CABG indicates coronary artery bypass graft surgery; PCP, personal care physician. This is an example of service assignment for coronary artery bypass surgery episode-based cost measures with triangles representing individual services that occur over time. The episode is initiated by coronary artery bypass surgery (“trigger service”) and attributed to the cardiac surgeon. Specific services related to the surgery are included pretrigger and posttrigger. Triangles above the line represent costs incurred by the attributed clinician, and triangles below the line represent costs incurred by other clinicians. The shaded triangles are assigned services, which are potentially related to the coronary artery bypass surgery, and can be performed by either the attributed cardiac surgeon or by other clinicians. The unshaded triangles, such as a podiatry evaluation, represent unrelated costs, which are not counted in the episode. The entire period encompassing all services assigned to the measure is the episode window.
A condition such as COPD can involve long-term treatment, disease exacerbations, and procedures. Each of these may lead to unique episode-based cost measures that identify different clinicians triggered by separate events. Long-term treatment of COPD may be attributed to the outpatient pulmonologist treating the patient, whereas a hospitalization can initiate a concurrent acute episode attributed to the hospitalist. Finally, a later lung reduction episode can lead to a surgical episode attributed to the surgeon. Some costs can be included in multiple measures; the cost of the COPD admission would be attributed to both the pulmonologist in long-term treatment measures and the hospitalist with the acute exacerbation measure. Other costs unrelated to COPD, such as cataract surgery, may not be assigned to any measure.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Duseja R, Andress J, Sandhu AT, et al. Development of Episode-Based Cost Measures for the US Medicare Merit-based Incentive Payment System. JAMA Health Forum. 2021;2(5):e210451. doi:10.1001/jamahealthforum.2021.0451
The Merit-based Incentive Payment System (MIPS), established as part of the Quality Payment Program, is a Medicare value-based payment program that evaluates clinicians' performance across 4 categories: quality, cost, promoting interoperability, and improvement activities. The cost category includes novel episode-based measures designed for targeted evaluation of the resource use of specific conditions. This report describes the development of episode-based cost measures and their role in the shift from volume-based to value-based purchasing.
Episode-based cost measures focus on resource use related to the treatment of a specific condition or procedure. The measures exclude health care costs unrelated to the condition or procedure of focus. The episode-based cost measures provide a nuanced examination of resource use that can be used alongside quality metrics to identify opportunities to improve the value by capturing costs that are clinically related to the care being delivered within a given patient-clinician relationship of care delivered to patients. These measures were developed with the input of clinical committees composed of over 320 clinicians from 127 specialty societies and stakeholder organizations. The MIPS program currently evaluates clinician cost category performance based on 2 population-based cost measures (Medicare spending per beneficiary and total per capita costs) in addition to 18 episode-based cost measures. Additional episode-based cost measures are currently under development.
Conclusions and Relevance
The transition to value-based payment requires an accurate assessment of clinician effect on health care quality and cost. The use of episode-based cost measures to assess clinician influence on health care costs for high-priority conditions and procedures is an important step. The Centers for Medicare & Medicaid Services is introducing MIPS Value Pathways that will align episode-based cost measures with related quality measures to further incentivize the transition from fee-for-service to value-based care.
In 2015, the US Congress enacted the Medicare Access and Children's Health Insurance Program Reauthorization Act (MACRA), which introduced the Quality Payment Program (QPP) into Medicare.1 This program rewards value and outcomes along 2 pathways: Advanced Alternative Payment Models (APMs) or the Merit-based Incentive Payment System (MIPS). Advanced APMs require organizations to take on substantial financial risk or participate in a medical home model organized under Centers for Medicare & Medicaid Services (CMS) Innovation Center authority. They must use certified electronic health record (EHR) technology and receive payment for covered professional services based on quality measures. The MIPS, by contrast, scores individual clinicians based on 4 performance categories: quality, cost, interoperability, and improvement activities. These categories determine a MIPS final score used to redistribute Medicare payments, with positive adjustments awarded to high-performing clinicians and negative adjustments withheld from poor performers. In 2019, 538 323 clinicians participated in MIPS as individuals or groups.2
Commenters have expressed concerns about using cost measures to evaluate clinician performance.3-8 First, since MIPS rewards lower costs, clinicians may face an incentive to choose healthier patients. Second, unless CMS balances these cost measures with aligned quality measures, clinicians may face perverse incentives to reduce costs by reducing the quality of care. Third, cost measures potentially penalize clinicians for costs and outcomes outside of their control. Fourth, unless clinicians receive actionable data to change their practice, the policy may not reduce inefficient care. This report describes the framework of MIPS cost measures, focusing on design features of episode-based cost measures (EBCMs) that address these problems.
To assess value, one needs to determine both quality and costs of care. If 2 clinicians treat similar patients differently for a given condition and have similar outcomes, the clinician treating the patient with lower spending is by definition delivering higher-value care. Poor care coordination, delivery of unnecessary services, and high complication rates all increase costs and can be signals of low-quality care.
Most MIPS-eligible clinicians with a cost performance category score are scored based on 2 population-based cost measures: Medicare Spending per Beneficiary (MSPB) and Total Per Capita Cost (TPCC).9,10 The MSPB evaluates clinicians providing inpatient care. The TPCC evaluates clinicians providing primary outpatient treatment on the total annual cost of care. These measures evaluate health care efficiency across a broad range of clinical conditions and thereby encourage shared accountability among clinicians. To complement the population-based measures, MACRA mandated the development of novel episode-based cost measures for MIPS. In 2021, an average of MSPB, TPCC, and 18 EBCM scores determine clinicians’ MIPS cost performance scores.
The effect of the MIPS and EBCMs on the cost of care remains to be seen because we are early in the MIPS experience. Penalties and rewards continue to rise, and the cost category has not yet reached its legislatively mandated 30% of the performance score. Bundled payment models may provide examples of the potential effect of evaluating episodes of care costs. Multiple episode-based bundled payment models have demonstrated a reduction in spending.11-14 However, this reduction in spending has not been universal,15,16 and there are essential differences between episode of care costs in bundled payment models and MIPS EBCMs. Motivated by CMS's effort to respond to the challenges described above, EBCMs only include costs related to the condition under evaluation and attribute to multiple clinicians involved in this care, promoting collaborative care.
Multiple stakeholder groups participate in EBCM development. First, a technical expert panel provides overarching input on the development process, measure calculation methods, and approaches to other challenges during development. It comprises experts in clinical care, payment policy, and performance measurement from specialty societies, academia, health care administration, and patient and caregiver representatives.
Second, clinical subcommitees provide detailed input on specific measures. Clinical subcommittees focus on clinical areas (eg, gastrointestinal disease). Members are nominated by a professional society or are self nominated. A broad membership captures diversity in geography, clinical specialty, training background (eg, advanced practice clinicians, nurses, physicians, and therapists), and practice type (eg, academic, nonacademic, urban, and rural). For example, the cardiovascular disease subcommittee included 39 clinicians across 17 subspecialties. Subcommittees select measures for development. In the first wave of measure development (May 2017-January 2018), the subcommittee provided input regarding all measure specifications.
In the second and third waves of development (April 2018-December 2018 and April 2019-December 2020, respectively), a measure-specific workgroup recommended detailed measure features. The subcommittee’s recommendations for specialties and relevant experience inform the workgroup composition. The workgroup is limited to approximately 15 clinicians, typically subcommittee members, including multiple clinicians who will have the episode attributed to them. The workgroup develops and iteratively refines measure specifications, carefully considering empirical analyses and stakeholder comments.
The Centers for Medicare & Medicaid Services field tests preliminary EBCM specifications as 1 of multiple avenues of stakeholder engagement prior to proposed regulation, before an EBCM is finalized for use in the MIPS program. Field testing includes posting preliminary measure specifications and providing measure scores to every clinician and clinician group that meets a minimum episode volume, usually 10 episodes. The Centers for Medicare & Medicaid Services also publishes comprehensive testing results, including analyses of reliability, validity, and potential effect on clinicians, patients, and cost.17
The Centers for Medicare & Medicaid Services has developed 24 EBCMs through this process (Table 1), involving around 400 clinicians from 151 specialty societies and stakeholder organizations. Overall, 18 EBCMs were included in the MIPS 2021 performance period. The Centers for Medicare & Medicaid Services may use the remaining measures in a future year. For the 18 EBCMs in MIPS, Table 2 provides descriptive statistics of measure coverage. Table 3 lists measure performance statistics, including the mean reliability. Although reliability evaluates measure-level precision, each individual measure contributes additional data to improve the overall precision of the cost category score, often comprised of multiple cost measures, which affects MIPS payment adjustment.
The EBCMs represent a clinician's contribution to Medicare expenditures related to the treatment of a specific condition for a patient (the episode group). The EBCMs attribute costs (defined as the expenditures by Medicare and patients for a given service) to clinicians responsible for treating the condition or performing the procedure. Our process involved selecting an EBCM for development; defining rules for attributing an instance of an episode to a clinician; defining rules for which services to assign to the episode; focusing the episode definition on a clinically homogenous patient population; and aligning the episode with existing quality measures.
An episode group focused on a specific clinical scenario requiring treatment. Episodes span a wide range of clinician-patient relationships, including acute inpatient conditions (eg, pneumonia hospitalization), chronic diseases (eg, diabetes management), and procedures (eg, hip replacement surgery). Criteria for selecting conditions for measure development include the share of Medicare spending, the number of clinicians responsible, the opportunity for improvement, the existence of concurrent quality measures, and the ability to construct a homogenous patient cohort.
The MIPS attributes EBCMs to clinicians based on their role in patient treatment. Attribution aims to be transparent and straightforward. If attribution rules are clear, clinicians can anticipate in real-time when CMS will attribute an episode to them to evaluate their resource use and improve their performance both immediately and in future practice.
Attribution rules vary by episode type. For acute inpatient episodes, CMS attributes the episode to clinician groups responsible for at least 30% of evaluation and management (E&M) codes billed during an admission. The E&M services that trigger attribution represent important clinical encounters that influence the course of care for acute medical admissions. For procedural episodes, CMS attributes the episode to clinicians billing for the procedure. For chronic condition episodes, CMS attributes the episode to clinician groups billing a pair of services indicating a clinician-patient relationship to treat or manage an ongoing condition.
A wide variety of clinicians (physicians, nurse practitioners, physician assistants, and therapists) with overlapping but distinct responsibilities deliver health care services. The MIPS cost measures encourage team-based care by permitting overlapping episodes defined around each clinician's role. For example, MIPS might include separate, parallel measures for surgeons and anesthesiologists that evaluate similar surgeries but hold surgeons and anesthesiologists responsible for different but overlapping sets of costs.
Each EBCM assigns specific services to the episode—ie, includes the cost of those services—if they are related to the care delivered by the attributed clinician; unrelated services are not assigned. Identifying related costs requires extensive input from clinicians with expertise in the condition and the clinician-patient relationship being measured.
Services eligible for assignment in an episode include all Medicare Part A and B spending, including diagnostic tests, treatments, rehabilitation, and durable medical equipment. To avoid penalizing clinicians working in higher-priced settings, EBCMs employ standardized payment amounts to adjust for cost differences due to geographic wage differences or policy-based payment adjustments such as payments for teaching hospitals. Methods to standardize and include Part D costs have been tested and may be implemented in the future.18,19
The EBCMs include costs directly related to treatment and secondary costs, such as relevant complications or readmissions. Many complications cannot be entirely prevented and are an expected component of the cost of care. However, the frequency of complications varies across clinicians.20,21 These measures are based on the principle that best practices may reduce most complications’ frequency or severity. Including the costs of these complications provides incentives to improve care quality and minimize such complications.
The EBCMs include related services whether delivered by the attributed clinician or another clinician. For example, the outpatient percutaneous coronary intervention (PCI) measure assigns costs from a repeated PCI 1 week later regardless of who performs it, which provides incentives for care coordination across different clinicians treating the same patient. Figure 1 displays an example of an EBCM definition.
Service assignment decisions specify a given window around a “trigger date” during which costs are counted. In the example coronary artery bypass graft episode in Figure 1, the cost of a pneumonia readmission is assigned within a short postsurgical duration, whereas the cost of wound care is included for a more extended period. Clinical expertise, empirical analyses of the frequency and costs of services over time, and patient and caregiver perspectives form the basis for service assignment decisions.9 We have demonstrated the service assignment process leads to reclassification of clinician performance and improves measure reliability (signal-to-noise ratio) by reducing noise while maintaining signal.22
A fair evaluation needs to compare decisions for similar patients. The EBCMs approach this problem by (1) limiting comparisons across clinicians who provided the same trigger services; (2) defining patient cohorts in clinically homogenous ways; (3) employing custom risk-adjustment; and (4) not assigning clinically unrelated costs.
The first method compares clinicians who perform the same services against each other to calculate scores. Because clinicians attributed an EBCM all filed the same trigger codes (combinations of service claims and diagnosis codes that initiate the episode), they presumably perform a similar role in patient treatment.
The second method uses patient exclusions to ensure patients are clinically similar. For example, the screening/surveillance colonoscopy measure excludes inpatient colonoscopies because these procedures are not typically performed for routine screening. Small patient subgroups with highly varied costs are also excluded, such as heart transplant patients in the elective PCI measure.
Third, each episode group has its own risk adjustment formula, designed to control comorbidity differences. The base model is the CMS Hierarchical Condition Category Model, which effectively predicts health care costs in the Medicare Advantage population.23 Clinical experts customize each model with additional variables that influence expected costs (average of 12 additional variables [range, 3-26] per measure). Also, some cost measures are stratified into mutually exclusive subgroups, each with a distinct risk adjustment model. For example, the stroke cost measure has subgroups for cerebral infarction and intracranial hemorrhage, so patients are only compared within each subgroup.
Finally, limiting measure costs to related services improves the clinical comparability of episodes. If a patient has a comorbidity that leads to unrelated services and increased costs, the measures will exclude these services.
Health care often involves teams of clinicians with overlapping responsibilities. A complex surgery may require multiple surgeons, whereas an acute medical admission may include treatment decisions by many clinicians. In MIPS, the EBCM costs are assigned to each clinician attributed the episode. There is no double-counting concern because, for any episode attributed to a particular clinician, a service or procedure cost is counted exactly once.
Attributing costs to all clinicians on the team aligns the incentives of the multiple clinicians driving care. Because health care is team-based, the entire team needs to be engaged in evaluation and improvement. The EBCMs that include different phases of care (eg, acute flare or long-term treatment), different care responsibilities (eg, medical treatment, surgical treatment, and rehabilitation), and related comorbidities can promote improved care coordination and higher-value care. Figure 2 presents a hypothetical example of overlapping EBCMs related to the treatment of chronic obstructive pulmonary disease (COPD). The possibility of overlapping episodes with relevant costs potentially included in multiple measures allows for a fair evaluation of resource use regardless of the number of clinicians involved. Each clinician is attributed costs related only to the care they are delivering and is compared only with clinicians providing similar care. This approach leads to more accurate comparisons than assigning the costs to a single responsible clinician, which would ignore differences in care systems across the country and the team-based delivery of health care.
For EBCMs to effectively provide incentives for efficient care, concurrent evaluations of the cost and quality of care are necessary. Cost measures intrinsically capture quality when high-quality care reduces cost. For example, clinicians with lower rates of rehospitalizations and complications tend to have lower costs.24-29 In some cases, though, cost and quality measures may provide opposing incentives. For example, higher-cost treatments may lead to improved outcomes that manifest beyond the horizon of an EBCM or improve quality of life without lowering medical costs.
In these cases, cost measures and quality measures must work together to capture the value of care and provide proper incentives. Quality and cost measures must be aligned to ensure that CMS rewards high-value care. An important criterion for selecting EBCMs for development is the availability of appropriate quality measures. Quality measures should capture the dimensions of care in which lower cost is not associated with improved quality. Suppose a higher-cost therapy is associated with improved functional status. In that case, a cost measure and an aligned functional status quality measure together reward clinicians for efficiently improving patients’ quality of life. Without such alignment, clinicians could improve their MIPS score by reducing costs at the expense of quality.
There are limitations to EBCMs meeting the challenges outlined at the start of this report.
First, additional clinician education regarding EBCMs is critical. Although EBCMs have been rigorously tested for their ability to account for patient complexity, the perception that cost measures have inadequate risk adjustment could inadvertently provide an incentive to selectively choose healthier patients for treatment. This concern is often related to gaps in claims data compared with more granular clinical data. It should be addressed with future research evaluating EBCM performance with linkage to other sources of data (such as clinical registries). However, we believe the measure-specific approach to risk model development and, more importantly, the counting of only clinically related costs through service assignment improves risk adjustment.
Second, there are a large number of measures in the MIPS program for clinicians and health systems to track, in addition to existing APMs. The MACRA mandated that EBCMs be constructed using Medicare claims data, which facilitates resource use measurement without additional reporting burden.1 Aligning measures between EBCMs and related APMs could help clinicians transition to such APMs. In addition, some EBCM concepts described here, such as our service assignment use, may be applied to future APM design.
Third, practice change requires timely reporting of measure performance. Measure attribution and performance scores are not delivered in real time. Currently, feedback reports are sent roughly 6 months after the end of the performance period. The more actionable clinicians find these reports, the higher the likelihood that these measures will improve performance.
Fourth, these EBCMs use a risk adjustment formula that does not universally include social risk variables. Without adjustment, clinicians may be penalized for cost differences from socioeconomic characteristics outside of a clinician's control. However, adjusting for such variables poses a problem if differences are owing to disparities in quality of care. In that case, social risk adjustment could have the unintended consequence of perpetuating inferior standards of care quality for disadvantaged populations.30 The EBCM testing indicates that including social risk variables in the risk adjustment models had a limited effect on clinician performance.31 Further evaluation of social risk variables across Medicare's value-based payment programs have been addressed in the Assistant Secretary for Planning and Evaluation reports.30,32
Fifth, mitigating the risk-of-care stinting and appropriately measuring care value requires EBCMs be balanced with quality measures that address meaningful performance gaps. Although there are more than 200 quality measures in MIPS (without Qualified Clinical Data Registry measures), only 41 assess outcomes or intermediate outcomes. The Centers for Medicare & Medicaid Services' Meaningful Measures Framework has focused greater attention on outcome measures, which could expand the inventory with which to assess quality outcomes. The Centers for Medicare & Medicaid Services has also developed a new framework for aligning cost and quality through the MIPS Value Pathways (MVPs), which we discuss in detail.
Sixth, as the number of EBCMs in MIPS increases, ongoing measure maintenance will be important to ensure specifications remain appropriate and guard against unintended consequences. This will require coding updates and harmonization with other cost or quality measures to improve alignment and the close monitoring of practice patterns and coding to identify potential gaming that may require measure respecification.
Although the first 2 development waves focused on frequent, high-cost procedures and acute conditions, wave 3 and beyond must expand to different types of care and diverse clinician roles. Chronic conditions pose particular challenges to account for complex disease interactions, changes in severity over long durations, and changes in a clinician’s role in disease treatment over time.
The MIPS was initially designed to maximize flexibility to accommodate the diverse range of clinicians in the program, including allowing clinicians to select from a large set of potential quality measures. Although this flexible approach promoted high MIPS participation, clinicians may be evaluated on the cost of treating a condition without concurrently being evaluated on the quality of that care. Clinicians providing similar types of care also may not be evaluated on the same measures and activities.
The Centers for Medicare & Medicaid Services has responded to these concerns, along with those regarding the complexity of MIPS, by developing MVPs. This new framework identifies an aligned set of specific cost measures, quality measures, and improvement activities that are important to the clinical practice of the clinician being evaluated, along with a foundational layer of measures that applies across all MVPs. This layer will include promoting interoperability measures (certified EHR technology use), and administrative claims-based quality measures focused on population health. In addition to better assessing the value of care, the MVP design better aligns with APMs to facilitate clinicians transitioning to such programs.
Overall, MIPS represents a transition away from pure fee-for-service medicine to value-based payment with incentives to provide coordinated, efficient, and high-quality care in the existing payment framework. The EBCMs are an essential tool to evaluate clinicians’ provision of high-value care.
Accepted for Publication: March 18, 2021.
Published: May 14, 2021. doi:10.1001/jamahealthforum.2021.0451
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Duseja R et al. JAMA Health Forum.
Corresponding Author: Reena Duseja, MD, Quality Measurement and Value Based Incentives Group, Centers for Medicare & Medicaid Services, 7500 Security Blvd, Office S3-10-03, Baltimore, MD 21244 (firstname.lastname@example.org).
Author Contributions: Dr Sandhu and Ms Lam had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Duseja, Andress, Sandhu, Bhattacharya, Lam, Nagavarapu, Nilasena, Choradia, Do, Bounds, Leoung, Swygard, MaCurdy.
Acquisition, analysis, or interpretation of data: Duseja, Sandhu, Bhattacharya, Lam, Nagavarapu, Choradia, Do, Feinberg, Bounds, Leoung, Luo, Swygard, Uwilingiyimana, MaCurdy.
Drafting of the manuscript: Duseja, Sandhu, Bhattacharya, Lam.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Bhattacharya, Lam, Choradia, Bounds, Leoung, Luo, MaCurdy.
Obtained funding: Bhattacharya, MaCurdy.
Administrative, technical, or material support: Duseja, Andress, Sandhu, Bhattacharya, Lam, Choradia, Do, Leoung, Luo, Swygard, Uwilingiyimana, MaCurdy.
Supervision: Duseja, Andress, Bhattacharya, Lam, Nagavarapu, MaCurdy.
Conflict of Interest Disclosures: Dr Sandhu reported grants from Centers for Medicare & Medicaid Services (CMS) during the conduct of the study. Dr Bhattacharya reported a contract (via Acumen, LLC) from CMS related to the work; and consulting fees from Acumen, LLC during the conduct of the study. Ms Lam reported grants from CMS during the conduct of the study. Dr Nagavarapu reported grants from CMS during the conduct of the study. Dr Choradia reported grants from CMS during the conduct of the study. Dr Do reported grants from CMS during the conduct of the study. Mr Bounds reported grants from CMS during the conduct of the study. Ms Leoung reported grants from the CMS during the conduct of the study. Ms Swygard reported grants from CMS during the conduct of the study. Ms Uwilingiyimana reported personal fees from Acumen LLC during the conduct of the study. Dr MaCurdy reported contracting fees from the US Food and Drug Administration contract during the conduct of the study. No other disclosures were reported.
Funding/Support: Episode-based cost measure development has been performed by Acumen LLC under Measure and Instrument Development and Support Contract No. HHSM-500-201313002I, Task Order HHSM-500-T0002 MACRA Episode Groups and Resource Use Measures, funded by CMS, an agency of the Department of Health and Human Services.
Role of the Funder/Sponsor: The role of CMS through this contract was in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.