The suggestions by Davidoff and Batalden¹ for strengthening reports on quality improvement offer useful guidance for those wishing to publish such work. Their rationale for providing this guidance stems from their perception that the failure to provide better information about local improvement efforts slows the spread of successful changes, limits the scrutiny of quality improvement work, and reduces the incentives to participate in such efforts. They are not the first to lament the variable quality of such reports on improvement. Yet Davidoff and Batalden focus primarily on the impact of enhancements in the reporting of quality improvement efforts on quality improvement practice. These efforts will also have an important impact on quality improvement research. And the benefits between research and practice are likely to be synergistic.

More thorough reporting of quality improvement is a first step toward greater academic respectability for these efforts. But a fuller dialogue about these methods and their epistemology is also critical. The goals of quality improvement practice are to enhance performance by setting aims, examining processes of care, testing changes in these processes, and implementing those changes that improve results.^2,3 Nolan has characterized quality improvement as “pragmatic science”, referring to its emphasis on using knowledge about how care is delivered to identify improvements and build better systems though the accretion of small changes.⁴ While the selected changes derive from relevant research as well as the knowledge that clinicians gain in treating their patients, there is a level of discomfort created among those who view this approach from the standards of evidence based health care.^5,6 However, valuing local knowledge does not negate the importance of research because quality improvement goals differ. Quality improvement efforts use evidence to identify changes and focus on implementing effective practices, not assessing whether they are effective. For example, a typical quality improvement project aimed at reducing postoperative infections does not assess which antibiotic is needed by the patient undergoing surgery. Rather, it tests ways of delivering the right antibiotic in a timely fashion using guidelines based on research assessing the efficacy of different antibiotics. Quality improvement research relies heavily on simple pre‐post designs, often in single sites without controls. But such research offers an important starting point for understanding workable approaches for implementing improvements. The goal of quality improvement research is weighted toward identifying how to implement effective changes, not assessing the efficacy of those changes.

There are several other areas where quality improvement research could be more closely associated with assessments of effective practice. Firstly, the focus of improvement research on identifying effective means for implementing new practices could help in the design of randomized controlled trials and other research assessing effective practice. Randomized controlled trials are widely accepted as the most reliable method for assessing effectiveness, but they were originally designed to test discrete interventions such as a medication. The success of randomized controlled trials performed on more complex interventions may be mediated by local contexts. Implementation of what appear to be sound practices is often impeded by unforeseen or difficult to control variables such as variations in staffing or organizational receptivity to change. Complex interventions in hospitals such as stroke units or medical emergency teams require careful orchestration between units and across disciplines. Complex interventions in the community to improve care for diabetes, asthma, and other chronic conditions often include multiple components such as guidelines, patient education, communication between care providers, and the development of registries and information systems. Such interventions target multiple caregivers as well as patients. This research poses logistical challenges that require constant attention. Increasing effort is being paid to improving the design of trials for such complex interventions,^7,8 but negative evaluations of complex interventions may occur in cases where such interventions are poorly defined or badly implemented, even if the underlying approach is sound.