Young infants with fever are at risk for serious bacterial infection, but no consensus exists on the optimal approach to diagnosis and treatment. Although the traditional recommendation is always to perform all sepsis tests, including lumbar puncture, and administer intravenous (IV) antibiotics until culture results are negative, recent studies suggest administering intramuscular (IM) ceftriaxone with outpatient follow-up or using laboratory and clinical data to exclude low-risk patients from hospitalization, further testing, and antibiotic treatment. A decision analysis model was used to evaluate six strategies for the diagnosis and treatment of infants aged 28 to 90 days with temperature ≥38.0°C. Data from the literature, data from a 1991 study of 503 febrile infants, and direct, short-term costs from the Children's Hospital of Philadelphia were used as model inputs. The model was run for a hypothetical cohort of 100 000 febrile infants who did not require admission for focal infection or for other reasons that clearly necessitated admission. The model included six strategies: (1) no intervention; (2) all sepsis tests (lumbar puncture, blood culture, urine culture, white blood cell count, and urinalysis) followed by hospitalization and IV antibiotics for all infants; (3) all sepsis tests followed by IM ceftriaxone and outpatient management for most infants; (4) blood and urine cultures with white blood cell count and urinalysis followed by either lumbar puncture and IV antibiotics for high-risk infants or outpatient management without antibiotics for low-risk infants; (5) white blood cell count and urinalysis followed by either lumbar puncture, blood and urine cultures, and IV antibiotics for high-risk infants or outpatient management without antibiotics for low risk infants; and (6) clinical judgment followed by either all sepsis tests and IV antibiotics for high-risk infants or outpatient management without antibiotics for low-risk infants. The two "all sepsis tests" strategies prevented the most cases of death or neurologic impairment, 78% (when IV antibiotics were used) and 76% (when IM ceftriaxone was used) of all potential cases. The most cost-effective strategy was to use all sepsis tests followed by IM ceftriaxone for all patients without meningitis, at an incremental cost of only $3900 per sequela prevented relative to no intervention. Strategies under which only those patients selected as high-risk by laboratory criteria received antibiotic treatment were less effective but incurred lower rates of antibiotic complications. Clinical judgment alone was the least clinically effective and the second least cost-effective strategy. The model's results depended most on assumptions about the effectiveness of IM ceftriaxone, the sensitivity of the white blood cell count, and the sensitivity of clinical judgment in identifying young infants with serious bacterial infection. Combining all sepsis tests with IM ceftriaxone has superior clinical and cost-effectiveness compared with other strategies for managing febrile infants in this model. Strategies that use selective antibiotic treatment based on laboratory tests or clinical judgment are acceptable when the sensitivity of the criterion used is high, but they do not surpass strategies that combine all sepsis tests and antibiotic treatment until the criterion's sensitivity is greater than 96%.
- Received June 17, 1991.
- Accepted November 18, 1991.
- Copyright © 1992 by the American Academy of Pediatrics