The patient half of the No Surprises Act arbitration story

On 11 August 2026 the Fifth Circuit, sitting en banc, vacated core rules for calculating the Qualifying Payment Amount, holding that ghost rates and the exclusion of incentive compensation produced artificially low benchmarks (Texas Medical Association v. HHS). Coverage framed it as a payment dispute between providers and insurers.

Under 45 CFR 149.140, a patient's cost-sharing for a surprise bill is computed from that same Qualifying Payment Amount. Here is what patients were actually charged, from the CMS Federal IDR public use files — the same download used for the widely reported dispute counts.

$415,740,263 of patient cost-sharing sits inside federal surprise-billing disputes in one half-year (2025 Q3–Q4). 111,790 line items show a patient owing more than $500; 51,714 more than $1,000.

The same care, the same city, wildly different bills

A cell is one service code, one metropolitan area, one service year, one place of service — the finest partition the file supports. Across 294 such cells for CPT 99284 (a high-severity emergency department visit) with at least 100 disputes each, patient cost-sharing varies 10.0x at the median cell between its own 10th and 90th percentile.

Metropolitan areaDisputes10th pct90th pctSpread
Colorado Springs, CO232$39$2,07852.8x
Odessa, TX482$21$81739.1x
Nacogdoches, TX257$37$1,25634.1x

Two patients in Colorado Springs, the same emergency visit code, the same service year: one owes $39, the other $2,078. The number their share is computed from is the number the Fifth Circuit just held was being calculated unlawfully.

Method

Source: CMS Federal IDR Public Use File, 2025 Q3 and Q4, qpa-and-offers table (3,007,933 line items; 97.9% carry a Cost-Sharing Amount value, 47.2% of those above zero). Dispersion is reported as the ratio of the 90th to the 10th percentile within a cell, which is robust to the single-record outliers that dominate max/min. Cells below 100 disputes are dropped.

What this does not show. It does not show that any plan miscalculated a QPA, and it cannot: the public file is de-identified, with no payer, provider, or dispute identifier, so no value can be attributed to any party. It does not cover bills that never reached a dispute. All figures are conditioned on reaching a determination, because CMS excludes ineligible and withdrawn line items from the file. Cross-plan variation is real and is part of what is measured here — the file carries no payer identifier, so it cannot be netted out.

Reproduce it

curl -O https://downloads.cms.gov/files/federal-idr-puf-2025-q3.zip
curl -O https://downloads.cms.gov/files/federal-idr-puf-2025-q4.zip
# Column: "Cost-Sharing Amount" in the qpa-and-offers table
# Cell:   (Service Code, Geographical Region, Year of Service, Place of Service Code)

Related measurement on the instability of the QPA itself — including two headline claims that were withdrawn when the checks pointed the other way — is available on request.

Disclosure. Medigami is a patient-side company and has a commercial interest in medical billing disputes. That is stated here rather than omitted, and it is why the method, the cell definition, and the limitations are all specified in full: the result can be reproduced or refuted from the public files without reference to us.

Press

Data, scripts, and methodology walkthrough available. Contact [email protected].