October 30, 2015
Andrew Slavitt, Acting Administrator
Center for Medicare & Medicaid Services
7500 Security Blvd.
Baltimore, MD 21244
Thank you for the opportunity to comment on the draft “Quality Measure Development Plan (MDP): Supporting the Transition to the Merit-based Incentive Payment System (MIPS) and Alternative Payment Models (APMs).” The National Committee for Quality Assurance (NCQA) commends you and your staff for this thoughtful document and strongly agrees with its overall direction for building a patient-centered measure portfolio for MIPS, APMs and other programs and payers. Prioritizing disparities, patient-centered measurement and patient-reported outcomes measures (PROMs), as well as measurement derived from clinical workflows, will play an especially pivotal role in advancing measurement, quality improvement and value-based payment.
The health care quality measurement field is and must continue evolving as our ability to collect data and measure performance improves over time. Our ultimate goal should be adaptive, personalized measurement that moves away from blunt thresholds and instead promotes care tailored to specific, individual patient needs, as well as patient populations. Measurement also must recognize the differing abilities of clinicians in different settings and payment models to influence measurement results.
We envision a system where measurement is adaptive to what is most important to an individual patient at that given point in time. Individual measures can be drawn from a bank of measures. However, the measure for each person will differ based on their needs, and for each person we would calculate an index of quality based on their specific metrics. In other words, we would apply the concept of item response theory and computer adaptive testing to quality measurement. While this approach lacks the transparency and desirable simplicity of, for example, ten measures compared across all clinicians, it reflects the underlying complexity of real-world individual patients and health care delivery.
MDP Strategic Vision
We strongly support your Strategic Vision for a patient-centered measure portfolio that addresses critical measurement gaps, facilitates alignment across public and private programs, and promotes efficient data collection, rapid cycle improvement, and meaningful, transparent and actionable information. The foundational principles’ goals of eliminating disparities, strengthening infrastructure and data systems, enabling local innovations and fostering learning organizations are all highly important. As measure developers, we look forward to working with you and other stakeholders to support these goals and principles.
The patient-centered portfolio you envision, importantly, must evolve over time, follow patients across the continuum of care, emphasize outcomes (including global and population-based measures), address patient experience, coordination and appropriate use, and promote multiple levels of accountability. It also must apply to all types of clinicians, account for all payment models, incorporate registries, stratify results by demographics to help address disparities, and support meaningful public reporting. Patient-centered measures are distinct from, and may conflict with, specialty-specific measures. A comprehensive patient-centric measure portfolio exists for diabetic patients, but not other populations.
Eliminating Disparities: Measurement is a powerful tool for tracking, reducing and ultimately eliminating disparities in care for disadvantaged populations. Our research shows that some plans and clinicians achieve high quality despite serving disadvantaged populations, and we applaud you for prioritizing this achievable goal. Disparities should be addressed in the MIPS Clinical Practice Improvement Activities (CPIAs), and will require clinicians to report data stratified by demographic information.
Strengthening Infrastructure & Data Systems: This is critical and we greatly appreciate the emphasis in the draft MDP on using data in electronic health records (EHRs) derived as much as possible from clinical workflows. Better infrastructure and data systems are particularly important for minimizing burden, maximizing access to meaningful clinical data for measuring and improving performance, and providing more timely feedback for rapid cycle improvement. However, we suggest that CMS look beyond data from just EHRs in attempting to glean accurate measures. Our experience indicates that EHRs are a source, but not the only source. Until interoperability is universal, we should look to quality reporting systems that aggregate several sources to get the most accurate assessment of quality.
Enabling Local Innovation & Fostering Learning Organizations: This is pivotal for meeting individual communities and clinicians where they are in the journey to improve quality and transform health care delivery. We work with a Learning Collaborative to develop standardized ways of utilizing electronic clinical data sources (ECDS) for depression management and treatment measures. We also work with a Learning Collaborative to develop standards and patient-identified goal measures for long-term services and supports. This approach will be extremely valuable for promptly filling measurement gaps. That said, it will be vital to ensure that local innovations are benchmarked for valid comparisons nationwide.
Measure Maintenance: In addition to these principles, we strongly urge you to keep in mind the significant effort required to maintain measures after initial development. Measure stewards must continually monitor the evidence for and impact of quality measures to update them as evidence evolves and, eventually, retire them when they show little room for further improvement.
Public Reporting: To have meaningful results for public reporting, it will be essential to test both the measures and how they are presented with beneficiaries. Their input will be critical in assuring that Physician Compare is truly useful for consumer decision making. In developing public report cards for other programs, we found that consumers need help understanding that measurements are objective assessments by neutral, independent third parties, not subjective reviews from individual consumers.
Low-Volume and Rural Clinicians: We see a great opportunity in the option to form virtual groups for low-volume and rural measurement purposes. Virtual groups help clinicians with lower volumes obtain sufficient sample sizes for valid measurement. They also start clinicians in moving up the continuum from traditional, independent fee-for-service practice toward more organized systems of care delivery, including Patient-Centered Medical Homes (PCMHs) and Patient-Centered Specialty Practices (PCSPs), APMs, Accountable Care Organizations (ACOs) and eventually to population-based payment models.
We also encourage you to accept data on all patients, rather than Medicare alone, which will help achieve adequate sample sizes for valid measurement for low-volume and rural clinicians.
Vital Signs: Core Metrics for Health and Health Care Progress: The Institute of Medicine’s report that urges a focus on a small number of measures is in fact a call for nesting measures into broader categories for more global assessments. The IOM’s measures are useful for population and community level assessments, but not for individual clinician assessments. New and existing measures nested into the IOM’s broader categories will still have utility in guiding clinical practice and quality improvement efforts at the patient and even population level.
We appreciate the interest in a parsimonious measure set to address the cacophony of measures and focus on a limited set of metrics. We also agree that measuring and managing population health is an essential goal. However, measures are most effective for improving quality when it is clear who is accountable for the results. Measures in the IOM’s framework have limited applicability to individual clinicians, especially in less organized practices and delivery systems. They become more relevant as clinicians move toward APMs and population-based payment models.
Also, the IOM’s parsimonious measure set would not address the needs of all populations, particularly for some of the most vulnerable populations with special needs whose care is and should be different, such as frail older adults. At worst, a limited measure set has potential to promote inappropriate care in efforts to score well on metrics intended for other populations. For the truly patient-centered portfolio the MDP envisions, measures must be specific to the needs of individuals.
Operational Requirements
Multi-Payer Applicability of Measures: We agree that a core measure set aligned across public and private payers to minimize burden must be a priority for the MPD. As partners in the Core Quality Measures Collaborative that released a proposed core measure set earlier this month, NCQA presented information relevant to the measures under consideration to the collaborative meetings. Given NCQA’s role as a leading measure developer, it is not surprising that 17 of 22 measures in the ACO/PCMH set are NCQA’s. Additionally, five of the 11 obstetrics and gynecology measures, and two each of the cardiology and HIV/ Hepatitis C measures, also are NCQA’s. We are glad to have participated in this worthwhile effort to ease burden on clinicians working with multiple payers.
Coordination and Sharing across Measure Developers: We look forward to working collaboratively with CMS and other developers and stakeholders to fill measurement gaps. In doing so, it will be critical to maintain high standards, guard against conflicts of interest and insist that all resulting measures are:
- Clinically important,
- Evidence-based,
- Transparent,
- Feasible,
- Valid and reliable,
- Actionable, and
- Rigorously audited to ensure accuracy.
Measures not meeting these criteria add burden but not value and will not help achieve MACRA’s goals.
We encourage CMS and other HHS agencies to ensure that measure development contracts do not include provisions that limit or slow data sharing. For full collaboration such contracts must instead explicitly allow data sharing across contracts. This will prevent avoidable duplication of efforts.
To further facilitate collaboration, online feedback mechanisms and forums that allow participation at users’ convenience may be more constructive than bi-weekly conference calls. Measure developers also would benefit from improved CMS feedback that includes blinded individual clinician results so we can analyze the data in order to strengthen and maintain the measures going forward. Also, the Agency for Healthcare Research & Quality’s National Quality Measures Clearinghouse already serves the measure library function for which the draft MDP calls.
Clinical Practice Guidelines: We greatly appreciate the draft MDP’s focus on the need for clinical guideline developers to address multiple chronic conditions. As a leading measure developer, NCQA relies on guidelines from other independent groups as the foundation of our work. Most existing guidelines address only single conditions and do not meet the needs of people with multiple chronic conditions for whom better quality measurement must be a top priority. Until evidence and guidelines are developed to support measures specific to multiple chronic condition patients, cross-cutting measures that apply to multiple patient populations with different chronic conditions can help. CMS’ “Potential Chronic Conditions Preferred Specialty Measure Set” addresses common concerns for Medicare beneficiaries with multiple chronic conditions and provides an excellent start.
Evidence Base for Non-Endorsed Measures: Measures must be evidence-based to have validity, and a prime reason for current measurement gaps often is lack of needed evidence to guide measurement. We are therefore pleased, as noted above, that the draft MDP focuses on strengthening infrastructure and data collection, which are both essential for developing evidence on appropriate care to fill measurement gaps. Strengthening infrastructure can help us utilize the rich trove of real-world clinical information in EHRs to develop some of this needed evidence and support broader measure development. We urge you to accelerate efforts with the Office of the National Coordinator for Health IT, vendors and clinicians alike to evolve current EHR standards so they do support these priorities.
Quality Domains & Priorities: Each of the five domains listed in MACRA – clinical care, safety, care coordination, patient and caregiver experience, population health and prevention – are vital.
We agree with your emphasis on improving coordination of new measure development, promoting harmonization of existing measures, achieving a set of highly effective measures that minimize burden, and providing useful information when publicly reported. We further agree on the importance of prioritizing outcomes measures, including PROMs and functional status measures, intermediate outcomes measures, and measures assessing diagnostic skills and clinical practice guideline adherence. However, measures of processes closely linked to outcomes will continue to be important, particularly for situations where clinicians have limited influence on outcomes.
Clinical Care: Clinical quality measures must be clinically important, evidence-based, transparent, feasible, valid and reliable, actionable, and rigorously audited for accuracy. The MDP’s call to strengthen infrastructure and use electronic data derived as much as possible from clinical workflows is critical for minimizing burden. However, EHRs and clinical workflows are not yet well-aligned with measurement and, importantly, do not incorporate PROMs. PROMs provide much more timely feedback than patient surveys conducted outside care settings. PROMs also can gauge whether treatments need adjustment and thus directly improve outcomes. Incorporating standardized PROMs is key to improving outcomes and delivering patient-centered care.
Specialty Societies: Specialty societies, like all other stakeholders, have a pivotal role in developing additional needed measures. Specialty societies tend to participate in traditional measure development avenues, but also can be engaged more creatively. We teamed with the American Academy of Pediatrics Quality Improvement Innovation Network of practicing clinician teams to vet pediatric measure framework and logic, assess specifications and recruit for field testing. We also conduct clinician focus groups and test measures within clinician field-test site EHRs. Post-development, we engage with specialty societies on how to use new measures, such as through the Maintenance of Certification process. These experiences taught us that early, meaningful clinician involvement through targeted vetting and field testing, as well as discussing measures’ value in programs, can improve clinician buy-in.
Safety: We encourage you to think broadly about what constitutes safety measures. For example, safety can include appropriate use assessments, as inappropriate use has great potential to cause harm. Safety also encompasses things like preventable hospitalizations for ambulatory care sensitive conditions that are treated more efficiently, effectively and safely in ambulatory settings.
Care Coordination: For care coordination, we urge you to not assess merely whether there is timely exchange of clinical information but also whether data are used to improve care across settings and clinicians. This will be especially important in distinguishing the different abilities in financial arrangements that create and strengthen shared accountability. We support your suggestion to also measure outcomes of successful coordination, such as admissions for ambulatory sensitive conditions as well as readmissions that apply to multiple members of care teams.
Patient & Caregiver Experience: For the most useful and timely patient and caregiver experience assessments, we need much more rapid and targeted feedback. Existing surveys generate weak response rates and much slower feedback than clinicians need. Existing surveys also go to randomized patient samples that do not provide information on different types of patients, such as vulnerable populations who may have different concerns or require more targeted approaches. Surveys now also do not assess critical areas, such as after-hours access, coordination, and cultural competency not relevant to most sampled patients. They furthermore are offered in only a few languages. Any patient experience tool used for MIPS should be in at least the top ten primary Medicare enrollee languages.
Instead of current surveys, we envision an approach using ongoing data collection methods linked to practice-based clinical data to target survey items to individuals based on their care needs. This would allow anonymous patient reporting, much more rapid and actionable feedback, and use of analytic methods to aggregate results over time, weight results back to represent the full patient population, and provide fair comparisons across clinicians. These measurement approaches, once developed best fit into the MIPS CPIA category, as patient feedback is essential for improve the experience of care.
Population Health & Prevention: We greatly appreciate the draft MDP’s emphasis that population health is a combination of factors, including some outside the traditional sphere of health care. The final MDP should make explicitly clear that the transition to paying for value requires clinicians to address all these factors, including those not traditionally considered to be in their purview.
Population-based measures can be used to calculate benchmarks and facilitate quality comparisons within and across payment models both nationally and within specific geographic regions. However, the specifications for population-based measures must be appropriate for each unit of accountability in order to work effectively. Population-based measures are generally not appropriate for individual clinicians in MIPS who have little ability to influence population outcomes.
Holding individual clinicians responsible for such outcomes when they lack shared accountability frameworks may create adverse incentives to avoid vulnerable and complex patients.
The Medicare Payment Advisory Commission suggested aggregating clinicians in a community or geographic region for measures that have limited meaning for individual clinicians.[1] While individual clinicians might wonder about their contribution to a geographic region’s performance, stakeholders need to understand the performance among the large number of clinicians outside coherent accountable entities. Local aggregation would allow comparisons of population-based measures among clinicians in a region under MIPS versus those in APMs, ACOs and MA plans. For non-aggregated assessments of individual clinicians, measures of processes proven to improve outcomes, like cancer screenings and good chronic care management, are appropriate.
Efficiency & Cost Reduction: For resource use measures, the challenge will be in assessing the total resource use over which practices have influence, and differentiating between individual clinicians and episodes of care. The resource use that clinicians influence will likely differ in MIPS versus APMs, and will likely increase as clinicians move along the continuum from traditional fee-for-service practices to better organized PCMH and PCSP practices, APMs, ACOs and population-based payment models. Clinicians in each model will have differing levels of influence on total spend, and need to be measured accordingly, while still holding the entire system accountable for outcomes, including total spend.
NCQA has experience linking quality metrics with resource use information for specific populations. This is the basis for our health plan-level Relative Resource Utilization measures, which cover five conditions that together account for over half of all health care spending. You could follow a similar approach for other conditions. For example, you could link quality and resource measures for non-preference-sensitive procedures where attribution is straightforward because the clinician is responsible across some span of time after the procedure. We also strongly support the draft MDP’s focus on appropriate use, according to guidance such as the Choosing Wisely® initiative.
We caution against bundling quality and resource use assessments for preference-sensitive conditions that clinicians could entice healthier patients to undergo to skew results. Delivering services, no matter how efficiently, to people who do not need them is not efficient care.
We agree that “balancing measures” are important for mitigating potential underuse associated with appropriateness measures, and we are now working on such measures. For example, we are working to assess the quality and efficacy of pain control in order to balance measures of potential opioid overuse.
We also are working to assess inappropriate cervical cancer screening in teens to balance measurement on appropriate cervical cancer screening in older women. We look forward to collaborating with you and other stakeholders as work on these and other balancing measures progresses.
Gap Analysis: We agree with the draft MDP’s approach for identifying current gaps in measurement, and prioritizing use of PROMs and processes closely linked to outcomes to fill them.
Applicability of Measures across Settings & Clinicians: This must be among the highest MDP priorities, given the large number of Medicare beneficiaries with multiple chronic conditions who are not served well by specialty-specific or condition-specific measures.
The best approach is a core measure set that allows comparison across MIPS, APMs, ACOs and Medicare Advantage. Core measures must be specified appropriately for the differing situations of individual clinicians, practice teams, ACOs and MA plans, yet aligned in concept and intent to allow meaningful comparisons. The measures should maximize use of patient-generated data about experience and outcomes and be meaningful to patients, clinicians and all other stakeholders alike. Measures also must be vigilantly maintained and evolve for new evidence, progress on measures with little room for further improvement, and improvements in measurement science and data collection methods.
Core set measures also can address a common set of domains, identified with multi-stakeholder input, such as those identified by the National Quality Strategy.[2] Many existing measures address possible core domains, such as preventive and evidence-based care, and could facilitate cross-program comparison. The core set can then support nesting of measures needed for other purposes. For example, groups of measures can be used as building blocks that are aggregated for each core quality domain. Domain scores meet the needs of consumers who prefer higher-level quality data, and other stakeholders seeking to focus on a reduced measure set.
Measures also are needed outside the core set for certain sub-sets of populations and types of clinicians. These include measures that specifically address the unique needs of vulnerable populations, including patients with functional and cognitive limitations and serious mental illness. They also include measures for specific medical specialties providing unique services not addressed in the core set. It is essential to have measures appropriate for all needed purposes, and to not discard measures that help clinicians improve value for the simplistic purpose of parsimony or fewer measures overall. Efforts to reduce burden should focus on selecting high-value, high-impact measures and efficient data collection, not arbitrary limits on the number of measures that have utility for improving quality.
Clinical Practice Improvement Activities: MACRA’s CPIA categories closely parallel standards in NCQA’s PCMH and PCSP programs, including expanded access, population management, care coordination and beneficiary engagement (for example on self-care support and shared decision making). MACRA also gives full CPIA credit to PCMHs and PCSPs as recognized by the Secretary. NCQA’s PCMH program is by far the largest nationwide, now recognizing more than 15% of all primary care clinicians. Our PCSP program also is to date the only one nationwide. Clinicians in NCQA’s programs or meeting similarly rigorous PCMH and PCSP standards should receive full CPIA credit. CMS should not let clinicians merely self-attest to PCMH and PCSP status, and instead must have confirmation of actual practice transformation by independent third parties with practice transformation expertise.
Our standards also can help assess whether practices qualify for partial CPIA credit. Must-pass elements, such as patient-centered appointment access, care planning and self-support, referral tracking and follow-up, may warrant higher credit. You could also give partial credit for work toward recognition, which will give clinicians a solid start towards full PCMHs or PCSPs transition. It also will promote MACRA’s goal of alignment across payers, since our programs are by far the most widely used by other payers and align with American Board of Pediatrics, American Board of Internal Medicine and the American Board of Family Medicine maintenance of certification policies.
You also should assess efforts to address disparities in the CPIAs. Coordination with other clinicians and non-clinical community supports can be pivotal in addressing disparities and is a PCMH priority.
Ensuring that CPIA data are stratified by demographic information will be particularly important for tracking progress in reducing and ultimately eliminating disparities.
We agree that the PHQ-9 depression tool has great potential as a measure. We are, in fact, now building on our existing PHQ-9 depression screening measure to also evaluate PHQ-9’s use in assessing either remission or need for adjustment of depression treatment over time. As data systems capturing PROMs like PHQ-9 become widespread, we will have more data to inform additional measures assessing the degree to which clinicians appropriately adjust individual patient’s treatment.
The CPIA’s beneficiary engagement domain can only be reported by patients. This includes patients’ confidence in their ability to manage their health and health care, participation in decision-making, alliance with the health care team, and choice and control over decisions. We envision this to include:
- Confidence in Self-care: Individual’s knowledge and confidence in managing their health and health care is associated with better control of chronic disease and potentially other outcomes such as well-being and functional status. Tools such as the Patient Activation Module and Confidence in Diabetes Self-Care scale have been validated and are commonly used in intervention studies. There is potential to use these PROMs in assessing beneficiary engagement.
- Shared Decision-making: Shared decision-making is often assessed via surveys that do not discern whether patients realize they have treatment options. Developing quality measures that capture information during the process of care for individuals with treatment options is therefore critical for more valid assessment. Examples include treatment for new or recurrent cancer as well as conditions where multiple treatment options include surgery or other procedures.
- Autonomy, Choice and Control: Additional aspects of engagement include patient sense of autonomy and control. Several surveys assess patients’ sense of choice and control in the long-term services and supports field such as the National Core Indicators and Money Follows the Person Quality of Life Survey. These could be a starting point to adapt the concept for use with multiple chronic condition patients who may face several treatment decisions.
- Therapeutic Alliance: This concept, explored in behavioral health, is the achievement of a collaborative relationship between the clinician and patient. Therapeutic alliance measures assess the degree to which patients are willing to engage and remain in a clinician’s treatment. In psychosocial therapy, there is a strong link between this and positive outcomes. Several therapeutic alliance PROMs have been used in research, and more efforts to evaluate them are needed.
Consideration for Electronic Specifications: New measure development must include specifications for using electronic data sources that are much richer than claims. We agree that the eCQM National Test Bed, National Testing Collaborative, AcademyHealth Electronic Data Methods Forum and NQF Incubator can support this priority. We further agree on the importance of prioritizing eCQM development that ensures patient relevance, improves measure quality, increases clinical data availability, accelerates development cycle times and drives innovation. We provide more detailed suggestions for this below on the draft MDP’s inquiry regarding ways to shorten the measure development time frame.
Challenges in Quality Measure Development
Engaging Patients in Measure Development: NCQA has a longstanding commitment to including patient advocates in the broad stakeholder group that supports our consensus-based measure development and maintenance process. It is particularly valuable to gain patient and family engagement beginning with the conceptualization and prioritization of measure opportunities.
Engagement continues with efforts to make sure initial specifications address quality problems that are important to patients, and to get input on how to present results to broad audiences.
Obtaining meaningful patient and patient advocate input requires planning and effort. Our methods of engaging patients and families include traditional methods like advisory panel participation as well as more innovative approaches like interactive computer surveys tailored to a patient audience. When involving patients on panels or in review of detailed technical specifications, more specific training in the measure development process may be helpful. Orientation and ground rules for patients and other panel members can help ensure that patient opinions are sought and honored. Having a trained patient advocate serve as a monitor or facilitator for the process may be helpful. It is furthermore important to ensure materials are understandable to a lay audience.
Logistics also are important. Working with patients may be more successful when done in person versus telephone calls or web-conferences. Online communities or other electronic social media tools may help to broaden participation. Because illness or other responsibilities may limit long-term participation, it may be helpful to have several patients involved or seek participation of different groups of patients at different stages. Patient advocacy groups and patient/family advisory councils also may provide an efficient approach to recruiting participants.
Reducing Clinician Burden: As discussed above, the most important effort to reduce clinician burden is to strengthen infrastructure and use EHR data derived from clinical workflows for measurement. However, EHRs and clinical workflows are not yet well-aligned with measurement and, importantly, do not incorporate PROMs. We urge you to accelerate efforts with the Office of the National Coordinator for Health IT, vendors and clinicians alike to evolve current EHR standards so they do support this priority.
Shortening the Measure Development Time Frame: NCQA has given much thought to developing measures more effectively and efficiently, particularly to reduce time and cost that is largely tied to testing. Making testing more efficient will be particularly challenging for new measures and measure concepts that are inherently aspirational and require re-engineering of clinical workflows, new data elements and analytical capacity. Such measures could influence what health systems and vendors include in EHRs and other data sources, and testing them requires radically different approaches.
Absence of data elements in data sources or “distance” from mainstream workflows should not deter development and testing of needed high-priority measures. Fortunately, the draft MDP can help “boot-strap” the availability of test beds and colloquia that measure developers need to support such measures. CMS should support a clearinghouse for sharing lessons learned though such measurement testing. The real-time information from these nascent initiatives, including actual testing experience, issues encountered, and lessons learned can help improve measure conceptualization, design, testing, transparency and evolution of a learning health system.
Importantly, bench testing does not produce the same results as “field testing” due to variability in EHR capabilities, variation in implementation among clinical sites and highly individualized workflows. Our experience specifying, collecting, and analyzing measurement data led us to implement automated bench testing, field testing as part of our specification process and independent audit of the collection and submission process. There is much more to ensuring accurate measures than just bench testing in a “national test lab” to result in nationally comparable measures. Current measure testing requirements therefore reveal limited information about the reliability or validity for broad scale use.
Many software vendors and applications developers make beta-versions available to a large group of users to try under real conditions and use that experience as part of beta testing. We did this recently in a Learning Collaborative to develop new depression measures using ECDS. CMS could consider a variety of incentives to encourage participation across a spectrum of entities and practices. CMS also could develop beta-use measure reporting and feedback mechanisms that mirror commonly used testing plans to provide developers and vendors with critical insight to improve measure specifications and EHR standards. Such an approach would be mutually beneficial to developers as well as early-adopters.
We also are eager to work with you and other stakeholders to develop measures in a rapid-cycle fashion in accordance with Lean principles. NCQA has participated in three annual CMS Lean Kaizen events and incorporated Lean principles into our measure development process as a result. We also have used Lean principles to improve other processes within our own organization, and evaluated how they can be applied to measure development. We have dedicated staff trained in Lean principles with experience applying those principles to achieve significant cost and staff time reductions by applying Lean to our health plan accreditation and education programs. The Lean process also strengthens cross-department collaboration and alliances with external partners who appreciate the commitment to efficiency and maintenance of high standards.
We recommend that you continue sponsoring Kaizen events, as CMS is uniquely positioned to bring together all relevant stakeholders. Applying Lean principles to the measure development process should yield comparable results in efficiency and strengthened collaboration. It thus is directly applicable in promoting the MDP goals of improving coordination of new measure development, promoting harmonization of existing measures, achieving a set of highly effective measures that minimize burden, and provide useful information when publicly reported.
Streamlining Data Acquisition for Measure Testing: We support your plan to leverage broader data sources for measure development and formation of a National Testing Collaborative to make data more accessible and less costly to acquire. This has great potential to help shorten the measure development process.
Identifying & Developing Meaningful Outcome Measures: The best outcome measures will be tailored to the individual by accounting for their particular health features. This is needed to help us move beyond blunt thresholds based on the general population to instead focus on thresholds that optimize care for individuals based on their own needs and goals.
One example is the Global Cardiovascular Risk (GCVR), a predictive model that creates cardiovascular disease risk scores from patient data in electronic systems. Current measures addressing diabetes and other cardiovascular risks set specific thresholds based on population targets (e.g., blood pressure of 140/90, A1c of 9.0). However, the greatest potential benefit for one individual may be tied to smoking cessation and for another may be bringing elevated systolic blood pressure of 200 down to 150 mm Hg. The foundation of GCVR scores is that they are sensitive to changes in individual patients’ risk profile, rather than achievement of static thresholds. The system uses electronic clinical data to gauge which care elements would register the most significant impact on outcomes for the individual patient.
Sorting volumes of data and recommendations down for individual patients is complex yet very worthwhile if results are meaningful to the patient, and therefore a more powerful motivation for behavior change. Giving patients an active voice in selecting treatment goals addresses a critical gap in improving patient outcomes.
Using large ECDS and innovative, whole-person quality measures such as GCVR will increase the accuracy, efficiency and timeliness of information needed to customize and improve care for individual patients.
Developing PROMs and Appropriate Use Measures: We strongly support the draft MDP’s focus on PROMs and appropriate use measures. Developing and strengthening PROMs is one of NCQA’s highest priorities. New electronic tools offer revolutionary opportunities to gather and use patient-generated data more efficiently and effectively for multiple purposes. We envision a system that:
- Captures patient information at relevant care process points using dynamic, algorithmically-selected approaches to choose appropriate items.
- Makes the data available for:
- Care teams – in partnership with patients – to use the data for care planning and to understand changes and improvement for individuals over time.
- Clinicians and organizations to focus improvement efforts.
- Accountability, such as public reporting or incentives.
- Strengthening the evidence base needed for performance measures are quality improvement.
CMS has already taken steps in this direction with measures that assess symptom improvement and functional status in the Meaningful Use and the Medicare Shared Savings Program. This will encourage use of PROMs in clinical practice and use of those measures to evaluate outcomes. We believe this work can be taken further with innovative approaches that use PROMs to engage patients in their care and self-management and to inform shared decision-making for complex populations. This is particularly important for patients with multiple chronic conditions or complex needs.
PROMs also can be used to develop prioritized patient-reported outcome measurement and goal attainment measurement. This method assesses a patient’s progress on standardized outcomes associated with their goals. Clinicians and patients identify one or more goals and select a standardized PROM questionnaire addressing a domain that best aligns with the outcome the patient identified as important. Over time patients are assessed on maintenance or improvement on the selected PROM domain that they have identified as most important. For example, one patient may identify pain as the outcome of most importance to them where another may identify sleep. Each patient has an individualized PROM measurement; however, at the population level you assess whether a clinician is helping patients to improve or maintain on the outcome identified as most important.
PROMs can similarly help measure individual patients’ goal attainment progress using quantitative scales. Clinicians and patients identify specific short-term goals important to the patient, define the expected, better or worse than expected outcome, and assign numerical weights to each of these outcomes. Over time they assess whether the outcomes are better, worse or as expected. For example, a patient may prioritize gardening or playing with grandchildren but has mobility problems that limit such activities. This approach thus measures the degree to which the individual’s goal is achieved.
PROMs can further be used to assess patients’ sense of safety in their home and community and ability to participate in their community. Improvement of outcomes in these domains can lead to better adherence to treatment and overall improved health. We are pleased that the draft MDP emphasizes PROMs, and hope the final MDP similarly emphasizes the need for EHRs to support PROMs and other measurement priorities directly from clinical workflows.
Appropriate Use Measures: We further agree with the draft MDP that Choosing Wisely® concepts can be a foundation for a better suite of appropriate use measures. We note, however, that appropriate use/overuse measures face three key challenges that require broader understanding.
- There is a narrow range of procedures and tests for which there is broad stakeholder consensus that a procedure can represent inappropriate care. In many areas, we simply lack the evidence base to state that a given intervention constitutes overuse.
- Overuse measures must account for symptoms and history that can make an otherwise questionable service appropriate for individual patients. However, information on whether individual patients have such factors is not in claims and requires extensive medical record review, which requires substantial resources. Once all appropriate exclusions are identified and removed, the number of cases that can be confidently categorized as overuse can be quite small.
- Skewed payments trump overuse measures. If clinicians can thrive under fee-for-service that rewards more services and complex care, overuse measures – no matter how good – will have minimal impact. Efforts to reduce over-valued codes and promote payment reforms that reward quality and efficiency are advancing, yet much more needs to be done.
Developing Measures That Promote Shared Accountability across Settings and Providers: Optimizing applicability of measures across settings and clinicians is essential to utility and efficient use of measures. However, current limitations on accessing patient data across settings and clinicians makes it difficult to follow a patient longitudinally. Some communities overcome this challenge. For example, Tulsa’s MyHealth Access Network, a health information exchange, contains claims, EMR, registry and other health care data, integrated at the person-level. Data tied to the individual is available across payers and clinician settings, and measures tied to patients can be aggregated to different levels. Similarly, Illinois’ Enterprise Data Warehouse houses all Medicaid claims plus other state agencies’ data, including vital records, early intervention, immunization registry information, etc. This lets them follow patients longitudinally and across settings for a comprehensive view of each patient’s care. NCQA’s health plan HEDIS measures allow data from multiple sources, including claims, pharmacy, laboratory, medical records (paper or electronic) and other supplemental databases.
Ideally, measures informed by comprehensive data are captured at the individual (patient) level, and can then be attributed by formula to whichever clinicians and health care systems are relevant. We also align measure specifications to allow a measure to apply to different settings from the one for which it was originally specified. This requires consideration of both how denominators are identified (e.g., an enrolled or identified population versus individual patients with a recent visits), as well as data sources accessible to different settings. NCQA has strong relationships, and can work with communities and integrated health care systems that have data systems that span health care settings.
Thank you again for inviting our comments. If you have any questions about our thoughts, please contact Paul Cotton, Director of Federal Affairs, at cotton@ncqa.org or (202) 955 5162.
Sincerely,
Margaret O’Kane,
President
[1] Medicare Payment Advisory Commission, Report to Congress: Medicare and the Health Care Delivery System, Chapter 3, Measuring Quality of Care in Medicare, June 2014.