I have been writing about the April 9 release of the National Provider Directory, and I have been explicit about giving CMS credit for getting it done.
This week, I want to do something that sounds harder but really is not: talk honestly about scope.
Every good piece of infrastructure is scoped. Scope is not a weakness; it is a discipline.
The NPD was built to do specific things, and it does them well. It was not built to do other things, and by design, it does not do them. Understanding that line is the most useful thing a payer leader can do in 2026.
What the NPD does
It publishes a FHIR-formatted national reference dataset of Medicare-enrolled providers, facilities, and endpoints. It aggregates identifiers across NPPES and Medicare enrollment into a single, federally maintained file. It makes that file freely available to any payer, provider, app, or researcher. It gives the industry a common anchor point for NPI validation and endpoint discovery.
That is a meaningful contribution, and it closes a gap that has existed for twenty years.
What the NPD does not do, by design
Here is where payer teams need to be clear-eyed, because these are the workflows that still live with the plan.
Today, the NPD is a snapshot of Medicare enrollment data reformatted into FHIR. The April 9 public-use bulk release contains approximately 7.44 million Practitioner records (the person-level FHIR resource) and 7.18 million PractitionerRole records (the link between a practitioner and an organization). Within that data, 71.30% of practitioners have no active organizational linkage at all, and 44.85% of PractitionerRole records are inactive historical affiliations. Birth dates are absent from every Practitioner record.
Accepting-new-patients status is absent. Languages spoken are present on roughly 2.8% of records, and GPS coverage on location records is 46.64%.
None of that is a criticism. It is an accurate description of what a federal reference dataset looks like when it aggregates what CMS already has, which is exactly what the NPD was designed to do.
What the NPD does not do, then, follows naturally:
- It does not attest. The September 1, 2026 attestation that a CEO, CFO, or COO will personally sign in HPMS is a legally binding declaration about the accuracy of the data your plan publishes, not about the federal reference file.
- It does not validate accepting-new-patients. When a plaintiff’s firm secret-shops a member’s search for a psychiatrist next year, the NPD will not have prevented that search from failing.
- It does not resolve identity across your internal systems. If your credentialing system says Dr. Smith practices at 123 Main Street and your claims system says 456 Oak Avenue, the NPD does not reconcile that.
- It does not cleanse data. It publishes what CMS has.
- It does not cover non-Medicare populations at scale. TRICARE, VA Community Care, FEHB, state Medicaid-specific data, ERISA plans, commercial directories; these remain outside its scope.
Why that scope is appropriate
A federal reference dataset should not be doing attestation, identity resolution, or data cleansing for individual payers. That work is too specific to each plan’s operational reality, and a federal system trying to do it at scale would be slower and less accurate than what well-built private infrastructure can deliver at the plan level.
The NPD is a distribution infrastructure. It closes the gap of “there is no common reference file.” It leaves untouched the gap of “the data entering the payer stack is broken.” Those are two different gaps, and both matter.
Next time, I will write about why Phase 2 in 2027 makes that second gap far more expensive for payers to leave open.
For MA executives working through what NPD-readiness actually means inside their organization, the September 1 attestation focuses the question. Between now and the deadline, the plans that are becoming clearer on the reconciliation layer are the ones that treat it as scoped, measurable work rather than an open-ended IT problem.
Datagence built Polus™ HCP for the reconciliation layer that the NPD intentionally does not cover. We have packaged the diagnostic side of that work as a 30-day NPD Submission Readiness Assessment: fixed scope, fixed fee; documents that drop directly into your September 1 attestation evidence package. Your team’s effort is one file export.
If you are the person whose name will be on the attestation, or the person whose work will be defending it, two paths:
Read the Assessment overview →
Or schedule a 30-minute conversation with me directly →
If you want to dive deeper