The Shift: From Fraud Detector to Gatekeeper
The Medicaid integrity industry has spent decades building fraud detectors, systems that identify bad payments after they've already gone out the door. The result: billions spent on recovery operations chasing money that should never have left the system in the first place.
Sentinel is built on a different premise: the data to prevent bad payments already exists. It's in the CMS Coverage Database. In the ICD-10 code set. In the NPI Registry. In seven years of national Medicaid claims history. The problem was never a lack of data, it was a lack of anyone cross-referencing that data against what MCOs actually do.
A gatekeeper doesn't wait for the fraud to happen. It stands at the gate with the federal evidence to ensure that every payment complies, every denial is justified, and every MCO is accountable, before a single improper dollar moves.
The Data Advantage: What the Gatekeeper Sees
Sentinel integrates five data layers that, together, create the most comprehensive Medicaid accountability infrastructure available. Each layer serves a specific gatekeeper function:
| Data Source | Gatekeeper Function |
|---|---|
| 150M+ Medicaid Claims (2018 to 2024) | Establishes behavioral baselines across 7 years. Every anomaly measured against the deepest independent claims intelligence in the country, not a sample, not an estimate. |
| 207,638 CMS Coverage Determinations | Validates every MCO denial against the federal rule that governs it. When the LCD says a service is covered and the MCO denied it, the gatekeeper proves non-compliance automatically. |
| 98,186 ICD-10-CM Diagnostic Codes | Catches upcoding and unbundling in real time by validating submitted codes against the clinical scenario. DRG gaming detected when observation downgrades reduce payments by 75% on services the LCD covers as inpatient. |
| 8,962+ Verified Provider NPIs | Identifies ghost providers, deactivated entities, and credentialing fraud before claims are submitted, not after they're paid. Taxonomy codes verify billing authorization for each service. |
| 35M+ PubMed Clinical Articles | Counters MCO medical necessity denials with peer-reviewed evidence. When an MCO claims a BH admission wasn't necessary, the gatekeeper attaches the literature proving accepted clinical standards. |
Why the Gatekeeper Model Matters Now
The Kansas Evidence
The October 2025 Kansas Medicaid Inspector General performance audit (154 pages, covering January 2021 through December 2023) documents exactly why the detection-only model has failed:
Between 2021 and 2024, the Kansas Medicaid IG conducted 280 investigations resulting in 2 prosecutions and less than $1,000 recovered. The IG identified $193 million in HCBS payments to potentially inactive beneficiaries. One MCO denied $1.8 billion in claims, 70% from hospitals. Prior authorization denial rates exceeded 50% at one MCO. Hospital claims represented just 6 to 7% of denied claim volume but 64 to 67% of all denied dollars.
The IG found that UHC owns its proprietary prior authorization criteria tool, a structural conflict of interest where the MCO that decides what gets paid also controls the criteria for payment decisions. KDHE's response to most findings: "acknowledged, no action taken."
The problem isn't a lack of audits. It's a lack of intelligence infrastructure that can turn audit findings into operational accountability. That's the gatekeeper's role.
What the Gatekeeper Changes
Detection Model (Industry Standard)
- Identifies anomalies after payment
- Flags statistical outliers in claims data
- Produces alerts that require manual investigation
- No federal coverage rule validation
- No MCO accountability metrics
- Recovery-focused: chasing money already lost
Gatekeeper Model (Sentinel)
- Validates compliance before payment processes
- Cross-references every claim against federal evidence
- Produces provable findings with evidence packages
- Every denial checked against CMS LCD/NCD rules
- MCO compliance scorecards grounded in federal data
- Prevention-focused: stopping bad payments at the gate
The Behavioral Health Integrity Crisis
Behavioral health is where the gatekeeper model provides the sharpest contrast with traditional fraud detection. Our national claims data shows a sustained decline in MCO behavioral health reimbursement. This isn't a coding problem, it's a structural misalignment between the capitation payments MCOs receive for BH coverage and the BH services they actually approve and pay for.
MCOs receive per-member-per-month capitation that includes behavioral health benefits. Yet MCO prior authorization denial rates for BH services consistently exceed rates for physical health services. BH providers face the most aggressive PA requirements, the narrowest network adequacy standards, and the heaviest medical necessity burden.
The result: MCOs collect capitation for BH coverage while systematically suppressing BH reimbursement. Sentinel quantifies this gap at the MCO level, the service category level, and the provider level. With federal data integration, we can now prove whether each BH denial complies with the LCD/NCD coverage rules, and when it doesn't, we can calculate the exact dollar amount the MCO retained in capitation for services it refused to deliver.
National Scalability: Built for Every State
Kansas has three MCOs. Nationally, Medicaid managed care operates in 40+ states with hundreds of MCOs managing $400+ billion in annual Medicaid spending. The fraud vectors, MCO accountability gaps, and oversight failures Sentinel identified in Kansas are structural, they exist in every managed care state because the incentive misalignments are identical.
The 36 fraud vectors are universal. The federal databases are national. The only state-specific component is the Layer 2 regulatory overlay, state MCO rules, fee schedules, and IG findings, and that can be built in weeks, not months.
The Bottom Line
The fraud detection industry asks: "What went wrong?"
The gatekeeper asks: "Should this payment go through?"
With 150M+ claims, four federal databases, and a 36-vector fraud prevention matrix mapped to IG audit findings, Sentinel has the data to answer that question, for every claim, every MCO, every state. That's not fraud detection. That's the gatekeeper.
Request a Medicaid integrity assessment for your state by reaching out through the contact page on sentinelintegritygroup.com.
Data Sources & Defensibility
| Source | Details |
|---|---|
| CMS Medicaid Provider Summary | 150M+ rows, all states, 2018 to 2024 |
| CMS Medicare Coverage Database | 207,638 LCD records + NCD database, released 04/02/2026 |
| CDC/CMS ICD-10-CM FY2026 | 98,186 diagnostic codes |
| CMS NPPES NPI Registry V2 | 8,962 providers across 13 target states |
| NCBI PubMed E-utilities | 35M+ biomedical articles |
| Kansas Medicaid IG Audit | October 2025, 154 pages, Jan 2021 to Dec 2023 |
| Sentinel Fraud Prevention Matrix | 36 vectors, 20 IG findings, 12 solution modules |
| KanCare MCO Provider Manuals | Sunflower, UHC Community Plan, Healthy Blue |