Hospital price transparency began as a data lineage test

CMS’s November 2019 rule told hospitals to publish five kinds of standard charges for every item and service. The numbers lived in different systems, some only as contract formulas. Later rules show the hard part was tracing each figure to its source.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

CMS’s November 2019 rule told hospitals to publish five kinds of standard charges for every item and service. The numbers lived in different systems, some only as contract formulas. Later rules show the hard part was tracing each figure to its source.

  1. The week On the afternoon of November 15, 2019, the Centers for Medicare & Medicaid Services (CMS) filed a final rule for public inspection that told every hospital in the United States to publish its prices.
  2. Sources: where each number lived A hospital chief information officer reading the rule in November 2019 would have found that it named data, not systems.
  3. Translation: from contract terms to a dollar figure The rule assumed that a negotiated charge was a number.
  4. Publication: the file, the display and whether anyone could find them The third part was the part the rule described in most detail.
  5. Accountability: who owned the result The fourth part was ownership.
  6. The strongest objections Three objections to the lineage reading deserve an answer.
  7. With hindsight The rule took effect on January 1, 2021.

The week

On the afternoon of November 15, 2019, the Centers for Medicare & Medicaid Services (CMS) filed a final rule for public inspection that told every hospital in the United States to publish its prices. The rule appeared in the Federal Register on November 27, 2019, at 84 FR 65524. It had been signed earlier in the month by the CMS Administrator and the Secretary of Health and Human Services. It would take effect on January 1, 2021.

The rule rested on section 2718(e) of the Public Health Service Act, added by the Affordable Care Act, which requires hospitals to make public a list of their standard charges each year. It also responded to Executive Order 13877 of June 24, 2019, on price and quality transparency. CMS had already updated its guidelines, effective January 1, 2019, to require hospitals to post a list of current standard charges online in a machine-readable format. CMS guidance had clarified that the posted information should represent standard charges as reflected in the hospital’s chargemaster, its own list prices. Many stakeholders, hospitals among them, told CMS that such lists did little to tell insured patients what they would pay.

The 2019 rule went much further. It defined five kinds of “standard charge”. The gross charge is the chargemaster price before discounts. The payer-specific negotiated charge is the rate negotiated with a named third-party payer and plan. The de-identified minimum and maximum negotiated charges are the lowest and highest rates negotiated across all payers. The discounted cash price applies to someone paying cash.

Hospitals had to publish all five, for all items and services, in a single machine-readable file, with billing codes such as CPT, HCPCS, DRG and NDC. The file had to follow a CMS naming convention, be free, require no account or personal information, be digitally searchable and be updated at least once a year. Each location with different charges had to publish its own. Separately, hospitals had to show prices for at least 300 “shoppable” services, including as many of 70 services named by CMS as they offered, in a consumer-friendly format. A hospital with an online price estimator tool meeting set conditions was deemed to meet that second requirement.

The rule applied to every hospital licensed or approved by a state, including those not paid under Medicare’s outpatient system. Federally owned or operated hospitals, such as those of the Department of Veterans Affairs and military treatment facilities, were deemed compliant. CMS could monitor through complaints, outside analyses and audits of hospital websites. It could then issue a warning, require a corrective action plan and impose a civil money penalty of up to $300 a day, publicized on a CMS website.

CMS estimated the work at 150 hours per hospital in the first year, across 6,002 hospitals, and 46 hours a year after that. Commenters objected that negotiated rates were trade secrets, that compelled disclosure raised First Amendment problems, and that some payer contracts contained non-disclosure clauses. Others argued that $300 a day was trivial for some hospitals or too low to compel compliance.

The same afternoon, the Departments of the Treasury, Labor and Health and Human Services filed a companion proposal, Transparency in Coverage. It would require group health plans and insurers to offer an online tool giving members estimates of their cost sharing. It would also require them to publish two machine-readable files, one of in-network negotiated rates and one of historical out-of-network allowed amounts. Comments were due by January 14, 2020.

The hospital price transparency rule was written as a publishing requirement. It told hospitals what to post, in what format, where and how often. Most of its regulatory text is about files, web pages, naming conventions and access. Read that way, the work belonged to the web team and the revenue cycle office, and the main risk was a missing link.

That reading missed where the difficulty lay. Each of the five standard charges came from a different place inside a hospital. Only one of them, the gross charge, came from a list the rule itself named: the chargemaster. The others could sit in payer contracts, contract management systems, billing systems and financial policy, and some existed only as formulas that produce a price when a claim is paid. Publishing them required something the rule did not mention: a traceable path from each published number back to its source. The discipline that provides such a path is data lineage, and building it is an architecture problem.

This article takes the problem in four parts that follow the path of a number. The first is sources: where each element of the file actually lived. The second is translation: how contract terms were turned into a dollar figure per item, per payer and per plan. The third is publication: the file, the consumer display and whether anyone could find them. The fourth is accountability: who owned the result and who would answer to CMS. A fifth section weighs the strongest objections. The hindsight section then shows how later rules pushed hospitals along this path.

The distinction matters to technology leaders because it decides who does the work and how it is funded. A publishing project gets a web budget and a deadline. A lineage problem needs data owners, integration work across systems that may never have been designed to talk to each other, and a process that repeats every year. A hospital that scoped the rule as the first kind of project and discovered it was the second would pay twice.

Sources: where each number lived

A hospital chief information officer reading the rule in November 2019 would have found that it named data, not systems. Mapping the five charges to systems was the first task, and the answer differed by element.

Gross charges and the chargemaster

The gross charge was the easy one. The rule defined it as the charge for an item or service as reflected on the hospital’s chargemaster, absent discounts. The chargemaster, also called the charge description master, is the master list of every item and service for which the hospital has set a charge. It already existed, and CMS guidance had made it the basis of the lists posted under the guidelines in effect from January 1, 2019.

Even here there were gaps. The rule’s definition of items and services included facility fees and the services of employed physicians and non-physician practitioners. Professional charges for employed clinicians may sit in a professional billing system separate from the hospital chargemaster. A hospital that exported its chargemaster alone could miss part of what the rule required.

Negotiated charges and the contract

The payer-specific negotiated charge was the hard one. It had to be shown for each item and service, by payer and by plan, in both inpatient and outpatient settings where applicable. Those rates are set in contracts negotiated by the managed care contracting team. A hospital may keep contract terms in a contract management system that models expected payment, or in documents and spreadsheets, with the payment logic loaded separately into the billing system.

None of those places was designed to produce a public list of rates by item. Contract management tools are built to estimate what a given claim should pay. Billing systems are built to produce claims. Neither is organized around a published table of every item crossed with every payer and plan. Building that table meant joining the chargemaster, the contract terms and the plan structure.

Payer and plan identity

The rule required each payer-specific negotiated charge to be clearly associated with the name of the third-party payer and the plan. That sounds like a label. It is a data mapping problem. A hospital billing system may organize payers by contract and financial class, the internal categories used to route claims and report revenue. Payers market their products under plan names that members recognize.

The two do not line up neatly. One contract may cover many plans. One plan may be paid under different contracts at different sites. Some plans reach a hospital through a network the payer rents from another company. To publish a rate against a plan name, the hospital needed a crosswalk from its contract structure to the payer’s plan names, and someone to maintain it as plans changed each year.

Inpatient and outpatient settings

The rule also required charges for each item in the inpatient setting and the outpatient department setting, as applicable. The same item can be paid differently in each. The rule’s own preamble noted that negotiated charges may be set for service packages, such as per diem rates and DRGs, that the chargemaster does not list. A file that showed one figure for an item without its setting could mislead a reader badly. Capturing the setting meant carrying another dimension through every step of the pipeline.

Derived values and cash prices

The de-identified minimum and maximum negotiated charges were derived values. They could only be computed once every payer-specific charge for an item was known. Any error in a single payer’s rate could change the minimum or maximum, so the derived columns inherited every upstream problem.

The discounted cash price came from financial policy. A hospital might set self-pay discounts as a percentage off the gross charge or as fixed prices for packages; the rule’s preamble mentioned self-pay package pricing. A hospital might also have no stated cash price for many items. Where a hospital offered no discounted cash price for a shoppable service, the rule required it to show the gross charge instead. That kept the requirement simple. It also meant a hospital without an item-level cash price policy would show self-pay patients its list prices.

Codes and descriptions

The file also had to carry any code the hospital used for billing the item, such as a CPT, HCPCS, DRG or NDC code. Codes sit in different places for different services. An inpatient stay may be grouped by DRG at billing time, while supplies may carry an internal code and drugs an NDC. A row in the price file therefore needed fields from several systems, each with its own update schedule.

Reference data that changes underneath

Billing code sets change on their own schedules. New codes are added, old ones retired and descriptions revised. Drug codes change as products enter and leave the market. A price file that joins rates to codes inherits all of that churn. The chargemaster, the contract model and the billing system may update their code tables at different times. If so, the same service can carry different codes in different sources on the day the file is built. The lineage map has to record which version of each reference table a release used.

Why the map mattered

The result was a dependency map with at least four systems of record and several owners. That map, more than any web page, determined whether a hospital could produce an accurate file and update it each year. Without it, the annual update became a fresh project every time, staffed by whoever remembered how the last one was done.

Translation: from contract terms to a dollar figure

The rule assumed that a negotiated charge was a number. A contract may not state a number for each item. It may state a method instead.

How contracts actually set payment

Contracts can set payment in several ways. The 2019 rule’s preamble recognized negotiated service packages such as per diem rates and DRGs, and the 2026 rule dealt with negotiated charges based on percentages or algorithms. A contract might pay a percentage of billed charges, a fixed amount per day of an inpatient stay, a case rate for an episode or a rate tied to the DRG. It might pay from a fee schedule for outpatient services. It might also layer exceptions on top, such as carve-outs for high-cost implants, outlier provisions for unusually expensive stays and different terms for different plans from the same payer.

For an item paid as a percentage of charges, the negotiated charge for that item can be computed as the percentage times the gross charge. For a service paid by DRG case rate, the price of an individual supply inside the stay is not separately negotiated at all. For a per diem stay, the total depends on the length of the stay, which is not known in advance. Each of those methods produces a dollar figure only when a real claim is adjudicated.

Choices that changed the answer

Translating those methods into the file forced decisions the rule did not settle in 2019. A hospital had to decide whether to publish a percentage, a computed dollar amount or a note that the item was paid inside a package. It had to decide how to show a case rate against the individual items it covered. It had to decide which plans of a payer to list separately when the terms differed only in small ways.

Hospitals free to answer these questions differently would produce files that were hard to compare, and later federal action asked for standardized, comparable data. A rule that defined the output as a dollar figure per item, without a method for converting non-dollar terms, guaranteed variation in the translation step.

Ancillary services

Translation for the shoppable services display added a further step. The rule required the negotiated charge for each shoppable service and for each ancillary service that customarily comes with it. A hospital had to decide which ancillaries belonged with each primary service, then translate each ancillary’s rate by payer and plan. A colonoscopy, for example, may involve a facility charge, anesthesia, pathology and supplies, each priced by a different contract term. The display is only as accurate as the weakest of those translations.

Service packages

The rule recognized that hospitals sometimes set a single charge for a group of items, which it called a service package. A package price can be published as one row. The difficulty comes when the package is defined in a contract but the chargemaster lists the items separately. The file then needs either the package, the items or both, with a way for the reader to tell which items are included. Each choice is defensible. Mixing them inconsistently across payers is not.

Where the translation should live

The architectural choice was where to put the translation logic. One option was a one-off spreadsheet built by a consultant or analyst each year. Another was to build the translation into the contract management system, so that the same logic that estimated expected payments also produced the published rates. A third, which later rules pushed toward, was to derive amounts from what payers actually paid, using remittance data.

Each option has a different failure mode. The spreadsheet is cheap and fragile, and nobody can reproduce it a year later. The contract model is consistent with internal expectations but only as good as the contract terms loaded into it, and those may be incomplete. Remittance data reflects reality but lags it, and it says nothing about services that were rarely billed. A hospital that chose deliberately among these, and documented the choice, was in a far better position than one that inherited whatever the first project produced.

Checking the translation against reality

Whatever method a hospital chose, it could check the result. A sample of paid claims for common services, compared with the published rates for the same payer and plan, shows whether the translation is close to what payers actually pay. Large differences point to missing contract terms, wrong plan mappings or a translation rule that does not fit how the contract is applied.

That check needs thresholds and an owner. A hospital might set a tolerance for the difference between published and paid amounts and require contracting staff to explain each case outside it. The results become part of the evidence for the release. They also become a quality signal for the contract management system itself, which benefits revenue cycle well beyond the transparency program.

Version control

Contracts change during the year, but the rule required only an annual update, with the date of the last update shown. A hospital therefore had to decide which contract version a release reflected and record it. If a rate changed mid-year, the published file would be out of date until the next release, and the hospital would need to show what was in force when the file was produced. Versioning contract terms and file releases together is basic configuration management, applied to pricing data.

The role of managed care contracting

Translation also changed a relationship inside the hospital. Contracting teams had negotiated rates in confidence, and commenters told CMS that some contracts contained clauses barring disclosure of negotiated rates. CMS answered that it had authority to define standard charges to include payer-specific negotiated charges, and it finalized the requirement.

Once the rule was final, the contracting team became a data supplier to the transparency program. It had to provide contract terms in a structured form, explain the methods and confirm the translated rates. The work differs from negotiating. A hospital that gives contracting staff the tools and time to do it is better placed to publish a file it can defend. One that treats contracting as a source to be raided once a year risks a file the contracting team itself cannot vouch for.

Publication: the file, the display and whether anyone could find them

The third part was the part the rule described in most detail. It had two outputs with different users.

The machine-readable file

The machine-readable file was meant for software, not people. The rule allowed formats such as JSON, XML and CSV and required a single file per hospital location with a set naming pattern. For a large hospital with many payers and plans, a file listing every item against every plan could be very large. The rule required the file to be digitally searchable and accessible without an account, a password or personal information.

Those conditions are simple to state and easy to undermine. A file can be posted but placed behind a form, a click-through or a page that search engines are told not to index. The 2019 rule said the information had to be accessible without barriers, but it did not list the technical patterns that would count as barriers. That gap became a later rulemaking topic, as the hindsight shows.

Engineering the file

Producing the file is a data engineering job. A pipeline extracts from each source, applies the translation rules, joins the results, validates them and writes the output in the required format. Each location with different charges needs its own output. The file has to show the date of the last update. The output then has to be published to a stable address that does not change between releases, so that anyone who found the file last year can find it again.

Validation should catch the obvious failures before publication. Examples include missing codes, negative or zero rates, payers with no plan names, minimums above maximums and rows missing a setting. Automated checks of this kind cost little to build and catch errors before an outside analyst does.

Integrity and privacy of the published data

The file contains no patient information by design. The rule defined minimum and maximum negotiated charges as de-identified and required access without personal information. The pipeline behind it, though, draws on systems that do hold patient data, especially if it uses claims or remittance records. Extracts should be built so that only rate-level data leaves the secure environment, with a review step that confirms no patient-level fields reach the output.

Integrity matters too. A public price file is a statement by the hospital. If the web server or content system that hosts it is compromised, an altered file could misstate prices under the hospital’s name. Publishing a checksum alongside the file and monitoring the hosted copy for unexpected changes are simple controls borrowed from software distribution.

The consumer-friendly display

The second output was meant for patients. Hospitals had to show prices for at least 300 shoppable services, defined as services a patient can schedule in advance. The list had to include as many of the 70 CMS-specified services as the hospital offered, plus hospital-selected services to reach the total. For each, the display had to show payer-specific negotiated charges, the discounted cash price, the de-identified minimum and maximum and the related ancillary services. It had to be searchable by service description, billing code and payer.

The display was a packaging problem layered on the lineage problem. A shoppable service, such as a scheduled imaging study, may involve a primary service plus ancillary items such as supplies, drugs and professional services. The display had to group those into a recognizable package. That required clinical and billing input on what a typical episode includes, which no single system held.

The display also required a plain-language description of each shoppable service and an indicator when a CMS-specified service was not offered. Writing plain-language descriptions for hundreds of services is clinical and editorial work, not data work. It is also the part patients actually read, which argues for testing it with patients rather than leaving it to the team that built the file.

The estimator option

The rule offered a shortcut. A hospital that maintained an online price estimator tool was deemed to meet the shoppable services requirement. The tool had to cover the required services and let a patient obtain an estimate of what they would owe.

An estimator answers a different question from the file. It estimates what one patient will pay, given their coverage and benefits, and so it needs eligibility and benefits data from the payer as well as the hospital’s rates. It can be accurate for the patient in front of it while revealing little about the underlying negotiated rates. Choosing the estimator path satisfied one requirement but left the file requirement, and the lineage problem behind it, untouched.

Findability as a requirement

The rule required the file to be displayed prominently on a publicly available website, clearly identified with the hospital location, and easily accessible without barriers. Those words describe outcomes, not mechanisms, so a hospital had to translate them into technical requirements for its own web team.

A reasonable translation would start with a stable address for each file and a link from a page reachable from the hospital’s home page. It would put no form or click-through in front of the download, and no instruction telling search engines to skip the page. It would also include a check, after each release, that the file can be found by searching and downloaded by a script. Web teams may add bot protection or indexing rules to pages for reasons unrelated to pricing. Those defaults can defeat the rule’s purpose without anyone intending it, so the transparency owner needs a say in how the hosting page is configured.

Two audiences, one source

The architectural point is that both outputs should come from the same source data. If the file is built from one extract and the estimator from another, the two will drift apart, and a patient or analyst who compares them will find contradictions. A single published dataset, from which both the file and the display are generated, is easier to defend and cheaper to update.

Accountability: who owned the result

The fourth part was ownership. The rule created a recurring, public, auditable obligation with no natural owner in the hospital’s structure.

Too many partial owners

Ownership could be split many ways. Revenue cycle might own the chargemaster and managed care contracting the rates. Finance might own the cash price policy, and health information management or coding the codes. IT might own the systems and the extract jobs, and marketing or digital teams the website. Compliance might own the regulatory relationship, and legal the contract confidentiality questions. Each would hold a piece, and none the whole.

A hospital that left the obligation spread across these functions would struggle on two counts. It would struggle to produce the file on time each year. And it would struggle to answer CMS if a warning notice arrived, because nobody could explain the file end to end.

What an accountable program looks like

A workable model names a single accountable executive, perhaps in revenue cycle or finance, with a data owner for each element of the file. It documents the lineage map from the first section and the translation choices from the second. It runs the annual update as a controlled release, with reconciliation checks before publication. It keeps the evidence an outside reviewer would ask for: which systems were used, which contract versions, which choices were made for non-dollar terms, and when the file was published.

Monitoring under the rule could come from complaints, outside analyses or CMS audits of websites. The first two meant that researchers, employers, payers and competitors could all find problems before CMS did. A hospital with a documented lineage could respond to such findings quickly and credibly. One without it would have to reconstruct its own process under deadline.

Answering CMS

The rule set out what would happen if CMS found a hospital out of compliance. CMS would generally first send a written warning notice. For a material violation it could request a corrective action plan, which had to describe the corrective actions and the timeframe for completing them and was subject to CMS review and approval. CMS could then monitor the hospital’s progress. A civil money penalty could follow if the hospital failed to respond to the request or to comply with the plan.

Every step in that sequence rewards a hospital that knows how its file was built. A credible corrective action plan names specific fixes and realistic dates. A hospital that cannot say which system produced a wrong figure cannot promise when it will be right. The lineage map, the translation choices and the validation results are, in effect, the hospital’s defense file.

Funding the annual cycle

CMS estimated that after the first year the work would take 46 hours per hospital annually. That estimate assumed the first year had built something reusable. A hospital that met the first deadline with a one-off extract could face much of the first-year effort again at the next update. The budget question for a CIO or chief financial officer was therefore whether to fund a repeatable pipeline in the first year or pay for a project every year.

The repeatable option also had a governance cost. Someone had to decide what events trigger a rebuild before the annual date: a new payer contract, a large chargemaster update, a new location. Those triggers belong in the contracting and revenue cycle calendars as well as in IT’s change schedule, because the business events come first.

Differences by type of hospital

The burden fell unevenly. A large academic medical center with dozens of payer contracts and thousands of items faced the biggest translation problem and the largest file. A critical access hospital with few contracts faced a smaller data problem but had little staff to solve it. CMS’s first-year estimate applied the same 150 hours to each of 6,002 very different hospitals.

Multi-hospital systems faced a structural question. The rule required each location with a different set of standard charges to publish separately, even under a single license. A system with shared contracts but location-specific chargemasters had to decide how to produce many consistent files from shared sources. Ambulatory surgery centers and physician practices that were not licensed hospitals fell outside the rule, which meant a health system could be publishing prices for some sites of care and not others.

Where the program should report

In our view, reporting lines shape behavior. A program that reports through IT is likely to optimize for delivery of the file on time. One that reports through compliance is likely to optimize for defensibility. One that reports through finance is likely to focus on what the disclosure reveals about rates. Each emphasis is useful and none is sufficient alone. A small steering group that includes all three, with a single accountable executive, is a reasonable compromise. Smaller hospitals can combine roles, but should still write down who decides on translation choices and who approves each release.

The payer side of the same data

The companion Transparency in Coverage proposal pointed to a further accountability question. If plans and insurers were also required to publish in-network negotiated rates, the same rates would appear in two places, published by two parties, translated by two methods. Discrepancies between a hospital’s file and a payer’s file would be visible to anyone who compared them. That gave hospitals another reason to own their lineage: when the two disagreed, the hospital would need to show where its number came from.

The strongest objections

Three objections to the lineage reading deserve an answer.

It was a pricing fight

The first is that the obstacle was strategic, not technical. Commenters, including hospital associations, argued that negotiated rates were proprietary, and one argued that every payer paying above the lowest rate would renegotiate down to it. Another warned that hospitals might treat the penalty as a business expense rather than an incentive to comply. If the obstacle was willingness, better data architecture would not change the outcome.

There is truth in this. A hospital that chose not to comply was making a strategic decision, and no lineage map would change it. But the objection explains only the hospitals that chose not to publish. A hospital that intended to comply could still produce a file that was incomplete or inconsistent, and for such a hospital the obstacle was the one this article describes.

The files were for intermediaries

The second objection is that patients would never read machine-readable files, so their quality mattered little. The point is partly right about patients. It is wrong about consequences. CMS could act on outside analyses of noncompliance under the rule. Academic researchers, employer organizations, health technology organizations and other users of the files were among those who answered CMS’s later request for information on the data. A poor file was a public record of the hospital’s data quality, open to anyone placed to use it against the hospital.

Small hospitals cannot afford lineage

The third objection is that a full lineage program is too much for a small hospital. The objection is fair if lineage means an enterprise data platform. It need not. For a small hospital with a handful of contracts, lineage can be a documented spreadsheet. It lists each element of the file, the system it came from, the person who supplied it and the method used to translate it. The scale changes. The discipline of knowing where each number came from does not.

Transparency would not move prices

A fourth objection questions the policy itself: if few patients shop for hospital care, price files would not change prices and the effort was wasted. That debate is about whether the rule was wise, and it remains open. It does not change the obligation. Once the rule took effect, hospitals had to comply whether or not the policy worked, and the cheapest way to comply year after year was to build the lineage once.

Where the argument is weakest

The lineage argument is weakest for services whose price cannot be known in advance at all, such as long inpatient stays paid per day with outlier terms. No amount of lineage turns an unknown length of stay into a firm price. The honest answer for those services is a published method plus historical amounts, and that is broadly where later rules went.

With hindsight

The rule took effect on January 1, 2021. CMS amended it in several later outpatient payment rules, each time pushing hospitals further from a publishing exercise toward a traceable data product.

The CY 2022 outpatient payment rule, published on November 16, 2021 and effective January 1, 2022, replaced the flat $300 daily penalty with one scaled by bed count. Hospitals with 30 or fewer beds still faced up to $300 a day. Those with 31 to 550 beds faced $10 per bed per day. Those with more than 550 beds faced up to $5,500 a day. The same rule required that the machine-readable file be accessible to automated searches and direct downloads. CMS described reports of hospitals using “blocking codes” such as NOINDEX and “rel canonical” tagging, techniques it said prevent commonly used web search engines from caching the web pages that link to the files.

CMS’s enforcement page lists the civil money penalty notices it has issued. The first two, dated June 7, 2022, went to Northside Hospital Atlanta and Northside Hospital Cherokee. As checked on November 14, 2026, the page listed notices through 2026 and had last been modified on June 11, 2026.

The CY 2024 rule, published on November 22, 2023, moved toward standardization. It required hospitals to make a good faith effort to ensure the data was true, accurate and complete, and to affirm this in the file. It required new data elements and a CMS template layout. It required a .txt file in the website’s root folder linking directly to the file, and a footer link to the page hosting it. It also let CMS work with health system officials when one or more hospitals in a system were noncompliant. The changes took effect on January 1, 2024, with later compliance dates for some elements.

On February 25, 2025, Executive Order 14221 directed the Secretaries of the Treasury, Labor and Health and Human Services to act, within 90 days, to require disclosure of the actual prices of items and services, not estimates. It also called for standardized, comparable pricing information and updated enforcement.

The CY 2026 rule, published on November 25, 2025, went to the source. Where a payer-specific negotiated charge is based on a percentage or algorithm, hospitals must now publish the 10th percentile, median and 90th percentile allowed amounts and a count of allowed amounts. They must calculate them from electronic remittance advice data, the EDI 835 transaction, or an equivalent source, using CMS’s method and lookback period. Hospitals must also encode an attestation and the name of the hospital’s chief executive officer, president or senior official designated to oversee the encoding of the data, and their Type 2 National Provider Identifiers. The policies took effect January 1, 2026, with enforcement from April 1, 2026. Under certain conditions, a hospital that admits a violation and waives its right to a hearing can have a penalty reduced by 35 percent.

The CY 2027 rule, published on November 6, 2026, did not amend the hospital price transparency regulations. It summarized a request for information on standardization and comparability that drew 185 pieces of correspondence, which CMS said it would consider for future rulemaking.

What held up is the shape of the problem. Each of the price transparency changes described here addressed lineage, findability or accountability: where the number comes from, whether anyone can find it, and who signs for it. What changed is the answer to the translation question. In 2019 the rule assumed a negotiated rate per item. By 2026 CMS was telling hospitals to derive many figures from what payers actually paid, and to name the executive who stands behind them.

Spotted an error? Tell us and we will check it and correct the article where needed.

  • price transparency
  • CMS
  • machine-readable files
  • data lineage
  • revenue cycle