A multi-component QI intervention led by a clinician champion resulted in a reduction in the number of radiographs ordered to evaluate low-back pain and an increase in the proportion of those ordered that were appropriate. In addition, there was no evidence that these imaging tests were replaced by available alternative imaging. To our knowledge this is the first evaluation of a clinician champion lead initiative that addresses thoracic and lumbar radiograph reduction in the ED setting using a multi-component QI intervention. By engaging clinicians through education, audit and feedback and providing them with a clinical decision support tool, the total number of unnecessary radiographs ordered decreased by almost 44% and of those that were ordered, the proportion that were appropriate increased by 10-fold. Over eight months we demonstrated through a multi-component intervention that this could be addressed and substantially reduced.
The results demonstrate some fluctuation in the appropriateness criteria on its trajectory of overall improvement and there may be a few causes. Our health system transitioned our hospital to a new electronic medical record the very end of July which could have made documentation more challenging the first month afterwards. Our hospital has a very significant presence of resident physicians, who rotate through the emergency department monthly necessitating repetitive in-servicing of quality initiatives. We also believe the quality of documentation improved overall from this initiative thus not having a true reverse linear relationship between appropriateness and absolute reduction.
Our results are consistent with previous studies which highlight evidence that QI interventions can address the issue of reducing image utilization. A QI intervention at a high-volume pediatric ED in California used similar methods to reduce abdominal radiographs for patients presenting with constipation [19]. Relying heavily on physician education in the form of lectures, and audit and feedback, they reduced the number of pediatric patients that received unnecessary abdominal radiographs by 50% over a 12-month period. Another QI intervention in a pediatric ED sought to reduce chest radiographs for patients presenting with bronchiolitis [20]. They employed physician education lectures and posted imaging guidelines at each computer station, and reduced radiographs for bronchiolitis patients by 44% at the end of a four-month period.
The dissemination of official guidelines alone may lead to successful outcomes in QI interventions. However, it has been shown that the addition of clinician champions led to higher rates of success. A literature review studying interventions that attempted to reduce low value care defined by Choosing Wisely found that interventions led by clinician champions reached their intended outcomes 71% of the time [21]. This underscores the importance of clinician-led initiatives when attempting to improve quality of care.
Concerned the reduction of radiographs could have the unintended consequence of an increase in CT scans of the thoracic and lumbar spine we assessed the absolute number ordered as well. As a level one trauma center that does a significant number of these CT scans, there were no demonstrated special cause variation around the median during our QI project. Further, the number of radiographs in the three months prior to the intervention demonstrate stable common cause variation.
The average physician to disposition time for patients presenting with back pain being discharged from our emergency department is approximately ninety minutes. A conventional radiograph from ordered to read time also averages approximately ninety minutes. This implies that by decreasing the utilization of radiographs, the patient throughput time improves dramatically which also has an outlying effect on the door to first provider time by freeing up space in the ED for patient evaluation and treatment.
Limitations
A limitation of this intervention is the lack of a control group. Without a control, we cannot be certain that this was not just a natural phenomenon that would have occurred without the intervention. ED volumes are dynamic and vary monthly due to multiple factors, and our study period was limited from June to January in an academic emergency department, making it possible that other factors could affect the number of radiographs ordered.
Low rates of radiograph appropriateness prior to the intervention could be due to poor documentation, but there was demonstrated consistent improvement at the same time of absolute reduction. Our institution changed from a legacy EMR (Quadramed) which utilized blank word type documents to EPIC the last week of July which may have had an impact on documentation thoroughness. It was not possible to calculate the rates of x-ray utilization in our diverse level one trauma emergency department automatically and would necessitate a substantial manual chart abstraction thus it was chosen to evaluate x-rays performed for appropriateness.
It is also possible that the observed impact of our intervention could be explained by the Hawthorn effect, in which one’s awareness they are being audited leads to changes in their behavior to improve their performance.
We did not measure the prolonged effects of this initiative to measure full sustainment but have used this opportunity to parlay additional clinician champion initiatives in other modalities leading to sustained awareness of overuse as a medical concern.