Statistical flags indicate unusual patterns — not proof of fraud or wrongdoing. Read our methodology

Guide

How Medicaid Fraud Works

Common schemes, red flags, and how data analysis can detect them.

The Scale of the Problem

Medicaid fraud is estimated to cost taxpayers $100 billion or more annually — roughly 10% of total program spending. The HHS Office of Inspector General (OIG) recovers less than 5% of that each year.

The problem is structural: Medicaid processes billions of claims per year through a fee-for-service model that pays providers for each service billed. This creates inherent incentives to over-bill. With 617,000+ active providers and 10,000+ billing codes, manual auditing can only catch a fraction of abuse.

That's why data-driven detection matters. By analyzing 227 million billing records at scale, patterns emerge that would be invisible to traditional auditing. Our analysis flagged 1,860 providers with unusual billing patterns worth investigating.

Common Fraud Schemes

👻

Phantom Billing

Billing for services, procedures, or supplies that were never actually provided to a patient. This is the most straightforward form of fraud — pure fabrication of claims.

Example: A provider submits claims for 50 home health visits that never occurred.

⬆️

Upcoding

Billing for a more expensive service or procedure than what was actually provided. By substituting higher-paying billing codes, providers can dramatically inflate their reimbursements.

Example: Billing a comprehensive office visit (99215) when only a basic visit (99213) occurred.

📦

Unbundling

Splitting services that should be billed as a single bundled code into multiple separate charges to increase total reimbursement.

Example: Billing lab tests individually instead of as a panel, charging 3× more than the bundled rate.

🤝

Kickbacks & Referral Schemes

Paying or receiving payment for patient referrals. Federal anti-kickback laws prohibit any form of compensation for referring Medicaid patients.

Example: A lab paying doctors $50 per patient referral for unnecessary blood tests.

🔄

Patient Churning

Scheduling unnecessary return visits or repeat services to generate additional billable claims. Patients may not realize they’re being used to pad billing.

Example: Requiring weekly visits for a condition that only needs monthly monitoring.

🎭

Identity Fraud

Using stolen patient identities or Medicaid numbers to bill for services. Sometimes involves enrolling fake patients or billing for deceased individuals.

Example: A provider billing under 200 patient IDs, many of whom have never visited the facility.

How Data Analysis Detects Fraud

Statistical Testing

We run 9 statistical tests that compare each provider's billing against national benchmarks for their specific procedure codes. Tests include cost outlier detection, billing swing analysis, new entrant monitoring, and rate outlier identification. Providers are flagged when their patterns deviate significantly from peers billing the same codes. Read full methodology →

Machine Learning

Our Random Forest ML model (AUC 0.77) was trained on 514 providers confirmed as fraudulent by the OIG Exclusion List. It scores every provider on how similar their billing patterns are to known fraud cases — catching subtle multi-feature patterns that rule-based tests might miss. Learn about the ML model →

Advanced Techniques

We also apply Benford's Law analysis (digit distribution anomalies), CUSUM change-point detection (sudden billing shifts), billing velocity analysis (impossible claim volumes), and peer similarity clustering. When multiple independent methods flag the same provider, the probability of a false positive drops dramatically.

What We Found

1,860

providers flagged

$226B

in flagged spending

0

on OIG exclusion list

24

investigations published

None of our flagged providers appear on the OIG's existing exclusion list — suggesting our analysis surfaces new suspicious activity that hasn't been investigated yet. Explore the findings in our 24 investigation articles.

Explore the Data Yourself

Search any provider, filter by state, or browse our investigations.

Related Guides