Feasibility of manual white blood cell counts as a predictor of neonatal sepsis in a low-resource setting

Abstract

Background: Manual white blood cell (WBC) differential counts as a predictor for neonatal sepsis development in a low-resource setting have not been thoroughly evaluated. We hypothesized that manual differentiation (specifically immature:total [I:T] neutrophil ratios) would be feasible and useful as an adjunct to predict early-onset neonatal sepsis (EONS). Secondarily, we hypothesized that vaccination with bacillus Calmette-Guérin (BCG) and oral polio vaccine (OPV) could alter WBC differential counts and thus might reduce its predictive performance.

Methods: We performed a prospective cohort study within a randomized trial, randomizing healthy, high-risk newborns admitted to the nursery at the national hospital in Guinea-Bissau 1:1 to BCG+OPV at admission or at discharge (usual practice). Thin capillary blood films were prepared at 2 d of age in a subset of 268 neonates. WBC counts were assessed by microscopy and neonates were followed up for sepsis development within 2 weeks.

Results: Ninety-eight percent (264/268) of smears provided interpretable reads. Of the 264 children, 136 had been randomized to receive BCG+OPV prior to sampling; the remaining 128 were vaccinated at discharge. The I:T ratio (average 0.017) was lower among children who did not develop clinical sepsis but did not predict sepsis (p=0.70). Only three children had an I:T ratio >0.2 (associated with a higher probability of clinical sepsis in previous studies) but did not develop sepsis. Immunization did not alter WBC composition.

Conclusions: Manual WBC differentials are feasible in low-resource settings. WBC differentials are not affected by standard newborn immunization. However, the I:T ratio had no value in predicting subsequent development of sepsis.

Keywords: BCG Vaccine; leukocyte count; neonatal sepsis; newborn infant; oral poliovirus vaccine; western africa.

Figure 1

Study process overview with target event time after birth in hours (mean).

Figure 2

WBC counts for healthy control newborns and newborns diagnosed with EONS or LONS. Boxplots indicate medians with first and third quartiles (25%, 75%). Whiskers extend no further than 1.5×interquartile range from the hinge.

Figure 3

I:T ratio calculated from WBC counts among healthy control newborns and newborns diagnosed with EONS or LONS. Boxplots indicate medians with first and third quartiles (25%, 75%). Whiskers extend no further than 1.5×interquartile range from the hinge.