Medical overuse, defined as delivery of a health care service for which the potential for harm exceeds the possible benefit, presents an ongoing challenge in emergency medicine . In particular, fear of missing a serious diagnosis may drive decisions to order unnecessary imaging, even when the ordering physician is aware that testing is unlikely to yield a positive result . In pediatrics, the relative radiation sensitivity and long life expectancy of children make the case for judicious use of imaging in trauma particularly compelling; nonetheless, substantial practice variation exists .
In this issue of CJEM, Beno et al. address the issue of overuse of CT imaging for pediatric trauma patients using the Model for Improvement . The Model for Improvement (Fig. 1) is one of the most foundational tools for practitioners of quality improvement (QI). Understanding this model and following its steps can help teams achieve substantial advances and avoid common pitfalls in their QI work. Developed by Associates in Process Improvement in 1996 and adopted by the Institute for Healthcare Improvement , the Model for Improvement begins with three consecutive questions. While they seem intuitive, there is much to be gained by giving them careful consideration.
The first question is simply, “What are we trying to accomplish?” At first glance, the question seems so basic as to not require significant contemplation. However, strong QI methods start with a deep understanding of a problem rather than starting with a proposed solution. A common QI pitfall is to craft a single solution to a perceived but insufficiently understood problem, and then to measure the effectiveness of that solution alone. This approach often results in cessation of improvement efforts after testing the proposed solution, regardless of whether the problem has been sufficiently addressed. Beno et al. begin their work by measuring their problem and benchmarking against published norms. They proceed to develop a deep understanding by conducting a root cause analysis, which revealed important improvement opportunities such as low staff familiarity with published decision rules, a trauma culture not aligned with observation instead of immediate imaging in low-risk patients, and a need for institutional guidelines. These learnings were then incorporated into a key driver diagram that informed subsequent steps. The improvement aim was carefully selected based on a review of recent cases and determining the proportion of patients at very low risk of intra-abdominal injury for whom CT could be safely avoided.
The second question in the Model for Improvement is “How will we know that a change is an improvement?” This refers to the selection of measures. Typically, a methodologically sound QI project will include a set of relevant process and outcome measures, as well as balancing measures to identify potential unintended consequences of change. In the article by Beno et al., the selected measures assess performance and rule out serious unintended consequences such as return visits and are displayed in statistical process control charts. This analytic approach has many advantages over simple before/after analysis, including illustration of background trends and variation, detection of special cause variation, and measurement of real-time progress toward the improvement aim.
The third question in the Model for Improvement is “What change can we make that will result in improvement?” Planning change is both a science and an art. Changes can be informed by published evidence, and they can be developed using proven tools such as key driver diagrams. They also may require creativity and strong change management skills. Use of embedded decision support and a forced-function requirement to provide an indication on the CT requisition by the authors of this study are established strategies that promote high reliability. The creation of a locally adapted evidence-based decision tool is an example of a creative change management strategy that undoubtably enhanced stakeholder engagement.
After addressing these three questions, the next step in the Model for Improvement is to initiate Plan–Do–Study–Act (PDSA) cycles. PDSA cycles support testing and learning from change to optimize performance. Each change may require multiple PDSA cycles to achieve acceptable performance prior to full implementation. The authors describe iterative refinement of changes during PDSA cycles. New tests of change are continued until aims are achieved, at which point sustainability planning embeds successful changes into the new system. The team was able to achieve the improvement aim of an absolute reduction of 20% in abdominal/pelvic CT imaging in pediatric trauma patients while increasing the CT yield for positive intra-abdominal injury from 38 to 64%, without the occurrence of any clinically significant missed intra-abdominal injuries. Sustainability planning, which included transitioning paper requisitions to a new electronic health record, incorporating the guidelines into Choosing Wisely criteria, and planning of ongoing education, was likely an important factor in the ability of the team to sustain improvements over 2 years.
Many emergency physicians believe that too many imaging tests are ordered in their EDs, and they report complex contributing factors including fear of missing a diagnosis, malpractice worries, time pressures, and local practice norms . These are systemic problems that are unlikely to be adequately ameliorated by projects that consist of only a single intervention. The potential harm of failing to address overuse of CT in pediatric patients is substantial. Using the Model for Improvement, QI teams are guided to take a problem-focused, measurement-based, and multifaceted approach that allows learning from change. Carefully executed quality improvement projects, such as the one described by Beno et al., will be essential in reducing the harms associated with over-testing in our emergency departments.