T-MSIS provider-level spending data, Jan 2018 – Dec 2024
This tool visualizes Medicaid provider spending data from the CMS Transformed Medicaid Statistical Information System (T-MSIS). The dataset covers January 2018 through December 2024 and contains provider-level billing records aggregated at the Provider (NPI) x HCPCS Code x Month level.
The underlying data comes from T-MSIS Analytic Files (TAF), which CMS publishes to support Medicaid research. Each row represents one provider's billing activity for a single procedure code in a single month.
CMS applies cell suppression to protect beneficiary privacy. Rows where a provider filed fewer than 12 claims for a procedure in a month are dropped from the dataset. This means the data skews toward higher-volume billing and may undercount rare or low-volume services.
All dollar figures are shown in nominal terms (not adjusted for inflation). When comparing across years, keep in mind that general price levels have risen — particularly during 2021–2024.
For each provider-procedure pair, we compute a Cost Index = (provider's cost per claim) / (procedure median cost per claim). A cost index of 2.0x means a provider charges twice the procedure median. This enables apples-to-apples comparison across different procedures with very different baseline costs.
Provider growth metrics split billing history at July 2021 (approximate midpoint). We compare the early period (pre-Jul 2021) total against the late period (Jul 2021+) to calculate spending growth %, cost-per-claim growth %, and volume growth %.
Providers are flagged as outliers when their cost index exceeds 2.0x (charging over twice the median) or falls below 0.5x (charging less than half). Only provider-procedure pairs with at least 100 claims and $10,000 in total payments are considered.
The raw dataset does not include provider location. We cross-reference each billing NPI against the NPPES (National Plan & Provider Enumeration System) full replacement file published by CMS. This gives us practice location state and provider name for the majority of billing NPIs. State-level spending totals are aggregated from all 227M rows where the billing NPI has a known state.
Beneficiary counts in this dataset are summed across months. The same individual receiving services in January and February is counted twice. Per-beneficiary cost metrics should be interpreted as approximations — they overestimate costs per unique patient.
HCPCS Level II code descriptions come from the NLM Clinical Tables API. Numeric CPT code descriptions are sourced from the AMA CPT reference. Dental D-codes use ADA CDT nomenclature. Some state-specific codes (W, X prefixes) may not have standardized descriptions.
November and December 2024 show spending approximately 21% and 67% below typical monthly levels, respectively. This is due to CMS reporting lag — states submit T-MSIS data with a 3–6 month delay, so the most recent months are incomplete.