ProviderScore Exclusion Screening

Description

Provider exclusion and sanction screening data, unioned from 6 distinct government source lists into a single table with a source_list discriminator column: OIG LEIE (federal healthcare-program exclusions), GSA SAM (federal procurement/contracting exclusions), FDA debarment (drug product application debarments), state medical/nursing/pharmacy board sanctions (13 states), state Medicaid program exclusions, and OFAC SDN (Treasury sanctions — not healthcare-specific, but a standard part of healthcare-vendor compliance screening). This is the same data underlying ProviderTrust/Streamline Verify-style compliance screening products. Columns not present in a given source's native schema are populated with an explicit typed NULL so the union is well-formed — see the per-source notes below for which columns are structurally absent (not just sparse) for each source_list.

  • Source: OIG (oig.hhs.gov), GSA SAM.gov, FDA, individual state medical/nursing/pharmacy boards, individual state Medicaid agencies, US Treasury OFAC.
  • Refresh cadence: varies by source — OIG LEIE refreshes monthly; GSA SAM, OFAC, and state lists refresh on their own publisher schedules.
  • Row count: 525,932 (2026-08-03 export). The per-source table below is from the 2026-07-15 export (528,383 rows) and has not been re-derived — the six sources re-ingest weekly, so re-measure before quoting a split.
  • Vintage: 2026-05-30 (max refresh-indicator column observed across the 6 sources at export time — gsa_sam_exclusions.created_at / state_board_sanctions.ingested_at).
  • Coverage: see the per-source breakdown below; NPI coverage varies sharply by source and should not be relied on as a universal join key.

Per-source row counts and NPI coverage (this export)

source_list Rows Has NPI Has state Notes
oig_leie 83,256 8,608 (10%) 57 distinct values (all 50 states + territories) Federal exclusions from Medicare/Medicaid/all federal healthcare programs.
sam 168,715 43,457 (26%) 65 distinct values (states, territories, "XX" placeholder for unknown/foreign) Federal procurement exclusions — broader than healthcare; includes non-provider entities.
fda_debarment 149 0 (0%) 0 (never populated) Drug-application debarments; no state/NPI concept in the source.
state_board 174,403 17,254 (10%) 13 states only: CO, CT, DE, FL, IL, KY, MD, MO, NY, PA, TN, TX, WA Medical/nursing/pharmacy board disciplinary actions. Largest single source by row count but narrowest geographic coverage — do not imply 50-state coverage.
state_medicaid 82,810 12,113 (15%) 38 distinct states State Medicaid program exclusions, aggregated from each state's own downloadable list.
ofac_sdn 19,050 0 (0%) 0 (never populated) Treasury sanctions list; not healthcare-specific, included for standard vendor-compliance screening bundles.

Table: provider_exclusions

Snowflake: HEALTHPARSE_DATA.PROVIDER_EXCLUSIONS.PROVIDER_EXCLUSIONS

Column Type Description Notes
source_list VARCHAR Which of the 6 source lists this row came from: oig_leie, sam, fda_debarment, state_board, state_medicaid, ofac_sdn. Always filter/group on this before comparing rows across sources — the other columns' semantics shift per source (see table above).
entity_name VARCHAR The excluded/sanctioned individual or entity's name. For oig_leie, this is business_name if present, else `first_name
npi VARCHAR National Provider Identifier, when the source list carries one (or one was matched post-import). Structurally absent (always null) for fda_debarment and ofac_sdn. Populated for only 10–26% of rows in the 4 sources that carry it — see coverage table above.
state VARCHAR Two-letter state code (or similar) associated with the record, when the source has a state concept. Structurally absent for fda_debarment and ofac_sdn. sam includes non-standard values (territory codes, "XX" placeholder) beyond the 50 states.
exclusion_type VARCHAR The specific exclusion/sanction category, semantics vary entirely by source_list: OIG LEIE statutory authority code (e.g. 1128a1, 1128b5); SAM classification (e.g. "Ineligible (Proceedings Completed)"); FDA debarment status/term (e.g. "Permanent", "25 Year"); state board action type (e.g. "Dismissed", "revocation"); always medicaid_exclusion for state_medicaid; OFAC sanctions program (e.g. "CUBA"). Do not treat this as a single controlled vocabulary — it's 6 different source vocabularies passed through unchanged.
action_date DATE Date of the exclusion/debarment/sanction action. Sanity-guarded at export time: any raw value outside 1960-01-01–(today+1 day) is set to NULL rather than passed through (a known source artifact: one state_board_sanctions row carried 2909-07-03 in the live DB).
reinstatement_date DATE Date the individual/entity was reinstated (no longer excluded), when applicable. Same sanity guard as action_date. Null does not necessarily mean "still excluded" — it can also mean the source simply doesn't track reinstatement for that record type.
source_url VARCHAR Link to the specific state/agency source page or dataset this record came from. Structurally absent for oig_leie and sam (no per-record URL in those source formats) and for ofac_sdn.
_dataset / _vintage / _exported_at VARCHAR See README. —

Example queries (Snowflake)

-- 1. Screen a name against all 6 sources at once ("has this person/entity
--    ever appeared on any federal or state exclusion list").
SELECT "source_list", "entity_name", "state", "exclusion_type", "action_date", "reinstatement_date"
FROM HEALTHPARSE_DATA.PROVIDER_EXCLUSIONS.PROVIDER_EXCLUSIONS
WHERE "entity_name" ILIKE '%Smith, John%'
ORDER BY "action_date" DESC NULLS LAST;

-- 2. Currently-active OIG LEIE exclusions (no reinstatement on record) by NPI.
SELECT "entity_name", "npi", "state", "exclusion_type", "action_date"
FROM HEALTHPARSE_DATA.PROVIDER_EXCLUSIONS.PROVIDER_EXCLUSIONS
WHERE "source_list" = 'oig_leie'
  AND "npi" IS NOT NULL
  AND "reinstatement_date" IS NULL
ORDER BY "action_date" DESC;

-- 3. Counts by source and state, to scope a compliance program to the
--    states where state-board coverage actually exists.
SELECT "source_list", "state", count(*) AS n
FROM HEALTHPARSE_DATA.PROVIDER_EXCLUSIONS.PROVIDER_EXCLUSIONS
WHERE "source_list" = 'state_board'
GROUP BY "source_list", "state"
ORDER BY n DESC;