How the No Surprises Act turned a medical bill into four data handoffs

The week

On December 21, 2020, Congress passed the Consolidated Appropriations Act, 2021, a very large year-end package that combined government funding with a second round of pandemic relief. The president signed it on December 27. Tucked inside was the No Surprises Act, the federal ban on most surprise medical bills that Congress had argued over for several years.

The typical case involved a patient who did everything right. She checked that her hospital was in her insurance network, showed up for surgery and went home to recover. Weeks later a bill arrived from an anesthesiologist she had never chosen and never met, who turned out to be outside her network. Stories like hers were common enough to become a political issue that both parties wanted to fix but could not agree on how.

A “surprise bill” usually came from something called balance billing. When a clinician is outside a patient’s insurance network, the health plan pays part of the charge and the clinician can bill the patient for the remaining balance. The patient often had no way to know this would happen, and no way to avoid it in an emergency.

Why it took years

Nobody in Congress defended surprise bills. The fight was over who would pay the difference once patients were taken out of the middle. One camp wanted a benchmark: out-of-network clinicians would be paid a rate tied to what plans typically paid in network. Hospital and physician groups opposed that, arguing it would let plans push rates down. Another camp wanted arbitration, with a neutral party settling each disputed payment. Plans and many employers worried arbitration would push costs up.

Competing bills passed committees in 2019 without becoming law. The version that passed in December 2020 was a compromise. It chose arbitration, but required the arbitrator to consider a benchmark called the qualifying payment amount alongside other factors. It also barred arbitrators from considering a provider’s billed charges or the rates paid by public programs such as Medicare and Medicaid.

What the law covers

The new law protects patients from balance billing in three main situations: most emergency care; certain out-of-network services delivered at in-network facilities, such as the anesthesiologist in the story above; and air ambulance transport. Instead of the patient being caught in the middle, payment disputes between the health plan and the provider go to an independent dispute resolution process, a form of arbitration. The patient pays only the cost sharing she would have paid in network.

The protections apply to private coverage: employer group health plans, including self-funded plans, and individual insurance. Medicare and Medicaid already limited balance billing under their own rules. Ground ambulances were left out, a notable gap; the law created an advisory committee to study them instead.

For non-emergency care at an in-network facility, the law allowed some out-of-network clinicians to ask patients to waive the protection in advance, through a written notice and consent. That option was not available for certain services patients rarely choose, such as anesthesiology, radiology, pathology and emergency medicine.

The parts beyond the ban

The law did several other things that mattered just as much in practice. It requires providers to give good-faith cost estimates to uninsured and self-pay patients before scheduled care. It adds duties to keep provider directories, the lists that tell patients which doctors are in network, accurate. And it calls for an “advanced explanation of benefits,” a personalized, coverage-specific estimate that insured patients would receive from their plan before a scheduled service.

Neighboring provisions in the same package added more. Patients in the middle of certain treatment would get up to 90 days of continued in-network coverage when their provider left the network. Insurance cards would have to show deductibles and out-of-pocket limits. A separate title on transparency barred certain contract clauses that kept price and quality data from plans.

The setting

The package arrived as hospitals were preparing for another transparency mandate. The federal hospital price transparency rule took effect on January 1, 2021, ten days after Congress voted. It required hospitals to publish their standard charges, including negotiated rates, in a machine-readable file and to show prices for a set of shoppable services. Revenue cycle teams were already stretched by the pandemic, and now faced two new layers of pricing obligations at once.

Most No Surprises Act provisions took effect on January 1, 2022. Congress left the detail to rulemaking by three federal departments: Health and Human Services (HHS), Labor and Treasury. That gap, between a clear promise to patients and the detailed rules about how hospitals, doctors and insurers would actually deliver it, is where the harder story of the law began.

Surprise billing was rightly framed as a consumer protection problem. Patients who did everything right still received large bills from clinicians they never chose. Nobody had to be persuaded that this was unfair. The remedy Congress wrote did more than prohibit a practice, though. It created a set of handoffs between organizations that rarely exchange data cleanly, and it made the patient’s experience depend on those handoffs working.

Executives outside health care finance often miss this. A law that sounds like a simple ban can, once it reaches the back office, turn into an information plumbing project spanning several companies. Read from 2026, the No Surprises Act contains four such integration problems, each with a different owner: the estimate, the directory, the dispute and the advanced explanation of benefits.

The four differ in who sends data to whom. The estimate moves information from clinicians’ systems to the patient, often across several provider organizations. The directory moves facts from providers to plans and on to patients. The dispute moves claim evidence from plans and providers to an arbitrator through a federal portal. The advanced explanation of benefits would move expected charges from providers to plans and coverage estimates from plans to patients, all before care happens. Each handoff fails in its own way, and each has fared differently since 2022.

The sections below set out the systems involved and take each handoff in turn. They then look at where the costs land, the habits of organizations that coped, and the questions that test whether the promise is being kept.

The systems behind a medical bill

The four handoffs connect systems that are not exotic, but each is usually owned by a different department, and sometimes by a different company.

The scheduling system is the calendar of the hospital or clinic. It records who is coming in, when, and for what. When a scheduler books a knee surgery, that booking is the first moment the organization knows care is going to happen.

The chargemaster is a hospital’s master price list, an enormous catalog with a price for every procedure, supply, drug and room. Very few people pay those list prices, because insurers negotiate their own rates, but the chargemaster is the starting point for billing.

The clinical order is the doctor’s instruction about what will actually be done: this procedure, these tests, this anesthesia. Orders change as doctors learn more about the patient, and each change can change the bill.

The revenue cycle is the industry’s name for everything that turns care into payment: registering the patient, checking insurance, coding the services, sending claims to the plan, and collecting what is owed. It is the financial assembly line of a health system.

Credentialing and contracting systems hold the facts about which clinicians work where, which insurance contracts are active, and whether a given doctor is “in network” for a given plan.

Finally, a data exchange is any arrangement where one organization’s computer system sends structured information to another’s and gets a usable answer back. When your bank app shows your credit card balance from a different bank, that is a data exchange. Health care has many of them, but they are often slower and less standardized than in banking.

The organizations on the other side

The clinicians at the center of surprise billing were often not employed by the hospital where they worked. Anesthesiology, emergency medicine, radiology and pathology were frequently staffed by separate physician groups, some of them large national staffing companies. Each had its own billing system, its own contracts with plans and its own network status. A single hospital stay could generate bills from four or five organizations.

On the payer side, the picture was just as fragmented. Fully insured plans are run by insurers regulated by the states. Self-funded employer plans are regulated federally under ERISA and usually administered by a third-party administrator, often using a network rented from another company. A self-funded plan’s directory, claims and dispute handling may therefore pass through two or three organizations before reaching the patient. Every extra hop is another handoff where data can go stale.

The No Surprises Act asks these systems to talk to each other in new ways, sometimes before the patient walks in the door. Those connections are why a law about fairness became a project about data.

The estimate: scheduling and billing meet before the visit

For uninsured and self-pay patients, providers and facilities must give a good-faith estimate of expected charges before scheduled care. On paper that is a document. Operationally it is a join between the scheduling system, the chargemaster, the clinical order and, when several clinicians are involved, the systems of other organizations that will deliver part of the care.

What the rules require

The departments set the timing in an interim final rule in the fall of 2021. When a service is scheduled at least three business days ahead, the estimate is due within one business day of scheduling. When it is scheduled at least ten business days ahead, the estimate is due within three business days. A patient who asks for an estimate without scheduling gets one within three business days. The estimate must list the expected items and services, with codes and expected charges.

The rules applied widely. They covered hospitals and surgery centers, but also independent physicians, therapists and other licensed clinicians who see self-pay patients. A psychotherapist seeing a client who pays out of pocket owes an estimate, which can cover a course of recurring sessions for up to a year. Many small practices had never produced anything like it.

A multi-provider estimate

Suppose a self-pay patient schedules a gallbladder removal at a community hospital. The surgeon is employed by the hospital, but the anesthesia group and the pathology lab that will examine the removed tissue are separate businesses. A full estimate needs the hospital’s facility charges, the surgeon’s fee, the anesthesia group’s fee and the lab’s fee. Those numbers live in at least three different billing systems. Somebody, or something, has to gather them.

The rule gave that job to the convening provider, the one who schedules the care, and required co-providers to send their portions. In practice the departments recognized that the data exchange did not exist. They said they would not enforce the requirement to include co-provider estimates until further rulemaking. The multi-provider estimate the law imagined was, from the start, a handoff the industry could not yet make.

Many organizations met the first requirement with templates and manual work, which was a reasonable start. A financial counselor would look up the procedure, pull typical charges from a spreadsheet, and send a letter. That works when volumes are low and procedures are routine.

When the order changes

The harder problem is that an estimate is only as good as the order behind it, and orders change. Take the case where the surgeon discovers during the operation that a more complex procedure is needed. For a patient quoted one figure at scheduling who receives a much larger bill after discharge, the old surprise has returned in a new form. The law gives uninsured patients a dispute process when the bill substantially exceeds the estimate. The rules set the threshold at $400 or more above the estimate for any one provider.

That patient-provider dispute process uses a separate selected dispute resolution entity. It gives the self-pay patient real recourse, but it also means each estimate is a commitment with consequences. An estimate that omits a likely component, such as pathology, is an invitation to a dispute.

The durable fix puts estimates on the scheduling data path, so the estimate is produced automatically from the same information the scheduler is already entering. It is generated from the scheduled procedure and its usual components, and refreshed automatically when the order changes. Patient financial counselors handle the exceptions instead of building every estimate by hand.

What an estimate engine needs

Automating estimates needs three assets. The first is a set of templates that map each scheduled procedure to its usual components: the facility charge, the professional fees, anesthesia time, likely lab and pathology work and common supplies. The second is pricing for each component, drawn from the chargemaster and, for self-pay patients, from the organization’s discount policy. The third is a feed of expected charges from the separate groups that will take part, which requires agreements with them about format and timing.

Each estimate also has to be kept. The rules require providers to retain copies, and a dispute will turn on what the estimate said. An estimate system that stores every version, with the order it was based on, gives the organization its evidence when a patient challenges a bill. One that produces letters and forgets them leaves the organization reconstructing its own promise.

How it differs by setting

The fix looks different by type of organization. A large health system can build estimates into its electronic health record and revenue cycle platform, drawing on historical claims for each procedure. An ambulatory surgery center has a narrower menu of procedures, so templates work well if they are kept current. An independent physician practice may rely on its practice management vendor’s estimate tool. A behavioral health practice mostly needs a clear, recurring estimate for a course of therapy. The obligation is the same; the right level of automation is not.

A useful measure is the share of final bills that land within a set tolerance of the estimate, by service line. If orthopedics hits its tolerance most of the time and cardiology misses often, leaders know where to look. That single number tells an executive more about the patient’s financial experience than any policy binder.

The directory: the plan publishes what the provider knows

When a patient wants to know whether a doctor is in network, she usually checks her health plan’s provider directory, a searchable list on the plan’s website or app. For years those directories were notoriously unreliable: doctors who had retired, moved or stopped taking the plan were still listed.

The law put new weight on provider directory accuracy. Plans must verify directory information on a regular cycle, and providers must send timely updates. Patients who rely on an inaccurate directory get some protection on cost sharing, so errors now carry a cost. Money attached to bad data tends to focus attention.

What the statute set

The statute was specific about timing. Plans must verify their directory information at least every 90 days. They must update the directory within two business days of receiving new information from a provider. When a patient asks by phone or electronically whether a provider is in network, the plan must answer within one business day. Providers, in turn, must submit updated information when their network status or details change.

The departments said they would not issue regulations on these provisions before they took effect. They expected plans and providers to comply in good faith with the statute in the meantime. That left each plan to design its own verification process, and each provider to respond to many different ones.

Who owns the facts

Directory errors are an old problem rooted in data ownership. The plan publishes the directory. The facts in it live in provider credentialing, scheduling and contracting systems: which clinicians practice where, which locations accept new patients, and which contracts are active. Each side has tended to blame the other’s data.

The arrangement works like a city restaurant guide that publishes opening hours only the restaurants know. If restaurants change their hours without telling the guide, diners find locked doors and blame the guide. The guide can call every restaurant every quarter, slowly and at some expense, or each restaurant can keep one accurate record and send changes out automatically.

Health systems that treat their own roster as a governed data product are better placed. A “data product” here simply means a dataset that is managed like something the organization sells: it has a defined owner, a quality standard and customers who depend on it. In this case that means one authoritative source for clinician, location and network status, an owner who answers for its accuracy, and an outbound feed to each plan.

The alternative is answering every plan’s attestation request by hand, with different staff giving different answers. A regional health system working with a dozen plans could receive a dozen different forms each cycle, each completed by whoever had time that week. A clinician who leaves should disappear from every plan’s directory on a predictable schedule, and someone should be able to prove it.

Measuring accuracy

Directory accuracy can be measured directly. A plan can sample listings each month and call the offices, as a patient would, to confirm location, network status and whether new patients are accepted. A health system can compare its roster with each plan’s published directory and count mismatches. Both measures are crude. Both are far better than assuming the attestation forms were filled in correctly. The mismatch count, tracked over time and by plan, shows whether the feed is working.

Continuity of care uses the same data

The roster also drives a neighboring obligation. When a provider leaves a plan’s network, patients in active treatment, such as those being treated for cancer or for a pregnancy, may keep in-network coverage for up to 90 days. The plan must notify those patients. It can do so only if it learns of the departure promptly and knows who was being treated. The same stale roster that misleads a new patient in the directory also leaves an existing patient uninformed.

The trade-offs

Shared industry utilities for provider data exist, letting a provider update once and have several plans draw from the same source. They reduce duplicate work but add another organization to the chain, with its own contract and data quality rules. A health system with a strong internal roster gains from them. One without a roster simply sends its errors to more places faster. For a self-funded employer, the practical lever is its contract with the administrator and rented network: who verifies, how often, and what happens when a patient relies on a wrong listing.

The dispute: arbitration became a back-office channel

With the patient protected, the plan and the out-of-network provider still have to settle payment. The law created an open negotiation period followed by independent dispute resolution through a federal portal. Independent dispute resolution is a form of arbitration: a neutral third party reviews both sides’ offers and picks a payment amount.

How the process runs

The sequence is rigid. The plan makes an initial payment or sends a denial. Either side can then open a 30-business-day negotiation. If that fails, either side has four business days to start the federal process. The parties select a certified arbitration entity, or one is assigned. Each side submits a single offer with supporting information. The entity must choose one offer or the other, a format often called baseball-style arbitration, because it rewards reasonable offers.

Each step has deadlines, and each submission needs claim-level evidence about the service, the qualifying payment amount and the factors the arbitrator may weigh. The qualifying payment amount is, roughly, a benchmark based on what the plan typically pays in-network providers for the same service in the same area. Much of the fighting over the law has been about how much weight that benchmark should carry.

Other details matter operationally. Similar claims can sometimes be batched into one dispute. Both parties pay an administrative fee, and the losing party pays the arbitration entity’s fee. After a decision, the party that started it must wait 90 days before bringing another dispute over the same items with the same counterparty. Every one of those rules creates a tracking requirement.

What happened when it opened

Design met reality hardest here. Dispute volumes ran far beyond what the departments had projected. The process was litigated repeatedly, including the Texas Medical Association cases over how arbitrators should weigh the qualifying payment amount, and operating rules changed more than once. A federal court in Texas struck down the original instruction that arbitrators presume the benchmark was the right payment. Later decisions vacated other provisions, and the departments paused parts of the process while they adjusted.

Reports the departments published on the process showed providers prevailing in most decided disputes. Many disputes were also found ineligible, because they were filed under the wrong process, outside a deadline or for coverage the law did not reach. Eligibility review became a large part of the arbitration entities’ work, which slowed decisions for everyone.

A standing workflow

For providers and plans, dispute handling became a standing workflow needing case management, document assembly and deadline tracking. Case management software is the kind of tool a law firm or insurance claims department uses: every dispute is a case, with a file, a status, an owner and a clock. Organizations that tried to run it from spreadsheets missed deadlines and lost cases on procedure. Losing on procedure, rather than on the merits, is the most avoidable kind of loss there is.

The handoff inside this workflow is the claim record. A provider needs to know, from the remittance it receives, whether the claim falls under the federal process or a state law. It needs the plan’s qualifying payment amount, which the plan must disclose with the initial payment. It needs the clinical documentation that supports its offer. Those facts sit in different systems, and the four-day window to start a dispute leaves no time to hunt for them.

State law adds a layer. Where a state has its own surprise billing law with a payment standard, it generally governs fully insured plans in that state instead of the federal process. Self-funded employer plans generally fall under the federal process, though some states let them opt in to the state system. Getting that routing wrong sends a dispute to the wrong forum, where it will be found ineligible after the deadline for the right one has passed.

How it differs by party

The burden differs sharply by party. Large physician staffing groups built dedicated dispute operations and filed at scale. Smaller independent groups often relied on billing companies or did not file at all, accepting the plan’s payment. Plans and administrators had to defend disputes and pay awards. For self-funded employers, awards came out of the plan’s own funds, often without the employer seeing the individual cases. Hospitals were less exposed for their own facility charges, which are usually in network, but felt the effects through their contracted physician groups.

The counter-argument

The strongest objection to calling this a data problem is that the dispute process is mainly about money. Plans argued that arbitration awards above the benchmark raised costs that would flow into premiums. Providers argued the benchmark itself was set too low, by a method that favored plans. On that reading, better case management only helps one side win a fight over price.

That is fair as far as it goes. The data point holds anyway. Whatever the right payment level, a party that cannot assemble evidence within the deadlines loses disputes it should win, or pays fees on disputes it should never have filed. Eligibility failures and missed deadlines are pure waste on both sides. Getting the data right does not settle the price argument, but it determines who can take part in it.

Patients are removed from the middle of the dispute, which was the intent. A provider whose dispute process is chaotic still feels pressure to recover revenue elsewhere, and patients can notice that in prices and collection practices.

The advanced explanation of benefits: the handoff that waited

The fourth piece was the most ambitious. For insured patients, providers would send expected charges to the patient’s plan, and the plan would return a coverage-specific estimate before scheduled care. The patient would learn what the hospital charges and also what her plan will pay and what she will owe, before the procedure.

That requires near real-time exchange of structured data between provider and payer systems. “Structured” means the information arrives in predictable fields a computer can read, such as procedure codes, dates and amounts, rather than as a scanned letter or a free-text email. In 2020 no standard form of that exchange existed. Without a standard, every hospital and every plan would have had to build custom connections to each other, an expensive tangle.

Why it stalled

The departments deferred enforcement while standards were worked out. In 2022 they issued a request for information on how the exchange should work, including which technical standards to use. Industry groups worked on standards based on FHIR, the modern format for exchanging health data through application programming interfaces. Agreeing on a standard was only the first step. Every provider’s scheduling system and every plan’s benefits engine would then have to implement it.

The advanced explanation of benefits also depended on the estimate for insured patients, the provider’s side of the exchange. That estimate needed the same multi-provider assembly that had already been deferred for uninsured patients. Two hard handoffs were stacked on top of each other, and neither could go live without the other.

The pattern is instructive. Where the law asked one organization to produce a document, the industry complied, with uneven quality. Where it asked two organizations to exchange structured data quickly, the industry was not ready, and regulators accepted that.

What preparing looks like

Some of the plumbing already exists. Providers routinely check a patient’s eligibility and benefits electronically before a visit, using standard transactions that return deductibles and coverage details. The missing piece is a structured way to send a full set of expected services and charges and receive a priced answer. An organization can prepare by making sure its scheduled services carry clean procedure codes, its components are mapped, and its contracted rates are loaded accurately for each plan. Those are the same inputs the self-pay estimate and the price transparency files need.

The substitutes

Patients did get partial substitutes. Hospital price transparency files and the plans’ own online cost estimators under the federal transparency in coverage rule gave insured patients some ability to look up prices. Those tools answer a general question about rates. They do not tell a specific patient, for a specific scheduled service, what she will owe after her deductible. The advanced explanation of benefits was meant to answer exactly that, and nothing else in the system does yet.

The same pattern applies beyond billing. Patient-facing promises that depend on exchange between organizations take longer to arrive than the statute suggests. Organizations that prepare the underlying data early, such as clean scheduling codes, accurate rosters and order-level pricing, carry less risk when the promise is finally enforced.

Bad estimates reach the bedside

The law’s costs show up in five places, starting with the patient who opens a bill that does not match her estimate.

Customers. A patient’s view of a hospital is shaped as much by the bill as by the bedside. An estimate that turns out badly wrong, or a directory listing that sends her to the wrong doctor, undoes a lot of good clinical work.

Money. Disputes lost on procedure are revenue left on the table. Manual estimates and hand-completed directory attestations consume staff time that could go elsewhere. Each of these costs is small per case and large across a year.

Regulators. The law created obligations to patients that regulators can check. A health system that cannot show how its estimates are produced, or when departed clinicians were removed from directories, is in a weak position if questions are asked. Enforcement is split: states enforce many provisions against providers and insurers, while federal agencies cover gaps and self-funded plans. A multistate system may face several regimes at once.

Staff. The people who absorb broken handoffs are financial counselors, schedulers and billing staff. When estimates are built by hand and directory forms are filled in from memory, those teams spend their days chasing information across departments and fielding calls from confused patients. Good data plumbing is, among other things, a way of making those jobs bearable.

One scenario ties these together. Say a mid-sized health system discovers that one of its surgeons left eight months ago but still appears in three plans’ directories. Several patients booked appointments expecting to see him, were redirected to a colleague who was out of network for their plan, and complained. Nobody was careless; each department assumed another had sent the update. The fix was naming one owner for the roster and sending changes to every plan from a single source.

Reputation. Surprise billing became a political issue because the stories were vivid. A health system that recreates the surprise through poor estimates risks becoming the next story.

Where the costs settle

The costs do not stay with the party that causes them. Arbitration awards paid by plans feed into premiums, which employers and employees pay. Staff time spent on manual estimates comes out of revenue cycle budgets already under pressure. A patient who receives an inaccurate estimate may delay care or pay late. Self-funded employers bear dispute outcomes directly, often without visibility into the cases. Each party has an interest in cleaner handoffs, even when they disagree about price.

Habits of the organizations that coped

Organizations that handled the law well share a few habits.

• They name an owner for each handoff. The estimate, the directory feed, the dispute workflow and the future advanced explanation of benefits each have a named leader, not a committee.
• They build on data they already have. Estimates come from scheduling and order data; directory feeds come from the credentialing system. Nothing is retyped.
• They measure outcomes the patient would recognize. How often the final bill landed near the estimate. How quickly a departed clinician vanished from directories. How many disputes were lost on deadlines.
• They treat new rules as variations on an old theme. Each new requirement is checked against the same clean data, rather than met with a fresh workaround.

Owners by organization

The natural owners vary. In a health system, the estimate usually belongs to patient access or the revenue cycle, the roster to credentialing or provider enrollment, and disputes to managed care contracting or the business office. In a physician group, one billing leader may own all three. At a health plan, directory accuracy belongs to network management, disputes to claims or payment integrity, and the coverage estimate to member services and benefits configuration. A self-funded employer owns none of the work directly, but owns the contract that decides who does.

The organizations that struggled did the opposite.

• Treating the good-faith estimate as a form to be filled in, rather than a promise that must survive changes to the order.
• Answering directory attestations by hand and hoping different staff give the same answer.
• Running arbitration from spreadsheets and email folders, then missing a deadline.
• Waiting for the advanced explanation of benefits to be enforced before cleaning up the pricing and scheduling data it will need.
• Assuming the dispute rules are settled, when litigation and rulemaking have changed them more than once.

Testing the promise to patients

An executive does not need to understand procedure codes to hold this work to account. These questions get to the substance quickly. The first two concern the estimate, the next two the directory, and the last the dispute. The advanced explanation of benefits is covered by the first: an organization that cannot generate an estimate from its own data will not be able to send one to a plan.

1. Who owns our good-faith estimates, and are they generated from scheduling and order data or built by hand?
2. When an order changes after scheduling, does the estimate update automatically, and how would we know if it did not?
3. Do we have one authoritative roster of clinicians, locations and network status, and who answers for its accuracy?
4. When a clinician leaves, how many days until every plan’s directory reflects it, and can we prove it?
5. Where do our payment disputes live, and if the answer is a spreadsheet, how many deadlines have we missed?

Answering them requires no technology background. Each one tests whether a promise made to patients is backed by data someone owns. The evidence should be routine reports: estimate accuracy by service line, roster change logs with dates sent to each plan, and a dispute register with deadlines met and missed.

With hindsight

The core protection held up. Most insured patients are no longer exposed to balance bills for emergency care, or from anesthesiologists, radiologists and similar clinicians they did not choose at in-network facilities. It is easy to forget how common those bills were, and how much fear they added to an already frightening experience. On its central promise, the law did what it set out to do.

The machinery around the protection was harder. The departments issued interim final rules in July and September 2021 and a final rule on the dispute process in August 2022. The dispute process was swamped and litigated, and its rules changed several times. The Texas Medical Association cases in federal court in Texas vacated several provisions, beginning with the presumption in favor of the qualifying payment amount in early 2022. The departments paused parts of the process in 2023 while they adjusted, and proposed a rule on dispute operations later that year. Providers won most decided disputes, which plans and employers argued was raising costs.

Good-faith estimates for self-pay patients became routine but uneven in quality. The requirement to include other providers’ charges in those estimates stayed under enforcement discretion, because the exchange between providers still did not exist. The advanced explanation of benefits was still unenforced well into 2026, with its implementing rule still on the federal regulatory agenda. A promise to give insured patients a personalized estimate before care, written into law at the end of 2020, was still waiting for the data exchange it depended on more than five years later.

That gap says something important. Laws can ban a practice quickly. They cannot make separate organizations’ computer systems talk to each other quickly, and regulators have shown they will wait when the industry is genuinely not ready.

Directories became a routine exchange between plans and providers, with verification cycles and attestations on a regular schedule. The statute did not create a single shared roster, so each plan still asks each provider for the same facts. Ground ambulances remained a gap in the federal law. The advisory committee the law created later recommended extending protections to them, and a number of states have acted on their own.

Set side by side, the record is clear. The ban held. The self-pay estimate held in form, though not in its multi-provider version. The directory duties became routine. The dispute process works, at far higher cost and friction than planned. The advanced explanation of benefits has not arrived. Success tracks the number of organizations each handoff involves: the fewer parties, the sooner the promise was kept.

Federal health policy has kept layering patient experience on data exchange, through price transparency files, information blocking rules and prior authorization interfaces. Hospital price transparency took effect in January 2021, and plans began publishing machine-readable rate files under the transparency in coverage rule in 2022. In January 2024 CMS finalized a rule requiring many plans in federal programs to build prior authorization interfaces. In February 2025 an executive order directed agencies to strengthen enforcement of price transparency. Each of these asks hospitals, doctors and plans to share structured information faster and more openly.

Each one looks, from the boardroom, like a separate compliance project. Underneath, most of them draw on the same few things: accurate prices, clean scheduling and order data, and a trustworthy roster of who works where. The transparency files, the estimates and the eventual coverage estimate all depend on knowing what a given service costs under a given contract.

Health systems that invested in clean provider, pricing and scheduling data found each new requirement less disruptive than the last. Those that met each rule with a one-off workaround kept paying for it again. The practical habit is this: when the next patient-facing rule arrives, ask first whether it can be met from data the organization already governs. If the honest answer is no, the problem lies deeper than the new rule. Asking that question costs nothing, and the answer usually shows where the next investment belongs.