Shorter or biomarker-guided antibiotic durations for common serious neonatal infections: a collection of non-inferiority meta-analyses.

Background: Unnecessary antibiotic prolongation promotes antimicrobial resistance. Evidence-based guidance on antibiotic durations for treating neonatal infections is lacking. Noninferiority meta-analyses evaluating shorter durations or biomarker-guided durations have not been conducted. Methods: We conducted systematic reviews and meta-analyses using a unified search strategy (MEDLINE, EMBASE, Cochrane Library, January 1990-December 2024) to determine whether short-course antibiotics are noninferior to "standard" courses for culture-positive sepsis, culture-negative sepsis, UTIs, uncomplicated/complicated meningitis, and fungal sepsis. Likewise, we determined whether biomarker-guided antibiotic durations are noninferior to "standard" courses for any sepsis. We included randomized controlled trials (RCTs) without language restriction comparing short-courses or biomarker-guided courses versus standard courses for treating the above neonatal infections. Minimum clinically important differences (MCIDs) for critical outcomes were pre-specified [mortality (28-day, in-hospital or 12-month corrected age) = +1%, culture-positive relapse = +3%, culture-negative relapse = +5%]. We evaluated risk of bias (Cochrane RoB 2.0) and certainty of evidence (CoE) using GRADE. Registration: PROSPERO CRD42023311895. Findings: We reviewed 146 full-text articles and included 26 in the review. For culture-positive sepsis (7 randomized controlled trials [RCTs]), 7-10-day courses were noninferior to longer courses for 28-day mortality [upper bound of 95% CI for risk difference +0.85%, below 1% MCID (low CoE)], in-hospital mortality [+0.51% (moderate CoE)], culture-positive relapse [+0.75%, below 3% MCID (low CoE)], culture-negative relapse [+4.7%, below 5% MCID (very low CoE)]. For culture-negative and -positive sepsis, biomarker-guided courses (5 RCTs) were noninferior to standard courses for mortality and relapses (all very low CoE). For culture-negative sepsis (6 RCTs), shorter courses (3-4 days) were noninferior for culture-positive relapses (very low CoE) but not noninferior for other outcomes (low to very low CoE). Conclusions could not be drawn for UTI (6 observational studies) or meningitis (1 RCT). There were no data for complicated meningitis and fungal sepsis. Interpretation: Compared to standard durations, 7-10-day antibiotic courses and biomarker-guided courses may be noninferior for treating culture-positive neonatal sepsis and any sepsis, respectively. Three-to-four-day courses may not be noninferior for culture-negative sepsis. Given the low to very low CoE, these conclusions are not definitive. No conclusions can be drawn for UTIs, meningitis, and fungal sepsis. Funding: Not funded.