Skip Navigation

Clinical Features that Discriminate Inhalational Anthrax from Other Acute Respiratory Illnesses

  1. Matthew J. Kuehnert1,
  2. Timothy J. Doyle5,
  3. Holly A. Hill1,
  4. Carolyn B. Bridges2,
  5. John A. Jernigan1,
  6. Peter M. Dull3,6,
  7. Dori B. Reissman4,
  8. David A. Ashford3, and
  9. Daniel B. Jernigan1
  1. 1Division of Healthcare Quality Promotion, National Center for Infectious Diseases, Atlanta, Georgia
  2. 2Division of Viral and Rickettsial Diseases, National Center for Infectious Diseases, Atlanta, Georgia
  3. 3Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, Atlanta, Georgia
  4. 4Bioterrorism Preparedness and Response Program, National Center for Infectious Diseases, Atlanta, Georgia
  5. 5Division of Public Health Surveillance and Informatics, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta, Georgia
  6. 6Division of Applied Public Health Training, Epidemic Intelligence Service Branch, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta, Georgia
  1. Reprints or correspondence: Dr. Matthew J. Kuehnert, Div. of Healthcare Quality Promotion, National Center for Infectious Diseases, 1600 Clifton Rd., Mailstop A-35, Atlanta, GA 30333 (mkuehnert{at}cdc.gov).
 
Next Section

Abstract

Inhalational anthrax (IA) is a rapidly progressive disease that frequently results in sepsis and death, and prompt recognition is critical. To distinguish IA from other causes of acute respiratory illness, patients who had IA were compared with patients in an ambulatory clinic who had influenza-like illness (ILI) and with hospitalized patients who had community-acquired pneumonia (CAP) at the initial health care visit. Compared with patients who had ILI, patients who had IA were more likely to have tachycardia, high hematocrit, and low albumin and sodium levels and were less likely to have myalgias, headache, and nasal symptoms. Scoring systems were devised to compare IA with ILI or CAP on the basis of strength of association. For ILI, a score of ⩾4 captured all 11 patients with IA and excluded 664 (96.1%) of 691 patients with ILI. Compared with patients who had CAP, patients with IA were more likely to have nausea or vomiting, tachycardia, high transaminase levels, low sodium levels, and normal white blood cell counts. For CAP, a score of ⩾3 captured 9 (81.8%) of 11 patients with IA and excluded 528 (81.2%) of 650 patients with CAP. In conclusion, selected clinical features of patients with IA differ from those of patients with ILI and are more similar to those of patients with CAP.

During the period of 4 October through 20 November 2001, 22 persons in the United States were reported to have infection caused by Bacillus anthracis [1]. B. anthracis spores cause disease by entering the body through breaks in the skin, by ingestion, or by inhalation, resulting in cutaneous, gastrointestinal, or pulmonary anthrax, respectively [2]. Pulmonary, or inhalational, anthrax (IA) is the most lethal form of the disease; after spread to mediastinal lymph nodes, bacteremia with sepsis, meningitis, and death often rapidly follows [3]. Of 11 patients with confirmed IA, all became critically ill, and 5 died [4, 5]. For these patients, nonspecific symptoms, including fever and cough, marked the early stage of illness, making it difficult to distinguish IA from other common causes of acute respiratory infection.

Direct exposure to envelopes containing B. anthracis spores or to spore-contaminated postal equipment was the probable mode of disease acquisition for 9 of the 11 patients with confirmed IA; for 2 others, the nature of exposure was unknown [6]. The possibility of future use of B. anthracis as a bioterrorist agent (delivered through the mail or via other new criminal methods of dissemination) makes it prudent to include IA in the differential diagnosis of acute respiratory illness if there are epidemiological risk factors for exposure. Seasonal outbreaks of acute respiratory illness may make identification of bioterrorism-related clusters of IA even more difficult, particularly if risk factors for exposure are unknown at the time of attack [7].

The clinical management of acute respiratory illnesses can vary widely depending on the etiology. The majority of acute respiratory illnesses for which an etiology can be determined are caused by viral pathogens (e.g., influenza virus, parainfluenza virus, respiratory syncytial virus, rhinovirus, and adenovirus) [8, 9]. These viral infections are grouped together under the term “influenza-like illness” (ILI) because of their similar presenting symptoms. ILI is generally regarded as upper respiratory in origin and has a self-limited course, although lower respiratory tract infection, including secondary bacterial infection, can occur [10–13]. In contrast, community-acquired pneumonia (CAP) requires medical management, including antimicrobial treatment and, in many instances, hospitalization; in adults, when an etiology can be determined, it is most often caused by Streptococcus pneumoniae or other bacterial pathogens [14]. Depending on disease stage at the initial health care visit, IA may have clinical features of either ILI or CAP. Rapid treatment and supportive management, including appropriate antimicrobial therapy, may be critical early in the clinical course [4]. Because rapid diagnostic tests for IA currently are not available, selected clinical features, in addition to evaluation of exposure history, may be important to aid clinicians in discriminating IA from other causes of acute respiratory illness at the initial visit.

We used existing databases to identify clinical features that differentiate IA from other causes of acute respiratory illness. Specifically, we compared signs, symptoms, and laboratory values for patients with IA, patients in an ambulatory clinic who had ILI, and patients who required hospitalization for CAP, to determine which clinical features were most useful in discriminating IA from other common causes of acute respiratory illness.

Previous SectionNext Section

Methods

Patients with IA were identified through clinician reports and active and passive surveillance by state and local public health authorities. The diagnosis of IA was confirmed by laboratory testing at the Centers for Disease Control and Prevention (CDC; Atlanta, GA) [15]. Data regarding signs, symptoms, and laboratory testing values (i.e., hematology and serum chemistry) measured at the time of initial health care visit were reviewed and extracted from patient medical records by the authors. Characteristics of patients who had IA were compared with those of patients who had ILI or CAP.

Data sources for comparison groups. Data for comparison with patients who had IA were retrospectively obtained from 2 independent databases. First, data for 2203 patients at an ambulatory clinic who had influenza or other causes of ILI were combined from 5 clinical trials designed to evaluate the effectiveness of zanamivir (Relenza; GlaxoSmithKline) for the treatment of influenza. The trials were conducted and funded by GlaxoSmithKline from 1994 through 1998. All enrolled patients had either subjective or objective evidence of fever (i.e., feverishness or a temperature of >37.8 °C) and ⩾2 of the following symptoms at the time of initial clinic visit: headache, myalgias, cough, or sore throat. The following variables were examined: temperature; heart rate; WBC count and differential; hematocrit; hemoglobin level; platelet count; serum levels of sodium, potassium, alanine aminotransferase (ALT), aspartate aminotransferase (AST), bilirubin, albumin, calcium, blood urea nitrogen (BUN), and creatinine; and the presence or absence of cough, sore throat, nasal symptoms (i.e., rhinorrhea or nasal congestion), headache, myalgias or arthralgias, and fatigue.

Respiratory specimens were obtained from all patients for culture for influenza virus, and they were obtained from selected patients for PCR. In addition, blood samples were obtained at the time of study enrollment and again 3–4 weeks later for serological testing for antibodies to influenza virus by hemagglutination inhibition. Persons with either a 4-fold increase in hemagglutination inhibition titer, a positive result of culture for influenza virus, or a positive PCR result were defined as having laboratory-confirmed influenza [16].

The second comparison database included 784 patients enrolled in the North American Pneumonia Etiology Study at 34 sites in the United States and Canada from 1995 through 1999 (funded by Pfizer and conducted in collaboration with the CDC). All patients enrolled in the study were >16 years of age and were hospitalized because of acute CAP confirmed by chest radiography. Blood cultures were performed for all patients in the study. Variables examined at the time of hospital admission included the following: temperature; heart rate; WBC count and differential; serum levels of sodium, ALT, AST, calcium, and BUN; and the presence or absence of cough, sore throat, nasal symptoms, headache, myalgias or arthralgias, abdominal pain, diarrhea, nausea or vomiting, pleuritic chest pain, dyspnea, and chills.

Data analysis. Data from the 3 sources were stripped of personal identifiers and combined for comparison. For control subjects, we included only patients aged ⩾40 years to more closely approximate the age distribution of patients with IA. On univariate analyses, 2-way comparisons were done between patients who had IA and each comparison group by Fisher's exact test. Patients who had IA were compared with all patients who had ILI and with subgroups of patients who had laboratory-confirmed influenza or some other ILI. Patients who had IA also were compared with all patients who had CAP and with subgroups of patients for whom blood cultures did or did not yield S. pneumoniae. Symptoms were analyzed as dichotomous variables (i.e., present or absent). Quantitative vital sign and laboratory measurements were analyzed as dichotomous variables by using cut points, as defined by generally accepted normal clinical ranges or published laboratory reference intervals and values (table 1). For hematocrit and hemoglobin level, sex-adjusted normal values were used [17, 18]. Tachycardia was defined as a heart rate of >100 beats/min.

Figure 1
View larger version:
Figure 1

Receiver operating characteristic curve displaying sensitivity versus specificity for scoring system comparing clinical features of patients with inhalational anthrax and those of ambulatory clinic patients with influenza-like illness. ●, Score.

Figure 2
View larger version:
Figure 2

Receiver operating characteristic curve displaying sensitivity versus specificity for scoring system comparing clinical features of patients with inhalational anthrax and those of hospitalized patients with community-acquired pneumonia. ●, Score.

Table 1
View larger version:
Table 1

Reference ranges against which all clinical laboratory measurements were compared in a study of inhalation anthrax versus other causes of acute respiratory illness.

Variables were included in multivariable analyses if they were found to be predictors (i.e., P < .10) on univariate analysis when compared with those of patients who had ILI or CAP. These variables were added to multivariable stepwise logistic regression models to generate a logistic discriminant function that maximally differentiated patients with IA from patients with ILI or CAP. Two separate models were run for each comparison with ILI and CAP; the first included only variables for which data were complete for a minimum of 9 patients with IA, and the second included variables for which data were complete for a minimum of 10 patients with IA. Variables for which data were missing for ⩾3 patients with IA were excluded from all logistic regression models. Analyses were done with SAS, version 8.02 for Windows (SAS Institute).

After logistic discriminant functions were generated for each comparison, a simple additive scoring system was designed by allocating risk factor points for each variable proportionate to strength of association for each variable identified as significant by either model. Because of extreme variability associated with the estimated coefficient for each variable in the model, the lower limit of the 95% CI of the OR was chosen to estimate the strength of association. A lower limit of the 95% CI of the OR of ⩾1.5 was taken to indicate a stronger association, whereas a value of <1.5 was considered to reflect a weaker association. Separate scoring systems were devised and applied for comparing IA with ILI and comparing IA with CAP. Receiver operating characteristic curves were used to describe sensitivity and specificity of classification for each scoring system [19].

Previous SectionNext Section

Results

Of 1352 patients in the combined database, 11 (0.8%) had IA, 691 (51%) had ILI, and 650 (48%) had CAP. Four hundred seventy-three patients with ILI (69%) had laboratory-confirmed influenza. Forty-three patients with CAP (6.6%) had pneumococcal bacteremia. Patients with ILI were younger than those with IA (median age, 49 vs. 56 years, respectively; P = .01); patients with IA did not significantly differ from patients with CAP with regard to age (median age, 56 vs. 68 years, respectively; P = .1). No significant differences were noted in sex distribution between patients with IA and those with ILI (7 [64%] of 11 vs. 357 [52%] of 691 were male, respectively; P = .63) or between patients with IA and those with CAP (7 [64%] of 11 vs. 390 [60%] of 650 were male, respectively; P = 1.0). Patients with IA were less likely to have reported current smoking than were patients with CAP (0 of 11 vs. 212 [33%] of 650, respectively; P = .02); the frequency of current smoking for patients with IA and those with ILI was not significantly different (0 of 11 vs. 129 [19%] of 691, respectively; P = .23). The median duration between symptom onset and the initial health care visit was longer for patients with IA than it was for those with ILI (median duration, 72 vs. 28 h, respectively; P < .05), but there was no difference for patients who had IA compared with those who had CAP (median duration, 72 vs. 120 h, respectively; P > .05).

Comparison by univariate analysis of signs and symptoms for patients with IA versus those with ILI or CAP is shown in table 2. In summary, patients with IA were significantly less likely than patients with ILI to have sore throat, nasal symptoms, headache, or myalgias; in contrast, presence of these symptoms was similar for IA and CAP groups. There was no significant difference in the presence of fever or cough between patients with IA and those with ILI or CAP. Tachycardia was more common among patients with IA than among those with either ILI or CAP. Patients with IA were significantly more likely than those with CAP to report nausea or vomiting or chest pain (e.g., chest discomfort or pleuritic chest pain).

Table 2
View larger version:
Table 2

Comparison of signs and symptoms in patients with inhalational anthrax (IA) versus those of ambulatory clinic patients with influenza-like illness (ILI) and hospitalized patients with community-acquired pneumonia (CAP).

Comparison by univariate analysis of laboratory findings for patients with IA versus those with ILI or CAP is shown in table 3. In summary, patients with IA were more likely to have an abnormally high WBC count than were patients with ILI, but they were significantly less likely to have a high WBC count than were patients with CAP. Patients who had IA were more likely to have high transaminase (i.e., ALT and AST) levels and low sodium levels than were patients with either ILI or CAP. Patients with IA also were significantly more likely to have high levels of BUN, low platelet counts, high serum bilirubin levels, low serum calcium levels, and low serum albumin levels than were patients with ILI. When patients with IA and the subset of patients with CAP and pneumococcal bacteremia were compared, the presence of tachycardia and hyponatremia no longer remained significant.

Table 3
View larger version:
Table 3

Comparison of laboratory findings for patients with inhalational anthrax (IA) versus ambulatory clinic patients with influenza-like illness (ILI) and hospitalized patients with community-acquired pneumonia (CAP).

On multivariable analyses comparing patients with IA and those with ILI, patients with IA were less likely to have myalgias, headache, or nasal symptoms and were more likely to have tachycardia, low serum sodium levels, high hematocrit or hemoglobin levels, and low serum albumin levels (table 4). A scoring system was devised by allocating risk factor points proportionate to strength of association to the 7 significant variables from the multivariable analyses comparing patients who had IA with those who had ILI. Tachycardia, low serum albumin levels, and lack of nasal symptoms were found to have a relatively stronger association and were assigned 2 points each, whereas low serum sodium levels, high hematocrit or hemoglobin levels, lack of headache, and lack of myalgias had a weaker association and were assigned 1 point each (table 5). Scores for IA and ILI were significantly different (median score for IA vs. ILI, 6.0 vs. 0; P < .0001). In a comparison of patients who had a score of ⩾4 versus those who had a score of <4, a score of ⩾4 was found to capture all 11 patients with IA (i.e., 100% sensitivity) while excluding 664 of 691 patients with ILI (i.e., 96.1% specificity; figure 1). Use of alternative score cut points did not improve sensitivity or specificity. The sensitivity and specificity did not change meaningfully when the analysis was restricted only to those patients without missing data for any variable or when restricted to the subset of patients who had laboratory-confirmed influenza.

Table 4
View larger version:
Table 4

Multivariable analysis comparing clinical features of patients who had inhalational anthrax (IA) with those of ambulatory clinic patients who had influenza-like illness.

Table 5
View larger version:
Table 5

Variables and score allocation used to construct scoring system comparing clinical features of patients who had inhalational anthrax with those of ambulatory clinic patients who had influenza-like illness.

On multivariable analysis comparing patients with IA and those with CAP, patients with IA were more likely to have nausea or vomiting, tachycardia, elevated transaminase levels, hyponatremia, and a normal WBC count (table 6). A second scoring system was devised by allocating risk factor points to the 5 significant variables on multivariable analyses comparing patients who had IA with those who had CAP. All 5 variables were determined to have approximately equal strength of association and thus were weighted 1 point each (table 7). Scores for patients with IA were significantly different from those for patients with CAP (median score for IA vs. CAP, 4.0 vs. 2.0; P < .0001). In a comparison of patients with a score ⩾2 versus those with a score of <2, a score of ⩾2 was found to capture all 11 patients with IA (i.e., 100% sensitivity) while excluding 310 of 650 patients with CAP (i.e., 48% specificity); when the cut point was changed to a score of ⩾3, sensitivity decreased to 82% but specificity increased to 81% (figure 2). Comparison of patients with IA and those with CAP and pneumococcal bacteremia separately decreased specificity further (33% for a score of ⩾2 and 70% for a score of ⩾3; receiver operating characteristic curve not shown). Sensitivity and specificity did not change meaningfully when the analysis comparing patients with IA versus those with CAP was restricted only to those patients without missing data for any variable.

Table 6
View larger version:
Table 6

Multivariable analysis comparing clinical features of patients who had inhalational anthrax (IA) with those of hospitalized patients who had community-acquired pneumonia.

Table 7
View larger version:
Table 7

Variables and score allocation used to construct scoring system comparing clinical features of patients who had inhalational anthrax with those of hospitalized patients who had community-acquired pneumonia.

Previous SectionNext Section

Discussion

IA has been described as a biphasic illness that is marked by an initial nonspecific phase characterized by fever, cough, and myalgias, followed by rapid progression to septic shock [20]. Until recently, assessment of risk for anthrax in the United States was based on travel history to areas of endemicity or exposure to certain imported or domestic animal products; patients with anthrax were extremely rare and not temporally clustered [20]. It is possible that, in a future, large-scale bioterrorist attack, which would resemble an outbreak of ILI or some other acute respiratory infection, clusters of persons with acute respiratory illness due to IA may seek care in emergency departments. In a setting in which the pretest probability of bioterrorism-related illness is elevated, in addition to use of epidemiological risk assessment, identification of a combination of vital signs, clinical history, and basic laboratory tests that discriminate IA from ILI with the highest possible sensitivity might be useful, depending on the need for high specificity. Clinical features have been used to differentiate other clusters of emerging or reemerging diseases by means of similar approaches, including use of discriminant functions, scoring systems, or flow charts [21, 22].

The majority of patients who had IA associated with this recent bioterrorist event had fever, cough, tachycardia, dyspnea, chills, fatigue, chest pain, and nausea or vomiting and lacked sore throat or nasal symptoms at the initial health care visit [4]. Comparison of patients who had IA with patients who had ILI revealed significant differences in the presence of myalgias, nasal symptoms (defined as rhinorrhea or nasal congestion), and tachycardia on initial presentation; hypoalbuminemia, hyponatremia, and hemoconcentration also were suggestive of IA. The clinical indicators that occurred more frequently in patients with IA than they did in patients with CAP included tachycardia, nausea or vomiting, elevated transaminase levels, hyponatremia, and normal WBC count. We developed a clinical scoring system that was able to discriminate all patients with IA from ∼96% of the patients from ambulatory clinics who had ILI; hospitalized patients with CAP were more difficult to differentiate.

Signs, symptoms, and laboratory abnormalities associated with IA that we identified in our analyses are consistent with the pathophysiology of B. anthracis infection via the inhalational route. After inhalation, B. anthracis spores are deposited in alveolar spaces and engulfed in macrophages, where germination to the vegetative form occurs. After transport to mediastinal lymph nodes, toxin production and continued replication results in hemorrhagic necrosis with fulminant bacteremia [2, 3, 20]. Thus, some symptoms suggestive of ILI, such as fever and cough, usually are present, but symptoms of upper respiratory infection, such as nasal congestion, generally are not noted. Later, clinical features of sepsis become more evident, as reflected by hypoalbuminemia, hyponatremia, and hemoconcentration.

Among the comparison groups we examined, univariate analyses indicated that the clinical features of IA most closely resembled those associated with the subset of patients with CAP who had pneumococcal bacteremia. For example, 80% of patients with IA and 53% of patients with pneumococcal bacteremia were hyponatremic. Hyponatremia has been associated with bacterial pulmonary processes; this abnormality has been most notably associated with legionnaires disease but also has been reported in patients with pneumococcal pneumonia [23, 24]. Hypoalbuminemia, hyponatremia, and hemoconcentration are consistent with movement of fluids from intravascular to extravascular fluid compartments, which occurs in systemic infection and sepsis [25]. Hyponatremia was recently reported in an infant with bioterrorism-associated cutaneous anthrax [26].

In evaluation of a patient with acute respiratory infection, health care providers are likely to first make a decision whether a patient has a disease process requiring outpatient care or hospitalization on the basis of vital signs, the presence or absence of key symptoms, and laboratory testing. After a decision to hospitalize the patient is made, the choice of empirical antimicrobials and more-extensive testing to further narrow the differential diagnosis are considered. In accordance with this model, the most logical approach was to compare patients who had IA with patients in ambulatory clinics who had ILI and then with hospitalized patients who had CAP.

We attempted to identify clinical features that, in a situation in which suspicion of B. anthracis is high, should prompt further evaluation and empirical treatment beyond that considered appropriate for acute respiratory illness in healthy persons. We did not analyze radiographic findings, because most ambulatory clinic patients with ILI would be unlikely to undergo chest radiography in the standard course of care, and patients with CAP would have abnormal results by definition. Rather, this algorithm was designed to assist in the clinician's decision to perform chest radiography and other diagnostic tests (e.g., blood cultures) when a patient presents with nonspecific upper respiratory tract infection and to help in discriminating IA from more-common causes of lower respiratory processes when the chest radiograph findings are abnormal, in a setting in which the suspicion for IA is high. Algorithms or prediction rules have been useful for improvement of pretest probability for chest radiography when CAP is suspected [27, 28.].

According to our assessment algorithm, in a situation in which risk for exposure to B. anthracis is believed to be high, if a patient in an ambulatory clinic being evaluated for acute respiratory infection has a score suggesting clinical features consistent with IA, simple diagnostic tests, such as chest radiography and blood cultures, must be considered as minimal evaluation to further investigate the possibility of B. anthracis infection. If a patient is being admitted to the hospital because of the possibility of CAP, in addition to the usual diagnostic tests, a score suggestive of clinical features consistent with IA might prompt further testing, such as CT. This approach also might include empirical treatment with antimicrobials effective against B. anthracis infection while awaiting diagnostic test results.

This study had 4 main limitations. First, our analyses were limited by the small number of patients who had IA and by the retrospective nature of the clinical study data for patients with ILI or CAP. Thus, not all laboratory measurements of interest were available for each comparison group, and we were unable to create a combined scoring system comparing IA, ILI, and CAP simultaneously. Second, use of clinical trial data may have introduced bias, because these patients may not represent the entire spectrum of illness for ILI and CAP because of restrictive study criteria for inclusion or exclusion. Furthermore, signs or symptoms required for study entry may have skewed differences found on analysis, although study treatment likely had no effect, because data used for analysis were collected before treatment for all comparison groups. Comparisons also may have been skewed by the underlying demographic characteristics of those with the greatest B. anthracis exposure—for example, postal workers—which may have affected such risk factors as age and smoking history. Third, we examined clinical features only at the time of initial presentation. Because IA is a dynamic and rapidly progressive disease, clinical features noted during the initial health care evaluation are probably quite variable, and their significance may vary dramatically depending on the time at which a patient seeks health care and in a pattern different from that of ILI and CAP. Finally, we were unable to validate our results, because there was no similar population of patients with IA that was appropriate for comparison; in addition, the small number of cases in the data set did not allow formal evaluation of model performance characteristics.

In summary, we conclude that the clinical features of patients with IA and of ambulatory clinic patients with ILI can be readily discriminated with high sensitivity and specificity. Because many clinical features of IA were more generally indicative of a septic process, results of comparison between patients with IA and hospitalized patients with CAP were less discriminating. Because the model constructed to compare patients who had IA with those who had CAP does not allow high simultaneous sensitivity and specificity, it may be of best use in conjunction with additional diagnostic testing. Both scoring models are most useful in conjunction with assessment of epidemiological risk or in settings of known exposure, in which clinical suspicion is high after a recognized bioterrorist event because of dissemination of B. anthracis spores.

Previous SectionNext Section

Acknowledgments

We first acknowledge the essential work performed by health care professionals, the academic community, and state and local health departments in response to this bioterrorist event. We also acknowledge the contributions of Oliver Keene, Jenny Colthart, and Michael Elliott of GlaxoSmithKline (Research Triangle Park, NC) and Jill Inverso of Pfizer (New York), for help with assembly of data and description of studies; and Richard Besser, Maureen Phelan, Brian Plikaytis, David Stephens, and Cynthia Whitney (Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, CDC), for help with data assembly and description and for guidance on analysis.

  • Received July 15, 2002.
  • Accepted November 3, 2002.
Previous Section
 

References

  1. ↵
    1. Centers for Disease Control and Prevention
    . Update: investigation of bioterrorism-related anthrax, 2001. MMWR Morb Mortal Wkly Rep 2001;50:1008-10.
    Medline
  2. ↵
    1. Swartz MN
    . Recognition and management of anthrax—an update. N Engl J Med 2001;345:1621-6.
    CrossRefMedlineWeb of Science
  3. ↵
    1. Dixon TC,
    2. Meselson M,
    3. Guillemin J,
    4. Hanna PC
    . Anthrax. N Engl J Med 1999;341:815-26.
    CrossRefMedlineWeb of Science
  4. ↵
    1. Jernigan JA,
    2. Stephens DS,
    3. Ashford DA,
    4. et al
    . Bioterrorism-related inhalational anthrax: the first 10 cases reported in the United States. Emerg Infect Dis 2001;7:933-44.
    MedlineWeb of Science
  5. ↵
    1. Barakat LA,
    2. Quentzel HL,
    3. Jernigan JA,
    4. et al
    . Fatal inhalational anthrax in a 94-year-old Connecticut woman. JAMA 2002;287:863-8.
    Abstract/FREE Full Text
  6. ↵
    1. Centers for Disease Control and Prevention
    . Update: investigation of bioterrorism-related anthrax—Connecticut, 2001. MMWR Morb Mortal Wkly Rep 2001;50:1077-9.
    Medline
  7. ↵
    1. Centers for Disease Control and Prevention
    . Notice to readers: considerations for distinguishing influenza-like illness from inhalational anthrax. MMWR Morb Mortal Wkly Rep 2001;50:984-6.
    Medline
  8. ↵
    1. Monto A,
    2. Ullman BM
    . Acute respiratory illness in an American community. JAMA 1974;227:164-9.
    Abstract/FREE Full Text
  9. ↵
    1. Makela MJ,
    2. Puhakka T,
    3. Ruuskanen O,
    4. et al
    . Viruses and bacteria in the etiology of the common cold. J Clin Microbiol 1998;36:539-42.
    Abstract/FREE Full Text
  10. ↵
    1. Sullivan KM
    . Health impact of influenza in the United States. Pharmacoeconomics 1996;9(Suppl 3):26-33.
    1. Zambon MC,
    2. Stockton JD,
    3. Clewley JP,
    4. Fleming DM
    . Contribution of influenza and respiratory syncytial virus to community cases of influenza-like illness: an observational study. Lancet 2001;358:1410-6.
    CrossRefMedlineWeb of Science
    1. Hall CB,
    2. Long CE,
    3. Schnabel C
    . Respiratory syncytial virus infections in previously healthy working adults. Clin Infect Dis 2001;33:792-6.
    Abstract/FREE Full Text
  13. ↵
    1. O'Brien KL,
    2. Walters MI,
    3. Sellman J,
    4. et al
    . Severe pneumococcal pneumonia in previously healthy children: the role of preceding influenza infection. Clin Infect Dis 2000;30:784-9.
    Abstract/FREE Full Text
  14. ↵
    1. Bartlett JG,
    2. Mundy LM
    . Community-acquired pneumonia. N Engl J Med 1995;333:1618-24.
    CrossRefMedlineWeb of Science
  15. ↵
    1. Centers for Disease Control and Prevention
    . Update: investigation of anthrax associated with intentional exposure and interim public health guidelines, October 2001. MMWR Morb Mortal Wkly Rep 2001;50:889-93.
    Medline
  16. ↵
    1. Lalezari J,
    2. Campion K,
    3. Keene O,
    4. Silagy C
    . Zanamivir for the treatment of influenza A and B infection in high-risk patients. Arch Intern Med 2001;161:212-7.
    Abstract/FREE Full Text
  17. ↵
    1. Braunwald E
    , editor. Harrison's principles of internal medicine. 1th ed. New York: McGraw-Hill; 1987. p. 1-5.
  18. ↵
    1. Goldman L,
    2. Bennett JC
    , editors. Cecil textbook of medicine 21st ed. Philadelphia: WB Saunders; 2000. p. 2299-308.
  19. ↵
    1. Hanley JA,
    2. McNeil BJ
    . The meaning and use of the area under a receiver operating characteristics (ROC) curve. Radiology 1982;143:29-36.
    Abstract/FREE Full Text
  20. ↵
    1. Brachman PS
    . Inhalational anthrax. Ann NY Acad Sci 1980;353:83-93.
    Medline
  21. ↵
    1. Moolenaar RL,
    2. Dalton C,
    3. Lipman HB,
    4. et al
    . Clinical features that differentiate hantavirus pulmonary syndrome from three other acute respiratory illnesses. Clin Infect Dis 1995;21:643-9.
    Abstract/FREE Full Text
  22. ↵
    1. Chapman LE,
    2. Ellis BA,
    3. Koster KE,
    4. et al
    . Discriminators between hantavirus-infected and -uninfected persons enrolled in a trial of intravenous ribavirin for presumptive hantavirus pulmonary syndrome. Clin Infect Dis 2002;34:293-304.
    Abstract/FREE Full Text
  23. ↵
    1. Sopena N,
    2. Sabria-Leal M,
    3. Pedro-Botet ML,
    4. et al
    . Comparative study of the clinical presentation of Legionella pneumonia and other community-acquired pneumonias. Chest 1998;113:1195-200.
    Abstract/FREE Full Text
  24. ↵
    1. Torres JM,
    2. Cardenas O,
    3. Vasquez A,
    4. Schlossberg D
    . Streptococcus pneumoniae bacteremia in a community hospital. Chest 1998;113:387-90.
    Abstract/FREE Full Text
  25. ↵
    1. Bone RC
    . The pathogenesis of sepsis. Ann Intern Med 1991;115:457-69.
    Abstract/FREE Full Text
  26. ↵
    1. Freedman A,
    2. Afonja O,
    3. Chang MW,
    4. et al
    . Cutaneous anthrax associated with microangiopathic hemolytic anemia and coagulopathy in a 7-month-old infant. JAMA 2002;287:869-74.
    Abstract/FREE Full Text
  27. ↵
    1. Fine MJ,
    2. Auble,
    3. Yealy DM,
    4. et al
    . A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336:243-50.
    CrossRefMedlineWeb of Science
  28. ↵
    1. Metlay JP,
    2. Kapoor WN,
    3. Fine MJ
    . Does this patient have community-acquired pneumonia? JAMA 1997;278:1440-5.
    Abstract/FREE Full Text
| Table of Contents

This Article

  1. Clin Infect Dis. (2003) 36 (3): 328-336. doi: 10.1086/346035
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 54 (5)
  1. Clinical Infectious Diseases
  1. Alert me to new issues

Most