Sepsis calculator for neonatal early onset sepsis - a systematic review and meta-analysis*

Background: Over investigation and overuse of empirical antibiotics is a concern in management of neonatal early onset sepsis (EOS) using the Centers for Disease Control and Prevention guidelines. "Sepsis calculator" is a risk-based prediction model for managing neonates at risk of EOS. Objective: To compare outcomes of neonatal EOS using of sepsis calculator versus conventional approach. Methods: A systematic review of randomized controlled trials (RCT) and non-RCTs reporting on outcomes after implementation of sepsis calculator for EOS for neonates >34-week gestation was conducted using the Cochrane methodology. Databases PubMed, CINAHL, Embase, Cochrane Central library and Google Scholar were searched in May 2019. Primary outcomes were antibiotics usage and laboratory tests for managing EOS. Secondary outcomes included hospital admissions and readmissions, blood culture positive EOS and mortality. The level of evidence (LOE) was summarized using the GRADE guidelines. Results: A total of 387 articles were retrieved after initial search. Six high quality non-RCTs fulfilled inclusion criteria. Meta-analysis (random effects model) showed that implementation of sepsis calculator was associated with reduced antibiotic usage [N = 172,385; OR = 0.22 (0.14-0.36); p < .00001; heterogeneity (I-2) = 97%, Number needed to treat (NNT): 22], laboratory tests [N = 168,432; OR = 0.14 (0.08-0.27); p < .00001; I-2 = 99%, NNT = 8], and admissions to neonatal unit [N = 16,628; OR = 0.24 (0.11-0.51); p = .0002; I-2 = 98%, NNT = 7]; LOE: moderate. There was no difference in mortality, culture positive EOS, and readmissions. Conclusion: Moderate quality evidence indicates that the implementation of a sepsis calculator was associated with reduced usage of antibiotics, laboratory tests and admission to neonatal unit with no increase in mortality and readmissions.