Accuracy of screening for inhalational anthrax after a bioterrorist attack.

BACKGROUND Bioterrorism using anthrax claimed five lives in the United States in 2001 and remains a potential public health threat. In the aftermath of a large-scale anthrax attack, mass screening to identify early inhalational anthrax may improve both the management of individual cases and the efficiency of health resource utilization. PURPOSE To develop the evidence base for outpatient anthrax screening protocols by quantifying differences in clinical presentation between inhalational anthrax and common viral respiratory tract infections. DESIGN Review, compilation, and data extraction from English-language case reports of inhalational anthrax and epidemiologic studies of influenza and other viral respiratory infections. DATA SOURCES 13 reports of 28 cases of inhalational anthrax from 1920 to 2001 and 5 studies reporting on the clinical features of 2762 cases of influenza and 1932 cases of noninfluenza viral respiratory disease. DATA SYNTHESIS Characterization of presenting clinical symptoms in anthrax and viral disease and calculation of likelihood ratios for the presence of selected clinical features. RESULTS Fever and cough do not reliably discriminate between inhalational anthrax and viral respiratory tract infection. Features suggestive of anthrax include the presence of nonheadache neurologic symptoms (positive likelihood ratio cannot be calculated), dyspnea (positive likelihood ratio, 5.3 [95% CI, 3.7 to 7.4]), nausea or vomiting (positive likelihood ratio, 5.1 [CI, 3.0 to 8.5]), and finding of any abnormality on lung auscultation (positive likelihood ratio, 8.1 [CI, 5.3 to 12.5]). In contrast, rhinorrhea (positive likelihood ratio, 0.2 [CI, 0.1 to 0.4]) and sore throat (positive likelihood ratio, 0.2 [CI, 0.1 to 0.5]) are more suggestive of viral respiratory tract infection. CONCLUSION Inhalational anthrax has characteristic clinical features that are distinct from those seen in common viral respiratory tract infections. Screening protocols based on these features may improve rapid identification of patients with presumptive inhalational anthrax in the setting of a large-scale anthrax attack.