Screening for severe acute respiratory syndrome in the emergency department

To the Editor:

The articles in the January 2004 issue of Annals by Chen et al,1., 2. Wang et al,³ and Su et al⁴ introduced novel scoring systems for predicting severe acute respiratory syndrome (SARS) in the emergency department (ED). These studies are indeed very useful first steps in developing diagnostic strategies for this new infectious disease. Nonetheless, I would like to discuss some areas of concern in interpreting and applying the results.

First, it should be noted that in all of these studies, patients were only included if they had documented fever (body temperature >38°C [>100.4°F]). Although Chen et al¹ and Su et al⁴ suggested that their scoring systems could be used in settings where mass screening for SARS is required, application will actually be limited to febrile patients with temperatures greater than 38°C (>100.4°F).

In mass screening, if patients with low-grade fever (37.5°C to 38°C [99.5°F to 100.4°F]) are not included, a substantial number of patients in the early stages of the disease will likely be missed. In a study from our ED-based SARS screening clinic during the outbreak in Hong Kong (which included afebrile patients), it was shown that 19% of SARS patients initially presented without fever, and that radiographic evidence of pneumonia often preceded fever.⁵ Therefore, a more useful tool for mass screening would have included fever or body temperature as a variable, thus allowing the evaluation of afebrile cases.²

Furthermore, the study of Wang et al³ actually enrolled only those cases that met the World Health Organization (WHO) criteria for suspected SARS. Although their scoring system was shown to have a sensitivity of 100%, this perfect figure will apply only specifically to those patients already screened as positive by the WHO criteria or case definition. Unfortunately, the WHO criteria for suspected SARS have been shown to have a sensitivity of only 26%.⁵ In practice, therefore, if 100 SARS patients presented to the ED for screening, only 26 would be correctly triaged to undergo the scoring system evaluation. Thus, relying on this strategy, despite the implementation of a near-perfect decision rule, will still result in an unacceptable number of missed cases.

Finally, the performance of chest radiography was not adequately examined in these studies. We believe that chest radiography will likely be the single most important screen for SARS in the ED setting. It has been shown to be the strongest predictor available in the ED, with an odds ratio of 17.4.⁶ During the outbreak, we performed screening chest radiography on virtually every patient with fever greater than 38°C (>100.4°F), regardless of symptomatology. It remains to be proven whether the scoring systems in these studies, given the limitations addressed by the authors, can significantly outperform chest radiography.

In addition to the multilobar infiltrates evaluated by Wang et al,³ it will be important also to analyze the predictive values for lobar, segmental, or even patchy infiltrative changes. A detailed time sequence of radiographic progression, in relationship to the progression of symptoms, may prove to be useful, as well.