Every number on this site traces to a public federal source and a retrieval date. This page describes exactly how the figures are produced, including the judgment calls and their limitations.
Since April 1, 2026, CMS requires every hospital to publish a machine-readable file of its standard charges (45 CFR Part 180, schema v3.0), including payer-specific negotiated rates and, where applicable, actual allowed amounts derived from 835 remittance data. We retrieve each hospital's file from the location it publishes in the required cms-hpt.txt pointer at its website root, parse it, and extract rates for a fixed basket of 107 common procedures (32 inpatient MS-DRGs, 75 outpatient CPT/HCPCS codes). The commercial median shown is the median of the hospital's published negotiated dollar rates across commercial plans, excluding Medicare Advantage, Medicaid, and other government program rates. For inpatient DRGs, only case-rate contract rows are compared; per-diem rows are excluded because they price a day, not an episode. Plan-row counts are always displayed: a median over 40 published plan rates is sturdier than one over 2.
Inpatient DRG benchmarks are each hospital's own average Medicare payment per DRG from CMS's Medicare inpatient claims data (2024, the most recent year published); where a hospital has too few Medicare cases to appear, the Colorado median for that DRG is used and flagged in the data. Maternity DRGs have no Medicare benchmark because Medicare rarely pays for deliveries. Outpatient benchmarks are the July 2026 OPPS Addendum B national rate, wage-adjusted to each hospital using its FY2026 wage index (60% labor share). Laboratory codes use the 2026 Clinical Lab Fee Schedule, which is national and not wage-adjusted.
Cost estimates multiply the hospital's published gross charge by its operating cost-to-charge ratio from the FY2026 CMS impact file. This is the standard cost-report method used by MedPAC and NASHP, applied at the hospital level. It is an estimate, always labeled as such, and is least precise for imaging and laboratory services where department-level markup differs most from the hospital average; those cells are marked with a tilde (~). Department-level ratios from full HCRIS worksheets are a planned upgrade. Critical access hospitals are not in the impact file (they are paid on cost) and currently show no cost estimate.
A small share of hospital-procedure cells (about 2%) publish commercial rates far below Medicare across their entire plan set, which indicates partial or component-level pricing in the hospital's file rather than a true episode price. Cells whose commercial median falls below 60% of the Medicare benchmark are flagged, excluded from cross-hospital charts and summary statistics, and shown in tables with a visible data flag. A second flag catches the opposite failure: cells where, for at least half the commercial plans, the negotiated rate equals the gross charge. That file has recorded the chargemaster price where a negotiated rate belongs (typical of percent-of-charges contracts written out by the hospital), and it is excluded the same way.
Payer-level figures (the payer scorecard and the insurer-gap finding) use fee-schedule, per-diem and percent-of-charges rows for procedure codes, and case rates only for MS-DRGs. A case rate attached to a CPT code is usually a bundled episode price shared across many codes, which makes it a different product from the fee-schedule rate beside it; hospital-level medians still include those rows, so a cell’s median can differ slightly from its payer breakdown.
Actual allowed amounts (medians and percentiles from remittance data) are used only when the hospital publishes a remittance count of 11 or more; masked or missing counts are suppressed. When a plan row has no negotiated dollar rate (a percentage-of-charges or algorithm contract) but an unsuppressed median allowed amount, that median stands in as the plan's price. Every parsed file produces an exclusion log: rows dropped for missing dollar values, malformed data, or algorithm-only pricing are counted, not hidden. Hospitals whose files cannot be used as published are named, with what we found and what we did about it, on the compliance log for each state.
2026-09-18 (Virginia): two corrections found while building Virginia, applied to every state. (1) Payer classification had treated the bare token “VA” as a Veterans Affairs marker, which in Virginia read “VA BCBS”, “BCBS of VA” and “Blue Cross Blue Shield of VA Anthem” as government payers and left Buchanan General Hospital with no commercial rate at all. Veterans Affairs payers are now matched by their actual spellings (Veterans, VA CCN, TriWest, ChampVA, Optum VA and the like), which also moves TriWest and VA Community Care rows that some files had labelled as commercial into the government class where they belong. (2) An MS-DRG rate whose methodology field says “other” but whose own algorithm text opens “Case rate ($…)” or “Fee schedule rate ($…)” — a stated dollar amount per stay — is read as a case rate; Bon Secours Mercy Health labels every Virginia DRG row this way, and no other state’s files carry the pattern. A bare “fee schedule” methodology on a DRG is still not accepted: tested on every state, it would have moved New Mexico from 367% to 290% and Colorado from 303% to 284%, which says those rows are not per-stay amounts. State medians moved by 0 to 12 points (New Mexico 355% to 367%, Indiana 256% to 259%; Colorado 303%, Kansas 240% and Washington 273% unchanged).
2026-09-17: a plan row published at 100 percent of billed charges under a percent-of-charges
methodology is now read as a non-contracted payer listed for completeness (out-of-state Medicaid programs,
life and supplement insurers, workers’ compensation carriers), not as a negotiated rate. Aspen Valley
Hospital confirmed this reading of its file on September 16 after our notice. Such rows stay in the
downloadable data as payer class non_contracted and are left out of every commercial comparison,
including the September 16 equals-gross test, which they had been tripping: a hospital with 110 placeholder
rows and 40 contracted ones was being read as unpriced. Aspen Valley, Mt. San Rafael and Weisbrod (Colorado),
Alta Vista Regional (New Mexico) and Community Hospital East (Indiana) now price from their contracted plans;
the notices to Alta Vista, Mt. San Rafael and Community East were withdrawn on the compliance log. Alta Vista
returns to the top of the New Mexico echocardiogram range at a $5,642 commercial median (its contracts run 72
to 85 percent of charges). State medians moved by 0 to 12 points (Colorado 305% to 303%, New Mexico 343% to
355%, Indiana 258% to 256%, Kansas 240% and Washington 273% unchanged).
2026-09-16 (Washington): where a plan row carries no negotiated dollar rate but the hospital publishes the median allowed amount CMS requires for percentage-of-charges and algorithm rates (with a count of 11 or more), that median is used as the plan's price. Applied to every state. Several Washington critical-access hospitals (WhidbeyHealth, Willapa Harbor, Peace Island, North Valley) publish only percentages and are priced entirely this way; the count of such plan rows is recorded per cell in the downloadable data. State medians moved by 0 to 6 points (Colorado 301% to 305%, New Mexico 341% to 343%, Kansas 234% to 240%, Indiana unchanged at 258%), and Grand River Hospital District in Colorado, previously unpriced, now carries 14 cells.
2026-09-16: two rules added while building Indiana, applied to every state. (1) Cells where the negotiated rate equals the gross charge for at least half the commercial plans are flagged and excluded; several rural hospitals had written their chargemaster into the negotiated column. (2) Payer-level figures use case rates only for MS-DRGs. State medians moved by 1 to 14 points (Colorado 304% to 301%, New Mexico 355% to 341%, Kansas 238% to 234%); figures quoted in The American Healthcare Conundrum Issue #20 are the pre-change values.
2026-08-25: the basket expanded from 41 to 107 procedures — every remaining code on CMS’s 70-service shoppable list plus nineteen high-volume MS-DRGs (stroke, AMI, cardiac valve surgery, coronary bypass, hip fracture, bariatric surgery, and others). All original codes were retained, so pre-expansion figures are unchanged; new codes simply appear as hospitals’ files are re-parsed. Every added DRG was verified two ways: its official title against CMS’s Medicare claims data, and its actual publication against a sample of Colorado hospital files. Missing something? Request a procedure or DRG and it will be triaged into the next refresh.
This is analysis of federally published data. It is not medical advice, legal advice, benefits advice, or a guarantee of what any patient will pay. Hospital-published files contain known quality problems; where the data cannot support a conclusion, no number is shown.
The published methodology and analysis code are open. Questions, corrections, and data problems: andrew@rexroadanalytics.com. Related publication: The American Healthcare Conundrum (americanhealthcareconundrum.com).