Abstract
BACKGROUND AND OBJECTIVES: Healthcare-associated infections (HAI) are a leading contributor to morbidity and mortality in hospitalised neonates. Diagnosing neonatal HAI is challenging owing to non-specific symptoms and lack of definitive diagnostic markers, contributing to high rates of inappropriate antibiotic use. This study evaluated the theoretical impact of implementing a bedside tool for decision-making on antibiotic length of therapy (LOT). METHODS: This prospective observational physician-blinded study consecutively enrolled patients with suspected HAI events at a large South African neonatal unit from September 2022 to September 2023. The antibiotic decision-making tool included an infection prediction score (NeoHoP), and a point-of-care C-reactive protein test (CRP) performed at HAI diagnosis and 24 h later. The theoretical impact of the tool on antibiotic LOT was calculated. RESULTS: We recruited 180 neonates with 214 episodes of suspected HAI, of which 22 (10.3%) were proven HAI, 56 (26.2%) were presumed HAI and 136 (63.6%) had HAI ruled out. The median observed antibiotic LOT was three days (9 days for proven HAI, 7 days for presumed HAI, and 3 days for no HAI). The antibiotic decision-making tool would theoretically reduce overall antibiotic LOT by 2 days (p < 0.001), particularly in neonates where HAI was subsequently excluded. CONCLUSION: We developed an antibiotic decision-making tool to support the clinical evaluation of suspected neonatal HAI and demonstrated a significant potential impact on reducing antibiotic LOT. Given increasing antibiotic resistance rates globally, this tool should be further evaluated to minimise unnecessary antibiotic use in hospitalised neonates.