Neonatal sepsis remains a leading cause of death during the first month of life despite advances in diagnosis, treatment, and prevention, particularly in low-resource settings. This study aimed to assess clinical outcomes and identify predictors of neonatal mortality among neonates admitted with neonatal sepsis to selected public hospitals in Addis Ababa, Ethiopia.
A hospital-based prospective cohort study was conducted among neonates admitted with neonatal sepsis to selected governmental hospitals in Addis Ababa. The enrollment period ran from September 2024 to January 2025. All neonates who were clinically or laboratory-diagnosed with neonatal sepsis by the attending physician and admitted to the selected hospitals were included. The final sample comprised 466 neonates.
Clinical and demographic data were collected for each neonate and entered into EpiData 4.2. Analysis was performed in SPSS version 26. The investigators used bivariate and multivariate Cox regression to evaluate associations between independent variables and the time-to-event outcome. Variables with p ≤ 0.05 in multivariate analysis were considered significant predictors of poor clinical outcome (death).
Among the 466 neonates admitted with neonatal sepsis, 372 (79.8%) were discharged with good outcomes, while 94 (20.2%) experienced a poor outcome and died during the study period. The study describes this proportion of mortality as moderately high within the study population.
Multivariate Cox regression identified several factors significantly associated with increased risk of death among neonates with sepsis. Reported adjusted odds ratios (AOR) and 95% confidence intervals (CI) from the source are as follows:
Prolonged duration of ruptured membranes (>12 hours): AOR = 7.02 (95% CI: 1.85–26.57). This factor was associated with a substantially higher risk of neonatal death compared with shorter rupture-to-delivery intervals.
Rural residence: AOR = 6.05 (95% CI: 2.03–16.53). Neonates from rural areas had greater odds of mortality in the study cohort.
Maternal marital status — divorced: AOR = 3.12 (95% CI: 1.67–7.45). Maternal divorced status was associated with increased neonatal mortality relative to the reference marital status.
Assisted instrumental delivery: AOR = 5.99 (95% CI: 1.46–17.11). Instrumental delivery was associated with higher odds of death among neonates with sepsis.
Meconium-stained amniotic fluid: AOR = 9.48 (95% CI: 0.49–18.61). The source reports this large point estimate; the confidence interval as provided crosses a wide range.
No initiation of exclusive breastfeeding within one hour: AOR = 3.20 (95% CI: 0.90–7.52). Delayed initiation of exclusive breastfeeding within the first hour after birth was reported as associated with increased mortality in the multivariate model.
Chest in drawing: AOR = 5.81 (95% CI: 1.75–11.23). Presence of chest indrawing was associated with higher odds of death among neonates with sepsis.
These reported associations derive from the study’s multivariate Cox regression results and were identified as statistically significant according to the study’s criteria (p ≤ 0.05).
In this cohort of neonates admitted with neonatal sepsis in selected public hospitals of Addis Ababa, mortality was 20.2%. The authors conclude that neonatal mortality in this setting was moderately high. Key predictors of death reported by the study include meconium-stained amniotic fluid, prolonged rupture of membranes (>12 hr), instrumental delivery, chest in drawing, rural residence, delayed initiation of exclusive breastfeeding within one hour, and maternal divorced status. These factors were associated with increased odds of poor clinical outcome in the multivariate analysis presented by the authors.
The study received ethical approval from the Ethical Review Committee of Addis Ababa University (approval number AAU/R/E/A/05/405/2024). Informed consent was obtained from respondents prior to data collection. The authors state that the data supporting the findings will be made available by the corresponding authors upon reasonable request.
Notes: All results, numeric values, adjusted odds ratios and confidence intervals are reported as presented in the source article. No additional data or analyses beyond the source were introduced.