CMS Releases Medicaid Data, But Context Is Key

The Centers for Medicare & Medicaid Services (CMS) is increasingly focused on addressing fraud, waste, and abuse within health programs, including Medicaid. Recent efforts emphasize collaboration between federal and state governments, building upon a foundation established in 2010 with the creation of the Center for Program Integrity (CPI). CPI shifted the approach from a “pay and chase” model to one that prioritizes data analytics for fraud detection and prevention, supporting states through training via the Medicaid Integrity Institute and providing access to comprehensive data sets .

On February 14, 2026, CMS released a provider-level spending data set intended to help identify unusual billing patterns. While this data offers potential insights, it’s crucial to understand its limitations to avoid drawing inaccurate conclusions.

What Does the Data Include?

The newly released data set includes the following information:

• National Provider Identifier (NPI) for the billing provider
• NPI of the servicing provider (individual or organization)
• Procedure code (HCPCS code)
• Month and year
• Number of beneficiaries seen
• Number of procedures delivered (claim count)
• Total amount paid for services

The data encompasses both fee-for-service and Medicaid managed care organization claims from 2018 to 2024.

What Does the Data Exclude?

Significant portions of Medicaid spending are not included in this data set. Notably, it excludes:

• Institutional records (e.g., hospital inpatient care, accounting for 37% of Medicaid spending)
• Information on prescription drug costs

the data lacks crucial context for evaluating service volume and spending, including:

• Enrollment: Medicaid eligibility and enrollment numbers fluctuate based on state policies, economic conditions, and demographics.
• Benefits and Coverage: State-specific benefit packages and eligibility criteria vary and can change over time.
• Payment Rates: Spending is influenced by state-determined payment rates, which consider local cost of living and access to care.
• Diagnoses: The data doesn’t indicate the medical conditions for which procedures were performed.
• Place of Service & Modifiers: Information about where services were delivered (in-person vs. Remote) and other relevant modifiers is absent.

Potential for Misinterpretation

Data analytics are valuable, but relying solely on this data set can lead to flawed conclusions. Several shortcomings should be considered:

• Procedure Comparability: Procedures are not always directly comparable. For example, “personal care” encompasses a wide range of service durations, while psychotherapy has specific codes for different visit lengths.
• Provider Comparability: Providers aren’t uniform. The data includes individual practitioners, group practices, and government agencies that both administer and deliver benefits.
• Data Quality Concerns: The data’s creation methodology and underlying quality are not fully transparent. The Transformed Medicaid Statistical Information System (T-MSIS), the data source, sometimes contains state-specific data issues. CMS maintains a data quality atlas to address these concerns, but it’s unclear how these issues were handled in the released data. In 2024 data, six states had unusable spending information, and sixteen had data of high concern.

Contextual Factors: The Impact of COVID-19

Changes in Medicaid spending and care utilization between 2018 and 2024 were significantly impacted by the COVID-19 pandemic. Increased enrollment during the continuous enrollment period and heightened awareness of behavioral health and long-term care needs led to increased service use and spending as states expanded coverage and adjusted payment rates.

Understanding these limitations is essential for accurate interpretation of the CMS data and effective program integrity efforts.