Discussion

This retrospective study of 11 patients with bioterrorism-related IA indicates that only 1 of 11 patients would have been evaluated and treated under interim CDC guidelines. The CDC interim guidelines (figure 1) recommend evaluation for patients with "history of exposure, or occupational/environmental risk with two to five day illness" (emphasis added) [9, p. 946] involving 8 symptoms and 1 clinical sign (fever). Perhaps because the guidelines were directed specifically at patients with known exposure or identified occupational/environmental risks, they require both epidemiological risk factors and a clinical presentation consistent with IA, a requirement that is the primary reason for the inability of the current guidelines to clearly identify patients in this outbreak for screening and treatment. A second reason is the failure to include persistent tachycardia as a clinical sign of the disease. The third reason is the guidelines' recommendation that clinical signs and/or epidemiological confirmation of exposure be present before antimicrobial treatment is administered to such patients. Each aspect deserves careful consideration.

Reliance on epidemiological plus clinical factors. The current CDC guidelines require a known "history of exposure or occupational/environmental risk with a two to five day illness" consistent with IA [9, p. 946]. This approach might be effective in a bioterrorism incident in which the at-risk population is more clearly identified. In contexts in which the risk factors for exposure are less certain, the CDC guidelines appear to be less useful. All 10 patients whom the CDC guidelines failed to identify for screening and treatment lacked a clearly known exposure or known occupational/environmental risk at the time of their initial clinical presentation. Clearly, the epidemiological factors suggesting an exposure to a bioterrorism agent evolve over the course of an investigation. In the Washington, D.C., anthrax bioterrorism attack of 2001, for example, at the time that the initial clinical diagnoses of IA were made for patients 3 and 4 (19 and 20 October 2001, respectively), epidemiological investigations to that point had limited the area of known exposure to people who had been on the fifth and sixth floors of the southeast wing of the Hart Senate Office Building on 15 October 2001 from 9 A.M. to 7 P.M. [14].

Only after the clinical diagnosis of IA was made in patient 3 on 19 October 2001 was it known that the Brentwood Postal Facility constituted an additional at-risk location and that others outside the Senate Office Building had been exposed. Patient 4 (a second Brentwood postal worker who presented on 20 October 2001) would have been identified by the current CDC guidelines. It is important to note that patients 3 and 4 presented to the same health care facility, where a clinical diagnosis of IA was made and treatment was instituted. Although confirmation of B. anthracis infection by PCR testing occurred on 21 October 2001, an internal communication network established within the hospital and its emergency department had communicated the clinical diagnosis and laboratory and imaging findings of IA in patient 3 (figure 2). Two additional Brentwood workers (patients 5 and 6) first presented at other health care facilities on 21 October 2001 but before the announcement that IA had been confirmed by PCR testing, because widespread media coverage of the confirmation of IA came via television and print coverage on the afternoon of 21 October 2001 and the morning of 22 October 2001, respectively [15, 16].

Several days later, an additional postal worker whose workplace received mail from Brentwood became ill and was later confirmed to have IA (patient 7). On 17 October 2001, a Brentwood postal worker (patient 5) presented to his primary care physician with symptoms of fever and myalgias, a temperature of 38.9°C, and a pulse rate of 79 beats/min. This case would not have been identified at the initial clinical visit either by epidemiological factors (because Brentwood was not known to have been exposed to B. anthracis on 17 October 2001) or by clinical findings, and, therefore, it would not have been identified by either the CDC interim guidelines or the proposed revised guidelines we have suggested.

Therefore, the absence of specific, confirmed epidemiological evidence of exposure should not, in and of itself, exclude consideration of IA if, in the context of a possible bioterrorist attack, the clinical presentation, laboratory findings, and radiographic studies raise significant clinical suspicion of the disease. Indeed, these data indicate that patients with ⩾5 clinical symptoms of IA and signs of fever and persistent tachycardia should be evaluated and treated if there is reasonable suspicion on the part of the clinician that there is or may have been exposure to an agent of bioterrorism, rather than waiting for definite confirmation of epidemiological risk.

Tachycardia as a clinical sign. The only clinical sign included in the current CDC guidelines is fever [9]. Seven of 11 patients had fever, but the mean presenting temperature was only 38.0°C, and all 3 patients who had temperatures of >38.5°C died. The clinical presentation of patients in the first phase of the disease more often included low-grade fever, as opposed to higher temperatures in the patients who presented in the advanced phase of the disease, all of whom died. Although previous reports and commentary [2, 17, 18] have noted the importance of tachycardia out of proportion to the clinical symptoms in these patients, tachycardia is not currently listed as a presenting sign in the CDC guidelines. All but one patient in this series had persistent tachycardia, with a mean presenting heart rate of 116 beats/min.

Reliance on clinical observation before confirmation of exposure. The CDC interim guidelines indicate that patients who do not have both a documented history of exposure or occupational/environmental risk and a 2–5 day illness consistent with IA should be "observed closely" and should be "provided antimicrobial prophylaxis if exposure is confirmed" [9, p. 946]. Among survivors in this small case series of IA, the mean interval from onset of symptoms to receipt of treatment with antibiotics known to be effective against B. anthracis was 4.7 days. Among those who died, the mean interval between onset of symptoms and receipt of appropriate antibiotic treatment was 5.8 days. This suggests that the window of opportunity for treating IA in symptomatic patients is narrow—perhaps as short as 24 h. Four patients who were not recognized to have IA upon first presentation to health care facilities were subsequently determined to have the disease. Two patients (patients 5 and 6) were not treated with antibiotics and later died. Both patients had presented to health care facilities in the second, fulminant stage of the disease and were then treated with antibiotics known to have activity against B. anthracis. Neither of these patients would have been identified by current CDC guidelines, and one (patient 5) would not have been identified by the revised guidelines. The 2 patients who were originally discharged from the hospital but survived (patients 7 and 8) had been given ciprofloxacin or levofloxacin therapy and had blood samples obtained for culture at the time of the first visit; however, neither patient would have been identified by the CDC guidelines for screening. Patient 7 was called back to the emergency department 17 h after being first seen when culture of a blood sample that had been obtained before antibiotic administration grew B. anthracis. He had only taken a single dose of ciprofloxacin. Patient 8 was provided oral levofloxacin upon initial discharge and survived, although a definitive diagnosis by blood culture and initiation of intravenously administered antibiotic therapy did not occur until 2 days later.

Although this is a small case series, the fact that 2 patients who received oral antibiotic therapy soon after onset of symptoms survived and 2 patients for whom antibiotic therapy was relatively delayed died suggests that interruption of the progression of the disease by appropriate antibiotic therapy at an early stage may be essential for successful treatment. All patients who presented with the advanced stage of the disease died despite receipt of aggressive therapy. Thus, this episode of bioterrorism-related IA suggests that a survival rate of 55% is attainable only when early identification of such patients is followed by aggressive and rapid treatment with antibiotics known to have activity against B. anthracis.

Suggested revisions to IA guidelines. On the basis of this review, the panel suggested revisions to the CDC interim guidelines (figure 3). Because of the inherent limitations in accurately and rapidly ascertaining those at risk for exposure to B. anthracis spores within the setting of bioterrorism (as illustrated by the outbreak of infection in Washington, D.C.), either epidemiological factors or a clinical presentation consistent with IA should provoke careful consideration of a clinical diagnosis of B. anthracis infection by appropriate laboratory and radiologic investigation of the disease, consisting, at a minimum, of blood cultures, chest radiography, and, when indicated, chest CT. Gram staining of the buffy coat and analysis of pleural effusions after thoracentesis may also be helpful. On the basis of review of these patients, a clinical presentation consistent with IA was defined by the panel as ⩾5 symptoms of the disease (fever/chills, fatigue, cough, dyspnea, nausea/vomiting, sweats, myalgias, chest discomfort, headache, abdominal pain, or confusion) plus clinical signs of fever and persistent tachycardia (heart rate, ⩾100 beats/min). In such cases, clinicians should consider whether the patient may have had a history of exposure to B. anthracis, including occupational and/or environmental risk. In addition to evaluating such patients, clinicians should alert appropriate public health authorities to their clinical suspicion of IA, so that appropriate epidemiological and public health investigations can be pursued.

Our analysis indicates that patients with either epidemiological exposure or clinical signs and symptoms of the disease should receive treatment for IA with appropriate antibiotics until blood culture results are negative and the patients' subsequent clinical course can be determined. However, we are not suggesting that patients with this constellation of clinical signs and symptoms be treated unless there is reasonable suspicion on the part of the treating physician that bioterrorism or other exposure to B. anthracis may have occurred.

One potential limitation of the proposed revised guidelines is the difficulty in differentiating patients with IA from those with influenza-like illnesses [10]. There is significant overlap of symptoms between the 2 entities, and patients with IA typically did not appear to be acutely ill during the initial phase of the disease. The seasonal and highly communicable nature of influenza-like illnesses is well known but is of no particular help in distinguishing individual patients. As noted by the CDC, rapid assays for influenza lack sufficient sensitivity (range, 45%–90%) and specificity (range, 60%–95%) to be of clinical usefulness in such settings [10].

The proposed guideline revisions would likely result in some patients who do not have IA being evaluated and treated until their blood culture results are known. How many patients would be treated, what complications they might face (including adverse reactions to antibiotics), and the potential cost of such screening and treatment are currently unknown. We are currently analyzing all patients presenting to the emergency department of a hospital that was one of the primary Washington, D.C.-area treatment facilities during the fall 2001 attack to determine how many patients would have been evaluated and treated under the suggested guideline revisions.

On the basis of these findings, we recommend that patients who have a clinical presentation consistent with IA presenting in a setting of known or suspected bioterrorism exposure receive antibiotic therapy, pending the findings of blood cultures, chest radiographs, and, when indicated, chest CTs. In all such cases, the treating clinician should consider the possibility of exposure to B. anthracis and obtain an appropriate occupational/environmental history. Close clinical follow-up for evidence of progression of the disease should also be arranged, and appropriate public health agencies should be informed regarding any patients for whom a clinical diagnosis of IA has been made or is suspected.

Clinicians faced with suspicion of IA in individual patients during such bioterrorist attacks may need to apply different thresholds for treatment than those recommended by public health guidelines, because understanding of the epidemiological risk factors associated with B. anthracis exposure necessarily evolves over time. The 2001 attack demonstrated the importance of rapid communication between front-line practitioners and public health agencies regarding this evolving understanding of the clinical presentation, risk factors for exposure, and appropriate treatment of IA. Improving communication systems between emergency department physicians, primary care physicians, infectious diseases specialists, and public health agencies should be a priority in future outbreaks of infection [19]. Because bioterrorism now looms as a distinct reality in our society, the intersection between public health guidelines and the application of clinical judgment regarding treatment decisions of potential victims of bioterrorism deserves continued consideration, study, and dedication of resources.