White Blood Cell Counts in Neonatal Early-Onset Sepsis

Abstract

Reply: We appreciate the letter from Resch et al and would like to respond to the concerns raised. They describe being “astonished” that previous studies have not been able to discriminate infected from uninfected neonates. And indeed, ours is the first large study to approach this question from the point of view of trying to determine which infants are likely not infected. Almost all other studies, including the referenced study,1 have primarily focused on the use of laboratory tests to identify the infected neonate. This is an important difference. Our hope was that if we could find a subset of infants evaluated for sepsis that were not infected at less than the typical 48- to 72-hour time frame of antibiotic treatment, broad benefits could ensue including shorter hospitalization, less invasive procedures (intravenous attempts) and perhaps most importantly, less antibiotic exposure. The retrospective analysis they describe attempted to evaluate whether the criteria we used held true for their population. There were, however, 2 critical differences. Their population included infants up to 72 hours of life. Our database only included infants evaluated for sepsis at <24 hours of age and almost always at time of birth. Most infants in this situation are asymptomatic and started on antibiotics for risk factors only. This is a very different population from infants who may become symptomatic at 24–72 hours of life. We made no attempt to extrapolate our data to any group of infants at >24 hours of age. Second, their infants had “at least one WBC” performed. As referred to in our study, a single, initial white blood cell count is fraught with poor sensitivity, specificity, positive and negative predictive values. In addition, we have no data on the use of white blood cell counts at 24–72 hours of age, only when obtained in the first hours of life and repeated in 8–12 hours. It is also of interest that their baseline infection rate is higher than reported in most reports (5.3%). Some of this may be due to the fact that 25% of their infections are caused by Ureaplasma urealyticum. This organism is rarely cultured for in the United States, and it is still undetermined whether treating this organism is associated with less or greater risks than not treating when it is identified. In conclusion, we would hope that our data are applied to the population described in the article only. Extrapolation to other ages is risky until they are specifically studied. Infants evaluated for infection in the first hours of life are a unique population in that they are almost always asymptomatic and treated with antibiotics due to maternal risk factors. The exact number of infants in this category is impossible to state accurately, but based on the limited published studies referred to in our article, it can be estimated to be well over one half million in the United States alone. Because so many infants are being exposed to antibiotics either from maternal treatment, treatment begun at birth or both, anything that might enable clinicians to decrease this exposure would be beneficial. We believe our study suggests that this may be possible in a subset of these infants. Kara Murphy, MD New Hampshire’s Hospital for Children Elliot Health System Department of Neonatology Manchester, NH Joel Weiner, MD University of Massachusetts Memorial Children’s Medical Center Division of Neonatology Worcester, MA