EBNEO commentary: Duration of Antibiotic Therapy for Gram-Negative Bloodstream Infections in the Neonatal Intensive Care Unit

MANUSCRIPT CITATION

Djordjevich CJ, Magers J, Cantey JB, Prusakov P, Sánchez PJ, Duration of
Antibiotic Therapy for Gram-Negative Bloodstream Infections in the Neonatal Intensive Care Unit, The Journal of Pediatrics (2026), doi: https://doi.org/10.1016/j.jpeds.2026.114993. PMID: 41554433

REVIEWED BY

Fabian J.S. VAN DER VELDEN, MD, PhD
Neonatal Unit, James Cook University Hospital, South Tees Hospitals NHS Foundation Trust, Middlesbrough, UK
Translational and Clinical Research Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne, UK
Fabian.vandervelden@nhs.net
ORCiD: 0000-0001-7864-4074

Prakash Kannan LOGANATHAN, MD
Neonatal Unit, James Cook University Hospital, South Tees Hospitals NHS Foundation Trust, Middlesbrough, UK
pkannanloganathan@nhs.net
ORCiD: 0000-0003-3717-8569

TYPE OF INVESTIGATION

Treatment, retrospective multicentre cohort study

QUESTION

In neonates with uncomplicated gram-negative bloodstream infections (GNBSI), both early and late-onset, is a short antibiotic course (≤8 days) safe and effective, compared to a long antibiotic course (≥9 days)?

METHODS

• Design: retrospective, observational, multicentre cohort study (USA, 2016-2022)
• Allocation: Eligible neonates were allocated to the shorter antibiotic course group if the course was ≤8 days, and the longer antibiotic course group if ≥9 days.
• Blinding: not applicable for this retrospective observational study
• Follow-up period:
o Treatment failure: 14 days after stopping antimicrobial therapy
o In-hospital mortality: 30 days after stopping antimicrobial therapy
o Subsequent fungaemia or GN multidrug-resistant organism emergence: 90 days or discharge, whichever is sooner, after stopping antimicrobial therapy
• Setting: 7 Neonatal Intensive Care Units across 2 healthcare systems in the USA
• Patients:
o Inclusion criteria
 Any neonate who received antibiotic therapy for uncomplicated GNBSI between January 2016 and May 2022 in NICUs in the USA
o Exclusion criteria
 Neonates with meningitis, osteomyelitis or endocarditis
 Receipt of vasopressor treatment
 Death prior to completion of full course of antibiotic therapy
 Polymicrobial bloodstream infections, bacterial contaminants, or gram-positive bloodstream infections
• Intervention/Exposure:
o Short antibiotic course group: ≤8 days
o Long antibiotic course group: ≥9 days
• Outcomes:
o Primary outcome: treatment failure, defined as recurrence of BSI with the same organism within 14 days of stopping antimicrobial therapy
o Secondary outcomes: onset of acute kidney injury (urine output <1mL/kg/h for 24h or rise in serum creatinine ≥1.5x baseline) during antimicrobial treatment; recurrence of BSI with any organism within 14 days of stopping antimicrobial treatment; subsequent occurrence of fungaemia; emergence of a gram-negative multidrug-resistant organism (MDRO); 30-day in-hospital mortality
• Analysis and Sample Size:
o Analysis: descriptive statistics were used for continuous demographic and clinical variable, reported as median with interquartile range (IQR). Categorical variables were summarised as frequency (%). Comparison of infants with central venous catheters who received short vs long antibiotic course was performed using chi-square tests.
o Sample size: 143 neonates
• Patient follow-up: 76 met eligibility criteria (53.1%), none of those included were lost to follow-up

MAIN RESULTS
There were no differences in baseline characteristics between the short and long antibiotic course groups, except the presence of central venous catheter. There was no recurrence of BSI with the same organism ≤14 days after discontinuation of antimicrobial therapy in the short antibiotic course group and 2 had recurrence in long antibiotic group. Of all isolated GN organisms, two were not susceptible to initial empirical antibiotic therapy. No neonates in the short course group and five infant in the long course group developed multidrug-resistant organism infection within the follow-up period.

Table 1

Table 2

CONCLUSION

Neonates of any gestational age or birthweight with uncomplicated gram-negative bloodstream infections may be candidates for a 7-day course antibiotic therapy provided their clinical manifestations of sepsis have resolved.

COMMENTARY

Gram-negative bloodstream infections (GNBSIs) are an important cause of morbidity and mortality in neonates(1, 2). Recognition and early instigation of appropriate antibiotic treatment is paramount to improve outcomes. The optimal duration of treatment for GNBSI remains uncertain, and prolonged courses may not be necessary for all infants. The aim should be to use the shortest effective course, balancing clinical recovery with the risks of antimicrobial resistance and the potential longer-term harms of unnecessary antibiotic exposure. Prolonged courses of antibiotics have been associated with increased risk of necrotising enterocolitis in premature neonates(3), an increased risk for allergic disease, and inflammatory bowel disease later in life(4). In parallel, antimicrobial resistance remains a growing concern, with reported resistance among neonatal E. coli isolates reaching 50–75% for amoxicillin, 15–30% for gentamicin, and 46% for cefotaxime(1, 5).

 

Although defined as ≤8 versus ≥9 days, the effective comparison was approximately 7 versus 14 days (short: median 7 days, 67% at 7 days; long: median 14 days; only 9 and 1 infant at the 8- and 9-day boundary). Within this contrast outcomes were broadly similar; treatment failure (same-organism recurrence within 14 days; 5% vs 0%) and subsequent multidrug-resistant gram-negative (MDRO) infection occurred only in the longer-course group. The design cannot recover a dose-response or inform intermediate durations.

 

An important limitation is confounding by indication. Infants in the longer-course group appeared clinically more complex, with higher central venous catheter use (68% vs 38%), all cases with multiple positive blood cultures (19% vs 0%), and greater exposure to broader-spectrum monotherapy such as cefepime or piperacillin-tazobactam (49% vs 10%), whereas gentamicin monotherapy was more common in the shorter-course group (36% vs 11%). These findings suggest that clinicians were already tailoring both antibiotic duration and spectrum according to perceived illness severity or complexity. Therefore, the higher MDRO rate in the longer-course group may reflect broader-spectrum exposure and underlying case mix rather than duration alone. This is reinforced by the MDRO definition itself—any-site detection within 90 days—with events occurring 10–82 days later and predominantly new E. coli ESBL isolates that are plausibly independent acquisitions.

 

Blood culture volume, sampling technique, and repeat-culture practice were not reported, and the methods do not explain how metastatic foci (bone, abscess) were excluded. Because lumbar puncture was not always performed at presentation, post-antibiotic or traumatic CSF sampling may have under-recognised meningitis. The study is underpowered, with no multivariable or propensity adjustment and no confidence intervals. Survivorship bias is also a concern: the 27 of 67 excluded infants who died before day 7 could not experience the outcome, and reaching ≥9 days itself required survival, favouring the longer-duration group on mortality. Importantly, the 7 NICUs sit within only 2 healthcare systems under shared stewardship, so this reflects closer to two practice cultures than seven independent centres; gentamicin monotherapy was viable only because local E. coli and Klebsiella susceptibility was 92–95%, against the 15–30% resistance cited in the literature, and centres with higher resistance could replicate neither strategy—narrowing external validity.

 

Taken together, these data support a hypothesis-generating comparison of approximately 7 versus 14 days in a selected, lower-acuity, vasopressor-excluded population managed within two health systems. Given the small sample, methodological limitations, and risk of bias, the findings justify a prospective trial rather than a change in practice, consistent with the authors’ own conclusion.

 

Artificial intelligence disclosure: ChatGPT (OpenAI; GPT-5.4 Thinking) was used to support drafting, language editing, and improvement of clarity in this commentary. The authors independently reviewed, critically appraised, revised, and approved the final content, and take full responsibility for the accuracy and interpretation of the work.

 

REFERENCES

1. Stoll BJ, Puopolo KM, Hansen NI, Sánchez PJ, Bell EF, Carlo WA, et al. Early-Onset Neonatal Sepsis 2015 to 2017, the Rise of Escherichia coli, and the Need for Novel Prevention Strategies. JAMA Pediatr 2020; 174 7:e200593.
2. Cailes B, Kortsalioudaki C, Buttery J, Pattnayak S, Greenough A, Matthes J, et al. Epidemiology of UK neonatal infections: the neonIN infection surveillance network. Archives of disease in childhood Fetal and neonatal edition 2018; 103 6:F547-f53.
3. Klerk DH, van Avezaath LK, Loeffen EAH, Hulscher JBF, Kooi EMW. Fetal-neonatal exposure to antibiotics and NEC development: A systematic review and meta-analysis. Front Pediatr 2022; 10:1102884.
4. Shekhar S, Petersen FC. The Dark Side of Antibiotics: Adverse Effects on the Infant Immune Defense Against Infection. Front Pediatr 2020; 8:544460.
5. Bergin SP, Thaden JT, Ericson JE, Cross H, Messina J, Clark RH, et al. Neonatal Escherichia coli Bloodstream Infections: Clinical Outcomes and Impact of Initial Antibiotic Therapy. Pediatr Infect Dis J 2015; 34 9:933-6.