According to results of a study, many office-based physicians do not routinely follow guidelines for management of febrile infants. Pantell and colleagues conducted a study among office practices to characterize the management, diseases, and outcomes related to febrile infants three months or younger; to develop a clinical prediction model to identify bacteremia and bacterial meningitis in these infants; and to compare actual office-based management with existing guidelines.

The study was conducted under the auspices of a practice-based research network of the American Academy of Pediatrics, the Pediatric Research in Office Settings (PROS). Eligible infants were three months or younger, had a temperature of at least 38°C (100.4°F), were not born prematurely, and were generally otherwise healthy. In this prospective cohort study, office staff and clinicians collected demographic and clinical data, recording the infant’s clinical appearance and management strategies. The primary outcome variable was occurrence of bacteremia and bacterial meningitis.

One prediction model used patient appearance alone, and two prediction models added white blood cell count or white blood cell count with urinalysis. A fourth prediction model used a blending of current guidelines. Based on the latter model, infants 30 days or younger and ill-appearing infants required a white blood cell count, blood culture, urine culture, urinalysis, cerebrospinal fluid analysis and culture, hospitalization, and antibiotic therapy; well-appearing infants older than 30 days required a white blood cell count and urinalysis. Finally, a fifth model used a tree-structured decision model derived from the analysis of the study’s sample.

The authors then calculated the sensitivity of these models on the basis of the number of infants with bacteremia/bacterial meningitis who would have been or were treated divided by the number of infants with bacteremia/bacterial meningitis. They also calculated a modified specificity based on the number of children not treated initially divided by all children without bacteremia/bacterial meningitis or other conditions requiring antibiotics.

Of 3,066 infants enrolled, 1,975 (64 percent) were managed as outpatients, and 125 had only a single office visit with no other medical follow-up. Children younger than one month were more likely to receive a white blood cell count or blood culture (83.0 versus 71.4 percent), to have a lumbar puncture (54.8 versus 25.6 percent), to begin immediate antibiotic treatment (68.2 versus 53.7 percent), and to be hospitalized (60.1 versus 27.3 percent). Other predictors of blood testing included temperature, appearance, care received after typical office hours, and Medicaid insurance.

Nearly one fourth of infants did not have their blood, urine, or cerebrospinal fluid tested; more than one half had their urine tested. Less than one half of infants were managed according to existing guidelines, regardless of age (see accompanying table). Bacteremia was present in 2.4 percent of infants with blood cultures, and bacterial meningitis was present in 0.5 percent of the entire sample; both conditions occurred with greatest preponderance in infants younger than one month.

According to logistic regression analysis, age and very ill appearance emerged as the strongest predictors of bacteremia and bacterial meningitis. To evaluate the predictive value of laboratory testing, focusing on children who had blood cultures, the authors found that white blood cell counts of less than 5,000 per mm³ (500 × 10⁹ per L) or at least 15,000 per mm³ (1,500 × 10⁹ per L) increased the predictive value. A predictive increase with urinalysis, although present, was not statistically significant. High-risk groups were younger, ill-appearing infants with a temperature of 38.6°C (101.5°F) or higher, while only 0.4 percent of well-appearing infants 25 days or older and a temperature of less than 38.6°C had bacteremia/bacterial meningitis.

Of the models analyzed, the PROS actual practice, which tested and treated younger, ill-appearing infants more aggressively, had equivalent sensitivity and specificity to current guidelines and involved fewer invasive procedures and hospitalizations than the guidelines. Practitioners missed only two cases of bacteremia, and both of these infants had a worsening course that led to immediate care the following day.

This study compared current guidelines for the management of febrile infants with actual, office-based strategies and found that in spite of broad discrepancies between guidelines and actual practice, office-based clinicians did as well as recommended guidelines while performing fewer tests and hospitalizing fewer infants. Available follow-up care is an essential component of these results.