As any fan of Monet can tell you, sometimes you can be too close to a picture to really see it. As general internists, we are all very close to the enterprise of quality measurement. Whether we practice inpatient or outpatient medicine, we are all subject to dozens of quality measures every day, and we may rarely, if ever, stop to think about where quality measurement came from or what it is intended to accomplish. The idea of quality measurement came of age in the 1980s, in part propelled by the structure-process-outcome (SPO) model of Avedis Donabedian.1 The SPO model is built upon the underlying assumption that quality of care is a concept too abstract and too multifaceted to be directly observable, but it can be inferred by measuring structures of care (the setting within which care is delivered), processes of care (discrete actions performed within the healthcare system), and outcomes of care (including adverse events, costs, patient satisfaction, and quality of life). Donabedian proposed that structure of care is a partial determinant of processes, which in turn are a partial determinant of outcomes.

Approximately 30 years into the quality measurement experiment, we are still struggling to apply this idea in ways that improve patient outcomes. As we increasingly use pay-for-performance to incentivize providers to focus on improving quality of care, we have learned that flawed quality measures can produce unintended consequences, such as promoting cherry-picking (eschewing difficult patients for those who are more likely to do well)2 or increasing the challenges for healthcare providers who care for patients of low socioeconomic status.3 It has also been observed that many quality measures are relatively blunt instruments, which may not fully incorporate all the nuanced clinical decision-making that characterizes excellent care, and which may therefore encourage prioritizing the quality measure over the patient’s needs or preferences.4 These well-recognized flaws have contributed to a call to re-imagine quality measurement. In a recent issue of JGIM, Asch and Kerr called for the development of a new generation of quality measures that would 1) be more closely linked to important clinical outcomes, 2) be more responsive to the preferences of patients themselves, and 3) incorporate the idea of value, namely not only the outcomes that are achieved but the relative cost of achieving them.5 It is clear that better quality measures would have greater potential to improve care and outcomes than current measures, with less potential for unintended consequences.

However, amidst all the pressure to create a next generation of quality measures, it should not be forgotten that the first generation of measures has already had a great impact on some areas of care. One prominent example is the provision of beta-blockers to patients after myocardial infarction, absent a strong justification to do otherwise. When the proportion of patients leaving the hospital after a myocardial infarction with a beta-blocker was introduced as a performance measure in 1996, only about 60 % of patients were receiving one. By 2007, performance on the measure had approached 100 %, and all hospitals were performing at similar levels, which led to the measure being retired, having served its purpose.6 The beta-blocker measure was therefore responsible for extending and improving the lives of millions of patients. This is a reminder to us all that while highly sophisticated quality measures are sometimes needed, some areas of care can be greatly improved by applying relatively basic quality measures. The key to applying this sort of “first-order” quality measurement may be in finding cases where the evidence is clear, the benefit is meaningful, and the complicating issues minimal—as they were with the beta-blocker measure.

In the current issue of JGIM, Frankel and Bishop bring some of these issues into sharp relief.7 They focus on the Physician Quality Reporting System (PQRS), a mandated quality reporting program managed by the Centers for Medicare and Medicaid Services (CMS). Comparing the corpus of PQRS measures between 2011 (its first year) and 2015 (the most recent year), the findings were simultaneously encouraging and troubling. The encouraging news is that over the 4-year period, there was movement away from an overwhelming focus on measuring process of care (decreasing from 85 to 66 % of measures), and a corresponding increased use of outcomes measures (from 13 to 29 %). Another encouraging finding is that the 2015 measures include added parameters for diseases managed by internal medicine subspecialties, in addition to the usual focus on primary care-managed conditions such as diabetes. However, there was also cause for concern. While measures were more broadly distributed across specialties in 2015 than in 2011, primary care remained the focus of more measures than any other clinical discipline—by far. Another cause for concern is that of the six National Quality Strategy (NQS) areas (effectiveness, safety, communication/care coordination, person-centered experience, efficiency/cost reduction, and population health), the great majority of PQRS measures continue to focus on the first three areas, with little or no change from 2011 to 2015. Overall, while some progress is being made, we still have a long way to go before we can truly say we are measuring all the care we deliver—whether by specialty or by different aspects of improving care. To a large extent, we are still measuring mostly what we have been measuring since the 1980s, namely the management of a few common conditions in primary care (e.g. diabetes), cancer screening in primary care, and outcomes of a few major surgeries. As any doctor (and many patients) can tell you, the spectrum of clinical care encompasses far more than that.

The question, then, is what areas of care would lend themselves to what I have called “first-order quality measurement”, which tends to work best when there is unequivocal evidence for a measure’s importance, together with little potential for unintended consequences. Frankel and Bishop make the point that PQRS is actually a leader in developing and improving quality measurement, and that PQRS already includes a more balanced approach to measurement than private sector efforts such as the Healthcare Effectiveness Data and Information Set (HEDIS). In this way, they argue, PQRS is already at the leading edge of making our quality measurement regime more balanced across medical specialties, among the components of the SPO system, and across the six NQS domains. The problem suggested by this study, however, is that progress is slow.

The question may be asked: how many beta-blocker-like cases are out there now, waiting to be the targets of first-order quality measures? And how many patients might benefit from these measures? We cannot know unless we look. Therefore, it is time to extend the steps PQRS has made toward measuring the care of a few conditions managed by specialists, such as inflammatory bowel disease, and add to this admirable (but modest) beginning. It is time to measure the care delivered by physicians outside internal medicine, such as dermatologists, neurologists, radiologists, and others. It is time to measure important care delivered by non-physicians, such as clinical pharmacy specialists, who are often entrusted with managing care for a complex condition,8 but are rarely subjected to (or perhaps honored with) performance measurement. Finally, to the extent that PQRS truly is ahead of the curve, it is time for others (like HEDIS) to follow their lead. The impetus to improve our existing quality measures is admirable and important. But we cannot solely focus on improving measurement of the areas we already measure, when so many other important areas of care have not been measured at all.