The State of Quality Improvement and Implementation Research: Expert Views: Workshop Summary (2007)

This session was open to all workshop attendees to discuss what is known about effective strategies for quality improvement research and spread. As starting points for discussion, everyone was asked to identify high-priority effective strategies as the basis for developing a spread agenda, and high-priority unanswered questions about quality improvement strategies as the basis for developing a research/evaluation agenda. No conclusions identifying top priorities were drawn during the session.

As discussed in the previous chapter, how a quality improvement intervention is implemented may be equally important as what quality improvement interventions “work.” However, little is known about which interventions are effective. It was mentioned that reminders for immunizations have been shown in the literature as a practice ready for widespread use, with the average number of patients receiving immunizations nearly doubling, as shown in 30– 40 RCTs. The field needs to better understand generalizability. Workshop attendees suggested ways studies could be more actionable and could provide greater insights to interventions. For example, audit-and-feedback mechanisms generally yield improvement, but vary widely in terms of magnitude of impact. To better understand this variation, it was suggested that head-to-head comparisons of

audit-and-feedback mechanisms be conducted instead of the current intervention group versus control group comparisons. It was also suggested that a process of implementing and evaluating effective quality improvement strategies be developed so that the implementing organization would have more knowledge of the intervention before attempting to spread to other organizations.

Current efforts around variations in health care were noted as attempting to narrow disparities. Instead, variations should be studied. Partnerships should form to help study why variations exist, help identify characteristics of variations, and help study the components of variations.

Addressing the more technical side of interventions, a few attendees called for a stronger technical infrastructure to support quality improvement efforts. It was noted that measures of health care must be standardized. Currently measures are defined in a variety of ways (e.g., some measures of breast cancer care include women aged 40–60, while others include women aged 35–55), making meta-analyses difficult to conduct because data cannot easily be aggregated, much like comparing apples to oranges. Others noted that a greater emphasis needs to be placed on developing outcomes measures that could be used to drive improvement efforts. Patient satisfaction measures also need to be developed and more widely incorporated in improvement efforts.

Another concern is the lack of comprehensive databases to use to make decisions about interventions. Such a tool could be very useful in sharing best strategies and lessons learned.

Another major focus is the need for knowledge about organizational change and cultural change in health care. The role of leadership is critical, but it is unclear how to engage top levels of leadership in quality improvement efforts. One response was to provide top leadership with motivation for incremental change. Implementing one or two small, moderately successful interventions could be the basis for widespread change. However, change does not occur only from the top, but must be integrated from levels throughout the entire organization. For example, the receptionist, a patient’s first point of contact, could be a key leader for change and should be as involved as the office manager.

Strategies for approaching change should also be considered. For example, finding ways to improve average providers may be more of a motivating factor for change than improving top perform-

ers, one participant said. It was noted that spread is often seen as a two-dimensional S-shaped curve¹ and is different from transformation. Perhaps cultural change in health care should not be viewed as an issue of spread, but rather transformation, requiring a multidimensional approach.

The ways in which culture and leadership are currently approached may need to be reassessed in order to drastically alter the way improvement is viewed. It was argued that widespread cultural change is unlikely to occur at this time due to current leaders’ resistance to change. A different generation of health care leaders may be needed to make the greatest strides. By making small, concrete steps, medical residents and students of health care could be trained to have different expectations of the culture of health care. However, others were not willing to sacrifice the current generation, stating that expectations must be changed now. Specifically, future leaders should be trained in an interdisciplinary manner and should be expected to engage other researchers such as behavioral, engineering, and organizational scientists. Partnerships should also be used to conceptualize how best to implement change.

One particular call for knowledge focused on quantifying opportunities for change, particularly for errors in care. The total costs to society of errors in health care, from all perspectives, are not well understood. The goal should be to minimize errors while recognizing that they cannot all be eliminated. Work should be done, an attendee commented, to change the culture of health care to one that can achieve this goal.

Comments were made supporting the need for cost-effectiveness analyses and benefit–cost analyses for performing quality improvement interventions. The ability to articulate the business case to both private and public payers will also become very important. In addition to informing decision makers, these analyses could help improve funding for quality improvement research. To attain this information, however, financial data for quality improvement research must be regularly collected and analyzed.

One person argued that health care is full of incompletely characterized problems. One way of determining the opportunity for

improvement would be to compare the current problem (e.g., take sampling of providers’ infection rates) to the theoretical limit (e.g., no infection rates). The economic value to society, including the patient’s perspective, should be assessed. What is not “perfect” is the opportunity for change. Thinking in this manner could help provide discipline in characterizing problems and finding solutions.