A New Strategy to Improve Quality: Title and subTitle BreakRewarding Actions Rather Than Measures

The focus on improving the quality of medical care in the United States through initiatives like public reporting and pay for performance is based on the belief that measuring quality of care is an essential first step in improving quality of care. Without measurement, it is implored, it will be impossible to know if the care clinicians deliver is good or bad. As a result, quality measurement has flourished and has been the foundation for quality improvement initiatives. Quality measures are publicly reported and perhaps influence consumer choice of physicians and hospitals and, therefore, create incentives to deliver high-quality care.¹ They are also used to determine clinician and hospital reimbursement.^{2 - 3}

Despite this focus on improving quality using measures thought to predict quality, the health care sector has made remarkably little progress toward that goal.⁴ Also, the rate of improvement is getting smaller over time and even patient safety has seen little progress.⁵

Why so little progress? Perhaps current quality improvement interventions are poorly conceived and communicated and complicated to deliver. Or, more likely, the measures used to codify quality of care are poor predictors of quality and do not reflect the underlying causes leading to variation in the measures. Perhaps, and most likely, quality is not even measurable; rather, it is highly specialized and practiced only in local environs where myriad relationships affect care delivery. This Commentary highlights limitations of the current approach to improving quality and proposes an alternative that rewards quality improvement action rather than changes in poorly associated quality measures.

Quality Measures Are Limited in Scope and Effect. An underlying assumption about quality improvement initiatives is that what is measured must matter not only in improving health systems and health care delivery, but ultimately in improving health. Without this assumption the connection between measurement and improvement unravels.

However, quality may be feasibly measured for only a narrow and discrete portion of clinical care. Even the provision of seemingly discrete clinical care processes can vary (ie, patients with congestive heart failure [CHF] at one hospital may differ from patients with CHF at another hospital). In the hospital setting, where quality measurement has focused on whether recommended care is delivered for a few common clinical conditions such as CHF, evidence suggests that current efforts that focus on a limited set of quality measures even in these discrete clinical situations have had a minimal effect on broader measures of quality.^{4 ,6 - 7}

One argument in favor of the limited measure sets currently used is that despite focusing the attention of clinicians and hospitals on a limited portion of the care they deliver, structural changes to care delivery may promote quality improvement outside of what is directly measured (“quality diffusion”). Hospitals that perform well on a small set of process measures appear to have better mortality rates than expected based on the documented direct effects of adherence with these measures.⁸ However, research in other health care settings has found that the magnitude of quality diffusion is small and limited to those that are very high performing at baseline.⁹

A potential solution to the limited-measures problem is to improve the measures. In fact, most efforts to improve prediction focus on finding new measures. Efforts are currently under way to expand quality measurement but it seems unlikely that this will advance the quality improvement agenda. There are 2 main reasons for this skepticism. First, there is currently little agreement about what the ultimate goal of improvement is. Is it to appropriately balance the benefits and risks of treatment options for individuals (and are reliable measures of benefits and risks for individuals even available)?¹⁰ To reduce errors? To reduce hospital mortality? Or to improve quality of life while dying? Perhaps the answer is all of these. But even if all these dimensions of high-quality care could be measured, it would be impossible to maximize all these stated goals. That is the second reason for skepticism: some of these disparate quality measures are likely synergistic, whereas others are more likely to be antagonistic. Efforts to perfectly quantify quality of care may be futile.

Quality Deficits Are Addressed Using Existing Knowledge. Current quality measures alone provide little guidance about how to improve quality. Rather, they are designed perhaps to identify problem areas leading to poor quality. Indeed, hospital measures of nosocomial infection rates are designed in part to trigger identification of quality problems, an important first step to improving quality. However, in the presence of significant uncertainty about how to improve quality, solely identifying problems may provide inadequate guidance and support to improve quality. Quality measures typically target individual care processes and thus rely on individuals to change their behavior based on existing knowledge. This discourages the discovery of new approaches to improve quality, lacks flexibility to adapt to new knowledge, and fails to address the fact that quality is locally determined by organizations.

Local Problems Are Addressed With National Solutions. Current approaches to quality improvement assume that there are universal or generalizable solutions that apply to all clinicians in all hospitals and that causes of low quality do not vary from hospital to hospital. The reality, however, is different. Investigations and root-cause analyses of problems leading to poor quality across hospitals have identified a wide variety of factors contributing to poor quality, and these factors are not the same across all hospitals. Addressing these problems locally, by finding local solutions, has been a successful strategy to improve quality. For example, despite following national guidelines to reduce central line–associated bloodstream infections, hospitals often continue to have high rates of these infections. However, local efforts focused on identifying and changing those factors leading to locally specific infection rates can lead to reduction in rates.¹¹ Clearly, measuring infection rates is a first step to identifying the existence of a problem. But too often, traditional quality improvement initiatives stop there. The solution to reducing central line–associated infections is not available from measurement alone. It requires local action and solutions to discern the core causes for the specific pattern of outcomes at local sites of care.

These lessons are not new but come from industry approaches to quality improvement.^{12 - 13} Approaches from industry have been adopted in health care, with some success. However, current quality improvement initiatives generally have failed to embrace these lessons.

Defining the Principles. Based on the shortcomings of current quality improvement initiatives, there are 2 principles that could guide reform of these initiatives. First, the focus of quality improvement initiatives should be on improving rather than measuring quality of care. This helps create a climate of openness that facilitates the identification and discussion of errors in day-to-day operations; it helps focus resources on identifying solutions to problems; and it acknowledges that quality improvement efforts are an ongoing process—that it is never time to stop looking for ways to improve quality of care. Second, quality improvement initiatives should be tied to local actions and local results rather than national norms. This acknowledges that quality improvement efforts are not generalizable and one solution does not fit all. Rather, each hospital must recognize that quality is a local phenomenon, dependent in part on the hospital's commitment to and culture of quality improvement. Emphasizing local norms, rather than national norms, also reduces the tolerance for mistakes because a zero-tolerance policy toward local errors pushes the theoretical limit of error rates further than a national benchmark for error rates.

Implementing Incentives to Improve Quality. Quality improvement incentives can be restructured based on these principles. Current incentives are based on measured performance and are benchmarked to national norms. An alternative approach is to tie incentives to the local process of improving quality of care rather than the results of quality measures. This could take the form of requiring local teams of quality improvement personnel to identify problems through investigation, identify solutions to these problems, implement solutions, and document local improvement. Measurement may still be a key part of such an initiative; for example, outcome measures such as an individual hospital's rate of hospital-acquired infections may be used to identify areas with quality deficits, but additional local measures must be tailored to each institution to reflect the local core causes leading to poor outcomes. If efforts are needed to encourage physicians to wash their hands more to reduce infection rates, then hand washing rates should be measured. If chlorhexidine washes are needed to reduce infection rates, then that process should be measured. The incentives to improve and the measurement of the core causes must be at the local sites and must tie together the investigation and learning process.

This approach is consistent with the principles defined above. First, linking incentives to local actions to improve quality takes the emphasis away from error rates and places it on solutions. Second, this approach highlights the continuous attention required to successfully deliver high-quality care. Providing sustainable and perfect care is difficult and requires continuous critical examination of the care delivered. Being satisfied with being benchmarked on national measures in the high range is a short-lived celebration. An approach of local, ongoing examination encourages hospitals to never stop trying or looking for ways to improve. After all, once one core cause is gone, another may become apparent. Third, this approach emphasizes the local nature of quality and thus promotes local (and feasible) solutions.

Improving quality of care has long been an important national priority. Despite large efforts, the effects of quality improvement initiatives on this goal have been modest. It is time to rethink the current approach to improving quality so that these efforts can generate a meaningful, robust, and sustained response. The proposed new model to improving quality of care might fit that bill.

The use of a collaborative and local approach to quality improvement is more frequent now than it used to be. Quality improvement collaboratives have been adopted in numerous settings; hospital groups have started to focus on creating an “improvement community” of hospitals to assist hospitals in learning their own practices and strategies that achieve reliable results. A logical next step is to tie current quality improvement incentives to this approach—pay based on participation in quality improvement efforts rather than simply comparing each other on measures that do not reflect the learning that is required to really improve care.