For decades, the faintest clinical murmur of neonatal early‑onset sepsis has sent a cascade of calculated panic through maternity wards: a heel‑prick culture drawn, an intravenous line inserted, and a broad‑spectrum antibiotic cocktail infused within the first hour of life. This standard of care, codified by the mantra that failing to treat a septic newborn could be lethal within hours, has saved countless lives. Yet a provocative review published in Pediatric Research now argues that twenty years of empirical antibiotic use have been shaped as much by fear and confounding as by clear‑cut evidence, and that the consequences of reflexive treatment are beginning to eclipse the benefits for many infants.
Early‑onset sepsis, usually defined as a bloodstream or meningeal infection in the first 72 hours of life, is predominantly caused by group B Streptococcus and Escherichia coli acquired during delivery. The difficulty lies in its nonspecific presentation: a baby who is mildly tachypneic, slightly lethargic, or just a degree cooler than expected may be brewing a catastrophic infection, or may simply be adapting to extrauterine life. Because the sensitivity of a single blood culture can be as low as fifty percent in neonates, clinicians have historically opted to treat first and ask questions later. The review by Flannery and Sauberan traces how this logic, combined with evolving maternal screening programs and medico‑legal pressures, has led to a situation where as many as one in ten newborns in some centers now receives antibiotics within days of birth, despite an incidence of confirmed sepsis that has plummeted to fewer than one case per thousand live births.
The authors dissect the concept of “choice” in empirical regimens, highlighting how ampicillin and gentamicin became the de facto standard not solely through randomized controlled trials, but through pharmacokinetic modeling in tiny populations and a desire for synergy against the most feared pathogens. They note that gentamicin dosing, in particular, has been plagued by decades of debate over peak and trough targets, with extended‑interval dosing gradually gaining favor only after pharmacodynamic studies showed that aminoglycosides kill bacteria in a concentration‑dependent manner and that a long post‑antibiotic effect allows for once‑daily administration. This shift, the review explains, was as much a story of therapeutic drug monitoring technology improving as it was of randomized evidence emerging, revealing how practice often runs ahead of definitive proof.
Confounding is the central thread woven through the analysis. The largest observational studies linking early antibiotic exposure to later outcomes — including asthma, obesity, inflammatory bowel disease, and even neurocognitive delays — are notoriously difficult to interpret because the very indication for antibiotics, suspicion of infection, is itself a marker of perinatal stress and inflammation. Flannery and Sauberan walk through the formidable statistical challenge of disentangling whether it is the drugs or the disease risk that shapes the gut microbiome’s nascent assembly. They highlight recent work using sibling‑matched cohorts and instrumental variable analyses that attempt to isolate the effect of antibiotics themselves, revealing that even short courses alter the intestinal metagenome, depleting keystone species like Bifidobacterium and enriching resistance genes that can persist for months.
Perhaps the most unsettling consequence illuminated in the review is the emergence of antimicrobial resistance in neonatal intensive care units. The very units that depend on ampicillin and gentamicin are now documenting rising minimum inhibitory concentrations among common Gram‑negative pathogens, a phenomenon the authors link directly to the sheer volume of antibiotic days administered. Through a meticulous dissection of twenty years of antibiograms, they show how gentamicin resistance in E. coli isolates from neonatal blood cultures has crept upward in a stepwise fashion, mirroring local prescribing rates. This silent evolution turns a predictable, low‑toxicity regimen into a gamble, pushing clinicians toward last‑line carbapenems for what were once easily treatable infections.
The review also introduces the concept of the “microbial scar,” a term gaining traction among neonatologists to describe the durable metabolic and immunologic fingerprints left by early‑life antibiotic disruption. Short‑chain fatty acid production, critical for regulatory T‑cell development and gut barrier integrity, can be suppressed for weeks after a three‑day course of empiric therapy. The authors cite emerging metabolomic data showing that neonates exposed to antibiotics exhibit persistently altered bile acid profiles and urinary metabolite signatures consistent with a state of chronic low‑grade intestinal inflammation, long after the drugs have been cleared. These molecular echoes, they argue, may represent the mechanistic bridge between a seemingly benign postnatal intervention and the chronic diseases that manifest years later.
Yet the review refuses to offer a simple algorithm for withholding antibiotics, acknowledging that the calculus at the bedside remains agonizingly imprecise. Instead, Flannery and Sauberan champion a risk‑stratification framework that integrates maternal colonization status, intrapartum antibiotic exposure, gestational age, and — crucially — serial physical examination using validated sepsis risk calculators. They point to pioneering centers that have implemented an “antibiotic time‑out” at 36 hours, leveraging negative blood cultures, normal C‑reactive protein trajectories, and clinical wellbeing to stop therapy, and have halved antibiotic use without missing a single infection. The key, they argue, is transforming a binary decision — treat or don’t treat — into a dynamic, multidisciplinary conversation that is revisited with each nursing shift.
As neonatal medicine stands at this inflection point, the review serves as both a rear‑view mirror and a call to action. Twenty years of data have taught us that the margin of safety for empirical antibiotics is narrower than once believed, and that the organisms we seek to eradicate are stealthily adapting to our every move. The next era will demand a fusion of rapid molecular diagnostics, machine‑learning‑driven risk prediction, and microbiome‑sparing therapeutic strategies to ensure that the legacy of saving lives today is not a generation burdened by the metabolic and immunologic fallout of our good intentions.
Subject of Research: The evolution, confounding factors, and health consequences of empiric antibiotic use for neonatal early-onset sepsis over a twenty-year period.
Article Title: Choice, confounding, and consequence: twenty years of empiric antibiotics for neonatal early-onset sepsis.
Article References:
Flannery, D.D., Sauberan, J.B. Choice, confounding, and consequence: twenty years of empiric antibiotics for neonatal early-onset sepsis.
Pediatr Res (2026). https://doi.org/10.1038/s41390-026-05284-x
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s41390-026-05284-x
Keywords: Neonatal early-onset sepsis, empiric antibiotics, antimicrobial resistance, neonatal microbiome, gut dysbiosis, therapeutic inertia, pharmacokinetic modeling, sepsis risk calculators, microbial scar, extended-interval gentamicin.
