Since the 1970s, the evaluation of a child younger than three years with a fever without a source has been problematic and controversial for physicians in the office setting. If the child is younger than one month or is toxic appearing, most physicians take an aggressive approach to management. The controversy exists for management of nontoxic-appearing children three to 36 months of age. Viral infection is the most common cause of fever in these children, but serious bacterial infection (SBI), which is difficult to rule out clinically, could be present. One study conducted in 1975 found that 3.2 percent of children four weeks to two years of age with fever without a source had bacteremia and were at risk of SBI.¹ It also found that white blood cell (WBC) count and the degree of fever were valuable predictors of bacteremia. Subsequently, studies in the 1980s and 1990s found that between 7 and 13 percent of children younger than three years without an evident source of fever had occult bacteremia, and that WBC count higher than 15,000 cells per mm³ (15 × 10⁹ per L) and fever of at least 102.2° F (39° C) correlated with higher risk.² Based on the results of these studies, practice guidelines for the management of fever without a source in infants and children 0 to 36 months of age were published, stimulating debate about management.² Questions were raised about the appropriateness of the guidelines, especially in the office setting, and several conflicting commentaries were published regarding their usefulness and whether they could or should be implemented.^3–6

At the time of the debates, Haemophilus influenzae type b (Hib) and Streptococcus pneumoniae were the most common isolates identified in occult bacteremia. The introduction of conjugate Hib vaccine and, subsequently, the conjugate pneumococcal vaccine resulted in a dramatic decrease of invasive disease caused by these pathogens, which called into question the continued need for the practice guidelines.² In this issue of AFP, Sur and colleagues undertake to develop an evidence-based approach to the management of fever without a source in children younger than 36 months of age in the era of universal Hib and pneumococcal vaccination.⁷ Although the rate of bacteremia (1.6 to 1.8 percent) has decreased significantly over time, the authors still recommend a cautious approach based on the potential adverse consequences of unrecognized and untreated SBI. Their algorithm offers two treatment options for nontoxic-appearing children with fever of at least 102.2° F; one option is more invasive than the other. The decision to give antibiotics is left to the discretion of the physician, but if administered, the authors recommend obtaining cerebral spinal fluid studies first. Their recommendations are based on evidence; however, there are still challenges to implementing these guidelines in the office.

One study found that reserving antibiotic therapy for culture-proved occult bacteremia in the post–Hibvaccine era was not associated with increased risk of developing an SBI.⁸ A separate cost/benefit analysis of the management of children with fever in the age of conjugate pneumococcal vaccine found that a complete blood count and selective blood culture and treatment using a WBC count cutoff of 15,000 cells per mm³ was cost-effective unless the pneumococcal bacteremia rate fell below 0.5 percent.⁹ Another study recently demonstrated that, in spite of a decline in pneumococcal bacteremia, other pathogens, including nonvaccine strains of S. pneumoniae, put children at risk of bacteremia.¹⁰ It also showed that WBC count only reliably predicts a higher risk of bacteremia in S. pneumoniae infection.¹⁰ The current low rate of bacteremia suggests that empiric testing and antibiotics may not be necessary. This could be reassuring for physicians in the office setting where it is neither practical nor feasible to draw blood, especially if the WBC count may no longer be a valuable predictor of risk. Similarly, it may not be feasible to give empiric therapy, especially if the recommendation is that cerebrospinal fluid studies be obtained before administering antibiotics.

Attempts to select out infants at low risk show promise, but they are not foolproof. One review on the use of rapid viral testing to distinguish febrile infants with viral illness from those with SBI found that infants with respiratory syncytial virus or influenza were at lower risk of SBI.¹¹ Rapid diagnosis of viral infection reduced the number of laboratory tests and radiography, and decreased antibiotic use, but the rate of urinary tract infection remained significant in infants with respiratory syncytial virus. However, there was insufficient evidence to change current clinical practice algorithms to reflect these findings.

The actual implementation of existing guidelines is not well known, and there are few data on what physicians actually do in their offices. Most of the data that are available were obtained in the emergency department setting. A 1998 survey examining the management of fever without a source in children younger than 36 months found that 59 percent of emergency room physicians, 45 percent of pediatricians, and 28 percent of family physicians gave empiric antibiotics to a child four months of age as recommended by the practice guidelines.¹² Overall, emergency physicians were most aggressive and family physicians were least aggressive in their management of fever without a source. Another study found that pediatric emergency department physicians were more likely to follow the guidelines than were general emergency department physicians.¹³

One question remains: how should physicians manage fever without a source in children three to 36 months of age? Some physicians believe that, although the incidence of occult bacteremia has significantly decreased, the number of patients with fever seen in the office will remain the same.⁶ The role of the physician will continue to be to identify the child with SBI, no matter how rarely it occurs. Continued surveillance for invasive pneumo-coccal disease and the causative serotypes is imperative. A population-based study of surveillance for pneumo-coccal serotypes causing invasive disease found that 48 percent were nonvaccine strains.^14,15 Other pathogens known to cause fever of unidentifiable source (e.g., Salmonella) may increasingly become a problem as well.¹⁰

In the end, guidelines are just that: guides to existing evidence. Each physician should interpret these data as they apply to his or her own practice. As suggested by Sur and colleagues, a cautious approach remains prudent because, if a child with occult bacteremia is missed, the consequences can be dire. As one commentary on practice guidelines points out, infectious disease experts tend to emphasize their experience with the occasional child who does poorly, whereas primary care physicians emphasize their experience with the hundreds of children who do well.¹⁶ Both perspectives are equally important; therefore, the debate continues.