“Not everything that counts can be counted, and not everything that can be counted counts.”
This aphorism has been deliciously, but, alas, incorrectly attributed to Albert Einstein (the saying actually has mixed origins, but credit properly might be given to sociologist William Bruce Cameron, writing in 1963).
But, whatever its provenance, the saying is particularly appropriate in describing the woeful lack of attention paid to the long-standing problem of diagnosis errors in the provision of health care services.
Last week academic researchers from Baylor and the University of Texas published important research estimating that one in 20 adults in the U.S., or roughly 12 million people every year, receive an error of diagnosis—a wrong, missed or delayed diagnosis—in ambulatory care.
This likely represents a conservative estimate of the incidence of such errors in ambulatory care and does not attempt to include inpatient hospital care or care provided in nursing homes and post-acute care facilities, such as rehab hospitals.
The news media correctly decided that this peer-reviewed finding deserved prominent attention—it was a lead story on “NBC Nightly News” and other national news programs.
It seems that attaching a large number to the prevalence of such errors provided the needed news hook to give the problem the attention it has long deserved. Surveys reveal that the public is worried as much about a misdiagnosis or missed diagnosis as any other quality and safety issue in health care.
Autopsy studies performed over time find that unacceptably high rates of diagnosis errors persist; similarly, diagnosis errors continue to represent a leading cause of medical malpractice suits.
But even without newsworthy body counts, the problem of diagnosis errors has been known to clinicians for decades, if largely ignored by stakeholders and policy-makers as a major quality and safety problem.
Is hospital consolidation creating new efficiencies or does it give health care providers clout over health care insurers? A well-publicized study published in Health Affairs last year by Robert Berenson, Paul Ginsburg, et. al said the latter: hospital consolidation has resulted in “growing provider market clout.”
The Berenson study’s key conclusion is that growing hospital clout has resulted in insurers not aggressively containing their claims payments, a view that will stun every patient who has had a health insurance company deny coverage for a procedure, prescription or preferred health care provider.
Because the Berenson study’s finding are counterintuitive to consumer experience, and because they have been widely discussed in publications ranging from Forbes to National Journal, the Center for Regulatory Effectiveness, a regulatory watchdog with extensive experience in analyzing federal health policies, undertook an analysis to see if the study complied with the Data Quality Act (DQA).
The DQA, administered by the White House Office of Management and Budget (OMB), sets standards for virtually all data disseminated by the agencies. Under the DQA, agencies may not use or rely on data in federal work products (reports, regulations) which don’t comply OMB’s government-wide Data Quality standards. Thus, unless the Health Affairs study complies with federal Data Quality standards, it is useless to Executive Branch policy officials.
The primary data source cited by the Berenson study as the basis for their conclusions regarding trends in relative clout between hospitals and health insurers is a well-respected, longitudinal tracking study which included interviews with heath care leaders from insurance companies, hospitals, and academia. The health care interviews, however, were only conducted in a single year following a change in longitudinal study’s methodology.
There is a consensus that measuring performance can be instrumental in improving value in U.S. health care. In particular clinical areas, such as cardiac and intensive care, measurement has been associated with important improvements in providers’ use of evidence-based strategies and patients’ health outcomes over the past two decades. Perhaps most important, measures have altered the culture of health care delivery for the better, with a growing acceptance that clinical practice can and should be objectively assessed.
Nevertheless, as we argue in the full-length version of this paper, substantial shortcomings in the quality of U.S. health care persist. Furthermore, the growth of performance measurement has been accompanied by increasing concerns about the scientific rigor, transparency, and limitations of available measure sets, and how measures should be used to provide proper incentives to improve performance.
The challenge is to recognize current limitations in how measures are used in order to build a much stronger infrastructure to support the goals of increased accountability, more informed patient choice, and quality improvement. In the following paper, we offer seven policy recommendations for achieving the potential of performance measurement.
1. Decisively move from measuring processes to outcomes.
There is growing interest in relying more on outcome measures and less on process measures, since outcome measures better reflect what patients and providers are interested in. Yet establishing valid outcome measures poses substantial challenges—including the need to riskadjust results to account for patients’ baseline health status and risk factors, assure data validity, recognize surveillance bias, and use sufficiently large sample sizes to permit correct inferences about performance.
2. Use quality measures strategically, adopting other quality improvement approaches where measures fall short.
While working to develop a broad set of outcome measures that can be the basis for attaining the goals of public accountability and information for consumer choice, Medicare should ensure that the use of performance measures supports quality improvement efforts to address important deficiencies in how care is provided, not only to Medicare beneficiaries but to all Americans. CMS’ current focus on reducing preventable rehospitalizations within 30 days of discharge represents a timely, strategic use of performance measurement to address an evident problem where there are demonstrated approaches to achieve successful improvement . Read more.
There is a plethora of health care quality data being pushed out to the public, yet no rules to assure the accuracy of what is being presented publicly. The health care industry lacks standards for how valid a quality measure should be before it is used in public reporting or pay-for-performance initiatives, although some standards have been proposed.
The NQF does a good job of reviewing and approving proposed measures presented to it, but lacks the authority to establish definitive quantitative standards that would apply broadly to purveyors of performance measures. However, as discussed earlier, many information brokers publically report provider performance without transparency and without meeting basic validity standards. Indeed, even CMS, which helps support NQF financially, has adopted measures for the Physician Quality Reporting System that have not undergone NQF review and approval. Congress now is considering “SGR repeal,” or sustainable growth rate legislation, that would have CMS work directly with specialty societies to develop measures and measurement standards, presumably without requiring NQF review and approval .
Without industry standards, payers, policy makers, and providers often become embroiled in a tug-of-war; with payers and policy-makers asserting that existing measures are good enough, and providers arguing they are not. Most often, neither side has data on how good the contested measures actually are. Most importantly, the public lacks valid information about quality, especially outcomes, and costs.
Indeed, most quality measurement efforts struggle to find measures that are scientifically sound yet feasible to implement with the limited resources available. Unfortunately, too often feasibility trumps sound science. In the absence of valid measures, bias in estimating the quality of care provided will likely increase in proportion to the risks and rewards associated with performance. The result is that the focus of health care organizations may change from improving care to “looking good” to attract business. Further, conscientious efforts to reduce measurement burden have significantly compromised the validity of many quality measures, making some nearly meaningless, or even misleading. Unfortunately, measurement bias often remains invisible because of limited reporting of data collection methods that produce the published results. In short, the measurement of quality in health care is neither standardized nor consistently accurate and reliable.
The unfortunate reality is that there is no body of expertise with responsibility for addressing the science of performance measurement. The National Quality Forum (NQF) comes closest, and while it addresses some scientific issues when deciding whether to endorse a proposed measure, NQF is not mandated to explore broader issues to advance the science of measure development, nor does it have the financial support or structure to do so.
An infrastructure is needed to gain national consensus on: what to measure, how to define the measures, how to collect the data and survey for events, what is the accuracy of EHRs as a source of performance, the cost-effectiveness of various measures, how to reduce the costs of data collection, how to define thresholds for measures regarding their accuracy, and how to prioritize the measures collected (informed by the relative value of the information collected and the costs of data collection).
Despite this broad research agenda, there is little research funding to advance the basic science of performance measurement. Given the anticipated broad use of measures throughout the health system, funding can be a public/private partnership modeled after the Patient-Centered Outcomes Research Institute or a federally-funded initiative, perhaps centered at AHRQ. Given budgetary constraints, finding the funding to support the science of measurement will be a challenge. Yet, the costs of misapplication of measures and incorrect judgments about performance are substantial.
Initiatives to promote performance measurement need to be accompanied by support to improve care. Quality measure data should not only be technically correct, but should be organized such that their dissemination is a resource to aid in quality improvement activities. As such, quality measurement should be viewed as just one component of a learning health care system that also includes advancing the science of quality improvement, building providers’ capacity to improve care, transparently reporting performance, and creating formal accountability systems.
There are several strategies to make quality measure data more actionable for quality improvement purposes. For example, for publicly reported outcome measures, CMS provides hospitals with lists of the patients who are included in the calculation. Since the outcomes may occur outside the hospital for mortality and for readmissions that are at other hospitals, this information is often beyond what the hospitals already have available to them. These data give providers the ability to investigate care provided to individual patients, which in turn can support a variety of quality improvement efforts.
Performance measurement has too often been plagued by inordinate focus on technical aspects of clinical care—ordering a particular test or prescribing from a class of medication—such that the patient’s perspective of the care received may be totally ignored. Moreover, many patients, even with successful treatment, too often feel disrespected. Patients care not only about the outcomes of care but also and their personal experience with care.
There is marked heterogeneity in the patient experience, and the quality of attention to patients’ needs and values can influence their course, whether or not short-term clinical outcomes are affected. Some patients have rapid recovery of function and strength, and minimal or no symptoms. Other patients may be markedly impaired, living with decreased function, substantial pain, and other symptoms, and with markedly diminished quality of life. It would be remiss to assume that these two groups of patients have similar outcomes just because they have avoided adverse clinical outcomes such as death or readmission.
In recommending a focus on measuring outcomes rather than care processes, we consider surveys or other approaches to obtaining the perspectives of patients on the care they receive to be an essential component of such outcomes. When designed and administered appropriately, patient experience surveys provide robust measures of quality, and can capture patient evaluation of care-focused communication with nurses and physicians . And while patient-reported measures appear to be correlated with better outcomes, we believe they are worth collecting and working to improve in their own right, whether or not better experiences are associated with improved clinical outcomes .
Historically, the physician has been viewed as the leader of medicine, with responsibility for the care and outcomes of patients; in iconic photographs and paintings, the physician is seen as a lone, heroic figure. Such a view has led to natural interest in the measurement of individual physicians’ performance. It is therefore not surprising that some information brokers, including the U.S. News and World Report and many city magazines like the Washingtonian, provide ratings of “top doctors,” often based mostly on reputation, warranted or not.
However, this focus on the individual is flawed for most measures of quality and presents substantial technical challenges. Systems-based care is emerging as a key value within health care and a vital component of high-quality care, while the notion that an individual health professional can be held accountable for the outcomes of patients in isolation from other health professionals and their work environment is becoming an outdated perspective. For example, better intensive care unit staffing sometimes mitigates the evidence that surgeons who perform more procedures achieve better outcomes .
The communication and coordination of services across providers is required to ensure that patients, many of whom have multiple conditions, are assisted through various health care settings . For some aspects of care, such as diagnosis errors and patient experience, measuring at the individual physician level might be considered. Nevertheless, focusing measurement on an individual runs counter to our goals in promoting teamwork and “systemness” as core health care delivery attributes.
While working to develop a broad set of outcome measures that can be the basis for attaining the goals of public accountability and information for consumer choice, Medicare should ensure that the use of performance measures supports quality improvement efforts to address important deficiencies in how care is provided, not only to Medicare beneficiaries but to all Americans.
CMS’ current focus on reducing preventable rehospitalizations within 30 days of discharge represents a timely, strategic use of performance measurement to address an evident problem where there are demonstrated approaches to achieve successful improvement . Physicians and hospital clinical staff, if not necessarily hospital financial officers, generally have responded quite positively to the challenge of reducing preventable readmissions.
CMS has complemented the statutory mandate to provide financial incentives to hospitals to reduce readmission rates by developing new service codes in the Medicare physician fee schedule that provide payment to community physicians to support their enhanced role in assuring better patient transitions out of the hospital in order to reduce the likelihood of readmission . CMS recently announced that after hovering between 18.5 percent and 19.5 percent for the past five years, the 30-day all-cause readmission rate for Medicare beneficiaries dropped to 17.8 percent in the final quarter of 2012 , simplying some early success with efforts to use performance measures as part of a broad quality improvement approach to improve a discrete and important quality and cost problem.
However, this Timely Analysis of Immediate Health Policy Issues 3“CMS’ current value-based purchasing efforts, requiring reporting on a raft of measures of varying usefulness and validity, should be replaced with the kind of strategic approach used in the national effort to reduce bloodstream infections.”approach is not without controversy.
Improvements have been modest, and some suggest that readmission rates are often outside the hospital’s control, so CMS’ new policy unfairly penalizes hospitals that treat patients who are the sickest . And while readmission in surgical patients is largely related to preventable complications, readmissions in medical patients can be related to socioeconomic status. Also, some have questioned the accuracy of CMS’ seemingly straightforward readmission rate measure, finding that some hospitals reduce both admissions and readmissions—a desirable result—yet do not impact the readmission rate calculation . And one of this paper’s authors (R. Berenson) has suggested a very different payment model that would reward hospital improvement rather than absolute performance, thereby addressing the reality that hospitals’ abilities to influence readmission rates do vary by factors outstside of their control .
There is growing interest in relying more on outcome measures and less on process measures, since outcome measures better reflect what patients and providers are interested in. Yet establishing valid outcome measures poses substantial challenges—including the need to riskadjust results to account for patients’ baseline health status and risk factors, assure data validity, recognize surveillance bias, and use sufficiently large sample sizes to permit correct inferences about performance. We believe the operational challenges of moving to producing accurate and reliable outcome measures, though daunting, are worth the effort to overcome.
Patients, payers, policy-makers, and providers all care about the end results of care—not the technical approaches that providers may adopt to achieve desired outcomes, and may well vary across different organizations. Public reporting and rewards for outcomes rather than processes of care should cause provider organizations to engage in broader approaches to quality improvement activities, ideally relying on rapid-learning through root cause analysis and teamwork rather than taking on a few conveniently available process measures that are actionable but often explain little of the variation in outcomes that exemplifies U.S. health care.
However, given the inherent limitations of administrative data, which are used primarily for payment purposes, and even clinical information in electronic health records (EHRs), consideration should be given to developing a national, standardized system for outcome reporting . A new outcome reporting system would not be simple or inexpensive, but current data systems may simply be insufficient to support accurate reporting of outcomes. An example is the National Health Care Safety Network system for reporting health care infections .