A Clinical guide to pediatric infectious disease


Outpatient Evaluation of Fever

Basic Issues in Pediatric Fever

Fever is a common presenting complaint in pediatrics. The pediatrician's concern about fever in a child can be summarized by the following questions:

  1. What is the chance, given this temperature elevation, that this child has an evolving serious bacterial infection?
  2. What should be the proper evaluation and empiric treatment concerning this possibility of an evolving serious bacterial infection?

These questions relates only to the well-appearing pediatric patient. Children who are toxic, hypotensive, have altered mental status or decreased peripheral perfusion should be managed aggressively regardless of initial laboratory evaluations.

The question regarding the management of the well-appearing, nontoxic, febrile infant remains one of the most controversial in pediatrics and emergency medicine. Proposed strategies are constantly revised and depend on the child's age, underlying condition of the child, and development of new vaccinations. One traditional approach is to divide the evaluation of the febrile child by age.

Neonate (0 to 28 days)

Epidemiology and Etiology

More than one half of all women have bacterial genital tract colonization, often with group B streptococcus. About one half of neonates born to colonized women themselves become colonized. Of these colonized infants, about 1% develop invasive disease. Risk factors for invasive neonatal disease include prematurity, maternal fever during delivery, and prolonged rupture of membranes.


There are certain patient populations in which the chance of a serious bacterial infection is high and the physical exam and laboratory evaluations nonspecific enough that a full workup and empiric antibiotics are always indicated. Such a patient is the child in the first month of life.


Neonates with rectal temperatures greater than 38°C (100.4°F) have an overall risk of 13% for serious bacterial infection.

The history, physical exam, and even complete laboratory evaluations do not separate these infants into “high-risk” (high statistical likelihood to have a serious bacterial infection) and “low-risk” categories.


Studies have suggested that even a negative full diagnostic workup, including complete blood count, urinalysis, stool Gram stain, and lumbar puncture, will miss a percentage of neonates who ultimately have a serious bacterial infection, including bacteremia and urinary tract infections.


Two regimens are accepted for empiric treatment of neonatal fever. Ampicillin is usually given to address the possibility of Listeria monocytogenes infection. The second agent given is usually a third-generation cephalosporin or gentamycin to cover gram-negative organisms.

Standard practice continues to be a full evaluation and admission for intravenous antibiotics pending results of blood, urine, and cerebrospinal fluid (CSF) cultures.

Infant (28 to 59 days)

For the child older than 1 month of age, efforts begin to define those children whose risk for a serious bacterial infection is such that they could be managed as outpatients and avoid the automatic admission with intravenous antibiotics.


The organisms in this population are similar to those found in the neonatal group, with group B streptococcus, gram-negative enteric organisms and occasionally Streptococcus pneumoniae being the major pathogens.




Definition of fever in this population is similar to that in the neonatal group.


In the early 1980s, the first attempt was made to develop diagnostic criteria. The Yale Observation Scale was entirely clinical and measured characteristics such as cry, social response, and hydration status. This was found to have a negative predictive value of about 80%; that is, up to 20% of those who satisfied criteria for discharge did indeed have a serious bacterial infection. This rate was not sensitive enough to satisfy most clinicians.

The Rochester Criteria in the late 1980s was the first to attempt to incorporate laboratory data in evaluation of the febrile infant. The Rochester Criteria used white blood cell (WBC) count, urinalysis, and stool WBC count as part of the protocol. Major criticisms of these criteria were that CSF analysis was not included and that the criteria did not include clinical impression as part of the evaluation process.

Next came the Baskin Study in 1992. In this protocol, infants 1 to 2 months of age with a rectal temperature of greater than 38°C (100.4°F) and a nontoxic appearance received a full laboratory evaluation, including lumbar puncture, complete blood count, urinalysis, and stool studies. Patients with unremarkable findings on these screening tests received intramuscular ceftriaxone before discharge with close follow-up. Although there were a number of positive blood, urine, and stool cultures, these children had negative cultures on follow-up evaluation, and no complications were noted as a result of outpatient management. Criticisms of the study included the use of a dipstick urine evaluation rather than a microscopic urinalysis and, most importantly, no control population. It was thought that this protocol essentially substituted intravenous antibiotic therapy with outpatient intramuscular ceftriaxone.

In 1993, Baker and colleagues developed the Philadelphia Protocol. In this study, febrile infants between 1 and 2 months of age all received a full diagnostic workup including lumbar puncture, urinalysis, and stool studies if diarrhea was present. A low-risk infant was defined as having a peripheral WBC count of less than 15,000/mm3, less than 10 WBCs per high-power field on urinalysis, a CSF with less than 8 WBCs per high-power field, and no evidence of a focal soft tissue infection. The criteria also included a peripheral blood band–to-neutrophil ratio of less than 0.2. These screening criteria were found to be 100% sensitive in identifying those children with serious bacterial illnesses, with a negative predictive value of 100%. Patients who had negative diagnostic evaluations were actually discharged home with no antibiotic therapy given. The negative predictive value was such that this study is cited today as proof that a select group of infants between the ages of 1 and 2 months can be managed as outpatients without intravenous or intramuscular antibiotics.




It has been suggested that no plan for outpatient management of febrile children in this age group is entirely without risk. Guidelines continue to be revised and new recommendations formulated, including those that give the option of deferring lumbar puncture in the correct low-risk setting.

The C-reactive protein (CRP) is an acute-phase reactant that is sometimes used as an indicator for evolving serious bacterial infections. Some studies have found the CRP to be a better indicator than the peripheral WBC count or absolute neutrophil count in predicting those febrile infants who truly have an evolving bacterial infection. A CRP of less than 5 mg/dL has been cited as the cutoff for effectively ruling out bacterial infection. Some investigators have suggested that a lumbar puncture can be deferred if the CRP is less than 5 mg/dL. Recent guidelines by Baraff and associates (2000) suggest that if lumbar puncture is deferred, empiric antibiotics should not be given because pretreatment of a child in whom the CSF has not been examined may make subsequent interpretation of the CSF difficult.

Low-Risk Criteria for Febrile Infants (30 to 59 days)

·   Serum WBCs: 5 to 15,000/m3, <1,500 bands/m3

·   Urinalysis: <10 WBCs/high-power field

·   When diarrhea present, <5 WBC/high-power field in stool

·   CSF: <8 WBC/high-power field, negative Gram stain

·   CRP: <5 mg/dL

Recent guidelines have suggested the following options: all febrile infants receive a complete blood count, urinalysis, CRP, and blood and urine culture. Options outlined in Table 7.1.

TABLE 7.1. Management of the Nontoxic Febrile Infants, 28 to 59 Days of Age

Option 1

1. “Low-risk” infant

2. Lumbar puncture obtained; consider no treatment or 50 mg/kg of intramuscular ceftriaxone

3. Close follow-up

Option 2

1. “Low-risk” infant

2. C-reactive protein < 5 mg/dL, lumbar puncture deferred, antibiotics not given

3. Close follow-up

Option 3

1. “High-risk” infant

2. C-reactive protein > 5 mg/dL, lumbar puncture required

3. Consider hospital admission, empiric antibiotics



Children (3 months to 36 months)

Epidemiology and Etiology

As in the case of the well-appearing febrile infant, there remains considerable debate about management of a febrile child aged 3 months to 3 years.

In the early 1970s, the first reports appeared of S. pneumoniae bacteremia in well-appearing febrile children. The reported rate of bacteremia in febrile, well-appearing children was approximately 4% and was labeled “benign” because most of these children had resolution of bacteremia without sequelae. After these early reports, numerous studies attempted to determine the actual morbidity associated with this “silent” bacteremia in children. The concern is that this bacteremia could result in one of the following:

  • Persistent fever
  • Progression to sepsis, especially in the case of Neisseria meningitidis(Fig. 7.1)
  • “Seeding” of secondary sites; the most feared site being the CSF (Fig. 7.2)

The widespread use of H. influenzae conjugate vaccine has essentially eliminated this pathogen. The organism currently causing 90% of the bacteremia in this age group is S. pneumoniae. Bacteria less frequently encountered include Salmonella species and N. meningitidis.


FIG. 7.1. Purpura fulminans in child with overwhelming Neisseria meningitidis infection (see color plate).


FIG. 7.2. Mechanism and potential sequelae of pediatric bacteremia.




Children with pneumococcal bacteremia often present with sudden onset of high fever. There are typically nontoxic in appearance, and no focus is found on physical examination.


About 2% to 3% of febrile children are bacteremic. In addition, about 3% of these bacteremia patients may progress to meningitis. Thus, the basic questions confronting the clinician while examining a well-appearing febrile child with no focus of infection are the following:

  1. How do we identify those children who are actually bacteremic at the time of evaluation?
  2. Does presumptive treatment in these children with either oral or intramuscular antibiotics actually alter the natural history of the bacteremia? Although it makes theoretical sense that bacteremic children identified early and presumptively treated have reduced progression to complications of bacteremia (i.e., meningitis), is this indeed the case?


In 1993, Baraff published guidelines for the management of fever in children. In this publication, it was suggested that when evaluating fever without a source in children, a complete blood count should be obtained. If the WBC count was more than 15,000/mm3, this represented an increased chance of bacteremia with the resultant chance (albeit small) of meningitis. It has been recommended to give antibiotic treatment to these children to prevent the chance of meningitis.

Baraff's guidelines were offered as suggestions, yet rapidly became the standard of care. Infants with fever without a source routinely had complete blood


counts and blood cultures obtained and empiric antibiotics given. The concern was that this abundance of antibiotics would generate increased bacterial resistance (a concern that has proved correct). In addition, there was no conclusive evidence that empiric antibiotics truly prevented the bacteremia progressing to meningitis. Over the next 10 years, many studies sought to address the issue of empiric treatment. The methodology and results of these studies continue to be debated.

In 1994, Fleisher and associates performed a multicenter study of more than 6,000 children. In this study, children 3 to 36 months of age with a temperature of more than 39°C (102.2°F) had a blood culture obtained and were given either a single dose of ceftriaxone or oral amoxicillin. The incidence of bacteremia was 2.8%. In the analysis of the data, the authors stated that fewer children who were treated with ceftriaxone developed meningitis than those who were given oral amoxicillin. This study continues to be frequently quoted as evidence that antibiotics given to febrile children will prevent meningitis.

In an accompanying editorial, Dr. Sara Long (1994) both praised the study and commented on its shortcomings. She commented that less than one fourth of all febrile children had examinations of the CSF. One of the patients with meningitis placed to the amoxicillin group actually had pneumococcus isolated from the CSF before therapy was begun; Dr. Long commented that this should have been reported as a misdiagnosis, not amoxicillin failure. Two of the ceftriaxone-treated children with bacteremia caused by H. influenzae type B and Salmonellaspecies had, on repeat visit, a CSF pleocytosis of 117 and 158 WBCs, respectively. Although this did not meet the preset requirements of more than 500 WBCs for the diagnosis of meningitis, Dr. Long suggested that most infectious disease specialists would indeed consider these two cases, in a setting of prior bacteremia, to be bacterial meningitis that had developed despite the administration of ceftriaxone. Dr. Long concluded that there were still no data that ceftriaxone, or any antibiotic, prevented the rare occurrence of pneumococcal meningitis.

In 1997, Rothrock and colleagues undertook a meta-analysis of several published studies to determine whether oral antibiotics given to febrile children prevented serious bacterial infections or meningitis in patients with S. pneumoniae bacteremia. Ten available studies with 656 total cases of S. pneumoniae occult bacteremia were identified. Patients who received oral antibiotics had fewer serious bacterial infections than untreated patients, 3.3% versus 9.7%. Meningitis developed in 0.8% of children in the oral antibiotic group and 2.7% of untreated children. The authors themselves concluded that the modest decreased risk for serious bacterial infection in patients with occult pneumococcal bacteremia was “insufficient evidence” to state that oral antibiotics prevented meningitis. Subsequent reviewers, citing this article, have pointed to the reduced number of cases of meningitis as proof that antibiotics were effective and warranted.

In 2000, Alpern and associates retrospectively reviewed 5,900 children 2 to 24 months of age with temperature higher than 39°C (102.2°F). Patients with positive cultures for pathogenic bacteria were evaluated and their outcomes determined. Occult


bacteremia was seen in 1.9%. Ninety-six percent of patients with pneumococcal bacteremia had resolution of their bacteremia without the use of parental antibiotics. The rate of meningitis or death was 0.03%. These numbers were significantly less than the rates of bacteremia and meningitis determined in prior studies. The authors concluded that the conjugate Haemophilus vaccine and the ability to monitor blood cultures continuously, which allows for early detection of bacteremia, resulted in low rates of adverse outcomes from occult bacteremia. The authors suggested that these findings should be considered in management strategies. They also stated that the development of a conjugate vaccine against S. pneumoniae may “ultimately lay the issue to rest.”

In December 2000, in anticipation of the approval of the conjugate pneumococcal vaccine, Baraff published revised guidelines for the management of fever without a source in infants and children. The author noted that the anticipated widespread use of conjugate pneumococcal vaccine might very well make the use of WBC count and blood cultures, with empiric antibiotic treatment becoming obsolete.

Baraff recommended raising the temperature threshold for obtaining screening WBC to 39.5°C (103.1°F). He also suggested the following strategy:

  1. In patients 3 years of age and older, no testing is required for temperatures lower than 39°C (102.2°F) because the risk for bacteremia is less than 1%.
  2. Complete blood counts and blood cultures should be obtained in children younger than 3 years of age with fever without a source and a temperature higher than 39.5°C (103.1°F). Patients who have fever without a source who are 3 to 24 months of age with a temperature higher than 39°C (102.2°F) should also have these tests because the risk for bacteremia in this age group is about 3%.
  3. Urinalysis and urine culture should be obtained in all febrile males 6 months of age and younger and in uncircumcised males 6 to 12 months of age. All females younger than 12 months of age should also receive urinalysis and culture.
  4. In patients whose WBC count is more than 15,000/m3or whose absolute neutrophil count is more than 10,000 mm3, the risk for pneumococcal bacteremia is about 8%. In these patients, the administration of empiric antibiotics could be considered. Ceftriaxone has been recommended, although many practitioners will continue to use oral amoxicillin.

Future Considerations

Management strategies are not intended to be rigidly applied in every circumstance. Physicians have the option to tailor their management on the basis of such factors as follow-up, the presence of a telephone, and the ability of a patient to return for reevaluation the following day. Emergency room physicians often treat more aggressively than a private practice pediatrician who knows that he or she will be able to speak with the family in 6 hours. In the era of conjugate vaccines for both H. influenzae and S. pneumoniae, there continues to be a turn away from rigid protocols involving complete blood cell counts, blood cultures, and empiric treatment.


Many physicians currently advocate a “no test, no treatment” strategy with careful watchful waiting, as was done in the early 1970s before the beginning of the “bacteremia bandwagon.” As the incidence of bacteremia decreases, the emphasis may be more on observation and close follow-up than on laboratory evaluation and empiric therapy.

Selected Readings

Alpern ER, Alessandrini EA, Bell LM, et al. Occult bacteremia from a pediatric emergency department: current prevalence, time to detection, and outcome. Pediatrics 2000;106(3):505–511.

Baker MD, Bell LM, Avner JR. The efficacy of routine outpatient management without antibiotics of fever in selected Infants. Pediatrics1999;103(3):627–631.

Baraff LJ. Management of fever without source in infants and children. Ann Emerg Med 2000;36(6):602–614.

Fleisher GR, Rosenberg N, Vinci R, et al. Intramuscular versus oral antibiotic therapy for the prevention of meningitis and other bacterial sequelae in young, febrile children at risk for occult bacteremia. J Pediatr 1994;124(4):504–512.

Issacman DJ, Burke BL. Utility of the serum c-reactive protein for detection of occult bacterial infection in children. Arch Pediatr Adolesc Med 2002;156(9):905–909.

Long SS. Antibiotic therapy in febrile children: “best laid schemes.” J Pediatr 1994;124(4):585–588.

Roth Rock SG, Green SM, Harper MB, et al. Parental vs. oral antibiotics in the prevention of serious bacterial infections in children withStreptococcus pneumoniae occult bacteremia: a meta-analysis. Acad Emerg Med 1998;5(6)599–606.