US Emergency Physicians Order Too Many Computed Tomography Scans—Or Do They?

Abstract

At this point, nearly everyone seems to be saying that US emergency physicians order too many computed tomography (CT) scans, exposing patients to substantial and unnecessary ionizing radiation and driving up the cost of medical care. In this issue of Annals, a study by Berdahl et al appears to provide further supporting evidence by showing that US emergency physicians order twice as many CTs as their Canadian counterparts (11.4% versus 5.9%) and that the rate of CT ordering is increasing faster in the United States (13% each year) compared with Ontario, Canada (10% each year). The assumption is that differences between health care systems result in lower, but still appropriate, CT ordering in Canada. If our Canadian colleagues are able to limit their use of CT, why can’t we? Is our frequent use of CT really excessive? Or does it represent the widespread and appropriate adoption of a powerful diagnostic tool for emergency medicine? There are limitations to the study by Berdahl et al, which restrict the conclusions we can draw from it. It does not determine whether US emergency physicians order too many CTs or Canadians order too few, or even whether the CT ordering rates are really that different. The shortcomings stem from (1) lack of data on patient outcomes necessary to assess the appropriateness of CT ordering; (2) differences between the US and Canadian emergency department (ED) populations (eg, US ED patients appear to be sicker); and (3) differences between heath care systems (eg, availability of CT). The data sets used in this article cannot be used to assess the appropriateness of CT ordering. Although they provide information on various aspects of care for a given ED visit, including test ordering, reason for visit, and basic demographics, they lack patient-level data about the diagnoses being considered by the treating physicians and patient outcomes during and particularly after the ED visit—information needed to assess the appropriateness of test ordering. Furthermore, by looking only at CT rates, the use of alternative imaging modalities (radiography for cervical spine trauma, lung scan for pulmonary embolism, and ultrasonography for appendicitis, especially in children) or management strategies (hospitalization for M