Bacteremia Workup

WBC count

The WBC count is the most widely studied laboratory parameter in occult bacteremia. The risk of occult bacteremia and occult pneumococcal bacteremia has been consistently found to increase with an increased WBC count. ^{[1, 2, 10, 12, 20, 53]} Randomized control trials, retrospective reviews, prospective cohorts, and meta-analyses have been performed. Many have used slightly different inclusion and exclusion criteria, age ranges, and fever cutoffs. A consistent trend has been that children aged 3-36 months with FWS and a WBC count of more than 15 per high-powered field (HPF) are at an increased risk for occult bacteremia. ^{[1, 2, 10, 12, 20, 53]}

Most young febrile children with increased WBC counts do not have underlying bacterial infections as a cause of fever. The goal of screening criteria and laboratory tests in evaluation of infants and young children with fever has been to determine which patients are at a low risk (ie, which patients can be safely managed as outpatients without antibiotic treatment). Thus, established screening criteria have been chosen to maximize sensitivity and negative predictive value (NPV) as the primary objective. ^[10] Subsequent studies have shown a WBC count of 15 per HPF to yield an NPV of 98-99% and a positive predictive value (PPV) of 5-6% in distinguishing occult bacteremia from benign or noninvasive causes of FWS. ^{[1, 35, 46]}

Table 5. Studies Evaluating the Established WBC More Than 15 per HPF Screen for Occult Bacteremia in FWS (Open Table in a new window)

 

Several studies have reassessed the use of the WBC count as a screen for bacterial infection and compared it with other laboratory markers. ^{[56, 57, 58, 59]} These were all prospective observational studies of infants and children who presented to the emergency department for evaluation of FWS. The ability to distinguish bacteremia and other serious invasive bacterial infections from noninvasive or benign infections based on WBC count was evaluated. The direct application of these results to the evaluation and treatment of occult bacteremia has some limitations.

This group of studies includes patients with bacteremia but also patients with other invasive bacterial infections, such as meningitis and sepsis. The results show relatively high rates of infection in the study populations. Previous studies have found a 1.5-2.3% prevalence of occult bacteremia in infants and young children with FWS. ^{[23, 21, 35]} However, the newer studies found an 11-38% prevalence of serious or invasive bacterial infections. ^{[56, 57, 58, 59]} These studies have clinical use in the context of occult bacteremia because they address the evaluation of febrile young children who have no focus of infection upon initial examination in an outpatient setting.

These studies have reported optimal screening values based on receiving operator characteristics (ROC) curves to determine the best balance of sensitivity and specificity. The results show an optimal cutoff for WBC count of 15-17 per HPF, yielding NPVs of 69-95% and PPVs of 30-69% in distinguishing invasive or serious bacterial infections from noninvasive or benign infections. ^{[56, 57, 58, 59]}

Table 6. Recent Studies Reevaluating WBC Count as a Screen in FWS (Open Table in a new window)

 

These studies and others have compared the test characteristics of WBC count with other laboratory tests in screening for occult bacterial infections. The results suggest that absolute neutrophil count (ANC), C-reactive protein (CRP) level, and procalcitonin (PCT) level have also been favorable test characteristics when screening for occult bacterial infections in infants and young children. See the Absolute Neutrophil Count calculator.

In several studies, these other laboratory tests were equal or superior to WBC count as screening tools, as discussed below. Currently, screening with WBC count remains the established standard, as set by guidelines published in 1993. ^[10]

ANC

ANC has also been evaluated as a screen for occult bacteremia; the risk of occult bacteremia increases with increases in ANC. ^[1] Although guidelines before the conjugate Hib vaccine did not recommend ANC as a screen for bacteremia, ^[10] more recent studies and guidelines suggest that an ANC higher than 7-10 has favorable screening characteristics.

ROC curves for ANC are equal to the WBC count; one analysis found that the screening characteristics of ANC remained significant when adjusting for other variables, such as WBC count, temperature, age, and YOS. ^[1] An ANC higher than 7,000-10,000 has a 76-82% sensitivity, a 74-78% specificity, a 7-8% PPV, and a 99% NPV for occult bacteremia. ^{[1, 46]} The ANC is related to cases of occult pneumococcal bacteremia as follows ^[1] :

• Less than 5,000 - 0%

• 5,000-9,000 - 1.4%

• 10,000-14,900 - 5.8%

• Greater than 15,000 - 12.2%

Table 7. ANC as a Screen for Occult Bacteremia ^{[1, 46]} (Open Table in a new window)

 

Band count

The absolute band count (ABC) has been found to have poor test characteristics as a screen for occult bacteremia and is not recommended as a screening test. ^{[1, 10]} In febrile children, the risk for occult bacteremia generally tends to increase with increasing ABC; however, no well-defined cutoff is recognized, ROC curve characteristics are poor compared with those of ANC and WBC count, and any changes in ABC are not significant when adjusting for other variables.

Elevated band counts have also been found in 21-29% of patients with culture-proven viral infections. ^[60] The ABC may be the most important component of the CBC counts for identifying meningococcal bacteremia, but the low overall prevalence limits its clinical use. The ABC (X 10³/mL) is related to cases of occult pneumococcal bacteremia as follows: ^[1]

• Less than 0.5 - 1.5%

• 0.5-0.99 - 1.7%

• 1-1.5 - 1.7%

• 1.5-1.9 - 5.2%

• Greater than 2 - 6.3%

Bandemia (band >15%) is related to cases of viral infections as follows: ^[60]

• Influenza A and B - 29%

• Enterovirus - 23%

• Respiratory syncytial virus - 22%

• Rotavirus - 22%

In most studies of bacteremia, infants younger than 3 months are considered separately. Groups in Rochester, Boston, and Philadelphia have established guidelines aimed at defining populations of infants who are at a low risk for bacterial infection. These guidelines were published in Pediatrics in 1993. Most of these guidelines use band count as part of the low-risk criteria. Low-risk band criteria according to these guidelines are as follows:

• Boston guideline - None

• Philadelphia guideline - Less than 0.2 band-to-neutrophil ratio

• Rochester guideline - Less than 1,500 ABC

• 1993 Pediatrics - Less than 1,000 ABC

Erythrocyte sedimentation rate

Numerous studies have evaluated erythrocyte sedimentation rate (ESR) as a marker for bacterial infection. Most studies were performed before widespread use of the conjugate Hib vaccine and included hospitalized patients and patients with focal infections. ^[1] These studies found that ESR had a better sensitivity than WBC count and similar specificity. One review found that the ESR did not predict occult bacteremia, and WBC count and ANC were more sensitive and specific. ^[1] Based on this information, ESR is not currently recommended as a screening test for occult bacteremia. ^{[1, 10]}

CRP level

CRP level is not currently an established standard screening test for occult bacteremia, as set by the guidelines published in 1993 in Pediatrics and Annals of Emergency Medicine. ^[10] Several studies performed before widespread use of conjugate Hib and pneumococcal vaccines found that the CRP level had better sensitivity than WBC count and similar specificity. However, an analysis in 1999 found that CRP level could not be used to predict occult bacteremia in young children. ^[40]

Several studies have reassessed CRP level as a screen for bacterial infection and compared it with other laboratory markers. ^{[56, 57, 58, 59, 61]} These were all prospective observational studies of infants and children who presented to the emergency department for evaluation of FWS. As discussed above, the application of these results to bacteremia is somewhat limited by the inclusion of other invasive infections and by the relatively high prevalence of infection in the study populations. However, these studies have clinical use in the context of occult bacteremia because they address the evaluation of febrile young children who have no focus of infection upon initial examination in an outpatient setting.

Recent studies have reported optimal screening values using ROC curves to determine the best balance of sensitivity and specificity. The results show an optimal cutoff for CRP level from 2.8-5, yielding NPVs of 81-98% and PPVs of 30-69% in distinguishing invasive or serious bacterial infections from noninvasive or benign infections. ^{[56, 57, 58, 59, 61]}

Table 8. Studies Reevaluating CRP level as a Screen in FWS (Open Table in a new window)

 

WBC count is currently the established standard laboratory screening test in young children with FWS. ^[10] Several of the studies above directly compared WBC count and CRP level as screening laboratory tests in febrile young children with FWS. In each of these comparisons, CRP level had NPVs and PPVs better than or equal to WBC count. ^{[56, 57, 58, 59]} Although one author concluded that CRP level did not have any advantage or additional value compared to WBC count, ^[59] CRP level screening for febrile children in the emergency department is a part of the established protocol at numerous medical centers. Potential strengths of CRP level screening include favorable test characteristics, timely availability of results, and an ability to perform tests reliably on a capillary blood sample.

The time course for changes in serum CRP levels after onset of inflammation and acute tissue injury is fairly well understood. The CRP level begins to increase within 6 hours, doubles every 8 hours, and peaks from 36-48 hours. ^[62] Based on this known delay between stimulus and CRP level response, some have been concerned that CRP level would have decreased sensitivity early in the course of an illness.

This issue was assessed in a few of the studies without a clear and consistent conclusion. In one study, children with a fever duration of less than 12 hours were analyzed separately, and ROC curves were created for each of the laboratory values studied. ^[56] The optimal cutoff for CRP level overall, including any duration of fever, was 2.8; the NPV was 81%, and the PPV was 69% in distinguishing invasive bacterial infection. The optimal CRP level cutoff in children with a fever of less than 12 hours was lower (1.9) and gave less optimal screening test characteristics; the NPV was 77%, and the PPV was 66%. In a smaller study, a CRP level cutoff of 7 was analyzed and was found to miss 3 patients with serious bacterial infections, all of whom had a fever duration of less than 8 hours. ^[57] These results support the concern that CRP level is lower and less useful as a screen early in an infection.

However, this finding is not universal. A third study separately analyzed patients with fever durations of less than and greater than 12 hours and found that, in both groups, CRP level has a similar optimal cutoff and similar favorable screening characteristics. ^[59] To complicate the results further, the first study above also analyzed WBC count in patients with a fever duration of less than 12 hours. In the first 12 hours of illness, the WBC count did not differ between invasive bacterial infections and other localized, benign, or viral infections. This suggests that laboratory screening in illnesses of short duration may be problematic, whether WBC count or CRP level is used.

Cytokines

Interleukin (IL)-1, IL-6, and tumor necrosis factor-α (TNF-α) all increase in the serum and cerebrospinal fluid (CSF) in gram-negative and gram-positive sepsis; the levels increase with the severity of illness. One review found that these levels also increase in bacteremia; sensitivity and PPV are similar to those of WBC count. ^[1] One prospective case control study found that IL-6 and TNF-α were not significantly different between study groups; however, IL-6 had screening test and ROC curve characteristics similar to those of WBC count and ANC. IL-6 as a test for occult bacteremia had a sensitivity of 88%, a specificity of 70%, a PPV of 7%, and an NPV of 99.6%. ^[46]

These cytokines have not been thoroughly investigated; they have marginal clinical use, unknown cost-effectiveness, and are not recommended as routine screening laboratory studies for occult bacteremia. ^[1]

Procalcitonin level

Several reviews have described what is currently known about PCT. ^{[63, 64, 65]} PCT is a prohormone of calcitonin. In studies, PCT levels increase rapidly in the serum following exposure to bacterial endotoxin. This increase begins at approximately 2-4 hours and is more rapid than that seen in CRP levels. How PCT level fits into the acute phase cascade is unclear, and the sites of production and function of PCT level are also unclear. PCT levels remain low in viral infections and in systemic inflammatory diseases such as systemic lupus erythematosus (SLE) and Crohn disease, but PCT levels significantly increase in bacterial infections and superinfections. PCT levels also increase in some nonbacterial diseases that involve major tissue injury (eg, major surgery, burns, cardiogenic shock, acute transplant rejection).

Numerous studies in ICU settings have assessed PCT levels. These seem to confirm the above findings and show that an increase in PCT level is directly correlated with an increased severity of infection. Serial PCT levels correlate well with response to treatment. PCT levels decrease with successful antibiotic treatment, and a persistent elevation of PCT levels correlates with poor outcomes in the ICU.

A few studies have assessed PCT level as a screen for bacterial infection and compared it with other laboratory markers, including WBC count and CRP. ^{[56, 58, 61]} These were all prospective observational studies of infants and children who presented to the emergency department for evaluation of FWS. The application of these results to bacteremia is somewhat limited by the inclusion of other invasive infections and by the relatively high prevalence of infection in the study populations. However, these studies have clinical use in the context of occult bacteremia because they address the evaluation of febrile young children who have no focus of infection upon initial examination in an outpatient setting.

These studies have reported optimal screening values based on ROC curves to determine the best balance of sensitivity and specificity. The results show an optimal cutoff for PCT level from 0.6-2, yielding NPVs of 90-99% and PPVs of 52-91% in distinguishing invasive or serious bacterial infections from noninvasive or benign infections. ^{[56, 58, 61]}

Table 9. Studies Evaluating PCT level as a Screen in FWS (Open Table in a new window)

 

In these studies, PCT level had favorable test characteristics when compared to WBC count and CRP level as a screen for serious or invasive bacterial infections. PCT level had better NPVs and PPVs than both WBC count and CRP level in each of these studies.

As mentioned above, laboratory screening in illnesses of short duration may be problematic. In one of these studies, children with a fever duration of less than 12 hours were analyzed separately, and ROC curves were created for each of the laboratory values studied. ^[56] WBC count had no use as a screening test for illness lasting less than 12 hours, and CRP level had a lower optimal cutoff value with lower predictive values as a screen in these recently onset illnesses. Analysis of PCT level screening in illness lasting less than 12 hours found an optimal cutoff value and screening characteristics that were similar to those found in illness of longer duration. This information fits with the known rapid increase in serum PCT level following a stimulus and suggests that PCT level may be useful as a screen for illnesses of short duration.

Table 10. Effect of Illness Duration - PCT level as a Screen in FWS ^[56] (Open Table in a new window)

 

Bacteremia is a concern because it can lead to focal bacterial infections, most importantly meningitis. In a prospective observational study of 59 infants and young children hospitalized with meningitis, serum PCT level was a perfect screen for bacterial meningitis. A PCT level cutoff of 2 had a 100% NPV and a 100% PPV in distinguishing bacterial meningitis from viral meningitis. ^[66] This suggests that PCT level may have use as a screen for bacteremia and for sequelae such as meningitis in young febrile children.

In summary, PCT level appears to be more sensitive and more specific for bacterial infection than are other laboratory values currently used as screening tests and has good results in illnesses of short duration. Other potential strengths include the need for a small amount of serum or plasma and the availability of a rapid qualitative colorimetric bedside PCT level test, which showed similar test characteristics when compared with the instrument-based laboratory PCT level test. ^[56]

Potential weaknesses of PCT level tests include cost (currently estimated at twice that of CRP level tests) ^[65] ; increased PCT levels found in some nonbacterial diseases as mentioned above; and current familiarity and availability limited to research laboratories. Also, studies of PCT level as a screening test have focused on patients in intensive care units or patients with serious, invasive, or focal infections. Currently, PCT level shows promise as a screening test in febrile infants and young children. Further study is needed to show more direct application to children with FWS, at risk for occult bacteremia, in the emergency department, or in the pediatric clinic setting.

Urinalysis

Evaluation of children with FWS often requires laboratory analysis to evaluate for UTI. Children with test results that suggest a UTI are generally treated for this focal infection and do not require further evaluation for occult bacteremia. Of children evaluated for FWS, approximately 7% of boys younger than 6 months and approximately 8% of girls younger than 1 year have a UTI. ^[11] All published guidelines for evaluation of FWS in infants younger than 1 month recommend a laboratory evaluation for UTI, and most guidelines also recommend urine studies in girls younger than 1-2 years and boys younger than 6 months. ^[10]

Although UTI is a separate topic and is not fully addressed here, traditional guidelines for urine studies in infants and children with FWS include urinalysis, microscopy, and urine culture. A negative screening test result is defined as fewer than 5-10 WBCs per HPF, no bacteria, and negative nitrite and leukocyte esterase. ^{[10, 14, 15, 16, 67]} Application of these guidelines revealed that, in infants and children, approximately 20% of UTIs established based on findings from a urine culture were not detected by the screening urinalysis. ^[11]

Studies using enhanced urinalysis (cell count by hemocytometer and urine Gram stain) and Gram stain of urine sediment showed 99-100% sensitivity and a 100% NPV for UTI. ^{[11, 68]} Improvement in sensitivity of urine studies has great potential for improving detection of systemic bacterial infection (SBI) in young febrile infants during the initial evaluation. ^[67]

Salmonella and stool studies

Salmonella bacteremia accounts for the second most common cause of pediatric bacteremia (see Causes), and the clinical and laboratory findings are different from those in pneumococcal bacteremia.

A WBC count is not a useful screening test because most infants and children with Salmonella bacteremia have a WBC count less than 15,000/μL, and only half of patients have a left shift of the WBC count differential. ^[1] Most patients who develop Salmonella bacteremia have gastroenteritis, and 6.5% of children younger than 1 year who have Salmonella gastroenteritis become bacteremic. ^[1] Because of this association, stool cultures are recommended for children with diarrhea. ^{[10, 11]}

The initial clinical application of low-risk criteria for infants younger than 3 months with FWS did not include a stool evaluation. However, numerous patients with Salmonella bacteremia were improperly identified as being at low risk by these guidelines, and current guidelines recommend a screening stool evaluation in young infants with diarrhea. Patients with fewer than 5 WBCs per HPF are considered at low risk for bacterial infection. ^{[10, 15, 16]}

N meningitidis

Meningococcus is also an uncommon cause of occult bacteremia, but the morbidity and mortality associated with meningococcemia are high (see Causes and Mortality/Morbidity). Laboratory findings in meningococcal bacteremia are also different from those in pneumococcal bacteremia.

Although the risk of pneumococcal bacteremia is directly related to increasing WBC counts, 6% of children with meningococcal bacteremia have a WBC count per HPF of fewer than 5. Overall, WBC counts and ANCs have not proved consistently useful in determining the risk of meningococcal infection. ^{[1, 40]}

The band count may be the most important component of the CBC count in meningococcus. ^[1] Approximately 60% of patients with meningococcal bacteremia have a band count of greater than 10%, and a retrospective review of FWS found that the band count was the only laboratory value that was significantly higher in patients with meningococcal bacteremia than in those without bacteremia. ^{[1, 40]} However, the clinical use of an elevated band count is limited because of the low overall prevalence of meningococcal bacteremia. The PPV of a band count greater than 10% is 0.06.

The use of plasma clearance rate (PCR) in the evaluation of occult meningococcal bacteremia has not been studied. In studies of known meningococcal disease, PCR is sensitive and specific and may be useful in detecting meningococcal bacteremia. ^[1]

CSF analysis

Infants and children with FWS may require a laboratory analysis to evaluate for meningitis. Febrile infants and children of any age who are toxic require a full sepsis evaluation, including CSF and empiric treatment with parenteral antibiotics. ^[10]

Guidelines by groups in Rochester, Boston, and Philadelphia for the treatment of infants younger than 3 months who have FWS all include screening CSF laboratory tests and a CSF culture; the guidelines published in Pediatrics in 1993 recommend that a CSF evaluation be performed in certain situations (see Medical Care). Negative screening test results were defined as fewer than 8-10 WBCs per HPF, no bacteria, and normal glucose and protein levels. ^{[10, 14, 15, 53]} Children with laboratory values suggesting meningitis should be treated for this focal infection. Evaluation and treatment for meningitis is a separate topic and is not fully addressed here.

A case-controlled study by Aronson et al that included 135 febrile infants with invasive bacterial infection (bacteremia without meningitis and or bacterial meningitis) reported that when compared to the Rochester criteria, the sensitivity of the modified Philadelphia criteria was higher but the specificity was lower. Since some infants with bacteremia were classified as low risk, the authors recommend close follow-up for infants discharged from the emergency department without CSF testing. ^[69]

Blood culture

A blood culture positive for known bacterial pathogens is the criterion standard used to define bacteremia.

Blood cultures should be obtained in infants and young children at risk for occult bacteremia. Blood cultures that are positive for single isolates of known pathogenic bacteria (see Causes) are generally considered to be true positive results; cultures that grow multiple isolates or nonpathogenic bacteria are considered contaminated. How fast the culture becomes positive for known bacterial pathogens is also useful in distinguishing pathogens from contaminants; true pathogens generally grow faster than contaminants, with most pathogens turning positive in less than 24 hours. ^{[1, 2]} The routine mean detection time for several pathogens are as follows ^[1] :

• S pneumoniae - 11-15 hours

• Salmonella species - 9-12 hours

• N meningitidis - 12-23 hours

Whether the quantity of colonies grown is useful in detecting occult bacteremia or in predicting prognosis is unclear. Occult pneumococcal bacteremia may yield fewer than 10 colony-forming units (CFU)/mL, which is lower than in focal disease. The yield in meningococcal infection widely varies, but one study found that patients with yields higher than 700 CFU/mL were at an increased risk for meningitis. ^[1]

Testing for respiratory syncytial virus and influenza may help in the evaluation of infants with symptoms typical of viral respiratory infection and fever. Positive test results for these viruses is associated with a lower risk of occult bacteremia and meningitis, although no significant different in bacterial UTI is seen. ^{[70, 71]} This could have implications for the use of empiric antibiotics and how aggressively a SBI is investigated, although how this testing would fit into one of the clinical algorithms is unclear.

The only imaging study routinely used in infants and children with FWS is chest radiography to evaluate for pneumonia. Consider pneumonia in febrile children with no other source of infection. Specific physical examination findings include grunting, flaring, retracting, rhonchi, wheezing, rales, and focal decreased breath sounds. These findings are 94-99% specific for pneumonia. ^[67] Obtain a chest radiograph as part of the evaluation of children with any of these findings; evaluation for pneumonia in febrile children without any of these findings is of very low yield. ^{[2, 20]}

Some studies suggest that pulse oximetry may be a more reliable predictor of pulmonary infections than is respiratory rate in infants and young children. One guideline recommends that chest radiography be used to evaluate for pneumonia if the patient's oxygen saturation is less than 95%. ^[2]

One study found that a subset of febrile children who did not have physical examination findings suggestive of pneumonia were at an increased risk for occult pneumonia. ^[72] Approximately 20% of febrile children younger than 5 years who had normal physical examination findings and WBC counts higher than 20,000/μL had chest radiographic findings consistent with pneumonia. This guideline recommends that a chest radiograph be obtained in febrile infants and children with signs and symptoms of pneumonia and in febrile infants and children without signs and symptoms of pneumonia who have WBC counts higher than 20,000/μL.

Procedures used in the workup include the following:

• Blood: Venipuncture is performed to obtain blood for a CBC count and blood cultures. This should be performed using a sterile technique to limit contamination. The recovery rate associated with blood cultures is improved with larger volumes of blood and a shorter period between the blood draw and incubation in the laboratory. ^[1] The recovery rate is 83% with a large volume of blood (6 mL) and is 60% with a small volume of blood (2 mL). The recovery rate is 95% after 2 hours between blood draw and incubation and is 70% after 3 hours between blood draw and incubation.

• Lumbar puncture: A lumbar puncture (LP) is performed to obtain CSF for cell count, glucose and protein levels, microscopy, and Gram stain and culture (see Lab Studies and Medical Care). This should be performed using a sterile technique to limit contamination. Children with bacteremia who have an LP may have an increased risk of meningitis, although this theory is controversial. ^[2]

• Urine specimen: Urine collection is performed for urinalysis, microscopy, Gram stain, cell count, and culture (see Lab Studies and Medical Care). Urine collection should be performed using a sterile technique to limit contamination. Suprapubic bladder aspiration and in-and-out bladder catheterization are best in young infants and children.