlap is context; it is not evidence that a payment caused or influenced a prescription.

Technical method

Identity and entity resolution

• Doctor attribution is NPI-centric. Publishable identity states are verified and high_confidence; ambiguous or conflicting identities are withheld rather than merged by name.
• Company normalization is not company resolution. A reported Open Payments entity becomes a canonical company only through an explicit, evidence-bearing, time-valid resolution.
• Product matching uses exact normalized brand matches, approved aliases, or manually verified mappings. Loose fuzzy matching is not a publishable relationship.
• Generic Part D claims are not assigned to a specific manufacturer without manufacturer-level evidence.

Amounts and products

Doctor-level General Payment totals are calculated once per payment record. A single payment can name multiple associated products, so product-associated amounts are explicitly non-additive; adding product totals together can double-count the same payment.

Negative statements

The site says “We did not find matching records in the selected dataset and year” only when the relevant source coverage is known to be complete for that analysis. An incomplete extract never becomes a zero, “never,” or “no overlap” conclusion.

Deterministic editorial templates

Verified Facts, search-status labels, and Questions to Ask come from reviewed, versioned templates populated only with validated fields. A general-purpose LLM does not write doctor-specific factual conclusions.

Loaded source releases

What this build is using

Known limitations

What these datasets do not prove

• Open Payments does not represent every source of a doctor’s income.
• Medicare Part D provider-and-drug data does not represent all prescriptions, commercial insurance, Medicaid in full, cash patients, or rows below applicable public-release thresholds.
• Different source datasets can refer to different years. The interface keeps those years explicit rather than visually implying they describe the same period.
• A named product, matching brand, or company–prescribing overlap does not establish causation or inappropriate care.
• Company ownership, licensing, marketing, distribution, and affiliate relationships can change over time; the site requires dated evidence rather than applying today’s company structure backward indefinitely.

Versions & change log

Current contract versions

Product contract

2.0

Methodology

2.1

Editorial templates

2.4

Schema

0.7.0

Current build

2026-08-12T23:34:59.370714+00:00

Change log

• Editorial 2.4: removed private-staging launch placeholders, froze the NPPES Type 1 provider-scope explanation, propagated the public corrections contact, and aligned Q&A company-overlap wording with the M5.4 exact-reporting-entity contract.
• Editorial 2.3: added canonical social metadata, conservative provider structured data, visible breadcrumbs, and production SEO/GEO release assets without changing evidence semantics.
• Editorial 2.2: updated Doctor Core same-company overlap wording to state exact reporting-entity evidence without ownership/manufacturer inference.
• Editorial 2.1: completed the M5 trust layer and made search-result negative wording dataset-specific.
• Methodology 2.1: productionized the M5.4 exact same-reporting-entity evidence gate, generic guard, and non-ownership interpretation for Analysis 2.
• Methodology 2.0: froze the two independent product relationships, non-causal wording, source-completeness rules, and evidence-bearing company resolution.

See how to report a possible matching or display error.