Medicare’s 2018 MIPS proposal added ways to pool clinicians for scoring while exempting more of them

On June 20, 2017, CMS proposed virtual groups and facility scores for MIPS while raising the threshold that excludes low-volume clinicians. In our reading, the two moves pulled against each other, and the unit of measurement stayed the program’s unresolved data problem.

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

On June 20, 2017, CMS proposed virtual groups and facility scores for MIPS while raising the threshold that excludes low-volume clinicians. In our reading, the two moves pulled against each other, and the unit of measurement stayed the program’s unresolved data problem.

  1. The week The Quality Payment Program has two tracks.
  2. Who is measured: a higher threshold and a two-step test The low-volume threshold is a gate.
  3. The unit of measurement: individuals, groups, pools and facilities The second part concerns who is scored together.
  4. What is measured: categories, weights and bonuses The third part concerns the content of the score.
  5. What the score buys: adjustments and the data behind them The fourth part concerns consequence.
  6. How the proposal reached different organizations The proposal applied one framework to very different organizations.
  7. Objections, in order It was a reasonable answer to one problem.
  8. What a health system’s quality and data leaders could check The four parts translate into checks any organization reporting clinician quality can run, under MIPS or a successor.
  9. With hindsight CMS finalized much of the June proposal in a rule published November 16, 2017. [3] The final rule adopted the $90,000 and 200-beneficiary low-volume threshold. [3] It kept cost at 10 percent of the 2020 final score rather than zero. [3] It finalized the small practice bonus and raised the complex patient bonus cap to 5 points. [3] It deferred the opt-in.

The week

On June 20, 2017, the Centers for Medicare & Medicaid Services issued a proposed rule for the second year of the Quality Payment Program. [1] The program was created by the Medicare Access and CHIP Reauthorization Act of 2015, known as MACRA. [2] The agency said its goal was to simplify the program, especially for small, independent and rural practices. [1]

The rule carried file code CMS-5522-P. The document was filed for public inspection at the Federal Register at 4:15 p.m. on June 20, 2017. [2] Comments were due by 5 p.m. on August 21, 2017. [2] The press release linked a fact sheet and the public inspection copy. [1]

The Quality Payment Program has two tracks. Eligible clinicians can take part through Advanced Alternative Payment Models or through the Merit-based Incentive Payment System, called MIPS. [2] The proposed rule addressed both. This article concerns its MIPS proposals. [2]

The first change concerned who would take part. In the first year, CMS had excluded clinicians and groups with Medicare Part B allowed charges of $30,000 or less, or 100 or fewer Part B patients. [2] The proposal raised that low-volume threshold to $90,000 or less in allowed charges, or 200 or fewer Part B-enrolled beneficiaries. [2]

CMS estimated the effect from claims data. The higher threshold would exclude about 134,000 additional clinicians from the roughly 700,000 who would have been eligible under the first-year threshold. [2] Almost half of the additionally excluded clinicians were in small practices. About 17 percent practiced in designated rural areas. [2] Under the proposal, CMS estimated about 572,000 eligible clinicians would be required to participate in MIPS in 2018. [2]

The second change concerned how clinicians could be grouped. The rule proposed to implement virtual groups, which MACRA had provided for. [2] A virtual group would combine two or more taxpayer identification numbers, each a solo practitioner or a group of 10 or fewer eligible clinicians. [2] Members did not need to share a location or specialty. Elections for 2018 would be due by December 1, 2017. [2]

The rule also announced facility-based measurement. Clinicians whose primary professional responsibilities were in a hospital could be scored on quality and cost using the facility’s performance in another value-based purchasing program. CMS proposed this from the 2018 performance period. [2] CMS had considered and deferred that option in the first year because of operational issues. [2]

Other proposals eased the scoring. The performance threshold for avoiding a negative adjustment would be 15 points. [2] Small practices would receive a 5-point bonus. Clinicians caring for complex patients would receive a bonus based on their patients’ average risk scores, capped at three points. [2] The cost category would be weighted at zero for the 2020 payment year. [2] Clinicians could keep using electronic health record technology certified to the 2014 Edition in 2018. [2] Small practices would get a new hardship exception from the advancing care information category. [2]

The stakes were set out in the impact analysis. Potential reductions in Part B payment were 5 percent in the 2020 payment year, rising to as high as 9 percent in later years. [2] The release said the proposal would advance the agency’s goals of regulatory relief and program simplification, and that CMS was making it easier for rural and small providers to participate. [1]

Every measurement system has to decide what it is measuring. A quality program for clinicians can score an individual, a practice, a voluntary pool or a hospital. Each choice changes who can be measured reliably and who carries the reporting work. The June 2017 MIPS proposal touched every one of those choices. [2]

Our argument is that the proposal moved Medicare toward measuring clinicians in pools, through virtual groups and facility scores. At the same time, it answered small practices’ burden by exempting more of them. [2] The two moves pulled in different directions. MedPAC later argued the individual clinician was the wrong unit for reliable measurement. [5] CMS’s later value pathways reorganized what clinicians report around specialties and conditions. [6]

We take the proposal in four parts:

  1. Who is measured: the low-volume threshold and how eligibility was determined.
  2. The unit of measurement: individuals, groups, virtual groups and facilities.
  3. What is measured: the categories, their weights and the bonuses.
  4. What the score buys: payment adjustments and the data behind them.

The first part decides the population. The second decides how that population is divided. The third decides the content of each score. The fourth decides the consequence. Together they cover the measurement system from entry to payment. After the four parts, we look at how the proposal reached different organizations, answer objections in order, and list what a health system’s quality and data leaders could check.

Some context helps. MACRA repealed the sustainable growth rate system, which had been meant to limit growth in fee schedule spending, and created MIPS. [5] MedPAC supported the repeal and the push toward alternative payment models. [5] MIPS carried forward elements of earlier Medicare quality and technology programs for clinicians. [5] The 2017 transition year had used a performance threshold of 3 points. [3] The June proposal set the course for the second year, when stakes and expectations would begin to rise. [2]

A note on sources and our reading. We describe the proposal from the document filed for public inspection on June 20, 2017, and CMS’s release that day. [1] [2] We describe what followed from the final rule, the 2018 budget law, MedPAC’s March 2018 report and CMS’s 2026 rule. [3] [4] [5] [6] Statements about what a rule requires of a particular clinician or organization are our reading. Organizations should confirm their obligations with counsel and CMS guidance.

Who is measured: a higher threshold and a two-step test

The low-volume threshold is a gate. Clinicians below it are excluded from MIPS for the year. [2] Raising the gate shrinks the measured population. The proposal raised it from $30,000 to $90,000 in Part B allowed charges, and from 100 to 200 Part B patients. [2]

Who the change removed

CMS’s estimate was specific. The higher threshold would exclude about 134,000 additional clinicians from about 700,000 who would otherwise have been eligible. [2] Almost half of them were in small practices. About 17 percent practiced in designated rural areas. [2]

The change altered the measured population’s mix. Before any exclusions, clinicians in small practices were 35 percent of clinicians billing Part B. [2] Under the first-year rules, they were about 27 percent of MIPS eligible clinicians. Under the proposal, they would be about 22 percent. [2] CMS said the change had little impact on rural clinicians, who made up 15 to 16 percent of the eligible population. [2]

Two tests joined by “or”

The threshold had two tests, joined by “or.” A clinician or group was excluded with allowed charges of $90,000 or less, or with 200 or fewer Part B-enrolled beneficiaries. [2] Meeting either test was enough to be excluded. [2]

In our reading, the “or” had a specific effect. A specialist who saw few Medicare patients but billed large amounts for each could fall below the patient test and be excluded. A primary care clinician who saw many patients for small amounts could fall below the dollar test. Each test caught a different kind of practice.

The final rule kept that structure. [3] The 2018 budget law later changed the basis of both tests to covered professional services, and CMS’s rule for 2027 refers to three low-volume criteria. [4] [6] The gate kept its shape while its inputs changed.

Status as a data product

Eligibility was not a form a clinician filled in. CMS determined it from claims. The determination period ran 24 months in two segments, each a 12-month window of claims. The first year had used a 60-day claims run-out, and the proposal would shorten it to 30 days. [2] The first segment let CMS tell clinicians their status in December, before the performance period. The second could exclude more clinicians during the year, but would not bring back anyone the first had excluded. [2]

For a practice, that meant eligibility arrived as a data feed from Medicare, not a decision made inside the practice. A group near the threshold could not be certain of its status until CMS told it. In our reading, the two-segment design turned eligibility into something to monitor across the year.

Who stayed measured

The clinicians left inside the gate were, by construction, those with more Medicare volume. CMS estimated about 572,000 would be required to participate in 2018. [2] It estimated that the proposed threshold would exclude 585,560 clinicians who did not exceed it, among eligible clinician types who were not newly enrolled. [2]

Those two figures describe a program measuring fewer clinicians than it excluded on volume alone. MedPAC later made a similar observation in its own terms. [5] For a measurement program, that matters. The scores describe the care of higher-volume clinicians, and say little about the rest.

An opt-in left for later

The proposal floated a future option. Clinicians excluded under a higher threshold might be allowed to opt in, for example if they met one threshold test but not the other. [2] CMS proposed this for future years, beginning with 2019. [2]

The final rule did not adopt it for 2019. CMS said it could not yet operationalize an opt-in in a low-burden way. [3] It also acknowledged a commenter’s concern that an opt-in could create a model where only high performers opt in. [3] That concern goes to the heart of measurement. A population that chooses whether to be measured will not be a fair sample.

Burden relief by exclusion

The threshold answered a real complaint. CMS said it had heard from small practices that challenges remained in their ability to participate. [2] Exclusion removes reporting work entirely. It also removes those clinicians from the program’s view.

In our reading, exclusion trades information for relief. Medicare would learn less about the quality of care delivered by small and low-volume clinicians. The clinicians themselves would lose the program’s incentives and its feedback. For a practice that wanted feedback, an exclusion was not a gift.

What this part asked of practices and systems

A health system with employed clinicians would need to track each clinician’s and each billing unit’s status. A clinician could fall below the threshold as an individual and still be measured as part of a group. The system would need claims data at both levels, reconciled to CMS’s determination. [2]

The unit of measurement: individuals, groups, pools and facilities

The second part concerns who is scored together. MIPS already allowed participation as an individual or as a group under one taxpayer identification number. [2] The proposal added a third way, the virtual group, and announced a fourth, facility-based measurement. [2]

Why the unit matters

A quality score is a rate. A rate built on a handful of patients swings with chance. Pooling patients across clinicians steadies the rate. MedPAC made the point in March 2018. It said many clinicians would not be evaluated at all, because as individuals they would lack enough cases for statistically reliable scores. [5] In our reading, every pooling option in the proposal was also a reliability measure, whatever its stated purpose.

Virtual groups

The proposed virtual group combined two or more taxpayer identification numbers. Each had to be a solo practitioner or a group of 10 or fewer eligible clinicians. [2] Members could be in different places and specialties. [2] They would be assessed as a group on all categories. [2]

The design came with logistics. Elections for 2018 were due December 1, 2017, with the process expected to open as early as mid-September. [2] CMS intended to offer technical assistance for the first two years. [2] A virtual group would need a representative, an agreement among members and a way to combine data.

The final rule recorded what commenters said that involved. A majority described virtual group formation as preparing health IT systems, training staff, sharing and aggregating data and coordinating workflows. [3] In our reading, those are the tasks of a small health information exchange, asked of practices chosen for being small.

The data plumbing of a virtual group

A virtual group is a data integration project with a quality report at the end. Members would need to agree on which measures to report, since the group is assessed as one. [2] They would need to combine records from different electronic systems, possibly from different vendors. They would need a way to submit through a registry or other mechanism that could handle data from several taxpayer numbers. [2] [3]

They would also need agreements. Members would decide who controls the combined data, who sees whose results and how disputes are settled. Sharing patient-level data across independent practices raises privacy questions. In our reading, members would want counsel to confirm the legal basis for each data flow before the first submission.

What reliability requires

Reliability is a property of a measure and the number of cases behind it. A rate based on few patients has wide uncertainty. Two clinicians with the same true quality can receive different scores by chance. Pooling more patients narrows that range. MedPAC called the small number of cases for individual clinicians a perennial issue in value-based purchasing for clinician services. [5]

The proposal did not set a minimum number of cases for reliability as a separate test. Its low-volume threshold was framed around burden, not statistics. [2] In our reading, the threshold did some of a reliability rule’s work by accident. It removed some clinicians whose scores would have rested on few cases. It did not ensure that those who remained had enough.

Facility-based measurement

The proposal also announced facility-based measurement. Clinicians whose primary professional responsibilities were in a hospital could have quality and cost scores drawn from the facility’s performance in another value-based purchasing program. [2] CMS had deferred the idea in the first year because of operational considerations. [2] It proposed to start with the 2018 performance period. [2]

The final rule adopted facility-based measurement but pushed its start back to the 2019 performance period. [3] For hospital-based clinicians, that meant their scores would eventually track their hospital’s, a unit far larger than any practice.

Two directions at once

The threshold and the pools pulled in different directions. The threshold removed small, low-volume clinicians from measurement. [2] Virtual groups invited small clinicians to join measurement in pools. [2] A small practice now chose among exclusion, individual reporting, its own group and a virtual group. In our reading, each choice yielded a score built on a different unit.

For Medicare, the result was a set of scores that were hard to compare. MedPAC later made that criticism. It said MIPS scores would not be comparable because each clinician’s score reflected a mix of different, self-chosen measures. [5] Different units compounded the problem.

The tension was understandable. The threshold responded to complaints about burden. The pools responded to the statute, which provided for virtual groups and facility-based options. [2] Each was defensible alone. Together they produced a program in which the measured population shrank while the ways of being measured multiplied. In our reading, that combination made the scores harder to interpret for patients, payers and the clinicians themselves.

One group score for many specialties

Group reporting under one taxpayer number pooled every clinician in the group. [2] For a multispecialty group or a health system, that meant one score covering primary care, surgery and specialties alike. Pooling steadied the rates. It also blurred differences among specialties.

In our reading, group reporting answered reliability by giving up detail. A health system that wanted to manage quality by service line would need its own internal measures, built from the same data, at a finer grain than its MIPS submission.

What this part asked of health systems

A health system had to decide how to report. Reporting as one group pooled all its clinicians. Reporting as individuals let strong performers stand alone. Facility-based scoring, once available, could tie hospital-based clinicians to the hospital’s results. [3] Each choice required data at the right level, and each had financial consequences under the payment adjustments.

What is measured: categories, weights and bonuses

The third part concerns the content of the score. MIPS combined four performance categories: quality, cost, improvement activities and advancing care information. [2] The proposal changed their weights and added bonuses. [2]

Cost set aside

The proposal would weight the cost category at zero for the 2020 payment year. Three categories would then carry the final score. [2] Removing cost meant the score would reflect reported activity, not spending.

Cost measures differ from quality measures in one useful way for this argument. CMS calculates them from administrative claims, so clinicians are not required to submit anything for them. [2] They add no reporting burden. Setting their weight to zero therefore did not ease reporting. In our reading, it removed the one category that measured every eligible clinician on the same claims-based footing, whatever unit or measures they chose.

The final rule declined that proposal. It kept the previously finalized 10 percent weight for cost in the 2020 payment year, with quality at 50 percent. [3] CMS said this would help clinicians understand cost scoring before its weight rose to 30 percent, as the statute then required for the 2021 payment year. [3]

Measures chosen by the measured

The proposal kept a feature of the first year: clinicians largely chose which quality measures to report. CMS proposed letting clinicians submit through as many mechanisms as needed, saying this would let them select the measures most meaningful to them. [2] Choice reduces burden and lets a specialist report relevant measures.

Choice also reduces comparability. MedPAC later wrote that MIPS scores would not be comparable among clinicians, because each score reflected a mix of different, self-chosen measures. [5] In our reading, measure choice and unit choice compounded each other. Two clinicians could differ in unit, measures and submission method at once.

How complete the data had to be

The proposal also kept the data completeness threshold at 50 percent for 2018, rather than raising it as previously planned. [2] For registry, record system and claims submissions, clinicians had to report on at least that share of eligible patients for a measure. [2] [3] CMS proposed to move to 60 percent for the 2021 payment year. [2] A rate built on half of the eligible patients, chosen by the submitter, can differ from the rate for all of them. The final rule declined to keep 50 percent. It retained the 60 percent threshold already set for the 2020 payment year, saying a larger sample reduces the likelihood of selection bias. [3]

Technology requirements eased

The proposal let clinicians keep using record systems certified to the 2014 Edition during 2018. [2] CMS said it had heard from stakeholders that few products were certified to the 2015 Edition, and that rushing the switch could pose risks. [2] It also proposed a hardship exception from the advancing care information category for small practices. [2]

For practices and vendors, the 2014 Edition extension bought time. It also extended the life of older software and the data formats it produced. In our reading, the measurement program here set the pace of technology upgrades, and chose a slower one.

The record system as a scored input

The advancing care information category scored how clinicians used certified record systems. The proposal made minor changes to its objectives and measures. It added an exclusion for the e-prescribing and health information exchange objectives. [2] It also proposed implementing 21st Century Cures Act provisions for hospital-based and ambulatory surgical center-based clinicians. [2]

The comments CMS summarized show the technology burden behind the category. Commenters noted the lack of products certified to the 2015 Edition. Some noted the burden of combining outputs from multiple certified systems. A few said the cost of switching was prohibitive for smaller practices. [2] In our reading, those comments describe a measurement program that depended on software the measured clinicians did not control.

For a health system running several record systems after acquisitions, combining outputs was a real task. Each system produced its own reports. A group submission needed one answer. The data team had to reconcile counts across systems before CMS ever saw them.

Bonuses as adjustments

The proposal added two bonuses. Small practices would receive 5 points. [2] CMS said that amount was one-third of the points needed to meet the 15-point threshold. [2] Clinicians caring for complex patients would receive a bonus equal to their patients’ average risk score, capped at three points. [2]

The final rule kept the small practice bonus and raised the complex patient bonus cap to 5 points. [3] It based that bonus on the dual eligibility ratio and average risk score. [3] CMS described the complex patient bonus as a short-term strategy while it worked on social risk. [3]

A bonus added to a final score is an adjustment, not a measurement. In our reading, it signaled that CMS recognized differences in practice setting that the measures themselves did not capture. It addressed those differences outside the measures.

The 15-point threshold

The performance threshold for 2018 was 15 points. [2] With the small practice bonus, CMS said a small practice could reach it by reporting two quality measures, or one quality measure and one improvement activity. [2] The threshold set the line between a negative adjustment and a neutral or positive one. A low threshold made avoiding a penalty easy. It also meant scores near the top would separate clinicians only weakly.

Improvement scoring

The proposal added improvement scoring, which MACRA required but the first year had not used. [2] For quality, CMS proposed comparing overall achievement with the prior year to calculate an improvement percent score. [2] Improvement scoring rewards change rather than level. It needs a prior-year score built on comparable data. A clinician who changed groups, or a group that changed its reporting choice, would complicate that comparison.

What the score buys: adjustments and the data behind them

The fourth part concerns consequence. MIPS adjusts Part B payments up or down based on the final score, two years after the performance year. [2] The proposal’s impact analysis described the stakes and the data CMS used to model them. [2]

The size of the adjustment

Potential reductions were 5 percent of Part B payments in the 2020 payment year, rising to as high as 9 percent in later years. [2] CMS noted that, on average, clinicians’ Medicare billings were about 23 percent of their revenue. [2] In our reading, that made the adjustment smaller in practice revenue than the headline percentages suggest.

Budget neutrality and the scaling factor

The adjustments were designed to balance. CMS modeled an exchange function meeting statutory requirements for a linear sliding scale, budget neutrality and an exceptional performance pool. [2] CMS has since explained that the scaling factor for positive adjustments depends on the total of negative ones. [6] When few clinicians score below the threshold, there is little money to redistribute.

A low threshold and generous bonuses mean few negative adjustments. In our reading, that combination pointed toward small positive adjustments for those above the threshold. Practices that invested heavily in reporting might see little return in 2020.

Two years from care to payment

The proposal described a performance period two years before the payment year. Care in 2019, for example, would set payments in 2021. [2] That lag lets CMS collect claims, receive submissions and calculate scores. It also separates the reward from the behavior it is meant to shape.

For a practice, the lag complicates planning. Staff changes, mergers and new systems can all happen between the year measured and the year paid. A group that changed its composition would be paid on the record of a group that no longer existed in the same form.

Data limits in the model

CMS was candid about its data. It assumed paid amounts were 80 percent of Part B allowed charges because of data limitations. [2] It assessed rural status by individual taxpayer and clinician identifiers and did not model a group definition of rural practices. [2] It used average risk scores from a 2015 public use file to model the complex patient bonus. [2]

Those assumptions are reasonable for a model. They also show that the program’s architects worked with imperfect data about the clinicians they measured. The units of analysis in the model did not always match the units of measurement in the program.

Small adjustments, real costs

The proposal’s design produced modest stakes for the 2020 payment year: a low threshold, generous bonuses and adjustments funded by penalties. [2] The reporting costs were not modest. MedPAC later cited CMS’s estimate of more than $1.3 billion in reporting burden in the first year. [5] MedPAC also argued that MIPS scores would be very high for most clinicians, limiting CMS’s ability to tell performance apart. [5]

In our reading, that imbalance shaped behavior. A practice that expects a small adjustment invests just enough to avoid the penalty. It reports the measures it does well on. That is rational for the practice and weak as measurement.

Feedback to clinicians

The proposal promised performance feedback, annually at first and more frequently in later years. [2] Feedback is the part of a measurement program that can change care. It depends on timely, reliable data. A score delivered long after the year it measures is hard to act on.

For a health system, the program’s feedback is one input among several. Its own quality data, refreshed monthly or weekly, can show problems sooner. The MIPS feedback then serves as a check: does the system’s internal calculation match CMS’s? Where the two differ, one possible cause is a data mapping problem, such as a clinician attributed to the wrong billing unit. Finding those mismatches early protects both the score and the payment.

How the proposal reached different organizations

The proposal applied one framework to very different organizations. The cases below are our reading of how its parts would fall, not findings about any organization.

Solo and small practices

A small practice under the new threshold was simply out. [2] One above it could report alone with a 5-point bonus, join a virtual group, or seek a hardship exception from the advancing care information category. [2] Each path had different data demands. A virtual group needed data sharing with practices the clinician might never have worked with. [3]

Health systems with employed clinicians

A health system may bill under one or a few taxpayer identification numbers. It would weigh group reporting against the prospect of facility-based scores for hospital-based clinicians. [2] [3] Its quality, revenue cycle and IT teams would share the work. Its data warehouse would need to reproduce CMS’s eligibility determinations and score estimates.

The system would also have to manage change. Clinicians join and leave during a year. Practices are acquired and moved onto new billing numbers. Each change can alter which unit a clinician is measured in and which claims count toward the threshold. [2] A health system that acquired practices in 2017 would want to model their status under both the old and the new billing arrangements.

Specialists with few, costly patients

The patient-count test could exclude specialists who billed large amounts for a small number of Medicare patients. [2] For such a specialist, exclusion removed the reporting work and the adjustment. Within a health system, that specialist’s data might still flow into a group submission if the group reported as a whole. [2] The system would need to know which applied.

Rural clinicians

CMS found the threshold change had little impact on rural clinicians’ share of the eligible population. [2] Rural practices that stayed eligible faced the same reporting tasks as others. The small practice bonus and virtual groups were their main relief. [2]

Practices in alternative payment models

Clinicians who reached qualifying status in Advanced APMs were excluded from MIPS and could earn a 5 percent lump sum. [2] CMS estimated those incentive payments at roughly $590 million to $800 million for the 2020 payment year. [2] For them, the MIPS changes mattered less than the APM rules.

Health IT vendors and registries

Vendors faced the 2014 Edition extension and new reporting options. [2] Registries and data submission vendors would need to support virtual groups, which pooled data across organizations. [2] In our reading, the proposal pushed integration work onto the technology market serving small practices.

Objections, in order

1. Exempting small practices was a reasonable answer to burden

It was a reasonable answer to one problem. CMS had heard that small practices struggled to participate. [2] MedPAC later cited CMS’s estimate of a reporting burden over $1.3 billion in the first year. [5] Our argument concerns the cost of the answer: an exempt clinician is unmeasured.

2. Virtual groups solved the reliability problem

Virtual groups could have helped. They were voluntary, and forming one required data sharing and coordination that commenters described as substantial. [3] MedPAC’s later design kept voluntary grouping but changed what groups were measured on, toward population-based measures. [5]

3. The opt-in would have restored the lost population

CMS itself raised the problem with that remedy. It said an opt-in could create a model where only high performers opt in. [3] A measured population that selects itself is not representative.

4. Bonuses distort scores

Bonuses do change scores. CMS framed the complex patient bonus as short-term while it studied social risk. [3] Our point is narrower. Bonuses show that the measures alone did not capture setting and patient mix. Pooling and risk adjustment address that inside the measures.

5. The 2018 rules were transitional, so their design did not matter

Transitional choices shaped later years. The 2018 budget law extended transition rules on cost weighting and thresholds through the fifth year. [4] The low-volume test persisted into CMS’s rule for 2027, published in November 2026, which uses three low-volume criteria. [6]

6. Facility-based scoring hides individual performance

It does, by design. A clinician scored on a hospital’s results is credited or penalized for work done by many others. [2] MedPAC’s view was that outcomes are determined by the combined efforts of many providers, so pooled measures fit them better. [5] Whether that is a flaw depends on what the score is meant to reward.

7. Medicare should measure outcomes, and the unit is secondary

Outcome measures are the stronger goal. In our reading, they also need more patients per unit to be reliable, because adverse outcomes are less frequent events than the processes of care. MedPAC paired its call for population-based measures with a call for voluntary groups large enough to support them. [5] In our reading, outcome measurement makes the unit question more pressing, not less.

What a health system’s quality and data leaders could check

The four parts translate into checks any organization reporting clinician quality can run, under MIPS or a successor.

  1. Can the organization reproduce CMS’s eligibility determination for each clinician and billing unit from its own claims data?
  2. For each reporting unit, how many patients sit behind each quality rate, and is that enough for a stable score?
  3. Which reporting choice, individual, group or facility, does each clinician fall under, and who decided?
  4. Are prior-year scores comparable, given changes in group membership or reporting choice?
  5. How long after the performance year does feedback arrive, and who acts on it?

The second check carries our argument. A measure built on too few patients tells a clinician little. Every unit decision is, underneath, a decision about how many patients a score rests on. [5] The same logic applies to internal dashboards and payer contracts. A health system that reports a clinician’s rate to a board or a health plan owes the reader the number of patients behind it.

With hindsight

CMS finalized much of the June proposal in a rule published November 16, 2017. [3] The final rule adopted the $90,000 and 200-beneficiary low-volume threshold. [3] It kept cost at 10 percent of the 2020 final score rather than zero. [3] It finalized the small practice bonus and raised the complex patient bonus cap to 5 points. [3] It deferred the opt-in. It adopted facility-based scoring but delayed it to the 2019 performance period. [3]

Congress then amended the program. The Bipartisan Budget Act of 2018 was approved February 9, 2018. [4] Section 51003 based the low-volume tests on covered professional services, for performance periods beginning on or after January 1, 2018. [4] It let the cost category count for 10 to 30 percent of the score for the second through fifth years. [4] It required gradual increases in the performance threshold for the third, fourth and fifth years. [4]

MedPAC went further. In its March 2018 report, the commission recommended that Congress eliminate MIPS. Commissioners voted 14 to 2. [5] It recommended a voluntary value program in which clinicians could be measured as part of voluntary groups on population-based measures. [5] It argued that individual clinicians would lack enough cases for reliable scores. It noted CMS’s estimate that over half of clinicians would be exempt. [5]

MedPAC sketched what it wanted instead. Clinicians would elect to be measured as part of a voluntary group, and groups would qualify for a value payment based on performance on a set of population-based measures. [5] It suggested the program could be phased in, and said the flaws of MIPS should not be replicated. [5] Its unit was the group, and its measures were about populations rather than self-chosen activities.

MedPAC also said CMS had used the low-volume threshold to exempt a significant share of clinicians. It noted that more Medicare-billing clinicians were exempted in the second year than the first. [5] The commission described MIPS as exempting more clinicians than would participate. [5]

Congress did not eliminate MIPS. By November 2026, CMS’s final rule for 2027 estimated 586,925 MIPS eligible clinicians, with about $51.70 billion in allowed charges, in the 2027 performance period. [6] The rule used a three-criterion low-volume test and kept the performance threshold at 75 points, as finalized for 2026. [6]

CMS moved the unit of measurement in a different direction from MedPAC’s. Its MIPS Value Pathways organize measures and activities around a specialty or condition. [6] The 2027 rule finalized, with a two-year delay, a policy requiring clinicians not reporting the APM pathway to report a selected value pathway from the 2031 performance period. [6] Traditional MIPS reporting remains available through the 2030 performance period. [6]

What held up from June 2017? The threshold endured, and the 2018 law changed its basis in statute. [3] [4] [6] Virtual groups and facility scores remained options, not the core. The 2027 rule finalized including virtual groups in value pathway reporting from the 2029 performance period. [6] MIPS itself outlived MedPAC’s recommendation to eliminate it. [5] [6] CMS’s answer, by our reading, was to standardize what clinicians report rather than to pool whom they are measured with. [6]

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

  • MIPS
  • Quality Payment Program
  • MACRA
  • quality measurement
  • virtual groups