OIG Exclusion Check (LEIE) — Bulk Roster Screening with Fuzzy Name + DOB and Audit Stamp

Monthly OIG exclusion screening for healthcare compliance teams: screen staff, provider and vendor rosters against the HHS-OIG LEIE excluded-providers list in one run, with fuzzy name matching, DOB/state disambiguation, and a SHA-256 list-version stamp on every row for your audit file.

NPI-only checks vs fuzzy name + DOB

	NPI-only LEIE checkers	This Actor
Coverage	only LEIE rows with an NPI (~11%)	every LEIE row, by name
Typos / swapped names / suffixes	missed	Jaro-Winkler + phonetic fold
Disambiguation	none	DOB + state
Decision	hit / no hit	excluded / review / clear with clear_basis caveats
Audit evidence	rarely	list last_modified, row count, SHA-256, thresholds

Decision rules

Batch-screen staff, provider, and vendor rosters against the HHS-OIG List of Excluded Individuals/Entities (LEIE), downloaded fresh from oig.hhs.gov/exclusions/downloadables/UPDATED.csv on every run. You get one audit row per roster line.

• Exact NPI match (check digit validated) → excluded
• Fuzzy name (typos, swapped first/last, punctuation, suffixes like JR/MD; businesses by token similarity) + DOB + state disambiguation
• excluded only on exact NPI or name ≥ hit threshold with exact DOB. Name-only similarity is always review, never excluded. A near-exact name (≥0.95) with a DOB mismatch is floored at review + DOB_MISMATCH, never clear. An NPI hit whose name scores <0.8 is review + NAME_DIFFERS, not excluded. Last names are compared with Jaro-Winkler plus a phonetic fold (C/K, PH/F, S/Z…) over the full list, so first-letter typos are not lost by blocking.
• Every row is stamped with list last_modified, row count, and SHA-256 of the file screened, plus screened_at and the thresholds used.

Example input

{"roster":[{"first":"James","last":"Dove","dob":"1961-04-02","state":"TX","own_id":"emp-1"},{"business":"Acme Home Health LLC","npi":"1234567893","own_id":"v-9"}]}

All input options

roster: [{first,last,middle?,dob?,npi?,state?,business?,own_id?}] or csvText (same headers, max 10 MB, max 50k rows). reviewThreshold (default 0.90) and hitThreshold (default 0.95). Street address/ZIP are omitted unless includeStreetAddress. Provider/staff/vendor identifiers only. Never upload patient data.

Output sample with audit stamp

{"own_id":"emp-1","match_status":"review","match_basis":"name_similarity_without_dob_confirmation","clear_basis":{"status":"review","basis":"name_similarity_without_dob_confirmation","note":"..."},"best_score":0.9053,
 "candidates":[{"score":0.9053,"leie_name":"JAMES L DOVE","dob_agreement":"not_compared","excl_type":"1128a1","excl_date":"...","state":"TX"}],
 "list_version":{"source":"HHS-OIG LEIE UPDATED.csv","last_modified":"Fri, 09 Oct 2026 12:51:03 GMT","rows":84093,"sha256":"7483d4…"},
 "screened_at":"2026-10-25T03:20:00Z","thresholds":{"review":0.9,"hit":0.95}}

Statuses: excluded | review | clear | invalid_input (not charged). Every row has clear_basis: for clear rows it lists fields_checked and caveats (e.g. MAY_BE_FALSE_NEGATIVE, NO_DOB_SUPPLIED, NPI_ONLY_WEAK).

Pricing (pay per event)

Event	Price
provider-screened (per roster row)	$0.003
potential-match (per row flagged review/excluded)	$0.01
Run start	$0.00005
Add DOB to cut review rows and cost.
$0.003 per provider screened, plus $0.01 per row flagged review/excluded. Common names with no DOB will usually be flagged review, so add DOB to cut noise and cost. Example: a 1k roster with no DOB and many common names can run $10+ (1,000 × $0.003 + ~700+ flagged × $0.01), not $3.30. A heavy-review roster approaches $0.013/row.

Limits & disclaimer

clear can be a false negative. Causes: name variants, maiden/married name changes (our test: a different surname is not caught), data-entry errors in your roster or the LEIE, the ~1% of LEIE individuals with no DOB, and the fact that ~89% of LEIE rows (75,169 of 84,093) have no NPI. An NPI-only clear therefore proves almost nothing. You remain responsible for exclusion screening under OIG guidance. This tool is not a substitute for the OIG online search, SAM.gov, or state Medicaid exclusion lists. Not legal advice and not an eligibility determination. clear means no candidate above threshold in the LEIE file version stamped on the row. It does not mean "never excluded", and it does not cover SAM.gov or state Medicaid lists. OIG advises verifying potential matches by SSN/EIN through its online search. Data: HHS Office of Inspector General, LEIE (public domain). Not affiliated with or endorsed by HHS-OIG. Throughput: about 10–15 rows/s (~40–50k rows/hr) after the list loads, because every name is scanned against the full list. Default timeout 3600 s; split rosters over ~40k rows.

More Actors by willed_nursery

• Healthcare credentialing pair: OIG LEIE Screener ↔ NPI Roster Validator — validate the same roster's NPIs against NPPES.
• KYB/lead: Maps Lead Enricher (Domain Contact Enricher)
• IP docket: USPTO Trademark Deadlines
• Financial risk: SEC Health Score