MACRA — Big Fix or Big Problem?

In January 2017, the Medicare Access and CHIP Reauthorization Act (MACRA) took effect, ushering in a new system for physician payment in Medicare. MACRA ended the Sustainable Growth Rate (SGR) method for updating physician fees in Medicare, a permanent “doc fix” relieving Congress of its annual duties to override substantial fee cuts that the SGR would have imposed. In its place, MACRA instituted the Merit-based Incentive Payment System (MIPS), which intends to reward clinicians for providing higher-quality and lower-cost care. Under the MIPS, the performance of providers (individual clinicians or their groups) is assessed on 4 dimensions: quality, clinical practice improvement, advancing care information, and resource use.(1) Poor performers receive negative adjustments to their Part B reimbursement rates, and high performers receive upward adjustments. If the adjustments are implemented as legislated, the MIPS will introduce wide variation in physician payment rates within the next 5 years (Table).

Table.

Anatomy and Pathology of MACRA

DESCRIPTION	PROBLEMATIC FEATURES
The Merit-based Incentive Payment System (MIPS)
Quality domain: Provider performance is assessed based on 6 quality measures. The scoring system for the quality domain encourages reporting on high-priority measures, limits gains from choosing “topped out” measures, and grades clinicians on a curve relative to others reporting on the same measure.	Performance is assessed on only 6 measures—far fewer than the 33 in Medicare ACO models. Providers can choose the measures from a set of about 300 measures, with no restrictions on the domains covered, except one must be an outcome or high-priority measure. Performance is judged against national benchmarks, and minimal adjustments for patients’ clinical and social characteristics have been advanced so far for most measures. Because practices much smaller than ACOs are exposed to the MIPS, patient risk is pooled to a lesser extent, causing greater differences in patient risk factors between providers.
Clinical Practice Improvement Activities (CPIAs) domain: Providers must submit data attesting to 2–4 CPIAs in any of 9 categories: expanded practice access, population management, care coordination, beneficiary engagement, patient safety and practice assessment, participation in an advanced payment model, achieving health equity, integrating behavioral and mental health, and emergency preparedness and response. Different activities carry different weights in the scoring of the CPIA domain.	Providers can choose activities from an approved list of over 90 CPIAs.
Advancing Care Information domain: Providers are assessed on their use of certified electronic health record technology and achievement of various functionalities, including protection of patient health information, electronic prescribing, patient electronic access, coordination of care through patient engagement, health information exchange, and public health and clinical data registry reporting.	Exemption of clinicians practicing in hospital outpatient departments from the Advancing Care Information domain could accelerate price-increasing consolidation between physicians and hospitals.(9,10)
Resource Use domain: For Medicare beneficiaries attributed to providers through a claims-based algorithm, providers are assessed on total Medicare spending per beneficiary and spending per episode for clinical episodes relevant to specific specialties.	Resource Use measures are given a weight of zero in the scoring initially, rising to a maximum of only 30% by 2021. The contribution of total per-beneficiary spending to the composite is further diluted by combining it with other episode-based measures. Thus, reductions in total spending translate into proportionately much smaller bonuses, and providers will have opportunities to obtain bonuses without substantially lowering total spending by focusing on episode-based measures and measures in the other three domains. Performance is judged against national benchmarks, with standard adjustments for patients’ clinical characteristics that may be inadequate. No adjustment for patients’ social characteristics has been advanced thus far.
Bonuses/penalties based on Composite Performance Score: The composite score is calculated as a weighted average of scores in the four domains. The Quality domain receives the greatest weight (50% in 2017). Providers with a composite score below a set threshold (and those failing to meet reporting requirements) receive a negative adjustment to their Part B reimbursement rates 2 years later, with the adjustment growing from −4% in 2019 to −9% in 2022 and beyond. Clinicians with scores above the threshold receive zero or positive adjustments, with higher scores meriting higher adjustments. Because of supplemental bonuses for the top quartile of performers and a budget neutrality provision that inflates base adjustments up to 3-fold if more clinicians are penalized than bonused, the maximum adjustment will be as high as 22% in 2019 and will grow to 37% in 2022 (base adjustment of 9% x 3 + 10% high-performer bonus).	The MIPS threatens to introduce wide variation in physician payment rates within the next 5 years (up to a 46% spread). MIPS bonuses are structured as rate increases rather than lump sum bonuses. Budget neutrality means that low performers cannot bonus even if they are improving. Assuming utilization stays constant, the MIPS effectively transfers payments from low performers to high performers.
MACRA Advanced Alternative Payment Model (APM) Participation Incentive
Clinicians are exempt from the MIPS if they are in an advanced APM, in which case they receive a lump sum bonus equal to 5% of their Part B revenue annually from 2019-2014, followed by favorable rates thereafter. Generally, an APM qualifies as advanced if it involves risk for spending in excess of a financial benchmark (downside risk), as in tracks 2 and 3 of the Medicare Shared Savings Program (MSSP). Because the most popular MSSP track (track 1) imposes no downside risk on ACOs in the first six years of participation, the vast majority of clinicians serving fee-for-service Medicare beneficiaries are exposed to the MIPS. Performance of track 1 MSSP ACOs in the MIPS is based on the 33 ACO program quality measures rather than 6 provider-selected measures.	The 5% bonus may be insufficient to induce ACOs in track 1 of the MSSP to enter an advanced APM because many ACOs are likely to perform well in the MIPS and any difference in bonuses may be insufficient to mitigate the risk of losses from assuming downside risk. The MIPS offers a new alternative that may be quite attractive to many ACOs or prospective ACOs—an opportunity for some to achieve equal or greater bonuses at a lower cost by reporting on far fewer measures and having control over measure selection. For ACOs taking downside risk (advanced APMs), the 5% bonus weakens incentives for ACOs to lower Part B spending because it grows with greater Part B fee-for-service revenue.

MACRA has been applauded by both sides of the political aisle, and the Centers for Medicare and Medicaid Services (CMS) was appropriately commended for its responsiveness to provider concerns in finalizing the rules. Bipartisan support and regulatory flexibility on implementation, however, do not change the fact that the basic design of the MIPS threatens to exacerbate health care disparities without clearly promoting lower spending or better quality.

There are 3 major problems with the MIPS. First, the incentives for clinicians to reduce the provision of services are very weak, and some features create incentives to do more rather than less. Resource use measures, such as total per-beneficiary Medicare spending, are given little weight in the scoring (Table). Thus, providers have ample opportunity to obtain rate increases without substantially lowering Medicare spending. Moreover, because MIPS bonuses are structured as rate increases rather than lump sums unrelated to Part B service volume, they create incentives for providers to deliver more services (more procedures, tests, imaging, office visits, hospitalizations, etc.)—particularly costly services, since a percentage applied to higher prices yields greater revenue increases.

Thus, it is likely that some providers will increase spending in response to higher rates achieved by performing well in domains unrelated to spending. Indeed, the net effect of the MIPS across all providers may be higher Medicare spending, depending on how clinicians trade off labor and leisure in determining service volume in response to rate changes.(2) Budget neutrality in the MIPS constrains average rate increases to be 0% but places no constraint on growth in service volume. In addition, practices will incur substantial costs from the MIPS’ reporting burden,(3) not to mention costs of hiring help to process its sheer complexity. Thus, with conflicting incentives not designed to reduce health care use appreciably, the net effect of the MIPS will almost certainly be cost-increasing.

Second, the design of the MIPS threatens to exacerbate health care disparities. Providers’ performance on quality and resource use measures will be judged against national benchmarks, and minimal adjustments for patients’ clinical and social characteristics have been advanced so far for many measures. The final rule promises some form of risk adjustment eventually, but anything short of a marked departure from standard adjustments in value-based purchasing programs will leave clinician performance unduly influenced by the characteristics of patients served, rather than the quality of care provided.(4,5)

There has been much debate over whether providers should bear the higher costs of performing well on quality measures for higher-risk patients, particularly when the higher risk stems from social factors. No matter one’s view on this debate, risk adjustment in the MIPS that only partially accounts for patient risk differences will effectively transfer resources from providers serving high-risk patients to those serving low-risk patients, whether providers bear the higher costs of improving quality for high-risk patients or incur the penalties from not doing so. Inadequate risk adjustment also establishes incentives for providers to attract lower-risk patients (favorable selection), diverting attention and resources away from improving the health of existing patients.

Third, the incentives for providers to improve quality are weak and easily gamed because providers can choose the few measures used to assess their performance from a broad set (Table). Despite a scoring system that grades clinicians on a curve relative to others reporting on the same measure, the limited number of measures and allowance of choice make it difficult to believe the MIPS will have a meaningful impact on patient outcomes or experiences.

Specifically, because gaming may be cheaper than improving quality, providers will have incentives to select measures on which they already score well relative to others. Providers serving lower-risk patients can choose measures where patient risk matters more, and gaming will be easier for better-resourced providers with more sophisticated information systems and analytic capabilities, potentially exacerbating disparities further. More generally, allowing providers discretion in measure selection establishes incentives to choose measures that minimize the costs of achieving a high score, whereas improvements that yield the greatest population health benefits are likely to be the costliest.

With a repeal of MACRA highly unlikely, what measures could policymakers take to limit the unintended consequences of the MIPS and attain its intended goals? Foremost should be strategies to encourage the continued growth of Medicare Advantage (MA) and alternative payment models (APMs) such as ACO models, which together cover over half of the Medicare population already, are both designed with stronger incentives to lower spending and improve quality and with stronger protections against favorable selection, have demonstrated promising results of late, and could deliver even greater value with further regulatory changes.(6,7)

Although a major goal of MACRA was to drive more providers into APMs, it provides a bonus (5% of Part B revenue) only to providers in advanced APMs, which generally require providers to assume downside risk for spending in excess of financial benchmarks. Providers in advanced APMs are exempt from the MIPS. Relative to expected bonuses under the MIPS, the 5% bonus may be insufficient for providers to assume downside risk, and it establishes no additional incentive to participate in non-advanced APMs (Table). MACRA may even erode participation in non-advanced ACO models—which are exposed to the MIPS and scored on all 33 ACO program quality measures—because the MIPS offers a new alternative affording many ACOs or prospective ACOs the opportunity to achieve equal or greater bonuses at a lower cost by reporting on fewer measures and having control over measure selection.

The creation of a new MSSP track in the final MACRA rule—one involving less downside risk but still qualifying as an advanced APM—was a step forward,(8) but further measures may be needed to expand APM participation substantially. For example, the size of MIPS bonuses could be cut, with the savings distributed as per-beneficiary payments to providers participating in APMs. In addition, advanced payments to support APM participation, as in the ACO Investment Model, could be expanded.

To strengthen incentives to lower spending, the link between MIPS or APM bonuses and the marginal incentives for service provision could be severed by restructuring bonuses as fixed per-beneficiary payments unrelated to the amount of care provided, with MIPS payments adjusted for performance. To mitigate the consequences of inadequate risk adjustment in the MIPS and APMs, these payments could be distributed disproportionately to providers serving high-risk patients, independent of performance, like care management fees in the Comprehensive Primary Care Plus model.

A host of smaller incremental tweaks are likely to be considered, such as adjusting performance for area-level sociodemographics. Fundamental flaws, however, deserve fundamental fixes. Short of getting rid of the MIPS, the best policy may be continuing the watered-down incentives CMS established for the MIPS transition period in response to provider concerns. Instead of Congress overriding the SGR annually, regulators could annually override the MIPS. At the very least, its flaws must be understood in mitigating its damage. If the ACO programs and other APMs have moved Medicare a couple steps forward, MACRA has it teetering on its heels. Whether the new administration and Congress build on Medicare’s pioneering progress in payment reform, or let it stumble backwards, remains to be seen.