Risk prediction of healthcare-associated neonatal sepsis in Amhara, Ethiopia: A prospective cohort study.

Background: Neonatal sepsis is a major contributor to neonatal mortality in low- and middle-income countries. Globally, rapid diagnosis and treatment are often a challenge, and standard antibiotic therapy is threatened by antimicrobial resistance (AMR). This study quantifies the predictors of healthcare-associated neonatal sepsis in Amhara, Ethiopia. Methods: This prospective cohort study recruited normal and low birth weight (LBW) neonates from two hospitals. Neonates with suspected healthcare-associated sepsis had blood cultures drawn to identify the organisms and AMR patterns. We used univariable and multivariable logistic regression to determine risk factors for culture-confirmed sepsis. Next, we performed marginal effects analysis to create a clinical tool for prospectively measuring newborn sepsis risk. Results: Twenty percent of 605 neonates developed healthcare-associated, culture-confirmed sepsis. 44.9% were resistant to first-line empiric therapy, and only 4% of those who failed first-line drugs had sepsis organisms susceptible to second-line drugs. Multivariable logistic regression identified five primary predictors: LBW (aOR 3.4), twin birth (aOR 4.7), maternal history of preterm births (aOR 2.7) or history of LBW births (aOR 2.6), low family income (aOR 1.7), and birth at the general hospital (aOR 2.5). There were significant differences in sepsis and mortality by facility and birthweight. Conclusions: Neonatal sepsis and AMR pose significant risk to our study population. Due to the high proportion of AMR and likelihood of treatment failure, developing effective facility-based sepsis prevention strategies is an urgent priority for advancing newborn survival in Amhara, Ethiopia and in similar global contexts.