The desire to improve the quality of health care has prompted the increased use of performance measures. These measures examine the outcomes of interventions for evidence of improved health and are used to hold providers accountable for the quality of health care. Yet tension exists regarding the capabilities of the current science of quality measurement. Physicians and researchers often question the validity of these data-for example, the use of hospital claims to measure qualityeven as policy makers, payers, and the public seek evidence of improved performance. This article discusses the current impasse in the field of quality measurement and what is needed to overcome this deadlock.T he desire to measure and improve the quality of health care is intensifying. Substantial shortcomings in the quality of care persist, causing needless patient harm and increasing health care costs.1,2 Still, there is little consensus on the best methods for measuring the quality of care for physicians, hospitals, or populations.Tension exists between scientists, who are dubious about the validity of many metrics, and policy makers, who have an obligation to protect the public. In the United Kingdom, for example, increased mortality following open-heart surgery among infants at Bristol Infirmary went undetected for more than four years before an external inquiry detected poor care.3 Tensions see-saw between the need for public protection on the one hand, and inaccurately identifying quality-of-care problems on the other, potentially levying unjust sanctions on clinicians and provider organizations and squandering public resources on unwarranted investigations.
Outcome MeasuresClinicians favor outcome measures, if those measures are valid and reliable enough to enable conclusions to be drawn about the quality of care. In addition, policy makers and the public have a natural affinity for outcome measures. In fact, much of health care reform worldwide is posited upon measuring and improving patient outcomes. 4 Yet scientists caution that differences in outcomes, such as death rates, are an imperfect reflection of the quality of care.
5David Shahian and colleagues compared the performance of four common vendors that provide measurement services in measuring riskadjusted in-hospital mortality (comparing how many patients died versus how many would be expected to die based on the severity of their condition), using the same data from hospital discharge records for each examination of the vendors' performance.5 They found variation in the proportion of hospital discharges included by each measurement method (range, 28-95 percent).The performance among vendors also varied: 43 percent of hospitals that showed higherthan-expected mortality by one vendor's method had lower-than-expected mortality by another. The authors concluded, "Efforts to use hospital-wide mortality rates to evaluate the quality of care must proceed cautiously…. Different approaches to measurement produced different