Authors’ Response

We thank Dr Bonadio for his commentary on the new American Academy of Pediatrics Clinical Practice Guideline. He expresses several concerns, which we will address one by one.

First, our inclusion of the PECARN study that generated low-risk criteria with only 10 cases of bacterial meningitis, noting that researchers in a subsequent study using only the criteria identified in the PECARN study missed infants with bacteremia and meningitis. Response: We would note that the PECARN study he cites identified only 3 predictor variables (urinalysis, ANC, procalcitonin) using recursive partitioning, but the clinical practice guideline relies on clinical appearance, level of temperature elevation, urinalysis, and combinations of 3 laboratory IMs (ANC, C-reactive protein, procalcitonin). Recursive partitioning analysis, the statistical technique used by PECARN and others investigating febrile infants,1,2 generally derives only 3 or 4 of the most powerful predictors because of a limited number of cases of bacterial meningitis and bacteremia. However, by using multivariate analysis, additional positive and negative predictors have been identified.2,3 Clinicians have access to multiple sources of historical and clinical information that allow for a more nuanced approach to each infant, as reflected in the opening statement of the Guideline. Moreover, although it is correct that 10 infants with meningitis is too small a sample (with too wide a confidence interval) on which to base an evidence-based guideline, the PECARN study was only one ofmultiple studies supplied by numerous investigators with large international, national, and regional data sets. There were many dozens of infants with bacterial meningitis and hundreds with bacteremia in our final analysis.

Second, the prospect of “partially treated” meningitis if an LP is not performed but antimicrobial therapy is initiated. Response: This problem is exacerbated by the difficulty obtaining interpretable CSF in young infants and caused considerable discussion. We acknowledge that the “risk” is real but sufficiently low to justify the key action statements as written.

Discharging an infant with possible “partially treated” meningitis at 24 to 36 hours because all culture results (ie, including CSF) are negative. Response: We presume concern for meningitis when interpretable CSF is not available would be based on a positive blood culture result, so termination of antimicrobial therapy at 24 to 36 hours would not apply. Moreover, the infant must be well or clinically improved. Certainly, if concern is high, we would expect clinicians to invoke the “variation” and “no other reason for hospitalization” clauses and continue hospitalization and treatment.

Third, UTI should be considered an “evident source of infection,” meeting exclusionary criteria for guideline applicability. Response: “Evident sources” include omphalitis and cellulitis, detected by inspection, and pneumonia, presumably signaled by an increased respiratory rate and or other physical findings. UTI is an unknown source until urine is analyzed, similar to meningitis being an unknown source until CSF is analyzed.

We hope the Febrile Infant Clinical Practice Guideline will be used as a starting point for addressing febrile infants on the basis of current evidence but not be considered an exclusive course of action. As noted by the multiple areas for research identified, this is a first step, not the final word.