Epidemiology of healthcare-associated bloodstream infection in South African neonatal units

Background: Reports of healthcare-associated bloodstream infection (HA-BSI) epidemiology in African neonatal units are limited.

Methods: We conducted a cross-sectional study (2017-2018) in nine neonatal units in the Western Cape Province, South Africa, including central, regional and district hospitals (416 beds) using laboratory and clinical records. Patient demographics, HA-BSI rates, pathogen spectrum, and hospital outcomes and empiric antibiotic coverage rates were determined.

Results: Over two years, 23,748 neonates were admitted with unit occupancy rates ranging from 79 to 93%. 485 HA-BSI episodes occurred, with median onset at 11 (IQR 7-24) days of life. Most HA-BSI episodes (348; 72%) affected very low birth weight neonates (< 1500 g). The overall HA-BSI rate was 2.0/1000 patient days. The highest HA-BSI rate was observed at the central unit with onsite surgery (3.8/1000 patient days). Crude HA-BSI mortality was 31.8% (154/485) with two-thirds of deaths occurring within three days of BSI onset. Higher mortality was observed for Gram-negative/fungal BSI compared to Gram-positive BSI (RR 1.5; 95%CI 1.1-2.0; p = 0.01) and very preterm neonates (gestation < 32 weeks) versus ≥ 32 weeks (RR 1.5; 95%CI 1.1-2.1; p = 0.01). Mean estimated empiric antibiotic coverage rates varied by unit type: 66-79% for piperacillin-tazobactam plus amikacin, 60-76% for meropenem and 84-92% for meropenem plus vancomycin.

Conclusion: Most HA-BSI events affected preterm neonates at the central hospital with onsite surgery. One-third of patients with HA-BSI died, with highest mortality in preterm infants and Gram-negative/fungal BSI. Empiric antibiotic regimens provide moderate coverage of circulating pathogens but require annual review given increasing carbapenem resistance rates.