Discussion

Because the spectre of bioterrorism moved from a possibility to a reality in the fall of 2001, many voices have called for a proactive approach to providing clinical guidelines for the diagnosis and treatment of IA, as well as of other agents that could potentially be used in a bioterrorist attack [1–7]. The study by Danzig [6] calls specifically for the development of such guidelines, as well as for the development of training scenarios for responding to outbreaks involving agents that might potentially be used in future bioterrorist attacks. Fortunately, as the 2001 outbreak showed, IA is a serious but treatable disease if it is detected and treated early in its course.

The development of screening and treatment protocols increases in importance when one considers the phenomenon of “reload,” as delineated by Danzig [6] in a penetrating and insightful study commissioned by the National Defense University for the Department of Defense. Reload refers to the likely ability of attackers using biological weapons to avoid detection and “stockpile or replenish resources that permit repeated attack” [6, p. 2]. Thus, although making a kilogram of weaponized anthrax in the 1–5 micron range presents certain technical challenges, as Danzig indicates, “it is not a significant challenge for a terrorist organization that can make a kilogram to make 10 or 100 kilograms” [6, p. 2]. The phenomenon of reload thus raises the unpleasant but distinct possibility of large-scale atmospheric distribution of anthrax spores in multiple cities, either simultaneously or sequentially. Without planning scenarios and screening and treatment protocols, the surge capacity of the health care system (and of emergency departments, in particular) would be overwhelmed. Simply stated, given the phenomenon of reload, without effective screening and treatment protocols, our health care system would go quickly from “reload” to “overload.” Our study evaluated existing screening and treatment protocols for patients who had possibly been exposed to IA by comparing them with a population of patients seen at one of the primary treatment facilities for patients with IA in the Washington, D.C., area after the October 2001 bioterrorism-related outbreak.

The criteria proposed by Mayer et al. [1] (the Inova Fairfax protocol) and by Hupert et al. [2] (the Cornell protocol) were each attempts to provide clear and concise screening mechanisms for the large number of patients likely to present for diagnosis and therapy after a bioterrorism-related outbreak. In an attempt to clarify the utility and cost of using the Inova Fairfax and Cornell protocols to screen and identify patients for treatment after possible exposure to Bacillus anthracis, we assessed the ability of these protocols to identify cases of IA in a busy emergency department during the 2 weeks following initial identification of patients with IA in the Washington, D.C., area. Both protocols identified the documented cases of IA seen at our institution, but the Inova Fairfax criteria would have screened only 4 patients at a cost of $1900, whereas the Cornell criteria would have screened 273 patients at a cost of $126,025. In the editorial accompanying the Cornell protocol, Sox [8] noted that “If a bioterrorism attack occurs in the meantime, we have a triage algorithm that seems safe, but is probably inefficient” (p. 1). Our study supports that statement, because 26% of patients with possible signs or symptoms of IA would have been screened by the Cornell protocol, compared with the ∼0.5% of patients who would have been screened by the Inova Fairfax screening guidelines.

Using a simple checklist of IA symptoms offers several advantages over the 3-tier algorithm. First, a checklist approach assures that each of the recognized symptoms of IA is assessed in each patient, and, therefore, it better fits the concepts of syndromic surveillance. Second, the checklist approach avoids the necessity of an algorithm with multiple branch-points and instead focuses on the simple issue of labeling each patient as having or not having ⩾5 symptoms plus fever and tachycardia. Third, this analysis indicates that the Inova Fairfax criteria are more accurate and cost-effective than the 3-tier algorithm, because only 2 additional patients would have been screened if fever and tachycardia had been included among the symptoms, and only 27 additional patients would have been screened if fever and tachycardia had been deleted completely from the list of clinical factors. In contrast, the Cornell protocol would have resulted in 271 patients being screened who did not have the disease, with charges of $126,025.

Although the primary purpose of our study was not to address the effect of occupational or environmental exposure on screening and treatment, we did find that 250 (22%) of the 1127 patients evaluated were either postal employees or worked in the Hart Senate Office Building (i.e., the site of the anthrax-laden letter addressed to Senator Daschle.) Of these patients, 239 had ⩽4 clinical symptoms of IA. However, 2 postal employees had ⩾5 symptoms plus fever and tachycardia and had PCR-confirmed diagnoses of IA. In this admittedly limited sample, these data suggest that the use of occupation alone as a screening condition would have performed similarly to use of either the Cornell protocol or the criterion of ⩾3 clinical symptoms of IA.

This study, like most, is subject to certain limitations. First, although it assesses the accuracy and cost among the population presenting to a specific emergency department, it is possible that these results may not apply to other populations. Second, because a rigorous patient questionnaire was not used proactively in the evaluation of patients with potential IA, it is possible that patients who appeared to be less seriously ill were questioned less extensively and may, in fact, have had more symptoms of IA than were captured in the medical record. If such questionnaires had been used, there might well have been more patients identified with ⩾5 symptoms of IA, thereby increasing the percentage of patients screened. This screening approach (i.e., use of the tachycardia and fever criteria) should not be applied to patients with a heart block or a pacemaker, those who have received β-blockers, or those who have ingested antipyretics within 6 h before presentation. Finally, we do not know whether the criteria of ⩾5 symptoms plus fever and tachycardia would apply in other outbreaks involving different strains of B. anthracis.

Notwithstanding these limitations, this study supports the concept that patients who may have been exposed to anthrax spores during this bioterrorism-related outbreak could have been safely screened for diagnosis and treatment on the basis of the presence or absence of ⩾5 clinical symptoms of IA plus the presence of fever (temperature, ⩾38°C) and tachycardia. This study also suggests that the proactive development of similar screening and treatment protocols that might be used in responding to a bioterrorism attack should be developed and disseminated to primary care physicians and emergency department physicians, as suggested by Danzig [6] and others.