|
Unmasking Anthrax and Smallpox
The authors show you how to apply some clinical assessment tools that may enhance your diagnostic accuracy when confronted with these illnesses caused by bioterrorism agents, which can all too easily be mistaken for common pulmonary or skin diseases.
By Carl H. Schultz, MD, and Christopher Eric McCoy, MD
Since the 2001 bioterrorism incident in the United States, when people were infected with anthrax spores delivered through the mail, physicians have worried about their ability to diagnose future cases of the illness. Diagnostic algorithms have been published to help clinicians differentiate inhalational anthrax from influenza or community-acquired pneumonia (CAP). However, it is unclear whether these diagnostic aids will successfully identify anthrax cases in real practice situations. In addition, the Centers for Disease Control and Prevention (CDC) have released a strategy for risk stratification in patients suspected of having smallpox. But does it work?
Using a case presentation format, this article will examine the challenges of diagnosing infection with anthrax or smallpox and evaluate what strategies, if any, can improve a physician’s diagnostic accuracy. We will compare syndromes and clinical presentations commonly seen in the emergency department with anthrax and smallpox cases that can mimic these clinical entities. Finally, we will discuss findings that suggest one of these two bioterrorism agents and put them into a real-world clinical context.
PULMONARY CASE #1
A 45-year-old man who works for the Internal Revenue Service presents to the emergency department complaining of cough, fever, headache, vomiting, and malaise for several days. He also reports shortness of breath but denies rhinorrhea. His vital signs are: heart rate, 115; blood pressure, 133/88; temperature, 102°F; respirations, 20, with an oxygen saturation of 98% on room air.
Physical examination finds the patient alert but in mild distress. His throat is slightly red. His heart rate is tachycardic without murmurs or rubs. His lungs show a questionable decreased breath sound in the right lower lobe with mild crackles. Abdominal and skin examinations are benign.
Laboratory test results are: white blood cell (WBC) count, 12,900/mm3 with 69% polymorphonuclear leukocytes and no bands; hemoglobin, 14 g/dl; hematocrit, 42%; and a platelet count within normal limits. Blood chemistry results are: sodium, 135 mEq/L; potassium, 4.1 mEq/L; chloride, 100 mEq/L; bicarbonate, 26 mEq/L; blood urea nitrogen (BUN), 16 mg/dl; creatinine, 0.9 mg/dl; and glucose, 105 mg/dl. Chest films reveal a right lower lobe infiltrate and mild left lower lobe atelectasis (see image below).
 |
Pulmonary case #1. Right lower lobe infiltrate and mild left lower lobe atelectasis.
|
This patient’s cough, fever, and lung infiltrate are consistent with a diagnosis of CAP. Discharging him home with a macrolide antibiotic is a reasonable choice.
PULMONARY CASE #2
A 53-year-old man who works for the New York Times presents to the emergency department complaining of fever, chills, sore throat, headache, and malaise for three days. He also reports a cough, dyspnea, night sweats, and vomiting for one day. His vital signs are: pulse, 110; blood pressure, 128/80; temperature, 99°F; respirations, 24, with an oxygen saturation of 93% on room air.
This patient is also alert but in mild to moderate distress with associated shortness of breath. The HEENT examination is unremarkable. His heart rate is tachycardic without murmurs, rubs, or gallops. Auscultation of the lungs reveals decreased breath sounds and rhonchi in the left base. Abdominal and skin examinations are normal.
Laboratory test results are: WBC count, 7500/mm3 with 76% polymorphonuclear leukocytes and 8% bands; hemoglobin, 15.5 g/dl; hematocrit, 46.9%; and a platelet count within normal limits. Blood chemistry results are: sodium, 138 mEq/L; potassium, 4.1 mEq/L; chloride, 102 mEq/L; bicarbonate, 26 mEq/L; BUN, 19 mg/dl; creatinine, 1 mg/dl; and glucose, 102 mg/dl. Chest films reveal infiltrates in multiple lobes, a normal mediastinum, an elevated diaphragm secondary to volume loss, and no pleural effusions (see image below).
 |
Pulmonary case #2. Infiltrates in multiple lobes, a normal mediastinum, an elevated diaphragm secondary to volume loss, and no pleural effusions.
Source: Centers for Disease Control and Prevention
|
This patient’s presentation is also consistent with CAP. Admitting him to the hospital and administering a macrolide with a third-generation cephalosporin would be a reasonable choice.
PULMONARY CASE #3
A 61-year-old man who works for the post office presents to the emergency department complaining of sweats, fatigue, myalgias, fevers, chills, headache, and vomiting for two days. He also reports a dry cough with no shortness of breath, rhinorrhea, or conjunctivitis. His vital signs are: pulse, 116; blood pressure, 141/90; temperature, 100.8°F; respirations, 18, with an oxygen saturation of 97% on room air.
The patient is alert and does not look particularly ill, although he appears mildly dehydrated. The HEENT examination is unremarkable. His heart rate is tachycardic without murmurs, rubs, or gallops. His lungs are clear to auscultation and percussion. Abdominal and skin examinations are unremarkable.
Laboratory test results are: WBC count, 9700/mm3 with 79% polymorphonuclear leukocytes and no bands; hemoglobin, 14.8 g/dl; hematocrit, 44.7%; and a platelet count within normal limits. Blood chemistry results are: sodium, 134 mEq/L; potassium, 3.4 mEq/L; chloride, 98 mEq/L; bicarbonate, 20 mEq/L; BUN, 23 mg/dl; creatinine, 1 mg/dl; and glucose, 105 mg/dl. Chest films reveal a wide mediastinum but are otherwise unremarkable (see image below).
 |
Pulmonary case #3. A wide mediastinum but no other remarkable findings.
Source: Centers for Disease Control and Prevention
|
A viral upper respiratory tract syndrome is a reasonable diagnosis for this patient. After intravenous hydration and antiemetics, he could be discharged home without antibiotics and advised to maintain close follow-up with his primary care physician.
PULMONARY CASE #4
An 89-year-old woman presents to the emergency department complaining of fever, cough, weakness, and muscle aches for three days. She denies chills, headache, vomiting, shortness of breath, or rhinorrhea. Her past medical history includes chronic obstructive pulmonary disease (COPD), hypertension, and renal insufficiency. She is retired and lives at home. Her vital signs are: pulse, 120; blood pressure, 106/50; temperature, 102.4°F; respirations, 18, with an oxygen saturation of 93% on room air.
On physical examination, the patient is alert and appears dehydrated. The HEENT examination finds dry mucous membranes. Her heart rate is tachycardic without murmurs, rubs, or gallops. Her lungs are clear to auscultation and percussion. Abdominal and skin examinations are unremarkable.
Laboratory test results are: WBC count, 8100/mm3 with 78% polymorphonuclear leukocytes and no bands; hemoglobin, 13.8 g/dl; hematocrit, 42.3%; and a platelet count within normal limits. Urinalysis was negative except for the presence of 3 to 5 WBCs per high-powered field and moderate bacteria. Blood chemistry results are: sodium, 134 mEq/L; potassium, 3.9 mEq/L; chloride, 102 mEq/L; bicarbonate, 25 mEq/L; BUN, 39 mg/dl; creatinine, 1.4 mg/dl; and glucose, 105 mg/dl. Chest films reveal a normal mediastinum and findings suggestive of COPD with possible blunting of the costovertebral angles (see image below).
 |
Pulmonary case #4. A normal mediastinum and findings suggestive of COPD with possible blunting of the costovertebral angles.
Source: Barakat LA, et al. (see Suggested Reading)
|
This patient’s diagnosis is not clear. Considering a diagnosis of viral syndrome versus dehydration would be a sensible approach. Admitting the patient, obtaining cultures, and administering oxygen and antibiotics that cover a possible urinary tract infection would be reasonable as well.
Differential diagnosis
These four patients represent scenarios that emergency medicine physicians expect to encounter on a daily basis. A reasonable differential diagnosis for all four cases includes pneumonia of either a viral, bacterial, or fungal etiology as well as influenza. However, in two cases—#2 and #4—the patient actually had inhalational anthrax. In case #1, the patient had pneumonia, and in case #3, viral syndrome.
Clinical assessment tool for anthrax
The emergency physicians who failed to diagnose anthrax in their patients are not unique. Patients presenting with influenza-like illness or CAP are extremely common, so recognizing a case of inhalational anthrax can be challenging. A few tools exist to aid in the diagnosis of anthrax, but they are limited in quality.
Kuehnert study. In 2003, Kuehnert and colleagues described a decision-making tool to aid in differentiating between inhalational anthrax and influenza. The researchers evaluated the differences in clinical presentations among 11 anthrax cases and 691 cases of influenza, creating a scoring system composed of clinical and laboratory data. A score of four had a reported sensitivity of 100% and a specificity of 91% (see box below).
A scoring system with 100% sensitivity means a health care provider should never miss a case of anthrax using this tool. So why hasn’t this scoring system been adopted in the United States or even world wide? Because the study has limitations.
First, the laboratory data contain albumin. In the emergency department, a serum albumin test is not commonly ordered for a patient presenting with signs of influenza or an upper respiratory tract infection.
Second, the data represent a derivation set. The associations are based on information collected retrospectively and then used to create the tool. There was no accompanying validation set. Unfortunately, taking the data from the derivation set and prospectively applying it to patients going forward in time was not possible, although it would have demonstrated the validity of the results. Because this was not done, it is unclear how well this tool would work on a different group of patients.
Third, the anthrax sample size was extremely small. A sample of only 11 cases makes it difficult to extrapolate the results to much larger and more diverse populations. Even if the sensitivity of 100% is correct for this patient group, the 95% confidence interval for a tool applied to just 11 patients is as low as 70%. In a different sample of 11 anthrax cases, the same scoring system may miss up to 30% of patients. In other words, a study with zero out of 11 cases missed is consistent with as many as 30% of the cases missed if applied to a group containing thousands of anthrax patients.
Finally, the definition of terms may have questionable clinical relevance. The researchers define hyponatremia as a serum sodium measurement of less than 135 mEq/L, which is technically correct. However, several patients in the study who were said to have hyponatremia actually had sodium levels of 134 mEq/L. Most emergency physicians would not consider a serum sodium measurement only one or two points below normal as clinically significant. As such, this decision tool contains significant flaws that limit its usefulness in helping to differentiate between anthrax and influenza.
In the second section of the same study, Kuehnert devised a scoring system to differentiate anthrax from pneumonia. Eleven cases of anthrax were compared with 650 cases of CAP. Once again, a scoring system was created using clinical and laboratory data. A score of three had a sensitivity of 82% and a specificity of 81% (see box below).
 |
This section suffered from significant limitations as well. Data points were not well defined. For example, a “normal” WBC count did not take into consideration the band count. By the researchers’ definition, a normal WBC count with 30% bands would still fall into the normal category. In spite of these oversimplifications, two cases of anthrax were missed and 122 cases of pneumonia were misclassified as anthrax using this tool. In addition, all clinical components of the tool had very wide 95% confidence intervals, with the lower limit for some as small as 1.8. An odds ratio with a lower limit of at least five would be needed to make this convincing. The reported wide confidence intervals for the odds ratios prevent this scoring system from being clinically useful.
Hupert study. Another study that attempted to distinguish anthrax from common diseases seen in clinical practice was done by Hupert and colleagues in 2003. The researchers compared 28 cases of anthrax to 2762 influenza cases and 1932 cases of other viral illnesses. Instead of a scoring system, they created likelihood ratios (LRs) to aid the diagnosis. For example, signs and symptoms favoring anthrax were dyspnea (positive [+] LR = 5.3), nausea/vomiting (+LR = 5.1), and abnormal lung examination (+LR = 8.1). Examination findings favoring viral illness were rhinorrhea (+LR = 0.2) and sore throat (+LR = 0.2).
This study was based on 28 cases of anthrax obtained from 13 different reports. Ten cases came from a single study, so the remaining papers averaged one-and-a-half cases per report. Reliance on multiple publications containing very few patients each weakens the overall power of the study. A second limitation was the wide confidence intervals. The authors reported that the presence of rhinorrhea implied a viral illness, which was found in only 14% of anthrax cases. However, the 95% confidence interval for this number is consistent with as many as 31% of anthrax cases having rhinorrhea. If the true number is closer to 31%, this implies that in another group of anthrax patients, as many as one in three may have rhinorrhea.
Probably the most important limitation of the study is the LR values obtained. Generally, an LR above 10 or below 0.1 must be obtained to suggest that a particular finding is significant enough to change management. None of the values reported in the study were in this range.
Kyriacou study. A third clinical assessment tool was reported by Kyriacou and colleagues in 2004. This group compared 47 cases of anthrax to 376 cases of pneumonia and influenza-like illness. Researchers examined characteristics from the patients’ history, physical examination, laboratory data, and chest films that might serve as indicators suggesting a diagnosis of anthrax. They used sensitivity, specificity, and LRs to identify significant differences between clinical entities.
Important findings suggesting anthrax included nausea and vomiting, diaphoresis, cyanosis, and altered mental status. However, the authors claim that a chest film demonstrating a widened mediastinum or a pleural effusion was best at differentiating anthrax from influenza or pneumonia, with a sensitivity of 100% and a specificity of 72%. Unfortunately, the LR for this comparison was only 3.5 for pneumonia, making its value less striking.
Interestingly, the article reports that the chest film findings were present on admission or developed during hospitalization, implying that not every patient presented with these findings. Emergency physicians are not given the opportunity to follow patients during their hospital stay. Therefore, these radiographic findings may have limited use to physicians seeing the patient during the first few hours of assessment.
Applying the tools to the four cases
What happens if these tools are applied to the four pulmonary cases described earlier? Can they be used in a rational manner to influence patient care? Let’s evaluate each case using one of the diagnostic tools we just discussed.
Case #1. This was a case of pneumonia. Applying the pneumonia versus anthrax rules from the Kuehnert study, this patient would have scored a two (one point each for tachycardia and nausea and vomiting), which is below the three points needed to diagnose anthrax. This tool would have accurately diagnosed pneumonia.
Case #2. This patient had anthrax. Using the pneumonia versus anthrax scoring system yields indeterminate results. If applied as strictly reported, the patient would receive one point each for a normal WBC, nausea and vomiting, and tachycardia. This produces a score of three, the number of points required to diagnose anthrax. However, this patient had 8% bands. The authors did not discuss what the presence of bands means in the setting of a normal WBC count. It is highly debatable whether a patient with 8% bands should be considered to have a normal WBC. The authors’ inadequate definition of a normal WBC leaves this patient’s real score in question.
Case #3. This patient had viral syndrome. He amassed five points, suggesting a diagnosis of anthrax (for no nasal symptoms, low sodium, and a wide mediastinum). Under this scoring system, the patient would have been misdiagnosed as having anthrax.
Case #4. This was another case of anthrax. Whether this patient receives a score of three or six points is in question. Six points could be given if two points each were allotted for tachycardia and absence of rhinorrhea and one point each for hyponatremia and absence of headache. However, after examining the patient’s chart more closely, it appears that the reported history may be incorrect. Although the patient said she had no rhinorrhea, she was using azelastine nasal spray and loratidine, medications for seasonal allergies and rhinorrhea. In addition, most emergency physicians would not consider a sodium level of 134 mg/dl as hyponatremia. Once again, the usefulness of this scoring system is questionable.
the anthrax verdict
So what is the verdict on clinical assessment tools used to help diagnose anthrax? Judging from the evidence above, without a known case of anthrax or a good reason to suspect it (such as an alert by the federal government or the CDC), there are no good clinical markers to identify anthrax. But when a case of anthrax is reported in your area or a high alert for bioterrorism exists, anthrax is a legitimate diagnosis. In these cases, findings suggestive of infection include chest films with a wide mediastinum or pleural effusion, altered mental status, and gastrointestinal symptoms. However, even these findings may be of limited diagnostic value.
rash CASE #1
A 15-year-old boy presents to the emergency department complaining of a fever for one day and lesions on his back and trunk that have grown worse over the last three days. The lesions are vesicular, at different stages of development, and have spread to his extremities, with very few on the palms and soles (see image below). The patient does not appear particularly toxic.
 |
Rash case #1. Vesicular lesions at various stages of development that started on the back and trunk and spread to the extremities, sparing the palms and soles.
Source: Centers for Disease Control and Prevention
|
rash CASE #2
A 10-year-old boy presents to the emergency department after several days of fever and malaise. He developed vesicular lesions on his face and extremities, which spread to his trunk, as well as many lesions on the palms and soles (see image below). He appears toxic.
 |
Rash case #2. Vesicular lesions that started on the face and extremities and spread to the trunk as well as the palms and soles.
Source: Centers for Disease Control and Prevention
|
rash CASE #3
A 27-year-old man presents to the emergency department with a fever of one day’s duration and vesicular lesions that started on his face and spread to his chest. All of the lesions evolved at approximately the same rate and did not spread to the extremities (see image below). The patient does not appear toxic.
 |
Rash case #3. Vesicular lesions that started on the face and spread to the chest but not to the extremities.
Source: Centers for Disease Control and Prevention
|
rash CASE #4
A 42-year-old woman presents to the emergency department after two days of fever and body aches. She developed vesicular lesions on her face, which subsequently spread to her extremities. The lesions are pruritic, do not appear on the palms and soles, and are at different stages of development (see images below). She does not appear toxic.
 |
 |
Rash case #4. Pruritic vesicular lesions at different stages of development that started on the face and spread to the extremities, sparing the palms and soles. |
Rash CASE #5
A 28-year-old woman presents to the emergency department with three days of fever to 101.5°F associated with sore throat, enlarged lymph nodes, and fatigue. She developed vesicular and pustular lesions on her hands that spread to her scalp, perineum, extremities, and chest, sparing the palms and soles. All lesions are in the same stage of development (see image below). The patient does not appear toxic.
 |
Rash case #5. Pustular vesicular lesions in the same stage of development that started on the hands and spread to the scalp, perineum, extremities, and chest, sparing the palms and soles.
Source: Marshfield Clinic, Marshfield, Wisconsin
|
differential diagnosis
The differential diagnosis for vesicular rashes includes entities representing multiple pathological categories. These include but are not limited to varicella (chicken pox), herpes simplex, herpes zoster, erythema multiforme, dermatitis herpetiformis, pemphigus vulgaris, and poison sumac, ivy, or oak. These diagnoses are all reasonable considerations for the five cases presented above. However, two of these patients were actually infected with biological agents that are rarely considered. Although their vesicular lesions are similar to those caused by the diseases just mentioned, they represent infection by potential biological weapons: the microbes from smallpox and monkeypox.
algorithm for smallpox from the cdc
The possibility is remote that an emergency physician would ever encounter a patient with a pox virus. But if this occurs, there is a resource that reliably stratifies the risk of smallpox versus other vesicular rashes. The CDC Web site contains an algorithm for evaluating patients who present with a vesicular rash that might be smallpox. The algorithm is called “The Acute, Generalized Vesicular, or Pustular Rash Illness Protocol” (see Suggested Reading). This diagnostic algorithm incorporates history and physical examination data into two categories identified as major and minor criteria. Major criteria include three elements and minor criteria include five elements. This approach is similar to the Jones Criteria for diagnosing rheumatic fever. While designed for smallpox, these criteria may provide some guidance in evaluating patients for monkeypox.
The three major criteria in the algorithm consist of the febrile prodrome, classic smallpox lesions, and lesions in the same stage of development. The febrile prodrome occurs one to four days before onset of the rash. Patients must have a fever of at least 101°F and at least one of the following: prostration, headache, backache, chills, vomiting, or severe abdominal pain. Classic smallpox lesions are deep seated, firm, round, well-circumscribed vesicles or pustules. As they evolve, they may become umbilicated or confluent. Lesions in the same stage of development refers to the appearance of vesicles or pustules on any one part of the body. To meet this criterion, all lesions in one anatomical area must appear identical (that is, be all vesicles or all pustules).
The five minor criteria are centrifugal distribution (greatest concentration of lesions on the face and distal extremities); first lesions on the oral mucosa or palate, face, or forearms; toxic or moribund appearance; slow evolution of lesions (rash evolves from macules to papules to pustules over days, each state lasting one to two days); and lesions on the palms and soles.
Patients presenting with a vesicular rash who might be infected with smallpox are evaluated for the presence of major and minor criteria. The CDC algorithm will then risk-stratify these patients into high, moderate, or low probability of having the disease based on the number of criteria found. Patients exhibiting all three major criteria are classified as having a high risk of smallpox infection. Those classified as moderate risk have the febrile prodrome and one other major criteria or the febrile prodrome and at least four minor criteria. Low-risk patients are those with no febrile prodrome or the febrile prodrome and less than four minor criteria.
After risk-stratifying the patient with a vesicular or pustular rash, the CDC protocol allows the emergency physician to develop an action plan based on this profile. For a high-risk patient, you should isolate the patient, consult experts in infectious disease or dermatology (or both), and alert local and state health departments immediately. Testing for smallpox will be conducted at the CDC. For moderate-risk patients, you should consult with an infectious disease expert or a dermatologist, test for varicella and herpes viruses, and consider other laboratory testing as clinically indicated. If no diagnosis is made, ensure the adequacy of the specimen and have the infectious disease or dermatology expert reevaluate the patient.
If smallpox cannot be ruled out, the patient should be classified as high risk. If a diagnosis other than smallpox is confirmed, the results should be reported to infection control. Low-risk patients may be sent home without an infectious disease or dermatology consultation. Tests for varicella or other conditions should be performed as indicated.
review of rash cases using cdc protocol
How well does the CDC protocol perform when applied to the five cases we presented? Each case is reevaluated below.
Case #1. This patient actually had chicken pox. Using the risk stratification method described above, the patient had no major and one minor criteria—rash on the palms and soles. He had no febrile prodrome, the lesions were in different stages of development, and they did not have the characteristic morphology of smallpox lesions. Also, the vesicles were centrally distributed and spread rapidly. Last, the patient did not appear toxic. He would be classified as low risk for smallpox with only one minor criterion.
Case #2. This patient had smallpox and demonstrated all three major and all five minor criteria. The presence of three major criteria classifies him as high risk for smallpox, regardless of how many minor criteria are present. This patient would require isolation with notification of appropriate consultants and authorities.
Case #3. Herpes simplex is the diagnosis for this patient, who exhibits one major and one minor criterion. The lesions are in the same stage of development and the first ones appeared on the face. He would be classified as low risk for smallpox, because he has one major and less than four minor criteria and does not have a febrile prodrome.
Case #4. The rash exhibited by this patient is erythema multiforme. She has no major and two minor criteria. The lesions first started on the face and are centrifugally distributed. She would be classified as low risk for smallpox.
Case #5. This patient was infected with monkeypox and has one major and three minor criteria. The lesions were in the same stages of development, centrifugally distributed, slow to progress, and started on the forearm. While the patient was febrile, she was not toxic and so did not meet the definition of the febrile prodrome. One could argue that the lesions were classic smallpox pustules. If so, this patient would have two major criteria and might qualify as moderate risk. However, with no febrile prodrome and so few pustules present, this risk assessment is incorrect. This patient should actually be classified as low risk.
Some cases of monkeypox are more virulent, and in these situations patients will have the febrile prodrome and classic smallpox lesions. As such, they will be clinically indistinguishable from smallpox and should be classified as high risk.
the smallpox verdict
To date, this CDC algorithm has been applied multiple times and has performed well. No cases have been classified as high risk, demonstrating 100% specificity, since no cases of smallpox have occurred. However, due to the lack of cases, the sensitivity of the protocol remains unknown. In each case, when physicians used the algorithm, patients were either classified as moderate or low risk.
Emergency physicians evaluating patients with suspected smallpox now have an algorithm that successfully stratifies them into high, moderate, and low risk. More important, there is also a plan of action to accompany each risk category. So far, this algorithm is the only one approved by the CDC to risk-stratify patients with suspected smallpox. For patients presenting with a vesicular or pustular rash, the CDC protocol is a readily available and useful resource that can be implemented in the emergency department.
raising awareness
Although there are still no reliable clinical markers to identify anthrax and the CDC algorithm has never been tried on a real case of smallpox, becoming familiar with these tools will help raise your awareness of bioterrorism agents that can mimic the diseases you see every day in your practice.
Suggested Reading
Barakat LA, et al.: Fatal inhalational anthrax in a 94-year-old Connecticut woman. JAMA 287(7):863, 2002.
Centers for Disease Control and Prevention: Update: investigation of bioterrorism-related inhalational anthrax—Connecticut, 2001. MMWR Morb Mortal Wkly Rep 50(47):1049, 2001.
Hupert N, et al.: Accuracy of screening for inhalational anthrax after a bioterrorist attack. Ann Intern Med 139(5 Pt 1):337, 2003.
Jernigan JA, et al.: Bioterrorism-related inhalational anthrax: the first 10 cases reported in the United States. Emerg Infect Dis 7(6):933, 2001.
Kuehnert MJ, et al.: Clinical features that discriminate inhalational anthrax from other acute respiratory illnesses. Clin Infect Dis 36(3):328, 2003.
Kyriacou DN, et al.: Clinical predictors of bioterrorism-related inhalational anthrax. Lancet 364(9432):449, 2004.
Roche KJ, et al.: Images in clinical medicine: cutaneous anthrax infection. N Engl J Med 345(22):1611, 2001.
US Department of Health and Human Services, Centers for Disease Control and Prevention: Evaluating patients for smallpox. Available at: www.bt.cdc.gov/agent/smallpox/diagnosis/pdf/spox-poster-full.pdf. Accessed November 5, 2008. |
|
