Pediatrics · Medical Literature
Individualised duration of antibiotic treatment in culture-negative early-onset sepsis in late-preterm and term-born neonates in Denmark (DURATION): a multicentre, open-label, randomised, controlled, non-inferiority trial.
Abstract
Suspicion of early-onset sepsis (EOS) is a major contributor to antibiotic use in late-preterm and term-born neonates. We aimed to evaluate whether a clinically guided, individualised approach to treatment duration could safely reduce antibiotic exposure compared with standard care. This nationwide, multicentre, open-label, randomised, controlled, non-inferiority trial included late-preterm and term-born neonates (aged 0-3 days, gestational age 35 weeks or more, birthweight 2000 g or more) with culture-negative early-onset sepsis (probable EOS), defined as continuation of antibiotic treatment beyond 36-48 h based on clinical signs of or maternal risk factors for infection in combination with elevated C-reactive protein and symptom onset less than 72 h after birth. Neonates were ineligible if they met the criteria for discontinuing antibiotics before 48 h. Participants were randomised (1:1) to individualised or standard treatment. In the individualised group, antibiotic treatment was discontinued after 24 h without clinical signs of infection, provided that C-reactive protein was declining to ≤30 mg/L. Standard treatment was 5-7 days. The coprimary outcomes were readmission due to bacterial infection (non-inferiority margin 4%) and total antibiotic duration (superiority assessment), performed on both intention-to-treat and per-protocol populations. Safety was assessed in all included neonates. This study was registered with ClinicalTrials.gov (NCT05329701) and is completed. Between April 22, 2022, and April 5, 2025, 493 (61%) of 811 eligible neonates were randomly assigned to individualised (n=246) or standard (n=247) duration. Antibiotic therapy was initiated at a median of 22·5 h (IQR 14·4-28·0) after birth in the individualised group and 24·5 h (16·6-28·0) in the standard group. Readmission due to bacterial infection occurred in two (1%) of 246 neonates in the individualised group and one (<1%) of 247 in the standard group (risk difference 0·4% [95% CI -1·5 to 2·6]; pnon-inferiority=0·0032). Median total antibiotic treatment within 28 days was 2·8 days (IQR 2·4-3·4) in the individualised group and 6·8 days (6·6-7·0) in the standard group (difference 4·0 days [95% CI 3·8-4·1]; p<0·0001). No serious adverse events occurred in the individualised group and one (<1%) occurred in the standard group (risk difference -0·4% [95% CI -2·3 to 1·1]). Per-protocol results were consistent. In neonates with probable EOS, a clinically guided, individualised treatment strategy was non-inferior to standard therapy with respect to infection-related readmission and reduced treatment duration to a median of 3 days. These findings support reduced antibiotic use through individualised treatment duration strategies. Copenhagen University Hospital Rigshospitalet Research Fund, Innovation Fund Denmark, and Greater Copenhagen Health Science Partners.
Clinical relevance
In late-preterm and term-born neonates with culture-negative (probable) early-onset sepsis, is a clinically guided, individualised antibiotic treatment duration strategy non-inferior to standard 5–7 day therapy with respect to infection-related readmission while also reducing total antibiotic exposure? __SNAPSHOT__ {"studyDesign":"Investigator-initiated, nationwide, multicentre, open-label, randomised, controlled, non-inferiority trial (co-primary superiority endpoint for antibiotic duration)","setting":"Denmark; all 17 Danish neonatal departments (nationwide pragmatic trial)","sampleSize":"n=493 (individualised: n=246, standard: n=247)","population":"Late-preterm and term-born neonates aged 0–3 days, gestational age ≥35 weeks, birthweight ≥2000 g, with culture-negative (probable) early-onset sepsis — defined as continuation of antibiotic treatment beyond 36–48 h based on clinical signs of or maternal risk factors for infection, in combination with elevated CRP, and symptom onset <72 h after birth. Neonates already meeting criteria for antibiotic discontinuation before 48 h were excluded.","intervention":"Individualised antibiotic treatment: discontinuation after 24 h without clinical signs of infection, provided CRP was declining to ≤30 mg/L","comparator":"Standard antibiotic treatment: fixed duration of 5–7 days","primaryOutcome":"Co-primary outcomes: (1) readmission due to bacterial infection (non-inferiority margin 4% risk difference); (2) total antibiotic treatment duration within 28 days (superiority)","followUp":"28 days (infection-related readmission also reported at 100 days per available figure data)","fundingSource":"Investigator-initiated; specific funding source not reported in the abstract — full text states investigator-initiated trial (no industry sponsorship identified)"}
Key findings
On the co-primary non-inferiority outcome, readmission due to bacterial infection occurred in 2 of 246 neonates (1%) in the individualised group versus 1 of 247 (<1%) in the standard group (risk difference +0.4% [95% CI −1.5 to +2.6]; p-non-inferiority = 0.0032). The upper bound of the 95% CI (+2.6%) fell within the pre-specified non-inferiority margin of 4%, confirming non-inferiority. On the co-primary superiority outcome, median total antibiotic treatment within 28 days was 2.8 days (IQR 2.4–3.4) in the individualised group versus 6.8 days (IQR 6.6–7.0) in the standard group — an absolute reduction of 4.0 days (95% CI 3.8–4.1; p<0.0001), confirming superiority. The relative reduction in antibiotic exposure was approximately 59% (calculated). No serious adverse events occurred in the individualised group; one serious adverse event (<1%) occurred in the standard group (risk difference −0.4% [95% CI −2.3 to +1.1]). Per-protocol analyses were consistent with intention-to-treat results. NNT to prevent one excess readmission is not calculable in a meaningful direction given near-zero event rates in both groups; NNH for a serious adverse event favoured individualised therapy based on the single event in the control arm, but confidence intervals crossed zero and the finding is not statistically significant.
Practice implications
The DURATION trial is the first RCT conducted in a high-income setting to demonstrate that a clinically guided, individualised antibiotic discontinuation strategy — anchored to symptom resolution and CRP ≤30 mg/L declining — is non-inferior to standard 5–7 day treatment for culture-negative probable EOS in late-preterm and term neonates, while also achieving a clinically and statistically significant reduction in antibiotic exposure of approximately 4 days. The findings support that the conventional 5–7 day standard of care for probable EOS is not evidence-based and can be safely shortened. The results apply specifically to late-preterm (≥35 weeks) and term neonates with birthweight ≥2000 g, culture-negative sepsis, and CRP dynamics meeting the specified stopping criteria; they cannot be extrapolated to extremely preterm infants, culture-positive EOS, or neonates failing to meet CRP de-escalation thresholds. Absolute readmission rates were very low in both arms (≤1%), which is clinically reassuring but also means the trial was not powered to detect rare safety events with precision. The open-label design introduces potential clinician behaviour bias, and Danish healthcare infrastructure may limit direct generalisability to other settings. Only the abstract was available for this summary; specific data on stratification, subgroup analyses, secondary outcomes beyond readmission and antibiotic duration, and the complete safety profile require full-text review. __LIMITATIONS__ Open-label design with no blinding of clinicians or outcome assessors, introducing risk of performance and detection bias,Small absolute event rates for the readmission endpoint (2 vs 1 events) limit precision; the trial was adequately powered for the pre-specified margin but cannot exclude rare adverse outcomes,Restriction to neonates ≥35 weeks gestation and ≥2000 g birthweight limits generalisability to more vulnerable preterm populations,Nationwide Danish setting with well-established intravenous-to-oral antibiotic switch protocols may not reflect standard practice in other high-income or low/middle-income countries,Prior RCT evidence for this specific question came exclusively from low/middle-income settings with low certainty; DURATION fills this gap but is a single-country trial,CRP as the sole biomarker-guided stopping criterion has limitations in sensitivity and specificity for neonatal bacterial infection; procalcitonin was not incorporated,The non-inferiority margin of 4% absolute risk difference for readmission, while clinically pre-specified, is a judgment call and its appropriateness may be debated,Long-term outcomes (microbiome disruption, antimicrobial resistance, atopic or immune sequelae) were not assessed within the reported trial period,Only the abstract was available for this review; full assessment of randomisation quality, allocation concealment, loss to follow-up, and secondary outcome data requires full-text access __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ In late-preterm and term neonates with culture-negative probable early-onset sepsis, individualised antibiotic discontinuation guided by clinical recovery and a declining CRP ≤30 mg/L is non-inferior to standard 5–7 day treatment and reduces antibiotic exposure by a median of 4 days (to approximately 3 days total). Readmission and serious adverse event rates were similarly very low in both groups. These results provide strong, first-of-its-kind RCT evidence from a high-income setting to support adoption of a biomarker- and clinically-guided individualised approach over fixed-duration therapy in this population.