ANTIBIOTIC PRESCRIBING PATTERN AND DIAGNOSIS OF BACTERIAL SEPSIS IN A NEONATAL INTENSIVE CARE UNIT IN BEIRA, MOZAMBIQUE

BACKGROUND: Neonatal sepsis is a significant contributor to neonatal morbidity and mortality globally, especially in low- and middle-income countries (LMICs). In these regions, the etiology of neonatal sepsis is characterised by a predominance of Gram-negative bacteria and concerning rates of resistance to first- and second-line empirical antibiotic therapies recommended by the World Health Organization (WHO). Understanding the local etiology of neonatal sepsis and antibiotic prescribing patterns is crucial for informing antibiotic stewardship initiatives. However, there is limited knowledge regarding these factors in Mozambique. METHODS: We conducted a retrospective observational study in the Neonatal Intensive Care Unit (NICU) of Beira Central Hospital (HCB) in Mozambique, including all admitted newborns between January 1, 2024, and May 31, 2024, for whom complete medical records were accessible. Data on demographic and clinical characteristics of the neonates, antibiotic prescribing patterns, blood culture procedures, and microbiological results were collected. FINDINGS: Among the 839 patients included in the study, 34.8% were preterms and 44.0% had a birth weight <2500g. The overall mortality was 27.1%. In less than half admitted patients maternal risk factors were known. During hospitalisation, 70.7% of admitted newborns received antibiotics, started for prophylaxis (50.6%), suspected sepsis (26.5%), or for other indication (22.9%). Median duration of antibiotic administration was 5 days. The most frequent prescribed empiric first-line was ampicillin+gentamicin (72.5%), as suggested by WHO, whereas the most frequent second-lines were represented by cefotaxime+amikacin (32.5%) and cefotaxime+ciprofloxacin (26.5%). Carbapenems were administered in 2.4% of patients, mostly as fourth-line. 268 neonates receiving antibiotics had a blood culture collected (45.2%), in 68.6% of cases before starting antibiotics. Gram-negative bacteria were most frequently isolated (72.6%), with a predominance of Klebsiella sp (82.0%). Among Gram-positive bacteria (26.1%), the only isolated pathogen was S. aureus (15.0%; other Gram-positives were considered contaminants); no S. agalactiae were found. Early-onset (EOS) and late-onset (LOS) sepsis were respectively 52.2% and 47.8%, with higher mortality in EOS (56.7% vs 29.1%), despite a similar pathogen distribution among the two groups. Antibiotic in-vitro resistance to WHO suggested first- and second empirical lines was found in respectively 89.4% and 90.1% of Gram-negative bacteria isolated, with a concomitant resistance to carbapenems of 45.0%. Outbreaks of Klebsiella sp were encountered during the study period, with an overall Klebsiella-associated mortality of 51.6%. Among the antibiotic combinations available, amikacin+ciprofloxacin appeared to be the most in-vitro sensitive therapy against Klebsiella-associated sepsis. CONCLUSION: The present study offers novel insights into the etiology of sepsis and the quality of current practices in the NICU of Beira, providing valuable information to further define first- and second-line antibiotic therapies in these settings. Moreover, in our knowledge, it represents the first study describing the antibiotic prescribing pattern in a NICU in Mozambique, contributing to the neonatal sepsis panorama in Sub-Saharan Africa.