Reducing unnecessary antibiotic exposure among neonates evaluated for suspected early-onset sepsis through an adapted risk-stratification and antimicrobial stewardship pathway in northwest Ethiopia: a prospective quality improvement study

Abstract

BACKGROUND: Early-onset sepsis (EOS) is a leading cause of neonatal morbidity and mortality and contributes substantially to empirical antibiotic use in neonatal intensive care units (NICUs). In low-resource settings, limited diagnostic capacity often results in overtreatment, increasing unnecessary antibiotic exposure and its associated risks. Evidence on context-adapted EOS risk assessment pathways in sub-Saharan Africa remains scarce. OBJECTIVE: To evaluate the impact of implementing an adapted EOS risk assessment pathway on antibiotic use and clinical outcomes among neonates admitted to NICUs in Northwest Ethiopia. METHOD: A prospective multicentre quasi experimental pre-post-implementation study was conducted in three tertiary NICUs in Northwest Ethiopia from January to April 2026. A total of 412 neonates aged ≤72 hours and born at ≥34 weeks' gestation with suspected EOS were included (pre-implementation, n=208; post-implementation, n=204). An adapted EOS risk assessment and antimicrobial stewardship pathway was implemented during the post-implementation phase. Differences between groups were assessed using χ², Fisher's exact, independent t or Mann-Whitney U tests, as appropriate. Logistic regression and Cox regression were used to evaluate empirical antibiotic use and 28-day mortality, respectively. Statistical significance was declared at p<0.05. RESULTS: Maternal and neonatal characteristics were similar between study phases. Empirical antibiotics intuition decreased from 80.8% to 55.4% (risk ratio (RR)=0.69, 95% CI 0.60 to 0.80; p<0001), prolonged antibiotics treatment (>5 days) decreased from 40.8% to 18.1% (RR=0.44, 95% CI 0.31 to 0.62; p=0.004), antibiotic discontinuation within 48 hours increased from 8.2% to 15.7% (RR=1.90, 95% CI 1.1=3.34; p=0.02). Complete blood count, C-reactive protein and blood culture testing also decreased (all p<0.001), and mean NICU safety was shorter after implementation (8.9±2.9 vs 8.1±2.5 days; p=0.002). In multivariable analysis, implementation of the EOS pathway was associated with lower odds of empirical antibiotic use (adjusted OR 0.29, 95% CI 0.19 to 0.46). No significant adverse clinical outcomes difference between study phases. 28-day mortality did not differ between groups (15.9% vs. 17.2%; p=0.7). CONCLUSIONS: Implementation of an adapted EOS risk assessments and stewardship pathway significantly reduced unnecessary antibiotic exposure and laboratory testing without observed increase in adverse clinical outcomes. These findings support the use of structured EOS risk assessment to improve antibiotic stewardship in NICU in resource-limiting settings.