Analysis of early-onset sepsis and antibiotic exposure among very preterm infants: a multicenter retrospective study (2022–2024)

Early-onset sepsis (EOS) is a major threat to very preterm infants (VPIs) (< 32 weeks gestation), with high incidence and mortality rates. Atypical clinical symptoms lead to widespread empiric antibiotic administration and subsequent excessive antimicrobial exposure. Real-world data on antibiotic consumption and inter-hospital variation among these infants are limited. Based on multicenter retrospective data, this study aimed to describe the current profile of EOS-associated antibiotic use and explore practice discrepancies across hospitals in a regional Chinese cohort. A multicenter retrospective observational study was performed using data from the Shenzhen Neonatal Network spanning 2022 to 2024, enrolling VPIs admitted to six NICUs. Collected variables included demographic information, EOS incidence, antibiotic utilization indicators (initiation rate, treatment duration, standardized antibiotic exposure), and clinical outcomes. Descriptive statistics were applied for data summarization, and Spearman’s rank correlation was calculated at the hospital level to preliminarily explore the association between antibiotic exposure and EOS incidence. Given the limited number of participating centers, correlation findings were interpreted descriptively only. A total of 1202 VPIs were enrolled, of whom 91.9% received early antibiotics within 3 postnatal days with a median antibiotic course of 7 days. The overall EOS incidence was 15.5% (95% CI: 13.5–17.6), and the ratio of early antibiotic recipients to confirmed EOS cases was 5.9:1, varying markedly across hospitals (range: 4.2–12.9). Considerable inter-hospital heterogeneity was observed across key indicators: early antibiotic initiation rate (83.5% – 100%), antibiotic continuation rate (54.9% – 100%), and standardized antibiotic exposure (5374–11738 antibiotic-days per 1000 VPIs). Hospital-level Spearman’s correlation revealed no statistically significant association between antibiotic exposure and EOS incidence ( r s = 0.058, P = 0.913). Findings from this regional cohort show relatively high early antibiotic exposure among VPIs that appears mismatched to the actual EOS burden, with substantial inter-hospital heterogeneity in prescribing. This pattern may align with inconsistent, non-standardized antibiotic management for EOS across centres. Our descriptive data characterise real-world antimicrobial use and offer preliminary evidence and testable hypotheses to inform future prospective research and the design of risk-stratified, individualized antimicrobial stewardship (AMS) strategies.