Data access / API + MCP

Access healthcare pricing intelligence via Vlada MCP.

US price data behind one MCP endpoint: what hospitals charge, what insurers actually pay, and the geographically localized Medicare benchmark and cost-to-charge that tell you whether a price is fair. Ask in plain language from Claude or any MCP client. Every answer comes back with its source file, vintage, and a replayable hash.

1 / Connect
bash
claude mcp add --transport http vlada https://mcp.vladahealth.com/mcp

Same Streamable-HTTP URL for Claude Desktop and Cursor. Open while in preview. The authenticated, multi-tenant endpoint is mcp-secure.vladahealth.com (sign in on connect), request access.

2 / The data

Hospital pricing

CMS Hospital Price Transparency (machine-readable standard-charge files)
5,040
Hospitals
53 states + territories
3.25B
Negotiated rates
payer-specific
174M
Gross charges
166M
Cash prices
Negotiated rates by payer

Raw hospital files name payers inconsistently; we normalize them to 701 parent companies (85% of rates mapped).

  • UnitedHealth Group286M rates / 4,154 hosp
  • CVS Health (Aetna)284M rates / 4,005 hosp
  • The Cigna Group200M rates / 3,479 hosp
  • BCBS Michigan161M rates / 1,942 hosp
  • Elevance Health (Anthem)134M rates / 2,366 hosp
  • Centene131M rates / 3,192 hosp
  • MultiPlan116M rates / 2,007 hosp
  • Humana87.6M rates / 3,213 hosp
  • Molina Healthcare51.4M rates / 1,746 hosp
  • Medica46.8M rates / 697 hosp
Negotiated rates by line of business

Each rate is classified by line of business so commercial is never compared against government plans by accident.

  • Commercial1.43B rates
  • Medicaid312M rates
  • Marketplace (ACA)108M rates
  • Workers' comp84.2M rates

Insurance pricing

Transparency in Coverage (payer machine-readable in-network rate files)
7.44B
Negotiated rates
36
Payers
commercial insurers
56
States
+ territories
9.4M
Providers (NPIs)
identity layer
Rates by payer

Each insurer's rate is keyed to a provider NPI and a state. Rates resolve through a 9.4M-NPI identity graph (Type 1 and Type 2).

  • Blue Cross Blue Shield (national)3.00B rates / 55 st
  • Highmark1.91B rates / 55 st
  • CareFirst BlueCross BlueShield1.20B rates / 55 st
  • Cigna730M rates / 53 st
  • Priority Health (Cigna network)130M rates / 53 st
  • Quartz101M rates / 21 st
  • Kaiser Permanente64.1M rates / 44 st
  • Geisinger Health Plan57.6M rates / 39 st
  • Premera Blue Cross51.4M rates / 24 st
  • Molina Healthcare37.2M rates / 43 st
  • Priority Health33.3M rates / 20 st
  • UPMC Health Plan33.1M rates / 17 st
Rates by methodology

Methodologies are not comparable, so the tools keep them separate (no blending a fee schedule with a percentage of charges).

  • Negotiated dollar3.88B rates
  • Fee schedule3.14B rates
  • Percentage of charges286M rates
  • Per diem68.3M rates
  • Derived30.1M rates

Medicare benchmarks & cost

CMS fee schedules + HCRIS hospital cost reports (public)

The public yardstick. Medicare publishes what it pays for every service, by setting, and hospitals file their actual cost. Together these turn any charge or rate into "X times Medicare" and into an estimated true cost.

Localized, not national

Every Medicare benchmark resolves to where care is delivered. Professional fees adjust to each Medicare locality (GPCI); facility payments and cost are keyed to the hospital. So "X times Medicare" is computed for the actual place of service, not a flat US average.

109
GPCI localities
all CMS localities, 53 states + territories, CY2025
31,913
ZIP / county to locality
ZIPs across 51 states + a national county map
6,949
Per-hospital cost & facility
hospitals, each located by state and CBSA
Price benchmarks, by setting

What Medicare pays. Each service uses the right schedule; they are never mixed.

  • Physician Fee Schedule (PFS)Office and professional services; the universal fair-price denominator92,128 rows
  • Outpatient (OPPS)Hospital outpatient facility payments, by APC18,613 rows
  • Ambulatory Surgery (ASC)Surgery-center procedures (separate system from OPPS)6,642 rows
  • Clinical Lab (CLFS)Lab tests (use instead of PFS for lab HCPCS)1,971 rows
  • Inpatient (IPPS / MS-DRG)Inpatient stays; DRG relative weights and length of stay1,528 rows
Cost-to-charge (HCRIS cost reports)

What care actually costs. The cost-to-charge ratio (CCR) turns a billed charge into an estimated real cost, whole-hospital or per service.

  • Whole-hospital cost-to-chargeOverall CCR per hospital (total cost / total charge), turns a billed charge into estimated cost81,368 rows / 6,949 hosp / FY2011-2024
  • Department cost-to-chargeService-specific CCR per Worksheet C cost center (CT, MRI, lab, OR ...)1.5M rows / 6,887 hosp
  • Code to cost-center crosswalkMaps each billing code to the cost center where its cost accrues, so the CCR is accurate per service28,328 rows
  • Outpatient volumeHospital outpatient service mix by Comprehensive APC892,793 rows

Coverage as of 2026-06-21. All three pillars share one provider, payer, and code spine, so a hospital's posted price, an insurer's negotiated rate, the Medicare benchmark, and the estimated cost line up on the same procedure.

3 / What you can ask

You talk to the data in natural language; the agent picks the right tool and returns a typed, sourced answer. The common questions:

What does an insurer pay a provider for a code?
Give a payer, an NPI, a billing code, and a state and get the negotiated rate the way it actually resolves through the network, with the Medicare benchmark for context.
get_negotiated_rate, compute_median_rate
What does a hospital charge, and is it high?
Pull a hospital's posted rate for a procedure, see the spread across hospitals, and compare it to Medicare and to the commercial rate for the same code.
get_hospital_chargemaster, compare_hospital_rates, compare_hospital_to_commercial
What's a fair, defensible price?
Turn any procedure into a reference-based price anchored to Medicare, with the math and sources behind it, ready to put in front of an actuary.
fair_rbp_price, fair_rbp_book
Where is care cheapest in a market?
Map the providers in a county, see who offers a procedure, and rank the options by price against the Medicare floor.
county_intelligence, county_price_options
Prove it.
Every number carries a hash; turn any answer into an auditable, re-runnable receipt that points at the public source file it came from.
prove_commercial_price, replay_*
Run my own query.
For power users: one read-only SQL SELECT over the tables your access allows, or the async lane for heavy pulls. Same provenance guarantees.
query_substrate, submit_query / poll_query / fetch_query

Discovery is built in: ask "what data and tools do you have?" and the agent answers from the live catalog. Full machine-readable index at /api/catalog.