The NPI gap in exclusion screening.
Nine in ten records on the federal exclusion list carry no NPI. A screen that matches on NPI alone can test the remaining one in ten, and it finds very little there.
Key figures
OIG LEIE, March 2026; HHS Medicaid provider spending, 2018–2024
- K.1
Of 82,749 records on the OIG List of Excluded Individuals/Entities, 8,482 (10.25%) carry an NPI.
- K.2
The NPI-bearing records cover 8,306 distinct NPIs. 1,280 of them appear in HHS Medicaid provider spending data for 2018 to 2024.
- K.3
A small number of excluded NPIs, fewer than 20, were paid for Medicaid services dated after their exclusion month: about 1.1% of the $724.2 million paid to excluded NPIs across all months.
- K.4
For NPIs excluded from 2019 to 2024 that had earlier Medicaid payments, the median gap between the last month paid and the exclusion month was 29 months.
How many excluded parties have an NPI on the exclusion list?
About one record in ten.
The HHS Office of Inspector General publishes the List of Excluded Individuals/Entities, the LEIE.1 The file we used, downloaded March 14, 2026, holds 82,749 records. Only 8,482 of them, 10.25%, carry an NPI. The other 74,267 identify the excluded party by name and the other fields on the record, with no NPI to match.
Claims carry NPIs, so it is tempting to screen by NPI alone. The arithmetic says what that screen can see: at most one exclusion record in ten. The rest pass through it untested, however carefully the match is run.
Exclusion records that carry an NPI
OIG LEIE, all records. Each square is 1% of the list.
- Record carries an NPI: 8,482 of 82,749 (10.25%)
- No NPI on the record: 74,267
Data table
| Measure | Count |
|---|---|
| All records | 82,749 |
| Records with an NPI | 8,482 (10.25%) |
| Records without an NPI | 74,267 |
| Distinct NPIs | 8,306 |
Did Medicaid pay excluded providers after their exclusion?
For the one record in ten that can be matched, rarely. The 8,482 NPI-bearing records cover 8,306 distinct NPIs. We matched them against HHS’s Medicaid Provider Spending by HCPCS data for service years 2018 to 2024,2 in both the billing and the servicing role. 1,280 appear.
Those NPIs were paid $724.2 million across all months. Almost all of it was for services dated in or before the month of exclusion. Fewer than 20 NPIs were paid for services dated after their exclusion month: about $8.0 million, deduplicated across roles, or 1.1% of the total.
We publish only these totals, and the count only as a range. The group is small enough that an exact count or a breakdown by state, specialty or year could help identify individual providers.
From the exclusion list to the payment record
Excluded NPIs matched to Medicaid provider spending data, service years 2018 to 2024. Each panel is scaled to its own first bar.
Data table
| Measure | Value |
|---|---|
| Distinct excluded NPIs on the list | 8,306 |
| Excluded NPIs appearing in the spending data | 1,280 |
| Excluded NPIs paid for services after the exclusion month (floor) | Fewer than 20 |
| Paid to excluded NPIs, all months | $724.2 million |
| Paid for services after the exclusion month, deduplicated | About $8.0 million (1.1%) |
Why a floor and not an estimate? The match used NPIs only, with no name matching and no state exclusion lists. It used the month a service was furnished, since the file has no paid date. And it counted only payments strictly after the exclusion month. Allowing a three-month lag, or using the earliest exclusion date instead of the latest, barely changes the result.
How long before exclusion does the Medicaid billing stop?
The median is 29 months. We looked at NPIs excluded between 2019 and 2024 that had Medicaid payments before their exclusion. Across those 929 NPIs, the median gap between the last month with a payment and the month of exclusion was 29 months.
That gap shapes what an exclusion screen can do. The list records the end of a matter. By the time a party appears on it, its Medicaid billing has usually stopped. A screen against the list keeps an excluded party from being paid again. Finding the next one requires reading the network while it is still billing.
What does name and entity resolution add to exclusion screening?
It tests the 90% of the list that an NPI cannot reach, and it looks past the enrolled entity to the people behind it. The same entity resolution and network mapping supports work for counsel.
- Matching by name. Excluded individuals and entities are matched by name and business name against provider registries, ownership and officer records.
- Every role a person holds. An excluded individual may surface as an owner, officer or billing agent of an entity whose own NPI is clean. An NPI match on the entity cannot see that.
- One entity across its names and numbers. Resolution joins the records an entity leaves across registries and filings, so a re-enrolled entity is recognized as the same one.
- People review the unclear matches. Names collide. A match is a lead until an investigator confirms it.
Exhibit A · One entity, two screens
Illustrative composite. No real provider depicted.
NPI-only screen: Entity NPI matched to the list
- NPIWithheld
- EntityNo match
- OfficersNot tested
- Billing agentNot tested
Result: Cleared
Name and entity resolution: Entity, people and agents matched to the list
- NPIWithheld
- EntityNo match
- OfficersOne officer matches an excluded individual whose record carries no NPI
- Billing agentShared with a second entity that officer controls
Result: Sent to an investigator for review
Marque resolves provider identities across 9.4 million records cross-referenced against federal registries, and prior exclusions are one of the ten signals scored for every entity in a payment integrity engagement. A name match is never treated as a finding on its own.
Caveats
- i.The count is a floor.
- NPI matching only, with no name matching and no state exclusion lists. A broader match would find more.
- ii.Service month, not paid date.
- The spending file dates payments by the month of service. We count only services dated strictly after the exclusion month.
- iii.Reinstated parties are absent.
- OIG removes reinstated parties from the list, so anyone excluded and reinstated before March 2026 is not counted. That is the conservative direction.
- iv.Small cells are missing.
- HHS drops provider, code and month cells with fewer than 12 claim lines or 12 beneficiaries, so low-volume payments to excluded NPIs may not appear.
- v.A payment is not a finding.
- These totals describe matches between two public files. They are not conclusions about any provider.
Methodology and sources
Note on method
Aggregate statistics only. No provider is named.
- Exclusion list
- HHS Office of Inspector General, List of Excluded Individuals/Entities (LEIE), downloaded March 14, 2026. 82,749 records. Four NPI records with a waiver date were dropped from the payment match.oig.hhs.gov/exclusions
- Payment data
- U.S. Department of Health and Human Services, “Medicaid Provider Spending by HCPCS,” HHS Open Data Platform, version 2026-02-09; file dated April 6, 2026; rows with a billing provider NPI; service months January 2018 to December 2024, 2024 preliminary.opendata.hhs.gov/datasets/medicaid-provider-spending
- Match
- Excluded NPIs matched to the billing and servicing NPI on each row. Where an NPI has several exclusion records, the latest exclusion date is used.
- After exclusion
- A row dated in a month strictly later than the exclusion month. Dollars are deduplicated where an NPI appears in both roles on the same row.
- Lead time
- For NPIs excluded from 2019 through 2024, months between the last month with a payment before exclusion and the exclusion month.
- Checks
- String-month and date-arithmetic queries give identical results. Every figure was also recomputed on a second, locally held copy carrying the same version label; dollar totals agree within about 6%.
- 1HHS Office of Inspector General, Exclusions program and LEIE downloadable database. File downloaded March 14, 2026.
- 2U.S. Department of Health and Human Services, Medicaid Provider Spending by HCPCS, HHS Open Data Platform, version 2026-02-09; file dated April 6, 2026; rows with a billing provider NPI; service months January 2018 to December 2024, 2024 preliminary.
Questions
What percentage of OIG exclusion records include an NPI?
In the LEIE file downloaded March 14, 2026, 8,482 of 82,749 records carry an NPI: 10.25%. The other 74,267 records identify the excluded party without one.
Can you screen providers against the LEIE by NPI alone?
You can, but an NPI-only screen can test only the roughly one record in ten that carries an NPI. It also checks only the enrolled entity, so an excluded individual who is an owner, officer or billing agent of an entity with a clean NPI passes untested. Matching by name and resolving entities is how the rest of the list gets tested.
Did Medicaid pay excluded providers after they were excluded?
Among NPI-bearing exclusion records, rarely. Of 8,306 excluded NPIs, 1,280 appear in HHS Medicaid provider spending data for 2018 to 2024, and fewer than 20 were paid for services dated after their exclusion month. Those payments total about $8.0 million, 1.1% of the $724.2 million paid to excluded NPIs across all months. The figure is a floor because it uses NPI matching only.
How long after a provider’s last Medicaid payment does exclusion usually come?
For 929 NPIs excluded between 2019 and 2024 that had earlier Medicaid payments, the median gap between the last month with a payment and the exclusion month was 29 months.
What is entity resolution in exclusion screening?
Entity resolution recognizes one person or organization across the names, numbers and records it appears under. In exclusion screening, it matches excluded parties by name against registries, ownership, officer and billing-agent records, so an excluded person is found in any role, with people reviewing the unclear matches.