rosterproof

Unified provider exclusion lists — OIG LEIE + SAM + FDA + state Medicaid

RosterProof unifies the federal OIG LEIE, the full SAM.gov exclusions extract, the FDA debarment and clinical-investigator disqualification lists, and state Medicaid exclusion lists into a single dataset: collected continuously, normalized into one schema, tracked with full change history, and resolved to NPIs with transparent confidence tiers and provenance on every derived fact. Delivered as native tables on the Snowflake Marketplace, so exclusion screening becomes a simple join in your own warehouse — no scrapers to babysit, no PDFs to retype, no blind spots between the federal list and the states.

346K+
exclusion records, history-tracked
Federal + 40 states + DC
OIG LEIE + SAM.gov + FDA lists plus state Medicaid exclusion lists
~54%
of state exclusions never reach the federal LEIE — see the full analysis
58K+
NPPES-verified NPI matches — screen claims by NPI, not name-matching
7.7K+
providers excluded in 2+ jurisdictions — under ACA §6501, one state's exclusion is every state's problem
Weekly
automated refresh with full change feed
TRY IT NOW — FREE

Screen your providers in 30 seconds

Paste NPIs, upload a spreadsheet, or add names — providers, staff, vendors, and companies. RosterProof checks every entry against the OIG LEIE, SAM.gov, the FDA lists and 40 state Medicaid exclusion lists, shows each hit with its source, status and match confidence, and gives you a report to download. No install, no Snowflake account — your list is checked in your browser against the live API.

Screen my providers →   or use the API

Always verify a hit against the source agency before adverse action.

WHY IT MATTERS

Screening only the federal list misses half the risk

States exclude providers for reasons that never propagate to the federal LEIE — and when they do propagate, it happens with months of lag. The exposure is concrete: payments involving excluded parties risk overpayment recoupment and Civil Monetary Penalties, and under ACA §6501 an exclusion in one state's Medicaid program is grounds for termination in every other state. RosterProof closes that gap for providers and vendors — billing companies, DME suppliers, staffing agencies appear on these lists as entities — with every record carrying its source, jurisdiction, exclusion window, and an NPI match with stated confidence, so a compliance analyst can defend every hit without asking anyone.

WHO IT'S FOR

Three audiences, one dataset

Providers & health systems

Screen staff, medical staff and vendors monthly against every federal and state list, run a claims lookback when a hit surfaces, and check owners and managing employees on enrollment and revalidation — a single join against your HR, credentialing and vendor-master tables.

Payers & managed care organizations

Network credentialing, monthly monitoring and pre-payment claims edits from one feed: paying an excluded provider means recoupment and penalties — 42 CFR 438.608 for Medicaid managed care, and FDR screening obligations for Medicare Advantage — and under ACA §6501 one state's exclusion is every state's problem.

Life sciences

HCP engagement screening — speakers, consultants and investigators paid by manufacturers must be screened against OIG and SAM — and clinical-trial investigator vetting against FDA debarment and disqualification, now unified with the exclusion lists in one join. How life-sciences teams use it →

RosterProof is the exclusion and owner-screening layer that feeds compliance, payment-integrity, and fraud programs — not a fraud detector.

Every format states publish. One schema you query.

CSVExcelPDFJSON endpointsJavaScript portalsEmail-delivered files

We normalize all of it into a single canonical schema, so your screening query never cares how a state chose to publish.

FROM THE DATA

Even the exclusions that reach the federal list arrive five months late

State exclusions that eventually make it to the LEIE take a median 156 days to get there — a quarter take more than fifteen months, and half never arrive at all. That, the specialty risk map, the watchlist of excluded providers practicing where no list can see them, and more: what the data says →

GET THE DATA

Three ways in, one dataset

Snowflake Marketplace

Live and free. In Snowsight → Marketplace, search “RosterProof” → Get → query the PUBLISHED_V1 views in minutes. No forms.

Get it on Snowflake →

Download the weekly snapshot

CSV or Parquet with a manifest and attribution. Enter an email, get a 15-minute link. No Snowflake account needed.

Download files →

API and provider screening

Check NPIs, names and organizations against every list. Free, 500 lookups per month.

Screen my providers →

All three come from the same weekly run and always match.

Need another way to get the data? support@rosterproof.com

Before you use the data: it is free for compliance screening under our Terms of Use. Verify every hit with the source agency before acting on it, and note that some state records (North Carolina, Texas) are non-commercial by their agencies' terms — see Attribution & Notices.

The federal gap, on the map

Share of each state's active exclusions that never appear on the federal LEIE. Live from the published dataset — refreshed weekly. Hover or tap a state. Full methodology and per-state table: The federal gap, measured.

Top ten: the biggest federal blind spots

The ten states where the largest share of active exclusions is missing from the federal LEIE — the compliance exposure a federal-only screen never sees.

How matching works: four tiers, strongest first

Every match carries its tier, evidence, and confidence. Unresolved records stay honestly unresolved.

T1
Source-published NPI, NPPES-verifiedthe excluding agency printed an NPI and the CMS NPPES registry confirms the identity
T2
Unique identity match (name+state)exactly one provider in the national registry shares the normalized name and state — including registered maiden/former/dba aliases
T3
License-number corroboratedthe state license number on the exclusion agrees with the registry
T4
AI-adjudicated, human review requireda large language model weighs the remaining evidence, writes its reasoning, and every verdict is flagged for human review