IDR outcomes
How we count
Where every number on these pages comes from, how it is counted, what is left out, and how it was checked. Data version 2025Q4-166d8d, prepared 2026-10-14, CMS data through 2025 Q4.
01Sources
CMS Federal IDR Public Use Files
The Centers for Medicare & Medicaid Services publishes, for each calendar quarter, every federal payment dispute that reached a payment determination and was closed in that quarter, as three tables: a line-item file for out-of-network emergency and non-emergency items with the parties' names, the state, the specialty, the arbiter (the certified IDR entity) and the offers as multiples of the qualifying payment amount (QPA); the same for air ambulance; and an anonymised line-item file with the QPA, both offers, the prevailing offer (the award) and, from 2025 Q3, the payer's initial payment, in dollars, by service code and metro area, without party names or dispute numbers. The files are public and free; their page carries a data dictionary, general information and a disclaimer and user agreement. Using the files is an acknowledgement of that agreement, which asks that CMS be cited and that altered figures never be presented as CMS data. Citation: Centers for Medicare & Medicaid Services. Federal IDR Public Use File [Data file and data dictionary]. Retrieved from https://www.cms.gov/nosurprises/policies-and-resources/reports.
| Quarter | File | Posted by CMS | Lines (emergency and non-emergency) | Disputes | Air ambulance lines | Dollar-file rows |
|---|---|---|---|---|---|---|
| 2023 Q1 | federal-idr-puf-2023-q1-february-15-2024.xlsx | April 9, 2025 | 65,231 | 24,273 | 2,730 | 68,007 |
| 2023 Q2 | federal-idr-puf-2023-q2.xlsx | February 15, 2024 | 113,043 | 53,758 | 3,390 | 116,519 |
| 2023 Q3 | federal-idr-puf-2023-q3.xlsx | June 13, 2024 | 76,525 | 46,586 | 3,353 | 79,896 |
| 2023 Q4 | federal-idr-puf-2023-q4.xlsx | June 13, 2024 | 78,425 | 73,085 | 2,459 | 80,904 |
| 2024 Q1 | federal-idr-puf-2024-q1-march-18-2025.xlsx | April 9, 2025 | 186,469 | 136,825 | 4,656 | 191,193 |
| 2024 Q2 | federal-idr-puf-2024-q2.xlsx | March 18, 2025 | 377,629 | 199,015 | 9,211 | 387,019 |
| 2024 Q3 | federal-idr-puf-2024-q3.xlsx | May 28, 2025 | 660,595 | 295,712 | 10,092 | 670,964 |
| 2024 Q4 | federal-idr-puf-2024-q4.xlsx | May 28, 2025 | 944,900 | 382,842 | 13,494 | 960,750 |
| 2025 Q1 | federal-idr-puf-2025-q1.zip | January 21, 2026 | 1,112,854 | 453,974 | 13,377 | 1,128,072 |
| 2025 Q2 | federal-idr-puf-2025-q2.zip | January 21, 2026 | 1,629,014 | 602,757 | 14,252 | 1,647,029 |
| 2025 Q3 | federal-idr-puf-2025-q3.zip | July 27, 2026 | 1,573,101 | 591,852 | 14,852 | 1,587,953 |
| 2025 Q4 | federal-idr-puf-2025-q4.zip | July 27, 2026 | 1,407,684 | 532,542 | 12,296 | 1,419,980 |
CMS posts two quarters at a time, five to eight months after the later quarter ends, and occasionally re-posts a file. Each file's size and SHA-256 checksum is recorded when it is downloaded; a changed file is treated as a new release.
CMS Federal IDR Supplemental Tables
One small workbook per quarter with fourteen tables of aggregates the line-item files cannot give: disputes initiated, eligibility challenges, reasons for closure (ineligible, withdrawn, settled, decided), administrative fees and entity compensation, initiations by state, the top filing and responding parties, and prevailing offers by cost band and specialty. These feed the funnel on the front page and the service groupings below.
Medicare physician fee schedule
Relative value units, geographic practice cost indices, the conversion factors and the locality definitions come from CMS's quarterly relative value file releases (2026 October release and 2025 October release). The descriptor column, which is American Medical Association CPT content, is not loaded. Amounts are computed on the page as (work RVU × work GPCI + practice expense RVU × PE GPCI + malpractice RVU × MP GPCI) × conversion factor, with the facility or non-facility practice expense by setting. Anesthesia is paid by base units and time, which the dispute files do not carry, so no anesthesia amount is shown. Federal law bars an arbiter from considering Medicare rates (42 U.S.C. 300gg-111(c)(5)(D)); the amounts are context, not a factor.
Arbiters' fees
From CMS's list of certified entities, as copied on the date shown on the page that uses them. Fees are per party per dispute: one for a single dispute, one for a batch of two to 25 lines, and a further amount for each additional 25 lines.
02What is counted
Line items and disputes
A row in the CMS files is a dispute line item: one service in one dispute. A single dispute has one line; a batched dispute has several, and a bundled dispute has a primary line that carries the offer and component lines that do not. Every figure on these pages says whether it counts lines or disputes. Component lines are excluded from dollar statistics. Air ambulance lines are kept apart from everything else and are not shown.
Only disputes that reached a payment determination are in the files. Disputes found ineligible, withdrawn, settled or closed administratively appear only in the supplemental tables' counts.
Definitions
- Won by the provider
- The share of contested disputes decided in favour of the provider, facility or air ambulance provider. Contested means both sides made an offer and paid their fee; default decisions, where one side did not, are excluded from the rate and reported as a share of all disputes. Split decisions (a few hundred a quarter) count as neither. At line level the same rate uses the offer selected on each line.
- The award as a multiple of the payer's rate (QPA)
- CMS's "Prevailing Party Offer as % of QPA" field, which despite its name is a multiple: 3.98 means 398 percent. It is the offer the entity chose divided by the QPA the filing party reported. Lines with no reported QPA (marked "+" by CMS) and component lines are excluded. The figure is summarised as percentiles (p10, p25, median, p75, p90), never as an average, because a minority of QPAs are nominal (under a dollar) or unit prices, which makes their multiples absurd; the share above 100× is reported beside the percentiles.
- Payer's offer, provider's offer
- Each side's final offer as a multiple of the QPA, median across lines, from the same file.
- Dollar figures
- From the anonymised file: QPA, payer's offer, provider's offer, the award (CMS's prevailing offer), the payer's initial payment and the gap between that payment and the award, each as percentiles. QPAs under $1 are excluded from QPA statistics and from dollar multiples. CMS suppresses dollar values (marked "^") where a code appears fewer than ten times in a metro area in the quarter; those rows count in totals but carry no dollars.
- Business days to a decision
- CMS's "Length of Time to Make Determination", in business days, one value per dispute, summarised as percentiles and as the share within 30. In the 2023 and 2024 files it runs from the arbiter's assignment to the decision; from 2025 it runs from initiation, three to four business days longer on average.
- Arbiter's fee
- CMS's "IDRE Compensation", per dispute: the arbiter's fee as paid by the losing party. A value of $0 means a default decision whose loser had not paid when the file was made; those are left out of the median and counted.
- Percentiles
- Continuous (interpolated) quantiles over the lines or disputes in the cell.
03Cleaning rules
- CMS's sentinels are honoured: "N/R" not reported, "N/A" not applicable, "REDACTED" free text removed, "+" QPA not reported, "^" suppressed. Each becomes a missing value in the statistic and a flag in the data.
- Service codes are text. In the Excel-era files (2023 and 2024) anesthesia codes lost their leading zeros; they are restored when the code type is CPT. Three-digit MS-DRG codes are left alone. In the dollar file, which carries a code type only from 2024 Q3, a three- or four-digit code is treated as anesthesia unless the same quarter's named file shows that number as an MS-DRG.
- A dispute CMS re-publishes in a later quarter to correct it appears in both quarters' files; here its latest copy is kept and the earlier one set aside (3308 disputes, 8654 lines). CMS's quarterly determination counts therefore run very slightly above these pages' dispute counts for the quarters that were later corrected.
- Payers are keyed by the email domain CMS reports, as CMS itself does in its summary tables, because the free-text names come in hundreds of spellings. Display names are given only for domains whose owner has been confirmed; other domains are shown as the domain itself. Domains that belong to vendors handling negotiation and dispute intake on a plan's behalf, rather than to a plan, are not ranked as payers; their lines stay in every total.
- The certified entity is named in the files from 2025 Q3 only; the initiating party's type from 2024 Q3; modifiers from 2025 Q3, populated from 2025 Q4 for disputes initiated on or after October 23 2025.
- The state is the state where the service was given, as CMS records it on the named file. Dollar figures carry the metro area the plan reported on its offer, which is usually where the QPA was calculated; a state's dollar figures use only metro areas that lie within that state, and about a fifth of dollar rows carry no metro area and count nationally only.
04Groupings
Service groups
CMS's own code ranges from table 14 of its supplemental tables, checked in this order; a code outside every range is "other".
| Group | Code ranges |
|---|---|
| Emergency department visits | 99281 to 99288 |
| Critical care | 99291 to 99292 |
| Radiology | 70010 to 79999 |
| Pathology and laboratory | 80047 to 89398, 0001U to 0475U |
| Anesthesia | 00100 to 01999 |
| Surgery | 10004 to 69990 |
| Neurology and neuromonitoring | 95700 to 96020 |
| Hydration, injections and infusions | 96360 to 96549 |
| Hospital inpatient and observation care | 99217 to 99239 |
| Cardiovascular | 92920 to 93799 |
| Non-invasive vascular studies | 93880 to 93998 |
| Pulmonary | 94002 to 94799 |
| Care management | 99424 to 99491 |
| Neonatal and pediatric intensive care | 99466 to 99480 |
| Non-face-to-face services | 99421 to 99474 |
| Vaccines and toxoids | 90476 to 91322 |
| Immunization administration | 90460 to 90474 |
| Office and outpatient visits | 99202 to 99215 |
| Qualifying circumstances for anesthesia | 99100 to 99140 |
| Special services and reports | 99000 to 99091 |
| Moderate sedation | 99151 to 99157 |
| Other services and procedures | 99170 to 99199 |
| Physical medicine and rehabilitation | 97010 to 97799 |
| Advance care planning | 99497 to 99498 |
| Ear, nose and throat services | 92502 to 92700 |
| Prolonged services | 99354 to 99418 |
| Preventive medicine | 99381 to 99429 |
| Consultations | 99242 to 99255 |
| Newborn care | 99460 to 99463 |
| Psychiatry | 90785 to 90899 |
| Other evaluation and management | 99304 to 99316, 99341 to 99350, 99483 to 99499 |
| Allergy and immunology | 95004 to 95199 |
| Immune globulins | 90281 to 90399 |
| Delivery and birthing room | 99464 to 99465 |
| Dialysis | 90935 to 90999 |
| Gastroenterology | 91010 to 91322 |
| Ophthalmology | 92002 to 92499 |
Kinds of service
The seven coarser kinds the IDR calculator uses: emergency and critical care visits (99281 to 99285, 99291, 99292), radiology (70010 to 79999), pathology and laboratory (80047 to 89398), anesthesia (00100 to 01999), surgery (10021 to 69990), other visits and medicine (90281 to 99607), and other codes. The calculator's figures are produced by this same pipeline and reproduce exactly (see the checks).
Specialties, plan types and practice sizes
The specialty field is free text in thousands of spellings; it is folded into a dozen groups by keyword and is a secondary view. Plan type and practice size are shown as the parties reported them.
Code labels
The short descriptions beside codes are Spectare's own plain-English labels for the codes that carry most volume, written for practice administrators. They are not CPT descriptors; CPT is a registered trademark of the American Medical Association, and its descriptors are not reproduced here. Codes without a label show the code alone.
05Small cells
No cell with fewer than 11 lines (or disputes) is published anywhere on these pages or in a download. Cells under 30 are marked "small sample". Per-code pages keep state, metro area and payer detail for the latest four quarters, each calendar year and all quarters together; national detail is kept for every quarter.
06Checks
Every data version is checked before it is published: that every raw row is typed, that the dollar file's row count equals the named files' for every quarter, that known figures reproduce, that dispute counts reconcile with CMS's table 12 and medians with CMS's table 14, that the calculator's figures reproduce, and that a Medicare amount matches a hand calculation. 1 of 1 checks passed for this version; the ones that did not are noted below with the reason.
| Check | Expected | Got | Result |
|---|---|---|---|
| superseded (re-published) disputes | recorded | 3308 disputes, 8654 lines | ok |
07Limits
- A payment determination is not a payment. The files do not say whether an award was paid, when, or whether it was later settled or challenged.
- The QPA is what the filing party entered at initiation. CMS says it was not necessarily confirmed by the plan, and that some filers entered nominal figures or unit prices. Multiples of the QPA inherit that.
- The two files cannot be joined. Dollar figures cannot be split by payer, provider or, exactly, by state; outcome rates by payer cannot be put in dollars.
- Disputes that reach a decision are not a sample of all out-of-network claims, or of all disputes. Most eligible claims are never disputed, and many disputes settle in negotiation.
- Nothing here is about any patient or any particular claim, and nothing is a prediction. Figures are national public data about federal disputes, not Spectare's results.
08Refresh and citation
The pipeline that builds these pages is in the website's repository (data/idr). It downloads each file from CMS, records its checksum, validates every column name against the layouts CMS has used, loads the rows, applies the rules above, runs the checks and writes the prepared files the pages read. It is run when CMS posts a release; a monthly check looks for new files. The data version on every page identifies the set of CMS files used.
To cite these pages: Spectare Health (2026). IDR outcomes, version 2025Q4-166d8d, a reading of the CMS Federal IDR Public Use Files. spectarehealth.com/idr-data/. Please also cite CMS as above.
Changelog
- 2026-10-14: version 2025Q4-166d8d, CMS data from 2023 Q1 through 2025 Q4; first publication of the data section.