CMS Medicaid Provider Spending Dataset

About the Dataset

The CMS Medicaid Provider Spending Dataset delivers 200+ million outpatient Medicaid claims aggregated by provider, procedure, and month across 50+ states — automatically updated when CMS publishes new T-MSIS (Transformed Medicaid Statistical Information System) data. This analytics-ready product enriches spending records with NPI provider demographics from NPPES and state-level data quality ratings from the CMS DQ Atlas. Designed for health policy analysts, Medicaid program administrators, and healthcare consultants who need provider-level Medicaid expenditure intelligence without months of data wrangling.

Coming Soon — This product is being prepared for marketplace listing. Request a Free Trial

Quick Access

Tables: PROVIDER_SPENDING, BILLING_PROVIDERS, SERVICING_PROVIDERS, HCPCS_CODES, STATE_QUALITY Sources: CMS T-MSIS Analytic Files, NPPES, CMS DQ Atlas Automatic Updates: Pipeline monitors CMS for new releases and loads data within days of publication Coverage: 50+ US states and territories, 2018-present (7+ years)

Overview

The CMS Medicaid Provider Spending Dataset provides comprehensive access to Medicaid outpatient spending data including:

• PROVIDER_SPENDING (spending) - 200M+ claims aggregated by billing provider NPI, servicing provider NPI, HCPCS procedure code, and month — with total paid, claims, and beneficiary counts

• BILLING_PROVIDERS (billing_providers) - 600K+ billing provider demographics including name, address, specialty taxonomy, and NPI assignment date

• SERVICING_PROVIDERS (servicing_providers) - 1.5M+ servicing provider demographics with the same NPI-enriched fields

• HCPCS_CODES (hcpcs_codes) - 20K+ HCPCS Level II and CPT procedure code descriptions from CMS reference files

• STATE_QUALITY (state_quality) - CMS DQ Atlas state-level data quality ratings for T-MSIS expenditure data across 54 states/territories

Metadata Tables

Every Dataplex data product includes these standard metadata tables:

Table	Purpose
FEEDS	Dataset catalog — available tables, descriptions, update dates
FEEDS_FILES	Batch load history with is_latest flag for data freshness
CHANGELOG	Change log — data loads, schema changes, corrections
DATA_DICTIONARY	Column descriptions for all tables

Entity Relationship Diagram

CMS Medicaid Provider Spending Entity Relationship Diagram

PROVIDER_SPENDING is the central fact table. Join to BILLING_PROVIDERS via BILLING_PROVIDER_NPI_NUM = NPI, to SERVICING_PROVIDERS via SERVICING_PROVIDER_NPI_NUM = NPI, and to HCPCS_CODES via HCPCS_CODE (LEFT JOIN — ~83% coverage). STATE_QUALITY joins to BILLING_PROVIDERS via STATE = STATE_CODE for quality-adjusted analysis. All tables link to FEEDS and FEEDS_FILES via feed_id and feeds_files_id for data lineage.

Medicaid Provider Spending Tables

Provider Spending (PROVIDER_SPENDING)

Medicaid outpatient claims aggregated by billing provider, servicing provider, HCPCS procedure code, and month. The core fact table with 200M+ rows spanning 7+ years of monthly data.

Key Features:

• 200M+ rows — one per billing NPI + servicing NPI + HCPCS code + month

• Total paid, total claims, unique beneficiaries per aggregation

• Computed metrics: average paid per claim and per beneficiary

• CLAIM_FROM_DATE enables native date arithmetic and time-series functions

• Privacy floor: minimum 12 beneficiaries per row (CMS suppression)

Column Reference

Column	Type	Description
BILLING_PROVIDER_NPI_NUM	VARCHAR	National Provider Identifier of the billing provider
SERVICING_PROVIDER_NPI_NUM	VARCHAR	National Provider Identifier of the servicing provider
HCPCS_CODE	VARCHAR	Healthcare Common Procedure Coding System Level II code
CLAIM_FROM_MONTH	VARCHAR	Month of claim service (YYYY-MM format)
CLAIM_FROM_DATE	DATE	First day of the claim month (enables date arithmetic)
TOTAL_UNIQUE_BENEFICIARIES	NUMBER	Number of unique Medicaid beneficiaries served (min 12 for privacy)
TOTAL_CLAIMS	NUMBER	Total number of Medicaid claims submitted
TOTAL_PAID	NUMBER	Total Medicaid amount paid in USD
AVG_PAID_PER_CLAIM	NUMBER	Average payment per claim (TOTAL_PAID / TOTAL_CLAIMS)
AVG_PAID_PER_BENEFICIARY	NUMBER	Average payment per beneficiary (TOTAL_PAID / TOTAL_UNIQUE_BENEFICIARIES)
feed_id	VARCHAR	FK to FEEDS — identifies which dataset this row belongs to
feeds_files_id	VARCHAR	FK to FEEDS_FILES — identifies which batch loaded this data
created_at	TIMESTAMP	When the source data was loaded into the warehouse
updated_at	TIMESTAMP	When dbt last rebuilt this table

---

Billing Providers (BILLING_PROVIDERS)

Billing provider demographics enriched with NPI data — 600K+ unique NPIs with name, address, specialty, and credentials.

Key Features:

• 600K+ unique billing NPIs

• NPI-linked demographics: name, address, phone, taxonomy code

• Entity type distinguishes individuals (1) from organizations (2)

• State field for geographic analysis (mostly 2-letter codes; ~0.01% non-standard from NPPES)

Column Reference

Column	Type	Description
NPI	VARCHAR	National Provider Identifier
ENTITY_TYPE	VARCHAR	1 = Individual provider, 2 = Organization
ORG_NAME	VARCHAR	Organization name (NULL for individual providers)
LAST_NAME	VARCHAR	Provider last name (NULL for organizations)
FIRST_NAME	VARCHAR	Provider first name (NULL for organizations)
MIDDLE_NAME	VARCHAR	Provider middle name
CREDENTIAL	VARCHAR	Provider credential (MD, DO, NP, etc.)
ADDRESS_LINE1	VARCHAR	Practice location street address
CITY	VARCHAR	Practice location city
STATE	VARCHAR	Practice location state (2-letter code)
ZIP	VARCHAR	Practice location ZIP code (5-digit)
PHONE	VARCHAR	Practice phone number
SEX	VARCHAR	Provider sex (individuals only)
TAXONOMY_CODE	VARCHAR	Primary healthcare provider taxonomy code
ENUMERATION_DATE	DATE	Date NPI was assigned
feed_id	VARCHAR	FK to FEEDS — identifies which dataset this row belongs to
feeds_files_id	VARCHAR	FK to FEEDS_FILES — identifies which batch loaded this data
created_at	TIMESTAMP	When the source data was loaded into the warehouse
updated_at	TIMESTAMP	When dbt last rebuilt this table

---

Servicing Providers (SERVICING_PROVIDERS)

Servicing provider demographics — 1.5M+ unique NPIs. Same schema as BILLING_PROVIDERS. The servicing provider is the individual or organization that actually performed the service.

Key Features:

• 1.5M+ unique servicing NPIs

• Same NPI-linked fields as BILLING_PROVIDERS

• Larger population because servicing includes both billing and non-billing providers

Column Reference

Same columns as BILLING_PROVIDERS above.

---

HCPCS Codes (HCPCS_CODES)

HCPCS Level II and CPT procedure/service code descriptions. Combines CMS Physician Fee Schedule (CPT codes) and CMS HCPCS quarterly file (Level II alpha-prefix codes).

Key Features:

• 20K+ procedure codes with descriptions

• Covers ~83% of distinct codes in PROVIDER_SPENDING

• Remaining ~17% are temporary codes, state-specific codes, and ADA dental codes — use LEFT JOIN

Column Reference

Column	Type	Description
HCPCS_CODE	VARCHAR	HCPCS Level II (alpha prefix A-V) or CPT (numeric) code
DESCRIPTION	VARCHAR	Short description of the procedure or service
feed_id	VARCHAR	FK to FEEDS — identifies which dataset this row belongs to
feeds_files_id	VARCHAR	FK to FEEDS_FILES — identifies which batch loaded this data
created_at	TIMESTAMP	When the source data was loaded into the warehouse
updated_at	TIMESTAMP	When dbt last rebuilt this table

---

State Quality (STATE_QUALITY)

CMS DQ Atlas state-level data quality ratings for T-MSIS spending data. CMS rates most states as "Unusable" under their stringent DQ Atlas methodology — this reflects CMS's quality standards, not that the spending data is invalid.

Key Features:

• 2,400+ quality assessments across states and topic areas

• OVERALL_QUALITY rollup across 4 spending-relevant topics

• Use to flag states with known CMS-identified reporting gaps

• Source: download.medicaid.gov DQ Atlas bulk CSV

Column Reference

Column	Type	Description
STATE_CODE	VARCHAR	Two-letter US state/territory code
STATE_NAME	VARCHAR	Full state/territory name
MEASURE_NAME	VARCHAR	CMS DQ Atlas measure name (e.g., 'IP Stays', 'Enrollment Counts')
TOPIC_AREA	VARCHAR	DQ Atlas topic area grouping
RATING	VARCHAR	Quality rating for this state + measure (Low/Medium/High Concern, Unusable)
OVERALL_QUALITY	VARCHAR	Worst-of quality rollup across spending-relevant DQ Atlas topics
feed_id	VARCHAR	FK to FEEDS — identifies which dataset this row belongs to
feeds_files_id	VARCHAR	FK to FEEDS_FILES — identifies which batch loaded this data
created_at	TIMESTAMP	When the source data was loaded into the warehouse
updated_at	TIMESTAMP	When dbt last rebuilt this table

Data Quality

Standardization

• All columns use UPPERCASE naming consistent with Snowflake conventions

• CLAIM_FROM_DATE computed from CLAIM_FROM_MONTH for native date arithmetic

• AVG_PAID_PER_CLAIM and AVG_PAID_PER_BENEFICIARY pre-computed for convenience

• Provider NPI fields validated as 10-digit identifiers

• STATE_QUALITY filtered to 4 spending-relevant topic areas with OVERALL_QUALITY rollup

• HCPCS_CODES merged from 3 CMS reference sources with deduplication

Data Freshness

Check when data was last updated:

Snowflake

SELECT source_name, data_period, row_count, created_at, is_latest
FROM DWV.FEEDS_FILES
WHERE is_latest = TRUE;

Databricks

How to Query CMS Medicaid Provider Spending Data

Platform Schema Reference

This dataset is available on both Snowflake and Databricks. Queries use schema-only references — the database is already set by the share or catalog context:

Platform	Schema	Example
Snowflake	DWV	DWV.PROVIDER_SPENDING
Databricks	cms_tmsis_provider_spending_dwv	cms_tmsis_provider_spending_dwv.provider_spending

Discover Available Data

Start with the FEEDS table to see what's available, and FEEDS_FILES to understand data freshness and load history.

Snowflake

-- Browse the data catalog
SELECT table_name, description, latest_data_period,
       latest_row_count, last_loaded_at, total_batches
FROM DWV.FEEDS
ORDER BY table_name;

Databricks

Snowflake

-- Check data freshness per source
-- is_latest = TRUE marks the current batch for each source
SELECT source_name, data_period, row_count, row_count_delta,
       created_at AS loaded_at, is_latest
FROM DWV.FEEDS_FILES
ORDER BY source_name, created_at DESC;

Databricks

Working with Data Lineage

Every data row links to FEEDS_FILES via feeds_files_id, which tells you exactly which batch loaded that data. Use this to filter to the current data version or trace any row back to its source load.

Snowflake

-- Get current spending data with batch lineage
SELECT
    ps.BILLING_PROVIDER_NPI_NUM,
    ps.HCPCS_CODE,
    ps.TOTAL_PAID,
    ps.TOTAL_CLAIMS,
    ff.data_period,
    ff.created_at AS data_loaded_at
FROM DWV.PROVIDER_SPENDING ps
JOIN DWV.FEEDS_FILES ff ON ps.feeds_files_id = ff.id
WHERE ff.is_latest = TRUE
LIMIT 20;

Databricks

Top Billing Providers by State

Identify the highest-spending Medicaid billing providers in any state.

Snowflake

SELECT bp.STATE, bp.NPI, bp.ORG_NAME, bp.LAST_NAME, bp.FIRST_NAME,
       SUM(ps.TOTAL_PAID) AS total_paid,
       SUM(ps.TOTAL_CLAIMS) AS total_claims
FROM DWV.PROVIDER_SPENDING ps
JOIN DWV.BILLING_PROVIDERS bp ON ps.BILLING_PROVIDER_NPI_NUM = bp.NPI
WHERE bp.STATE = 'CA'
GROUP BY 1, 2, 3, 4, 5
ORDER BY total_paid DESC
LIMIT 25;

Databricks

Monthly Medicaid Spending Trends

Track Medicaid spending and provider participation over 7+ years of monthly data, including pandemic impact periods.

Snowflake

SELECT CLAIM_FROM_MONTH,
       COUNT(DISTINCT BILLING_PROVIDER_NPI_NUM) AS unique_providers,
       SUM(TOTAL_CLAIMS) AS total_claims,
       SUM(TOTAL_PAID) AS total_paid
FROM DWV.PROVIDER_SPENDING
GROUP BY 1
ORDER BY 1;

Databricks

Spending by HCPCS Procedure Category

Analyze which procedures drive the most Medicaid spending.

Snowflake

SELECT ps.HCPCS_CODE, h.DESCRIPTION,
       SUM(ps.TOTAL_PAID) AS total_paid,
       SUM(ps.TOTAL_CLAIMS) AS total_claims,
       SUM(ps.TOTAL_UNIQUE_BENEFICIARIES) AS total_beneficiaries
FROM DWV.PROVIDER_SPENDING ps
LEFT JOIN DWV.HCPCS_CODES h ON ps.HCPCS_CODE = h.HCPCS_CODE
GROUP BY 1, 2
ORDER BY total_paid DESC
LIMIT 50;

Databricks

State Quality-Adjusted Spending Analysis

Cross-reference spending totals with CMS DQ Atlas quality ratings to identify states where data quality may affect analysis reliability.

Snowflake

SELECT bp.STATE, sq.OVERALL_QUALITY,
       SUM(ps.TOTAL_PAID) AS total_paid,
       COUNT(DISTINCT ps.BILLING_PROVIDER_NPI_NUM) AS unique_providers
FROM DWV.PROVIDER_SPENDING ps
JOIN DWV.BILLING_PROVIDERS bp ON ps.BILLING_PROVIDER_NPI_NUM = bp.NPI
LEFT JOIN (
    SELECT DISTINCT STATE_CODE, OVERALL_QUALITY
    FROM DWV.STATE_QUALITY
) sq ON bp.STATE = sq.STATE_CODE
GROUP BY 1, 2
ORDER BY total_paid DESC;

Databricks

Billing Outlier Detection

Find providers billing significantly above state medians for specific procedures — useful for fraud detection and billing compliance.

Snowflake

WITH state_medians AS (
    SELECT bp.STATE, ps.HCPCS_CODE,
           MEDIAN(ps.AVG_PAID_PER_CLAIM) AS median_per_claim
    FROM DWV.PROVIDER_SPENDING ps
    JOIN DWV.BILLING_PROVIDERS bp ON ps.BILLING_PROVIDER_NPI_NUM = bp.NPI
    GROUP BY 1, 2
    HAVING median_per_claim > 1.00
)
SELECT ps.BILLING_PROVIDER_NPI_NUM, bp.STATE, ps.HCPCS_CODE,
       h.DESCRIPTION,
       ps.AVG_PAID_PER_CLAIM, sm.median_per_claim,
       ps.AVG_PAID_PER_CLAIM / sm.median_per_claim AS ratio_to_median
FROM DWV.PROVIDER_SPENDING ps
JOIN DWV.BILLING_PROVIDERS bp ON ps.BILLING_PROVIDER_NPI_NUM = bp.NPI
JOIN state_medians sm ON bp.STATE = sm.STATE AND ps.HCPCS_CODE = sm.HCPCS_CODE
LEFT JOIN DWV.HCPCS_CODES h ON ps.HCPCS_CODE = h.HCPCS_CODE
WHERE ps.AVG_PAID_PER_CLAIM > sm.median_per_claim * 3
ORDER BY ratio_to_median DESC
LIMIT 100;

Tracking Data Changes Over Time

FEEDS_FILES records every batch load with row_count_delta showing what changed. Use this to monitor source data updates.

Snowflake

SELECT source_name, data_period, row_count, row_count_delta,
       created_at AS loaded_at, is_latest
FROM DWV.FEEDS_FILES
WHERE source_name = 'spending'
ORDER BY created_at;

Databricks

Who Uses This Data

Common Use Cases

• Medicaid billing outlier detection — Identify providers billing significantly above state medians by procedure code and geography, flagging potential fraud or billing errors

• Medicaid spending trend analysis — Track spending patterns across 84 months (2018-2024) including COVID-19 pandemic impact on Medicaid utilization and costs

• Cross-state Medicaid comparisons — Compare per-provider and per-procedure spending across 50+ states with quality-adjusted analysis using CMS DQ Atlas ratings

• Provider network analysis — Map billing-to-servicing provider relationships to identify coverage gaps, network concentration, and referral patterns

• HCPCS procedure cost benchmarking — Benchmark procedure-level costs against state and national medians for rate-setting and contract negotiation

• Grant writing and policy research — Access structured Medicaid expenditure data for FQHC grant applications, state Medicaid program evaluations, and federal policy analysis

Related Datasets and Research

This dataset pairs well with:

• CMS NPPES Provider Dataset — Extended NPI provider attributes including additional practice locations, other names, and full taxonomy classification details

• HRSA Healthcare Resources — County-level healthcare workforce and shortage area data for correlating spending patterns with provider supply

• CMS Data Feeds Dataset — Medicare physician spending comparisons via the Medicare Provider Utilization and Payment Data feeds

Frequently Asked Questions

What states are included in the T-MSIS provider spending data? All 50 US states plus DC, Puerto Rico, US Virgin Islands, and Guam — 54 jurisdictions total. Coverage varies by year as states onboarded to T-MSIS between 2015-2020. By 2020, all states report through T-MSIS.

How often is the data updated? CMS publishes new T-MSIS spending data approximately annually. Our pipeline automatically monitors for new releases and loads updated data within days of publication — no manual intervention needed. Coverage currently spans 2018-present and grows with each CMS release.

What does "Unusable" mean in STATE_QUALITY? CMS's DQ Atlas applies stringent quality thresholds to T-MSIS data. Most states (~50 of 54) receive an "Unusable" rating on at least one spending-related measure. This reflects CMS's internal quality standards — it does not mean the spending data is invalid or should be discarded. Use OVERALL_QUALITY to identify states with known reporting gaps and adjust analysis accordingly.

Why do only ~83% of HCPCS codes have descriptions? The HCPCS_CODES table combines CMS's Physician Fee Schedule (CPT codes) and HCPCS Level II quarterly file. The remaining ~17% are temporary local codes (e.g., T-codes, S-codes), state-specific codes, and ADA dental codes not published in national CMS reference files. Always use LEFT JOIN when joining PROVIDER_SPENDING to HCPCS_CODES.

Ready to access CMS Medicaid Provider Spending data?

Platform	Action
Snowflake	Coming soon — listing in preparation
Databricks	Free Trial

Questions? Contact our team for a walkthrough.