Reply

Abstract

We thank Lwin and Bragonier for their letter and for their interest in our study. We would like to address the 3 points that they raise. First, “Important data such as CSF cell counts and whether patients developed other features of bacterial meningitis (BM) later in admission were not included.” Our gold standard for defining BM was the growth of a pathogen in the cerebrospinal fluid (CSF). Two investigators independently reviewed records of any patient with CSF growth. As stated in the article,The 2 investigators determined if the bacteria in the CSF represented a pathogen or a contaminant based on the specific organism, time to positivity, and detailed analysis of the patient's management by hospital physicians. In the event of a disagreement for any data point, the 2 investigators reached a consensus. In addition, crosscheck was performed between this database and microbiology laboratory records identifying all children of any age diagnosed with BM over the study period. In this context, CSF cell counts would not be helpful or accurate in determining whether or not a child had BM. Assuming no growth of organisms in the CSF, CSF pleocytosis could simply represent aseptic meningitis or a traumatic lumbar puncture. Further, for hospitalized infants, our methods would have detected a positive CSF culture at any point during the hospitalization. Second, “Data regarding when CSF was obtained in relation to antibiotic therapy was also not included. This would have been useful in validating their conclusion, as diagnoses of BM could have been missed if based primarily on culture results in infants who received antibiotics prior to lumbar puncture.” We apologize that this was not clearer in the article, but it is our prevailing practice in the emergency department to administer antibiotics after the lumbar puncture is performed. There were 73 infants who had their lumbar punctures performed in the inpatient setting; of these, 6 were treated for possible BM. Each infant had CSF pleocytosis and 5 of them failed to meet low-risk criteria. The 1 patient who met low-risk criteria presented to the emergency department with a chief complaint of vomiting and fever, looked well upon initial presentation, and remained well while in the hospital. The CSF studies were as follows: white blood cell count, 32; red blood cells, 8150, protein, 76; and glucose, 40. In addition, stool studies were positive for norovirus, to which the inpatient medical team attributed the fever and vomiting. The inpatient team documented their clinical suspicion for BM was extremely low given the CSF profile and the infant's well appearance, and the infant was treated with less than a full course of intravenous antibiotics. Third, “Finally, there was a lack of direct comparison between high and low risk groups.” We report that 45.6% of patients (542) met the low-risk criteria and none had BM. By extrapolation, 53.5% patients (646) had 1 or more factors placing them at high risk and 1 of these infants had BM. Further, the authors of the letter state, “without clarification, the paper does not appear to provide sufficiently strong evidence to induce a paradigm shift in current management of this population group.” Thank you for giving us the opportunity to provide that clarification. In the report, we are careful not to overstate our conclusions by providing this limitation: “Given the low overall incidence of BM, the sample size was relatively small.” Importantly, these data do not prove that febrile infants 29-56 days of age meeting low-risk criteria are at substantially lower risk for BM compared with others. However, this nearly 7-year experience demonstrates that febrile infants 29-56 days old cared for in a tertiary care referral center who otherwise met low-risk criteria are at very low risk for BM. A multicenter study including hundreds of thousands of infants would be necessary to determine the true prevalence of BM in low-risk patients compared with others and to assess the impact of excluding CSF analysis as a variable used to define low risk. Absent such a study, clinicians must continue to weigh the risks (lumbar puncture failure rates, traumatic lumbar punctures leading to unnecessary hospitalizations, increased durations of stay in the emergency department, and increased costs) and benefits (earlier detection of BM) of routinely performing the LP in otherwise low-risk febrile infants greater than 28 days of age. Febrile infants and lumbar punctures: unravelling the evidenceThe Journal of PediatricsVol. 195PreviewScarfone et al published a retrospective study that identified the incidence of bacterial meningitis in 1188 febrile infants aged 29-56 days of age undergoing sepsis screening.1 A secondary outcome was to determine whether those infants with bacterial meningitis met “low-risk” criteria. Forty (3.4%) infants had “contaminant” cerebrospinal fluid (CSF) cultures. One infant (0.08%) was diagnosed with bacterial meningitis with positive CSF culture. Analysis of 401 infants showed that 45.6% met low-risk criteria. Full-Text PDF