About This Project
Mildly Concerning is a public, data-driven platform for objective exploration of Medicaid provider spending patterns.
Data Source
HHS Open Data — Medicaid Provider Spending
Published by the U.S. Department of Health & Human Services (HHS) via the Centers for Medicare & Medicaid Services (CMS).
- Underlying System
- Transformed Medicaid Statistical Information System (T-MSIS)
- Time Coverage
- January 2018 – December 2024
- Granularity
- Provider (NPI) × HCPCS Code × Month
- Claim Types
- Outpatient and professional claims with valid HCPCS codes
- Coverage
- Fee-for-service, managed care, and CHIP
Dataset Fields
| Column | Type | Description |
|---|---|---|
| BILLING_PROVIDER_NPI_NUM | String | National Provider Identifier of the billing provider |
| SERVICING_PROVIDER_NPI_NUM | String | National Provider Identifier of the servicing provider |
| HCPCS_CODE | String | Healthcare Common Procedure Coding System code for the service |
| CLAIM_FROM_MONTH | Date | Month for which claims are aggregated (YYYY-MM-01 format) |
| TOTAL_UNIQUE_BENEFICIARIES | Integer | Count of unique beneficiaries for this provider/procedure/month |
| TOTAL_CLAIMS | Integer | Total number of claims for this provider/procedure/month |
| TOTAL_PAID | Float | Total amount paid by Medicaid (in USD) |
Reference Data & Code Mappings
To provide human-readable context for medical codes, this platform incorporates the following public reference datasets:
HCPCS / CPT Code Descriptions
8,222 procedure code descriptions mapping HCPCS/CPT codes to human-readable labels (e.g., 99213 → “Office/outpatient visit, est patient”).
- Source
- CMS Healthcare Common Procedure Coding System (CPT-4)
- Usage
- Procedure descriptions on all pages displaying HCPCS codes
ICD-10-CM Diagnosis Code Descriptions
74,260 diagnosis code descriptions from the FY2025 ICD-10-CM release. While the Medicaid spending dataset does not include diagnosis codes directly, this reference is available for contextual lookup.
- Version
- FY2025 (October 2024 release)
NPI Registry (Provider Details)
The CMS NPPES NPI registry contains provider names, specialties, and practice addresses. Integrating this data would enable provider name display, specialty-based analysis, and geographic (state-level) views.
- Source
- CMS NPPES NPI Registry
- Status
- Future enhancement
Data Limitations
- Cell suppression: Rows with fewer than 12 total claims are excluded for beneficiary privacy protection.
- No diagnosis codes: The dataset does not include ICD-10 diagnosis codes, so clinical context is not available.
- No beneficiary-level data: Only aggregate counts are provided. No individual patient information is included.
- No risk adjustment: Spending figures are not adjusted for patient acuity or clinical complexity.
- State submissions: Data quality depends on individual state submissions to CMS through the T-MSIS system.
Methodology
All derived metrics on this platform are purely descriptive:
- Avg Paid per Claim = TOTAL_PAID / TOTAL_CLAIMS
- Claims per Beneficiary = TOTAL_CLAIMS / TOTAL_UNIQUE_BENEFICIARIES
- Paid per Beneficiary = TOTAL_PAID / TOTAL_UNIQUE_BENEFICIARIES
- Peer percentiles are computed within each HCPCS code group, comparing a provider against all other providers billing the same code nationally.
- Behavior Deviation Index (BDI) is a composite statistical score (0–100) reflecting price deviation, volume deviation, revenue concentration, and temporal change intensity. It reflects statistical variation only and does not indicate wrongdoing.
Philosophy
This platform does not accuse, infer intent, or evaluate clinical appropriateness. It surfaces distribution, variation, trends, and statistical deviation.
The data speaks. Users interpret.
Contact & Feedback
Message or reply to @no_stp_on_snek on X.