Young, febrile infants: a 30-year odyssey ends where it started.

Abstract

THIRTY YEARS AGO, SENIOR PEDIATRIC RESIDENTS ASsigned to the emergency department (ED) at Johns Hopkins Children’s Center identified management of febrile infants younger than 2 months as one of the clinical problems they found most vexing. The residents believed that their clinical judgment was demeaned by the prevailing notion that all young, febrile infants need a complete evaluation for invasive bacterial disease, regardless of clinical appearance. They knew that pediatricians in practice caring for febrile infants performed fewer tests and hospitalized less. This dissonance between the rules in academic medical centers and the behavior in office practices prompted a study to demonstrate whether the residents could, in fact, identify which febrile infants had invasive bacterial disease, requiring hospitalization and intravenous antibiotic treatment, and which did not. Of the 61 infants enrolled during the 8 months of the study, one 17-day-old infant considered to be well appearing by the residents had group B streptococcal bacteremia. The study reinforced the notion that clinical judgment (at least of senior residents) could not reliably identify all infants with invasive bacterial disease. Subsequent studies performed in EDs over the next several years reached the same conclusion, leading in 1984 to the following “rules of management” regarding febrile infants aged 0 to 8 weeks: “1. All patients should be admitted to the hospital. 2. Assessment as ‘ill’ predicts the presence of a significant diagnosis. 3. Assessment as ‘well’ does not exclude the presence of a significant diagnosis. 4. Performance of laboratory tests in an outpatient is inappropriate. 5. Performance of blood culture in an outpatient is inappropriate.” While the rules minimized the chance of failing to treat an infant with unrecognized, potentially serious bacterial disease, the extensive hospitalization, testing, and treatment resulted in their own set of complications and cost. The search for predictors of serious illness to identify infants who needed treatment gave way to an alternative research strategy in 1985; ie, to identify febrile infants who do not need treatment. Various “low-risk” criteria were proposed, each set coming close to but not achieving complete safety. With the availability of ceftriaxone, a third strategy emerged in the early 1990s; ie, to identify infants who do not appear to need treatment—and treat them anyway, as outpatients. This strategy reduced hospitalizations, costs, and hospital-associated morbidity but continued to involve extensive testing and included widespread administration of a broad-spectrum, longer-acting antibiotic. During the 2 decades during which these various strategies were studied in EDs and guidelines were generated, dissent about rules persisted, even in the centers that generated the rules. What effect did the guidelines have on behavior in office practices? In this issue of THE JOURNAL, Pantell et al share the long-awaited results of a collaborative study conducted in pediatric practices using the American Academy of Pediatrics Pediatric Research in Office Settings (PROS) network. It appears that little has changed since the survey in 1973. The practitioners ordered fewer tests for their febrile infant patients than guidelines recommend, yet the “miss” rate of infants with bacteremia or bacterial meningitis was very low. How can the apparent differences between what is considered necessary in guidelines generated in academic center EDs and the good outcomes in office practices that have not adopted the guidelines be reconciled? Several possible explanations exist. First, it is possible that the sickest infants bypass offices and go to EDs (either directly or at the instruction of a private pediatrician), thereby reducing the incidence of bacteremia and meningitis in offices. Differences in population composition could also be operating to reduce the incidence of bacteremia and meningitis in the practice settings (mostly white, suburban families) compared with the patient population of inner-city EDs. The higher rate of hospitalization and treatment of infants with Medicaid insurance coverage suggests that the practitioners may have taken population differences into account. However, the authors of the PROS study point out that the incidence of serious bacterial illness in the offices was not lower than those reported from EDs, so distribution of disease does not appear to be the major factor. Second, it is probable that practitioners use a mechanism not as available in EDs as in office practices: the op-