Reducing Unnecessary ‘Admission’ Chest X-rays: An Initiative to Minimize Low-Value Care

Introduction Internal medicine admission services often request a baseline admission chest X-ray (CXR) for patients already admitted to the emergency department (ED) and who are waiting for inpatient beds, despite rarely providing clinical value. Adverse consequences of such CXRs include unnecessary radiation exposure, cost, time, and false positives, which can trigger a diagnostic cascade. Extraneous CXRs performed on already-admitted ED patients can delay inpatient transfer, thereby increasing boarding and crowding, which in turn may affect mortality and satisfaction. In 2016, our ED and internal medicine hospitalist services implemented guidelines (reflecting those of the American College of Radiology) to reduce unnecessary admission CXRs. All relevant providers were educated on the guideline. The primary aim of this study was to determine if there were changes in the percentage of patients with pre-admission and admission CXRs following guideline implementation. Our secondary aim was to determine which patient characteristics predict getting a CXR. Methods All ED and internal medicine hospitalist providers were educated once about the guideline. We performed a retrospective analysis of pre- vs. post-guideline data. Patients were included if admitted to the internal medicine service during those timeframes with an admission diagnosis unrelated to the cardiac or pulmonary systems. A CXR performed during ED evaluation prior to the admission disposition time was recorded as “pre-admission,” and if performed after disposition time it was recorded as “admission.” A CXR was "unwarranted" if the admission diagnosis did not suggest a CXR was necessary. The numerator was the number of unnecessary admission CXRs ordered on patients with diagnoses unrelated to the cardiac or pulmonary systems (minus those with a pre-admission CXR); the denominator was the number of such admissions (minus those with a pre-admission CXR). Variables of interest that might influence whether a CXR was ordered were age, gender, respiratory rate ≥20, cardiac- or pulmonary-related chief complaint, ED diagnosis category, or past medical history. Results Among admitted patients with diagnoses that did not suggest a CXR was warranted, there was no change in the percentage of admission CXRs (21.7% to 25.6%, p = 0.2678), whereas the percentage with pre-admission CXRs decreased (66.6% to 60.7%, p = 0.0152). This decrease was driven by fewer CXRs being performed on patients whose chief complaint did not suggest one was indicated (p = .0121). In multivariate analysis, risk factors for an unwarranted CXR were age >40 (risk ratio (RR) = 2.9) and past medical history of cardiovascular disease (e.g., myocardial infarction, atrial fibrillation), renal disease, or hyperkalemia. Conclusion This educational initiative was not associated with the intended decrease in ordering unwarranted admission CXRs among ED boarding patients, though there was an unanticipated decrease in pre-admission CXRs. This decrease was driven by fewer CXRs being performed on patients whose chief complaint did not suggest one was indicated. Organizations interested in reducing processes with little clinical value might adopt a similar program while emphasizing the lack of benefit to admitted patients through iterative educational programs on hospital admitting services.