569 - Shorter Antibiotic Durations for Common Serious Neonatal Infections: An Umbrella of Systematic Reviews, Noninferiority Meta-analyses and GRADE-based Recommendations

Publication Number: 569.6912

Sourabh Dutta, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, Chandigarh, India; Nandkishor S. Kabra, Surya Hospitals Mumbai, Mumbai, Maharashtra, India; Shiv Sajan Saini, PGIMER Chandigarh, CHANDIGARH, Chandigarh, India; Rajendra Prasad. Anne, Kasturba Medical College Manipal, Manipal, Karnataka, India; Sandeep Kadam, KEM hospital, Pune City, Maharashtra, India; Monisha Rameshbabu, Chettinad Hospital and Research Institute, Chennai, Tamil Nadu, India; Supreet Khurana, Government Medical College & Hospital Chandigarh, Chandigarh, Chandigarh, India; Sai Kiran, Fernandez Hospital, Hyderabad, Telangana, India

Background: Evidence-based guidance for antibiotic durations to treat common serious neonatal infections is lacking. Noninferiority meta-analyses comparing antibiotic durations have not been conducted.

Objective: To conduct systematic reviews and aggregate data noninferiority meta-analyses of a group of serious neonatal infections.

Design/Methods: We concurrently answered several questions using a unified search strategy (MEDLINE, EMBASE and Cochrane library: Jan 1990 to Apr 2024) to determine whether shorter antibiotic courses were noninferior to longer courses to treat neonatal blood culture-positive sepsis, culture-negative sepsis, urinary tract infection (UTI), uncomplicated and complicated meningitis, and fungal sepsis. Minimum Important Differences (MIDs) of outcomes (mortality, relapse) were decided a priori. 95% CIs of risk differences (RD) were interpreted vis-à-vis the MIDs for deciding noninferiority. We assessed risk of bias by Cochrane RoB 2·0 and for noninferiority. We followed GRADE for certainty of evidence (CoE) and recommendations. Registration: PROSPERO CRD42023311895.

Results: Of 16,267 records identified, 138 full-text articles were retrieved and separated question-wise. For blood-culture positive sepsis [7 randomized controlled trials (RCTs)], shorter courses (7-10 days) were noninferior for death by day 28 and by hospital discharge [upper bound of the 95% CI (UBCI) of RD= +0·78% and +0·88%, respectively, below pre-specified MID of +1% (moderate to high CoE)] and culture-proven relapse [UBCI of RD= +0·97%, below the MID of +3% (moderate CoE)]. For suspect sepsis, biomarker-guided courses (6 RCTs) were noninferior for death by day 28 and hospital discharge [UBCI of RD of both outcomes= +0·26%, below the MID of 1% (very low CoE)] and culture-proven relapse [UBCI of RD= +0·28%, below the MID of +3% (very low CoE)]. For culture-negative sepsis (6 RCTs), shorter courses (3-4 days) were noninferior for culture-proven relapse [UBCI of RD= +1·51%, below the MID of +3% (very low CoE)] and probable relapse [UBCI of RD= +3·9%, below the MID of +5% (very low CoE)], but not mortality (very low CoE). Noninferiority of short courses could not be shown for UTI (very low CoE, 5 observational studies) or uncomplicated meningitis (1 RCT). Data were too inadequate for complicated meningitis and fungal sepsis to draw conclusions.

Conclusion(s): We make a weak recommendation in favor of shorter courses for culture-proven sepsis and biomarker-guided courses; a weak recommendation for either shorter or standard courses for culture-negative sepsis; and strong recommendations against short courses for UTI or meningitis.

Figure 1
PRISMA diagram

Figure 2
Per protocol analysis: Forest plots of comparison between short duration versus standard duration of antibiotics
Outcome: Mortality by hospital discharge


Figure 3
Per protocol analysis: Forest plots of comparison between short duration versus standard duration of antibiotics
Outcome: Culture-proven relapse of sepsis/meningitis