Management of fever without source in children: Changing times.

Management of the acutely febrile infant or toddler is one of the most common clinical problems that presents to paediatricians and family physicians. In most cases, a careful history and physical examination reveals findings of an uncomplicated focal infection. The etiology of such infections will be presumed with no laboratory testing (eg, upper respiratory tract infection that is presumed to be caused by a virus, or otitis media that is presumed to be caused by a bacteria). The recommended treatment will be based on the clinical findings only. Occasionally, however, there will be no apparent source for a young child’s fever and so the etiology is ‘occult’. Such children will appear well or only somewhat ill. In the majority of such cases, the cause of the fever is a virus and the illness will pass without complications. However, in a minority of cases, there will be a bacterial cause for the fever (eg, urinary tract infection, bacteremia or meningitis). In cases of fever without source, the physician must first decide which children need laboratory tests (and what tests to do) and then decide what immediate treatment to recommend. The choices range from supportive measures only to admission to hospital and treatment with parenteral antibiotics. In this issue of the Journal, Drs Girodias and Bailey ( pages 76–82) have reviewed much of the pertinent literature about this issue and present their own algorithm from the Hopital Sainte-Justine in Montreal. Their review is timely because of the virtual elimination of Haemophilus influenzae type b in Canada (1) and the potential for reduction in Streptococcus pneumoniae disease with the availability of the pneumococcal conjugate vaccine. The clinical and laboratory evaluations of children with fever without source remain imperfect tools to detect accurately those with bacteremia at the initial visit. Girodias and Bailey discuss the importance of careful clinical evaluation and highlight the use of validated clinical scores, including the Yale Acute Illness Observation Score for children from three to 36 months of age (2) and the Rochester Criteria for infants 60 days of age or younger (3). It is right to promote the use of these scores for at least two reasons. The first is that they do have reasonable predictive value (see the following paragraph for more on this). The second is that they are valuable teaching tools to help clinicians in training learn to use explicit and reproducible criteria to describe the appearance of a febrile child and to conclude whether the child appears ‘well’ or ‘ill’. What will be the impact of routine use of pneumococcal conjugate vaccine on the management of fever without source? The only Canadian jurisdictions with routine infant immunization programs for the pneumococcal conjugate vaccine are Alberta and Nunavut, and so it may be some time until the impact of this vaccine is known across Canada. In recent years, 1.5% to 2% of children with fever without source have been found to be bacteremic and 83% to 92% of these cases are caused by S pneumoniae (4,5). The importance of S pneumoniae as a cause of fever without source will eventually decline, but because nearly all children with fever without source do not have a bacterial infection, pneumococcal conjugate vaccine use should not be expected to decrease the number of children presenting with fever without source. The results of a recent cost effectiveness analysis that evaluated cases of meningitis prevented and life-years saved concluded that if the rate of occult bacteremia falls below 0.5%, strategies to automatically perform blood cultures and leukocyte counts on febrile infants (and provide antibiotics to all those tested or to a subset with elevated leukocyte counts) will no longer be appropriate. The only appropriate strategy will be to use clinical judgement (using the Yale Acute Illness Observation Score) to determine whom to perform blood cultures on and to treat with empiric antibiotics (6). Girodias and Bailey highlight the contradictory recommendations of the numerous published clinical guidelines, meta-analyses and decision analyses about how to manage fever without source most appropriately. They contend that the reason for these contradictions is that the scientific data are contradictory and unclear. However, contradictions in the data alone are not sufficient to explain inconsistent practices and at least two other factors must be considered. The first is that there are differences in the degree of uncertainty that different physicians will accept in the management of children with fever without source. The risk that meningitis or another serious complication would develop in a mildly ill child with fever without source was low (approximately 3%) in the era of H influenzae type b infections and is even lower now, because S pneumoniae bacteremia is much less likely than H influenzae type b to progress to a focal invasive infection such as meningitis (7,8). For a physician in an emergency department, evaluating a febrile child whom he or she has never seen before, and for whom appropriate follow-up cannot be assured, the concern about the risk of an adverse outcome to the illness is likely higher than that for a physician in office practice evaluating a child well known to them and for whom follow-up can be assured (9,10). Thus, emergency department physicians are justifiably more likely to investigate children with fever without source and treat them with antibiotics than are office-based physicians. A second, related factor is that there are differences between the aims of management by individual physicians for individual children and the objectives of guidelines and decision-analysis models. One study has shown that the primary aim of paediatricians in the clinical setting is early detection of meningitis, more so than the prevention of meningitis or other focal infections (11). This highlights that the current literature does not adequately address this important aim and that further work to develop rapid and accurate diagnostic tests is needed. Girodias and Bailey conclude by describing their practice in the large emergency department at Hopital Sainte-Justine. Their approach differs little from the guideline of Baraff et al (12) except that Girodias and Bailey’s approach is less explicit about the information obtained and used to decide whether to give antibiotics and/or whether to admit to hospital. It would be of great interest to know how universally their guideline is applied in their department, to know what proportion of children with fever without source are treated with antibiotics and to know what follow-up arrangements are typically made for children not admitted to hospital. Such data from a Canadian centre would be a valuable contribution.