|
Pay-For-Performance: The MedPAC Perspective
Medicare payment systems are neutral and sometimes negative toward quality of care. The Medicare Payment Advisory Commission (MedPAC) has recommended that Congress build incentives for quality into Medicare’s payment systems for hospitals, physicians, home health agencies, facilities that treat dialysis patients, and Medicare Advantage plans. In this Commentary we describe the rationale for the recommendations, criteria for determining which settings are ready, program design principles, and potential measures.
IN JUNE 2003 THE MEDICARE PAYMENT Advisory Commission (MedPAC) recommended that Medicare build financial incentives for quality into its provider payments.1 Over the next two years, MedPAC developed and applied criteria for determining which care settings had sufficient quality measures to incorporate these incentives. The commission has since recommended that a portion of Medicare payments to hospitals, physicians, home health agencies, providers that treat dialysis patients, and Medicare Advantage (MA) plans be based on quality. What led MedPAC, as the federal agency charged with advising Congress on Medicare payment issues, to make these recommendations?
Evidence of unsafe care. Evidence continues to mount that large numbers of beneficiaries do not receive necessary care and sometimes receive unsafe care.2 Yet Medicare spent $281 billion in 2003 with little to no incentive in the payment system to improve quality. Because health spending continues to rise faster than the cost of other goods and services, private and public payers must ensure the value of additional spending for health. Resource-based payment system. Medicare’s primary payment mechanism—fee-for-service—is resource-based: It pays more for the quantity of services provided than it does for the quality of those services. High-quality and low-quality providers receive the same payment amount. In some cases—for example, when a patient develops an infection due to medical care—Medicare pays more to a provider whose care resulted in a lower-quality outcome: The provider is paid for the initial care and then again for the care needed to treat the infection. Concerned about quality, particularly given rapidly rising costs, MedPAC determined that Medicare could no longer afford for its payment system to be neutral toward quality. In fact, many commissioners saw the payment system as a central part of the problem. Although incentives for quality might not reduce costs, MedPAC believed that Medicare should, at a minimum, get the best value possible for the dollars it was spending. MedPAC determined that these incentives should build on the infrastructure of measures, data collection, and feedback tools already established by the Centers for Medicare and Medicaid Services (CMS). The CMS Medicare Compare initiative provides useful information for consumer choice and the infrastructure necessary for pay-for-performance (P4P) reporting. Further, the Quality Improvement Organization (QIO) program could be used to assist providers in improving performance in the focus areas of the P4P program.
Based in part on the experiences of private-sector initiatives, MedPAC developed (1) criteria for determining whether the measures and measurement activities for each provider setting were sufficient for distinguishing between high- and low-quality performance; and (2) design principles for a P4P program. These criteria and design principles do not address the full range of implementation issues. However, the commission used them to identify settings in which to implement the P4P program and to create a general policy approach. Evidence-based measures should be available. The measures should be accepted by independent quality experts and familiar to providers. Although few individual measures are perfectly valid or reliable, they should identify real differences in provider quality. Data collection and analysis should not be unduly burdensome. To minimize the burden, the CMS should base quality measures on data it now collects, wherever possible. The need for additional information should be balanced against the value of the information to the provider, patients, and Medicare. Appropriate risk adjustment is important when comparing quality. It is even more critical when dollars are attached to those comparisons. The program could use measures, such as process, structure, or patient-reported experience-of-care measures that, in general, are not affected by the complexity of a patient’s condition. Risk adjustment is critical for outcome measures. Yet no standards now define "adequate" adjustment. Addressing this concern requires balancing the burden of data collection with the level of accuracy of the risk adjustment. Quality measures should be improvable by providers. This criterion has several dimensions. First, the measures should apply to a broad range of care and providers. The greater the proportion of providers whose care is measured, the broader the impact on beneficiaries. It is also important to measure a broad range of types of care each setting delivers. Measures focused on specific conditions are available in most settings, but to capture a broad range of care in each setting, measures of safe practices, use of patient registries, or patients’ perceptions of care that apply to all types of patients could also be used. Second, the measures should capture aspects of care that the providers can affect. Attributing beneficiaries’ care to a specific provider might be challenging because beneficiaries often see multiple providers in one episode. This is particularly salient in measuring physician quality. MedPAC discussed various attribution rules used in the private sector but did not make a recommendation on this point. Third, the measures should be related to aspects of quality that most need improvement. The measure sets should continually evolve to focus on clinical areas where the opportunity for improvement is greatest.
Rewarding providers. To improve quality for the most beneficiaries, providers should be rewarded for attaining or exceeding certain benchmarks and improving at certain levels. Providers already at high levels will be rewarded for those efforts. Those who score low at baseline will have an incentive to improve. Funding set-aside. The P4P program should be funded by setting aside a small proportion of current payments—1–2 percent—and redistributing it based on quality. The first dimension of this principle is whether the policy should be funded by withholding the dollars or whether new spending is necessary. Through a separate process, MedPAC evaluates the adequacy of payment levels every year when it recommends payment updates for providers. The commission determined that the initiative should be funded within current levels of spending. The primary rationale was to shift the incentives of payment, not the level. The second dimension is whether the incentive is large enough to encourage provider change, or whether it is too disruptive. Evidence on this dimension is limited. One survey of private-sector efforts found that purchasers reported needing to provide incentives of 5–20 percent for physicians and 1–4 percent for hospitals.3 Yet it is difficult to know what portion of overall payment these percentages represent. Because Medicare payment often accounts for a higher percentage of any one provider’s total revenue than is the case for payment from a single private payer, a smaller percentage of Medicare’s payment might encourage change. In the CMS Premier hospital demonstration, results show improvement in all conditions in the first four quarters of the demonstration, in anticipation of financial rewards of either 1 or 2 percent for those in the upper rankings.4 Neither of these efforts funds the rewards through current payments. In a program that is budget-neutral, smaller incentives can be more powerful, as providers perceive the dollars as potential income forgone.5 The much smaller 0.4 percent incentive for hospitals in the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) was designed to encourage data reporting as a condition for receiving a full update. It resulted in nearly universal hospital reporting on certain process measures. Others have suggested that if the dollars are withheld, even a 1–2 percent set-aside could harm providers at low levels of quality. This concern was one rationale for suggesting that improvement from low levels should also be rewarded. Given limited evidence on the "right" level and to ensure minimal disruption for beneficiaries and providers, MedPAC chose to recommend that 1–2 percent be set aside, at least initially. It expects the percentage to increase as Medicare and providers gain more experience with P4P. Continual evolution of measures. After Medicare chooses an initial measure set, the CMS will need to alter, add, and drop measures and ensure that research is under way to create or validate others. A single entity could help coordinate public and private efforts and, based on the advice of quality experts, make recommendations on measures. The National Quality Forum (NQF) has brought together stakeholders to build consensus around measure sets. However, it is also important to consider the critical role that other measure developers and consensus-building organizations may have. The process should be open for all to participate and to receive independent, stable funding.
Using the criteria described above, the commission formally recommended that P4P be implemented for hospitals, physicians, home health agencies, MA plans, and dialysis facilities and physicians who treat dialysis patients.6 The CMS already collects information on many measures of quality for all of these settings that could be used as a "starter set"; however, some new information would also be needed. Hospitals. Twenty-two hospital process measures for treatment of acute myocardial infarction (AMI), heart failure, pneumonia, and surgical infection prevention are now used or endorsed by various organizations.7 Almost all inpatient acute care hospitals and 200 critical access hospitals report to the CMS on ten of these; 70 percent of hospitals report on seventeen of them. These measures are also used by the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) and are endorsed by the NQF. Risk adjustment is not necessary for these measures. The commission also suggests two widely endorsed risk-adjusted mortality measures: for AMI and coronary artery bypass graft (CABG). Adverse-event rates could also be useful, but current measures need further risk adjustment to ensure that the hospital is held accountable for adverse events that are "preventable." Other sets that could be used within a few years are a set of NQF-endorsed safe practices used by the Leapfrog Group’s purchaser initiative and a survey of patients’ perceptions of care jointly developed by the CMS and the Agency for Healthcare Research and Quality (AHRQ). Physicians. Measuring the quality of physician care is more complex than measuring quality in other settings because of the lack of a data infrastructure, the wide variety of specialties, and the sheer number of physicians. One of MedPAC’s goals was to create a more positive return on investment for physicians’ purchase of information technology (IT) systems. Measures of physician office systems’ ability to track patient care exist, and they apply to almost all physicians. In addition, some condition-specific clinical measures are available for many generalists and some specialists.8 However, data on these measures are not broadly available. The only data physicians now provide to the CMS are claims data. These factors led MedPAC to recommend a two-step implementation strategy for physicians. In the first two or three years of a P4P program, quality of physician care would be measured, and they would be rewarded for their quality-improving activities associated with IT use.9 Such measures would include indicators of whether physicians used IT systems for identifying patients with chronic conditions and tracking their care. The purpose of this strategy is not to measure the purchase of IT, such as electronic health records (EHRs), but to assess whether physicians could perform certain functions associated with quality outcomes. Simple patient management systems could be used to show compliance. This flexibility would make it more likely that all types and sizes of physician practices could participate in the program. The National Committee for Quality Assurance (NCQA), which uses these types of measures in its physician recognition program, requires physician offices to show evidence of how the IT system was used for compliance. The CMS could build on those efforts. Second, two or three years after the program begins, a set of measures of clinically appropriate processes of care should be added. Condition-specific process measures are not yet available for every type of patient or physician but are available for a broad range of conditions for which beneficiaries see physicians. Providing a date certain for inclusion of these measures would act as an incentive for all specialties to develop measures. The two-step process also would ensure that physicians have the systems necessary to improve on the measures. These types of measures can be derived from claims, medical record abstraction, flow sheets, or EHRs. The majority of physician offices do not use EHRs or flow sheets, and medical record abstraction can be expensive. Therefore, the commission suggests that at least initially, these process measures be derived from claims. Although claims-based process measures are not available for every type of condition or specialty, if linked with prescription and lab values, they are available for many conditions of importance to Medicare beneficiaries.10 Home health care. Since the CMS’s first public report on home health agency quality in November 2003, indicators have shown small, consistent improvements. Moving toward buying outcomes of care instead of inputs to care could be particularly powerful in home health care because there is so little evidence about the "right" amount or type of inputs in this setting. For home health agencies, many outcome measures from the CMS Outcomes-Based Quality Indicators (OBQI) set are promising. They are well accepted and risk-adjusted and pose no additional data collection burden. Agencies have been reporting them to the CMS since 1999, as part of patient assessments. To ensure valid comparisons among agencies with different patient populations, the data include information on patients’ prognosis, functional status at the start and completion of care, multiple diagnoses, caregiver status, obesity, smoking, and behavioral and cognitive status. A subset of these measures could be used as the starter set for a P4P program in the home health setting.11 Care for end-stage renal dialysis patients. The CMS has provided the public with quality information on dialysis facilities since 1999. Improvement has occurred, yet some beneficiaries still receive inadequate dialysis and anemia management. To encourage collaboration between physicians and facilities, MedPAC recommended that both be included in P4P for dialysis patients.12 Because information on these measures is already reported on claims, they pose no additional data collection burden, but the CMS should consider the need for risk adjustment. Dialysis facilities already report to the CMS information on patient characteristics that affect these measures. Indicators of nutritional management, vascular access care, bone disease management, and adherence to clinical guidelines on the use of vitamin D analogues could also be used. Medicare Advantage plans. As a condition of participating in Medicare, MA plans must submit standardized quality data to the CMS. Scores on these measures are publicly available. They show much variation among plans and the opportunity for more improvement. An additional benefit to implementing P4P for MA plans is their ability to coordinate efforts across providers. All MA plans report data on some measures; however, some plans only report on a subset. Health maintenance organizations (HMOs) report on all measures, including outcome and process measures in the Health Plan Employer Data and Information Set (HEDIS), CAHPS, and the Health Outcomes Survey (HOS). Private fee-for-service and preferred provider organization (PPO) plans report on both surveys but report only a subset of HEDIS measures (those available through claims data). These measures are familiar to the providers and have been shaped by researchers, users, and providers themselves for some time. HMOs have been reporting on HEDIS measures to the NCQA since 1993. The CMS began requiring the collection of HEDIS and the member survey data pursuant to the Balanced Budget Act (BBA) of 1997. Using these measures in a P4P program would pose no additional data collection burden on MA plans. The HEDIS set does not require risk adjustment. Adjustment is available for the CAHPS measures of patient experience, based on the correlation between certain demographic factors and patient satisfaction.
MedPAC focused its current P4P recommendations on improving quality, an important and necessary shift in reimbursement policy. However, P4P incentives could also be aimed at improving efficiency. MedPAC is evaluating the potential to include measures of both quality and resource use to encourage greater efficiency. Improving care coordination across settings is also an important goal for the health care system.13 Adding measures such as the quality of transfers across settings or patient functioning a year after a hospitalization could broaden P4P to encourage providers to better coordinate care across settings and over time. Addressing the myriad issues raised by any type of P4P program is a challenge. Such issues include how rewards should be distributed; the number, type, and relative weights of the measures; and the continual evolution of measures. Despite these limitations and challenges, the costs of not proceeding—costs measured by potentially avoidable illness and injury as well as spending on care that does not improve patients’ health—outweigh the potential for unintended negative consequences. The payments of the largest single purchaser of health care should not continue to be neutral or negative toward the quality of care.
Karen Milgate (kmilgate{at}medpac.gov) is a principal policy analyst and Sharon Cheng is a senior policy analyst at the Medicare Payment Advisory Commission (MedPAC) in Washington, D.C. The authors acknowledge contributions to this work from Glenn Hackbarth, Nancy Ray, and Scott Harrison. They also give special thanks to Mark Miller for his guidance and support in developing this paper.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Rationale For Adding Quality...
Criteria For Measures
