Fever is a common presenting complaint among pediatric patients, accounting for approximately 20% of emergency department (ED) visits by children.1,2 Hence, management of the febrile child is a challenge faced by emergency physicians on a daily basis. Despite the fact that the vast majority of children with fever have self-limited viral illnesses,3 there is a finite number who may harbor serious bacterial illnesses (SBIs), and, in many cases, these patients are clinically indistinguishable from the rest. The emergency physician's challenge is to identify and treat those children who have SBIs while avoiding overtreatment with antibiotics of those without SBIs, thereby limiting the propagation of antimicrobial resistance. Making this distinction is particularly difficult early in the course of a febrile illness. In addition, this decision process is often conducted in the setting of a family with "fever phobia." Many myths regarding fever exist among the general public, and these misconceptions are often reinforced by the mixed messages that we in the medical community provide. Assessing the risk of SBI to an individual patient, selectively making reasonable diagnostic and therapeutic interventions, and simultaneously reassuring and educating families regarding appropriate concern for fever can make what appears to be a routine common complaint an important and challenging encounter.
Some instances of fever in children require simple decision making. When a child with fever has an evident source of infection, such as acute otitis media or acute gastroenteritis, decisions are relatively straightforward: treat the source and manage the patient's condition appropriately. In the case of the febrile patient with an underlying medical condition (such as sickle-cell disease) or indwelling hardware (such as a central venous catheter), diagnostic investigations and empiric therapy are usually protocoldriven. These circumstances place the patient at greater risk for SBI, and more aggressive management is apropos. This conservative approach extends to the youngest infants (less than 2-3 months of age), who have yet to develop a fully competent immune response. Finally, any patient who appears "toxic" demands a comprehensive search for the source of fever and empiric broad-spectrum antibiotic coverage until the clinical picture clears. This is true whether the patient is 45 days or 45 years of age.
Like the child in our vignette, however, it is the febrile pediatric patient without a readily identifiable source of infection, an unremarkable medical history, and a nontoxic appearance who can be the most challenging. What is this patient's risk of SBI? Are there laboratory tests that can guide us in pinpointing those at risk? Who should receive antibiotics? And what is an appropriate disposition and followup plan for these patients?
The story of occult infection in children is an evolving one, and practice has changed over the past 30 years. Much of the initial literature regarding fever in the 3-year-and-under age group focused primarily on the identification of clinically inapparent infection in the form of "occult bacteremia" and the effort to prevent the potentially serious sequelae of bacteremia, such as meningitis, osteomyelitis, or pneumonia. Early investigations predated the availability of broad-spectrum parenteral antibiotics such as ceftriaxone, technology enabling continuous monitoring and detection of microorganisms in culture media, and development and widespread implementation of immunizations against the more common pathogens. As the landscape of occult infection in children has changed, more recent literature has attempted to take these factors into account, modifying recommendations and expanding the focus to include newer resistant organisms and the identification of other "occult" infections, such as urinary tract infection (UTI) or pneumonia.
Many ED physicians predicate their approach to febrile children on the practice guidelines outlined in a landmark 1993 article that appeared simultaneously in both Pediatrics and the Annals of Emergency Medicine. Because of their prominent display in the journals published by the American Academy of Pediatrics and the American College of Emergency Physicians, these guidelines had a certain voice of authority and quickly became a de facto standard of practice. A panel of experts chosen by the primary author performed a review of the existing literature at that time and arrived at recommendations on how to approach children of various ages with fever. These guidelines included two recommended options in the pursuit of occult bacteremia for children 3 to 36 months of age with a fever of 39°C (102.2°F) or greater without an identifiable source of infection: 1) obtain a blood culture and administer empiric treatment with parenteral antibiotics (ceftriaxone) pending culture results in all children meeting the above criteria; or 2) selectively culture and treat those whose white blood cell count (WBC) exceeds 15,000 muL. In addition, urine culture obtained by catheterization or suprapubic aspiration was recommended for all boys less than 6 months of age and all girls younger than 24 months.4-5
Historically, much of the medical literature that laid the groundwork for this approach to occult infection in children originated in the 1970s and ‘80s and was a patchwork of sometimes flawed and inconsistent data. Initial reports simply described bacteremia rates as they varied by patient age and height of fever and characterized the primary offending organisms in a variety of population samples.6-19 Most were gathered from patients seen in emergency departments and outpatient clinics, not in private practitioners' offices, a fact that injected a healthy dose of selection bias. None of the studies that applied a temperature threshold for initiating a fever workup accounted for prior use of antipyretics or subjective parental reports of fever in assessing bacteremia risk. Thus initial estimates of occult bacteremia rates in children less than 36 months of age were likely overstated and did not represent true prevalence data.20-22 Nonetheless, they laid the groundwork for subsequent efforts to find and stop bacteremia in its tracks.
Furthermore, the 1993 guidelines are the result of a meta-analysis of existing studies, so they are only as good as the studies upon which they are based. The inclusion criteria (age, height of fever, etc.), the laboratory tests performed, the degree of WBC elevation associated with bacteremia, and the use of empiric antibiotics varied from study to study, making comparative analysis problematic. Some investigations lumped patients who had an identifiable source of infection (such as otitis media or pneumonia) with those without an apparent source on exam, which inevitably confounds interpretation of the results. Patients with a presumed bacterial source of infection would be expected to have a greater rate of bacteremia, and they would likely receive antibiotic therapy regardless of their WBC. In fact, because most of these studies did not randomly treat or not treat children with antibiotics, the group of children who were treated often already had one of the outcomes of interest and thus had a lower probability of subsequently developing a new focus of infection, biasing the outcomes of these studies in favor of antibiotic treatment.23
The other presumption of the guidelines is that therapy with antibiotics (oral or parenteral) is effective in preventing sequelae, particularly meningitis, and it is not clear that this has been proven.24-33 After the guidelines appeared, editorials written by prominent pediatric infectious disease specialists warned against the blanket use of ceftriaxone as a panacea.23,34-36 In some of the studies promoting expectant antibiotic treatment, for example, recommendations were based upon the outcomes of the subset of patients with positive blood cultures and not the population of febrile children at risk for bacteremia as a whole. Naturally, few would quibble about treating patients with demonstrated bacteremia; but the issue—particularly for the emergency physician—remains to reliably identify which patients have bacteremia and selectively treating those. Even in the best case scenario, blood culture results are not available for 12-24 hours after they are obtained and are, therefore, not helpful in front-end decision making. To date, no readily available laboratory test(s), including the WBC, has been discovered that consistently and accurately positively predicts the presence of bacteremia. So, for the physician confronting the child with fever in real time, the question remains: who, if anyone, do you treat expectantly?
After the 1993 guidelines appeared, several surveys of the practicing medical community were circulated to assess their impact. It rapidly became clear that many emergency physicians were either unaware of the guidelines or actively chose not to follow them.37-38 This was true not only for pediatric emergency physicians, but for general emergency physicians and primary care practitioners as well.39-40 Further, those who were aware of and invoked the guidelines did not always apply them consistently.41 The use of ceftriaxone became widespread, in many instances indiscriminate and not in accordance with the published guidelines—the proverbial hammer for every nail that presented itself. This may have stemmed from the option, suggested by the 1993 guidelines, to treat everyone at risk (i.e., with a fever greater than 102.2°F without an obvious source). But many physicians obtained screening laboratories on patients, disregarded the (normal) results, and administered ceftriaxone anyway. While this strategy may provide an immediate sense of security for the emergency physician, who may feel that s/he has limited his or her personal liability and protected the patient in giving ceftriaxone, s/he may simultaneously be tying the hands of his or her partners in primary care and painting us all into the corner of antibiotic resistance in the long run. As was pointed out by early critics of ceftriaxone use in the emergency department, once this long-acting, broad spectrum, blood-brain barrier-crossing antibiotic has been administered, the parents and the primary care physician providing follow-up evaluation are robbed of their abilities to assess the child's clinical condition or need for continuing therapy.34-35 And even if we grant that one or two doses of parenteral ceftriaxone are effective in treating bacteremia, two doses of ceftriaxone would be inadequate to treat meningitis if the child had already seeded the meninges. Clearly there is no easy or right answer to the question, to treat or not to treat?
Complicating the picture is the falling prevalence of SBIs as immunizations against the more common offending organisms—Haemophilus influenzae type B (HIB) and Streptococcus pneumoniae—have been developed and implemented on a widespread basis.42-44 As rates of bacteremia and invasive infections due to these agents decline and, concomitantly, as the levels of resistance to our current antibiotics rise (witness the prolific emergence of MRSA and drug-resistant S. pneumoniae), management strategies we learned during our training years have become outdated and may no longer apply. The landscape of fever in children is constantly evolving, and the emergency physician must adapt his or her approach accordingly. This is not always easy, as old habits die hard. A recent study by Cox et al. highlighted that physicians tend to adhere to published guidelines or algorithms they were exposed to during their residency training, despite the appearance of newer or contradictory findings in the medical literature.45 Though it is difficult to reconsider what was once dispensed as gospel, it is incumbent upon practicing physicians to modify their approach to the febrile child as new data and therapies emerge.
Fortunately, the guidelines have been appropriately revisited and modified to reflect the current situation.46-52 While some current investigators persist in the attempt to build a better mousetrap for predicting SBI than the WBC (the absolute neutrophil count [ANC], C-reactive protein [CRP], and various cytokines have been posited as more appropriate substitutes),53-62 these newer laboratory indices are rapidly becoming weapons in search of a war. Vaccination effectiveness has led several commentators to suggest that the search for occult bacteremia may already have become the medical equivalent of tilting at windmills.48,50-51 Hence, the emphasis in more recent literature on fever in this 3-to-36-month age group is on detecting other sources of occult infection, such as UTI.63-69
1. "The girl's urine dipstick was nitrite- and leukocyte esterase-negative so I sent her home without antibiotics and canceled her urine culture."
2. "I know his white blood cell count was only 12,000/L, but with a fever of 103°F, I felt it best to give him a shot of ceftriaxone before discharging him."
3. "She's got minimal upper respiratory symptoms, but her RSV is positive, so I've got my source of infection."
4. "The child's white blood cell count came back from the lab at 18,500/L, but he looked great and Mom was anxious to go home, so I discharged him without antibiotics. I hope he's OK."
5. "Billy has a raging left otitis media, but I don't think that fully explains his fever of 104.2°F. I'm sending a CBC on him."
6. "Josè and his family just moved to the States and don't have a PCP yet. Despite his fever, he looked >pretty good, so I told them they need to find themselves a doctor and sent them on their way."
7. "I know Keisha has sickle-cell disease, but her temperature is only 101°F and everybody in her family has got a cold right now. I think we're safe calling this thing a virus and having her follow up on Monday."
8. "Two-year-old Joey was transferred here for fever, a WBC of 22,000/L, and belly pain, but his urine is clean and his abdominal CT was negative. He looks good now after IV fluids and some acetaminophen. I think he's just got a virus."
9. "Mom says 18-month-old Sandra's fever at home was 103°F and she looked just terrible, but here in the ED she's playful and afebrile. No way she has a bacterial infection."
10. "I'm waiting on the CBC. If his WBC is up, I'm going to tap this kid."
Evidence-based medicine requires a critical appraisal of the literature based upon study methodology and number of subjects. Not all references are equally robust. The findings of a large, prospective, randomized, and blinded trial should carry more weight than a case report.
To help the reader judge the strength of each reference, pertinent information about the study, such as the type of study and the number of patients in the study, are included in bold type following the reference, where available.