Medicare's 2015 imaging rule began a program tying a clinician's pay to another's data; CMS paused it in 2023

The physician fee schedule rule published November 16, 2015 set up the first part of Medicare's appropriate use criteria program for advanced imaging. The statute tied the imaging provider's payment to the ordering clinician's decision-support data, and that hand-off is where CMS later stopped.

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

The physician fee schedule rule published November 16, 2015 set up the first part of Medicare's appropriate use criteria program for advanced imaging. The statute tied the imaging provider's payment to the ordering clinician's decision-support data, and that hand-off is where CMS later stopped.

  1. The week Our argument is that the program's central design problem was a data hand-off, not the criteria.
  2. The schedule ran ahead of the systems In our reading, the rule ordered the work the way a systems team would: content first, then the tool, then the interface to billing.
  3. The criteria came from provider-led entities In our reading, the definition put the decision about what counts as appropriate imaging in the hands of organizations led by clinicians, with CMS as the qualifier rather than the author.
  4. The consultation lived in the ordering workflow The statute required the ordering professional to consult criteria through a qualified clinical decision support mechanism. [2] It required the ordering professional to give the furnishing professional the consultation information. [2] Those two steps are where the program met the clinical workflow.
  5. The claim was where the program stopped In our reading, the program's data design failed here.
  6. What remains is decision support without the mandate A fair objection is that the criteria, not the data path, were the real problem.
  7. With hindsight For health system leaders, the lasting lesson is about data obligations that cross organizational lines.

The week

On November 16, 2015, the Federal Register published the Centers for Medicare & Medicaid Services' final rule with comment period on the physician fee schedule for calendar year 2016. [1] Among its parts was section G, "Appropriate Use Criteria for Advanced Diagnostic Imaging Services." [1]

The section implemented part of a law passed the year before. [1] Section 218(b) of the Protecting Access to Medicare Act of 2014, approved April 1, 2014, added a new section 1834(q) to the Social Security Act. [1] [2] Its title was "Recognizing Appropriate Use Criteria for Certain Imaging Services." [1]

The rule described four major components of the program, each with its own date in the statute. [1] The first was establishing appropriate use criteria, or AUC, by November 15, 2015. [1] The second was mechanisms for consulting AUC, by April 1, 2016. [1] The third was consultation by ordering professionals and reporting by furnishing professionals, by January 1, 2017. [1] The fourth was annual identification of outlier ordering professionals for services furnished after January 1, 2017. [1]

The rule said it outlined the initial component of the program and CMS's plan for the rest. [1] It addressed mainly the first component: the process for establishing AUC, along with related definitions. [1] It added a new regulation, 42 CFR 414.94, titled "Appropriate use criteria for advanced diagnostic imaging services." [1]

The rule described AUC as criteria that link a specific clinical condition or presentation, one or more services, and an assessment of the appropriateness of the service. [1] It said the goal of the program was to promote evidence-based use of advanced diagnostic imaging, improving quality of care and reducing inappropriate imaging. [1]

Under the rule, criteria would come from provider-led entities. [1] Entities would apply to CMS to become qualified to develop, modify or endorse AUC. [1] Applications would be due by January 1 each year, and CMS would post a list of qualified entities by the following June 30. [1] At that point, all AUC developed or endorsed by a qualified entity would become specified AUC. [1]

CMS finalized January 1, 2016 as the deadline for the first applications. [1] It changed its proposed six-year cycle to a five-year cycle of approval. [1] Under the example it gave, an entity qualified for a cycle beginning July 2016 would apply for requalification by January 2021. [1]

The rule addressed the later components only in outline. [1] It said CMS did not intend to require ordering professionals to meet the consultation requirement by January 1, 2017. [1] It said CMS did not expect to have approved clinical decision support mechanisms until approximately summer of 2017. [1]

It also introduced the idea of priority clinical areas. [1] CMS said it would identify such areas through later rulemaking. [1] It proposed to identify outlier ordering professionals from within those areas. [1] It said this would give clinicians a chance to become familiar with AUC in those areas before their claims fed outlier calculations. [1]

The statute behind the rule was specific about payment. [2] For an applicable imaging service furnished in an applicable setting and paid under an applicable payment system, payment could be made only if the claim included certain information. [2] That information named the decision support mechanism the ordering professional consulted, whether the order adhered to the criteria, and the ordering professional's national provider identifier. [2]

The November 2015 rule set up the part of the program that everyone could agree on: where the criteria would come from. [1] The harder parts were in the statute, waiting. [2] The statute required that the professional who furnished an imaging study be paid only if the claim carried information about a decision made by someone else, the ordering professional. [2]

Our argument is that the program's central design problem was a data hand-off, not the criteria. The criteria could be qualified, listed and specified. [1] The consultation could happen in a decision support tool. [3] The hard part was moving the record of that consultation from the ordering clinician's system to the furnishing provider's claim, and having Medicare check it. [4] CMS later said that real-time claims-based reporting requirement was an insurmountable barrier. [4]

For health system leaders, the program is a case study in data obligations that cross organizational lines. A payment rule that depends on another party's data creates work in systems neither party fully controls. It also creates a question every compliance officer should ask early: can the payer actually tell which claims the rule applies to?

We take it in five parts, in the order the data would have moved:

  1. The schedule: what the statute required and when, and what the 2015 rule did and deferred.
  2. The criteria: who decided what counted as appropriate, and how.
  3. The consultation: the decision support mechanism and its place in the ordering workflow.
  4. The claim: the reporting requirement, and why CMS could not apply it reliably.
  5. What remains: decision support without a federal mandate, and the lessons for other data hand-offs.

The schedule ran ahead of the systems

Section 1834(q) set dates that left little room. [1] [2] It required the Secretary to specify applicable AUC by November 15, 2015, and to publish an initial list of qualified decision support mechanisms by April 1, 2016. [1] [2] It required consultation and reporting beginning January 1, 2017. [2] It required prior authorization for outlier ordering professionals beginning January 1, 2020. [2]

The final rule appeared in the Federal Register on November 16, 2015. [1] It addressed the first component and said CMS would develop the others through further rulemaking. [1]

What the rule deferred, and why

The rule was candid about the remaining dates. [1] It said CMS did not intend to require ordering professionals to meet the consultation requirement by January 1, 2017. [1] Responding to comments about the timeline, CMS said it agreed the statutory timeline was difficult to meet. [1] It said consultation with decision support mechanisms would not be required on January 1, 2017, because it did not expect to have approved mechanisms by then. [1]

The rule explained the sequence it intended. [1] It said it was important first to establish, through notice and comment, the process for specifying AUC and the mechanisms through which ordering providers would access them. [1] It said it anticipated adopting claims-based reporting policies in the 2017 and 2018 rulemaking cycles. [1]

In our reading, the rule ordered the work the way a systems team would: content first, then the tool, then the interface to billing. That order made sense. It also meant the riskiest part, the claim, came last.

The statute's exceptions

The statute exempted three categories from consultation and reporting. [2] They were imaging ordered for an individual with an emergency medical condition, imaging ordered for an inpatient and paid under Part A, and orders by professionals granted a significant hardship exemption. [2] The statute gave as an example of hardship a professional in a rural area without sufficient Internet access. [2]

It limited the program to three payment systems: the physician fee schedule, the hospital outpatient prospective payment system, and the ambulatory surgical center payment system. [1] [2] Those limits defined which claims the reporting rule would touch. They became central to the program's later difficulty. [4]

Why the dates mattered to providers

Dates in a statute shape budgets. A health system that read section 1834(q) in 2015 saw a reporting requirement with a 2017 start and a prior authorization consequence from 2020. [2] Planning for decision support integration, vendor selection and staff training had to begin before CMS had defined the mechanism requirements. [1]

The 2015 rule gave some relief by saying consultation would not be required on January 1, 2017. [1] It did not give a replacement date. [1] In our reading, that left providers to invest against an uncertain schedule, a pattern that continued through later years of the program. [3] [4]

A schedule that kept moving

The 2015 rule was the first of several that adjusted the program's timing. [1] [3] [4] The 2017 rule said that, at the earliest, the first qualified decision support mechanisms would be specified on June 30, 2017. [3] It said CMS expected furnishing professionals to begin reporting on January 1, 2018. [3] It said CMS was announcing that date so that physicians and others could prepare, while further rulemaking would follow. [3]

The 2024 rule records what came after. [4] It says an educational and operations testing period for claims-based reporting began in calendar year 2020. [4] It says the 2022 rule announced an intention to begin the payment penalty phase by returning non-compliant claims. [4] The penalty phase did not begin before the 2024 rule paused the program. [4]

In our reading, the repeated deferrals were themselves a signal. A requirement whose start date moves each year is telling planners that a dependency is unresolved. In this program, CMS later identified that dependency as the claims system. [4]

What the deferrals cost

Each deferral reset decisions that providers had already made. A health system that bought or built a decision support integration for a January 1, 2017 start had a tool and no requirement. [1] One that waited risked a compressed schedule if the next date held. In our reading, the safest course was to build the parts with clinical value on their own and to hold back the claims-side work until CMS settled the edits.

The criteria came from provider-led entities

The 2015 rule's main work was defining who could produce the criteria. [1] The statute called for AUC developed or endorsed by national professional medical specialty societies or other provider-led entities. [2] The rule set out how such entities would become qualified. [1]

The qualification process

Entities that believed they met the definition of a provider-led entity would apply to CMS. [1] Their applications had to document adherence to each qualification requirement and state how the entity met the definition. [1] The rule noted that the application was not a CMS form; the applicant created it. [1]

CMS would post the list of qualified entities by June 30 each year. [1] At that point, every AUC the entity had developed or endorsed would become specified AUC for the program. [1] The 2017 rule later recorded that the first list of qualified entities was posted at the end of June 2016. [3]

The definition debate

The rule discussed whether clinicians who take part in developing criteria should themselves count as provider-led entities. [1] Some commenters suggested they should. [1] CMS disagreed. [1] It said the development process would typically be embedded in a larger organization, and that the statute focused on the structure of the organization as a whole. [1]

CMS said it did not believe its modified definition would limit the AUC market or the participation of third parties, such as radiology benefit managers, in developing criteria. [1] It said third parties might collaborate with qualified entities. [1]

In our reading, the definition put the decision about what counts as appropriate imaging in the hands of organizations led by clinicians, with CMS as the qualifier rather than the author. That choice spread the clinical judgment across several sources. It also meant a provider could face criteria from more than one qualified entity, delivered through different tools.

Two approaches to rolling out criteria

The rule described a choice the field faced. [1] One view favored starting with as comprehensive a library of criteria as possible. [1] The rule linked this view to the experience of practitioners in the Medicare Imaging Demonstration, who found no criteria for their patient's condition about 40 percent of the time. [1] The other view favored focusing on a few priority clinical areas at a time, to avoid alert fatigue. [1]

CMS said there was merit to both approaches. [1] It said the focused approach might work better for a large health system that produces its own criteria, and the comprehensive approach for a smaller practice adopting a complete system developed elsewhere. [1] It said a successful program would allow flexibility. [1]

A health system as its own criteria source

The 2015 rule anticipated that large health systems might become qualified provider-led entities themselves. [1] It said professional medical societies and large health systems that become qualified could get local feedback at the outset and build alternatives into their criteria. [1] It said local entities could themselves become qualified to develop, modify or endorse criteria. [1]

For a health system with its own imaging guidelines, this was an opportunity and an obligation. Qualification would let its clinicians use criteria they had written. [1] It would also bring the system's guideline process under CMS's requirements for evidence review and transparency. [1] And it would require reapplication on the five-year cycle the 2015 rule set. [1]

In our reading, the choice turned on governance capacity. A system that already ran a committee to write and maintain imaging guidelines, with documented evidence review, was close to the qualification standard. A system without one would find it cheaper to adopt criteria from a qualified society through a commercial decision support tool.

Priority clinical areas

The rule combined the two approaches. [1] Large libraries of criteria could become specified across conditions. [1] Outlier identification, the step that would lead to prior authorization, would draw only on priority clinical areas. [1]

The 2017 rule named the first priority clinical areas. [3] It finalized a list that included coronary artery disease, suspected pulmonary embolism, headache, hip pain, low back pain, shoulder pain, cancer of the lung, and cervical or neck pain. [3] It removed chest pain, abdominal pain, suspected stroke and altered mental status from the proposed list. [3]

In our reading, the priority areas were a sensible way to limit the stakes while clinicians learned the tools. They did not change the reporting requirement, which in the statute applied to applicable imaging services generally, subject to its exceptions. [2]

The consultation lived in the ordering workflow

The statute required the ordering professional to consult criteria through a qualified clinical decision support mechanism. [2] It required the ordering professional to give the furnishing professional the consultation information. [2] Those two steps are where the program met the clinical workflow.

What a decision support mechanism was meant to be

The 2017 rule described decision support mechanisms as the electronic portals through which clinicians would access criteria during the patient workup. [3] It said they could be standalone applications requiring direct entry of patient information. [3] It said they might be more effective when they automatically drew in patient characteristics, laboratory results and comorbidities from electronic health records. [3]

The statute allowed flexibility in where the mechanism sat. [3] The 2017 rule noted that qualified mechanisms could be within certified EHR technology, private sector mechanisms independent of it, or mechanisms established by the Secretary. [3]

The 2015 rule had already stressed integration. [1] It said there was general agreement that criteria, and the decision support mechanisms through which providers access them, must be integrated into the clinical workflow and facilitate evidence-based care. [1] It said CMS would work with stakeholders on standards for certified health IT that could enable interoperability of criteria across systems. [1]

Alert fatigue and clinician time

The 2015 rule took clinician time seriously. [1] It recounted that some practitioners in the Medicare Imaging Demonstration had worked through a decision support tool only to find no criteria for their patient's condition. [1] It described the view that too many low-evidence alerts create alert fatigue. [1]

The rule described an alternative. [1] Rather than searching through a tool, clinicians would enter the patient's condition and see a message saying whether criteria existed. [1] In our reading, that design puts the burden on the tool's knowledge of the clinical picture, which depends on how well it draws data from the EHR. [3]

For a chief medical information officer, three measures decide whether clinicians accept the tool. They are how often a consultation finds applicable criteria, how many clicks it adds to an order, and how often clinicians override its result. The rules did not set targets for any of them. [1] [3] They are local decisions with local evidence.

Small practices and the hardship exception

The statute's hardship exception named a rural professional without sufficient Internet access as an example. [2] It made the exemption case-by-case and subject to annual renewal. [2] In our reading, that design placed the burden of showing hardship on the practice, year after year.

Small practices also faced the furnishing problem in a sharper form. A small imaging center can receive orders from referring clinicians on several different systems. Each referral source would need to send consultation data in a form the center could use. [2] The center had no authority over any of them.

The data an order had to carry

The consultation produced three pieces of information the statute required on the claim. [2] They were the identity of the mechanism consulted, whether the order adhered to the criteria, did not adhere, or had no applicable criteria, and the ordering professional's national provider identifier. [2]

In our reading, those items describe a data record that had to travel. It would be created in the ordering clinician's EHR or decision support tool. It would ride with the order to the imaging provider, which might be a hospital department, a freestanding center or a practice in a different health system. It would then have to land, intact, on the furnishing provider's claim.

Standards the program leaned on

The 2015 rule tied the program's success to health IT standards. [1] It said CMS would work with stakeholders to understand how to ensure that appropriate mechanisms were available, particularly regarding standards for certified health IT, including EHRs, that could enable interoperability of criteria across systems. [1]

In our reading, that sentence names the gap the program never closed. Interoperability of criteria, the content, is one problem. Interoperability of the consultation result, the record that must reach a claim, is another. An EHR can display criteria from a qualified source and still have no standard way to send the result with an outbound order to an unaffiliated imaging center.

Who held the record at each step

The 2015 rule observed that the statute distinguished between the ordering and furnishing professional. [1] It said the professional who orders an imaging service is usually not the one who bills Medicare for the test. [1] That observation is the root of the program's operational problem.

Consider the systems involved. The ordering clinician's EHR generates the order. A decision support tool, possibly from another vendor, produces the consultation result. An interface, or a fax, or a referral portal carries the order to the imaging provider. The imaging provider's registration and scheduling system receives it. Its billing system builds the claim. Each hand-off is a place the consultation data can be lost or altered.

Within one integrated health system, the hand-offs can be engineered. Across organizations, each party controls only its own part. In our reading, the program asked the furnishing provider to answer for data it did not create, through a chain it did not own.

What a health system would have built

A health system preparing for the reporting requirement had to make choices on both sides. As an ordering organization, it needed a qualified mechanism in its clinicians' workflow, and a way to send the consultation result with each order. [3] As a furnishing organization, it needed to receive that result from outside referrers and carry it onto claims. [2]

The furnishing side also needed a decision rule for orders that arrived without the information. [4] It could contact the ordering professional, decline to furnish the service, or risk a claim denial. [4] Each choice had a cost: staff time, patient delay, or revenue at risk.

The claim was where the program stopped

CMS's 2024 rule explained why the program could not proceed as written. [4] It said that while each component had presented unique challenges, the greatest challenge was fully operationalizing the real-time claims-based reporting requirement. [4]

The edit problem

The rule described the core difficulty. [4] The program applied only to imaging in certain settings and paid under certain systems. [2] [4] To apply it, Medicare's claims system had to identify which claims were subject to it, automatically, without manual action by the providers submitting them. [4]

The rule said the existing claims processing system did not have the capacity to fully automate that distinction. [4] It said this meant the system could not ensure that claims not subject to the reporting requirement would not be improperly denied. [4] It said there was a significant risk that full implementation of the payment penalty phase would result in inappropriate claim denials. [4]

The split-setting example

The rule gave an example. [4] An imaging service can be furnished in two settings, only one of which is an applicable setting. [4] When the technical component is furnished in a critical access hospital, which is not an applicable setting, the whole service is outside the consultation requirement. [4] Neither the technical nor the professional component claim needs consultation information. [4]

The rule said there was no way in real-time claims processing to identify that the professional component claim was for a service not furnished in an applicable setting. [4] For it to be paid, the furnishing professional would have to append a modifier marking it as exempt. [4]

The modifier alternative and its burden

CMS said it had considered requiring modifiers on claims that were not subject to the program. [4] It said this would add burden on furnishing professionals, including freestanding and hospital-based imaging facilities. [4] They would have to append information even to claims outside the program, so that Medicare could tell which claims were in. [4]

The rule also described the ordering side's problem. [4] Ordering professionals would need to know, at the time of the order, where each imaging service would be furnished and under which payment system it would be paid. [4] Only then could they tell whether consultation was required. [4]

In our reading, the program's data design failed here. The rule's applicability depended on facts that were not known when the data had to be created. The ordering clinician created the consultation record before the setting and payment system were settled. The claims system could not determine applicability after the fact.

The accuracy risk

The 2024 rule added a further concern. [4] It said risks around reporting accuracy were inherent to the program as the statute prescribed it. [4] Those risks fell on furnishing providers, sometimes based on the conduct of ordering professionals with whom they had little or no affiliation. [4]

It said that by appending information supplied by ordering professionals, furnishing providers would be attesting to its credibility and accuracy. [4] It said they might find themselves subject to audits or post-payment review. [4] In our reading, this is the compliance cost of a data hand-off across organizations: liability attaches to the party that submits the data, not the party that created it.

Why a modifier is a data-quality problem

A modifier looks like a small field on a claim. In our reading, in this program it would have carried a legal assertion: that the service was outside the program. [4] The furnishing provider would have to determine that fact from information about the setting and payment system, and record it correctly on each claim. [4]

In our reading, every manual flag of this kind creates three risks. It can be omitted, causing a wrongful denial. It can be applied wrongly, causing an improper payment. And it can be applied inconsistently across billing staff, creating audit exposure. CMS's rule named the added burden on furnishing professionals as one reason not to rely on it. [4]

Returning versus denying

The difference between returning a claim and denying it matters to a revenue cycle team. A returned claim can be corrected and resubmitted. A denied claim leads to appeals or write-offs. CMS said its plan had been to start with returns. [4] It said the statute's wording, that payment may only be made if the claim includes the information, meant returns would eventually have to become denials. [4]

In our reading, CMS was unwilling to move to denials without an automated way to tell which claims the rule covered. That position protected providers from wrongful denials. It also left the program without a working enforcement step.

What CMS had tried first

The 2024 rule recounted the history. [4] It said CMS had solicited public comment on claims reporting in the 2017 rulemaking cycle, and further suggestions in the 2018, 2019 and 2022 cycles. [4] It quoted its own 2017 statement that it had to evaluate options given the vast number of claims affected and the limitations of the legacy claims processing system. [4]

It said that in calendar year 2020 CMS had begun an educational and operations testing period for claims-based reporting, and that the program was still in that phase. [4] It said its intention, announced in the 2022 rule, had been to begin the payment penalty phase by returning rather than denying non-compliant claims. [4] It said the statute required that payment be made only if the claim included the information, so returns would eventually have to become denials. [4]

What remains is decision support without the mandate

The 2024 rule finalized a pause of the program for reevaluation and rescinded the regulations at 42 CFR 414.94. [4] It reserved the section for future use. [4] The current eCFR shows section 414.94 as reserved. [5]

What CMS said about decision support itself

The rule separated the tool from the mandate. [4] It said clinical decision support predates the statute and, given its utility when accessed and used appropriately, CMS expected it to continue to be used to improve decision-making and quality of care. [4]

It said CMS would continue efforts to identify a workable approach and would propose any such approach through later rulemaking. [4] It said that would include implementing any amendments Congress might make to the statutory provisions. [4] It did not propose a timeframe for resuming implementation. [4]

The strongest case against our argument

A fair objection is that the criteria, not the data path, were the real problem. Clinicians found gaps in criteria during the demonstration, and alert fatigue was a known risk. [1] On this view, the program would have struggled even with perfect claims data.

The 2024 rule does not support that reading as the reason for the pause. [4] It identified the claims-based reporting requirement as the greatest challenge and the insurmountable barrier. [4] It said each component had its own challenges, but it singled out reporting. [4] We take CMS at its word on its own reasoning.

A second objection is that the program succeeded in part, since it pushed decision support into ordering workflows. We have no source that measures adoption, and we make no claim either way. A few commenters on the 2024 rule told CMS of the time, money and staff training they had spent implementing consultation. [4] That tells us investment occurred, not what it achieved.

Where the costs fell

The program's costs fell on parties outside CMS. A few commenters on the 2024 rule described the time, effort, money and staff training resources they had spent implementing consultation in their health systems or practices. [4] Several commenters representing EHR vendors, decision support developers and a criteria developer raised strong concerns about a pause. [4] They warned of a chilling effect on innovation in Medicare and across health care, including Medicare Advantage and commercial plans. [4]

CMS acknowledged these concerns and still paused. [4] It said rescinding the regulations and reserving the section was the most appropriate and straightforward option, given the need for clarity on the program's status. [4] In our reading, that weighing reflects the asymmetry of the data problem: the cost of wrongful denials across Medicare claims outweighed the sunk cost of preparation.

A third objection is that the fault lay with the statute's design rather than with the data path, since Congress wrote the claim requirement. [2] We agree that the requirement came from the statute. [2] [4] Our point is that the statute's design was a data design. It specified which data elements had to appear on which claim, created by which party. [2] When CMS said it could not implement that design reliably, it was describing a data problem in its own claims system. [4] The distinction matters for anyone drafting or commenting on a future rule: the data path deserves as much scrutiny as the clinical content.

Lessons for other data hand-offs

The program is closed for now, but its structure recurs. Prior authorization requirements, quality reporting and referral rules can all depend on data one party creates and another submits. In our reading, the AUC program offers four tests to apply to any such rule.

  1. Can the payer tell when the rule applies? If applicability depends on facts the payer's system cannot see, the rule will need manual flags.
  2. Is the data known when it must be created? If the creator cannot know whether the rule applies, it must create the data every time, or risk gaps.
  3. Who attests to the data? If the submitting party attests to another's data, it carries audit risk for conduct it does not control.
  4. What happens on a miss? If a missing element leads to denial rather than correction, the cost of a broken hand-off falls on the furnishing party and the patient.

What to keep from the preparation

Some of the work providers did for this program has value without it. A structured order that carries the clinical indication, the ordering clinician's identifier and supporting data helps the imaging provider protocol the study and helps any later review. Integration of decision support into the ordering screen helps clinicians choose studies, which CMS itself expected to continue. [4]

Other work has no use without the mandate. Claims logic that appended consultation codes, and staff procedures for chasing missing consultation data from referrers, served only the reporting requirement. [2] [4] In our reading, those can be retired or left dormant, with documentation of how they worked in case a revised program returns. [4]

What we would tell a health system today

Keep the decision support tools that clinicians find useful, on their own merits. [4] CMS itself expected decision support to continue. [4] Evaluate them as clinical tools, not as compliance tools for a rule that is reserved. [5]

Keep the interface work that carries structured order information between ordering and furnishing organizations. That plumbing serves referrals, prior authorization and quality reporting, whatever happens to this program.

Before committing to any new payment rule that depends on another party's data, map the chain end to end. Name the system and the person at each step. Ask whether the payer can apply the rule automatically. If the answer is no, plan for the manual work, or for the rule to change.

For organizations that both order and furnish imaging, run the exercise from both sides. As the ordering party, confirm what your systems would send and in what format. As the furnishing party, confirm what you would need to receive from referrers you do not control, and what you would do when it does not arrive. The AUC program showed that the second question is the harder one, and that its answer sits in contracts, interfaces and front-desk procedures as much as in software. [4]

With hindsight

The program the November 2015 rule began did not reach its payment phase. [4] CMS published the next component a year later, in the calendar year 2017 physician fee schedule final rule of November 15, 2016. [3] That rule set requirements for qualified clinical decision support mechanisms and named the first priority clinical areas. [3] It said the first qualified mechanisms would be specified on June 30, 2017 at the earliest. [3] It said CMS expected furnishing professionals to begin reporting January 1, 2018. [3]

According to CMS's 2024 rule, an educational and operations testing period for claims-based reporting began in calendar year 2020. [4] The rule said the program was still in that phase in 2023. [4] It said the 2022 rule had announced an intention to begin the penalty phase by returning, rather than denying, claims without consultation information. [4]

The calendar year 2024 physician fee schedule final rule, published November 16, 2023, finalized a pause of the program for reevaluation. [4] It rescinded the regulations at 42 CFR 414.94 and reserved the section. [4] The current eCFR text, as checked on 2026-11-16, shows section 414.94 as reserved. [5]

The 2024 rule said the real-time claims-based reporting requirement presented an insurmountable barrier to fully operationalizing the program. [4] It said Medicare's claims processing system could not fully automate the distinction between imaging claims that were and were not subject to the reporting requirement. [4] It said full implementation of the penalty phase risked inappropriate claim denials. [4]

The rule said CMS would continue to look for a workable approach and would propose any such approach through later rulemaking, including any changes Congress made to the statute. [4] We found no later rule restoring section 414.94 in the eCFR text we checked. [5]

The 2024 rule anticipated that Congress might amend the statutory provisions, and said CMS would implement any such amendments through rulemaking. [4] We did not review the current text of section 1834(q) for this article and make no claim about whether it has changed since. [2] [4]

What held up from November 2015 was the content side of the program. [1] The process for qualifying provider-led entities ran on the schedule the 2015 rule described; the 2017 rule recorded the first list posted at the end of June 2016. [3] CMS's 2024 rule spoke of decision support as useful when accessed and used appropriately, and expected it to continue in clinical practice. [4]

Section 414.94 had carried the program's definitions and processes from the 2015 rule onward, through amendments in later fee schedule rules, including the 2017 rule's additions on priority clinical areas and decision support mechanisms. [1] [3] In our reading, its removal left the qualification process for provider-led entities and decision support mechanisms without a regulatory home. [4] [5]

What did not hold up was the statute's payment mechanism. [2] [4] It made one professional's payment depend on information created by another, in a claims system that could not tell when the requirement applied. [4] CMS's own account places the failure there. [4]

For health system leaders, the lasting lesson is about data obligations that cross organizational lines. The AUC program's criteria and tools could be built inside one organization. [1] [3] Its reporting could not be applied reliably across many, and that is where it stopped. [4]

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

  • appropriate use criteria
  • clinical decision support
  • advanced imaging
  • PAMA
  • Medicare claims
  • EHR integration