What an Antibiotic Leaves Behind: Testing Whether a Preventive Medicine Changed the Bacteria in Premature Babies' Lungs and Guts

Two follow-on studies from the AZTEC trial, which measured antibiotic-resistant bacteria in the airway and the gut of very premature babies

In a large British trial, 799 very premature babies were given either the antibiotic azithromycin or a dummy medicine for ten days. Two follow-up studies then looked for antibiotic-resistant bacteria in their airways and stool. The antibiotic did not prevent the lung disease it was meant to prevent, and it left a measurable mark on their bacteria — most clearly in the gut.

Antibiotics raise a second question, harder to answer than "did it help?" — what do they leave behind in the body's bacteria? Because half the babies received a placebo and nobody knew which, these studies could answer it cleanly: they found more of one resistance gene in the airway, and in the gut less variety of bacteria, less of the helpful bacterium Bifidobacterium, and slightly more resistance genes overall.

Why doctors hoped an antibiotic would protect premature lungs

Babies born more than ten weeks early often develop a chronic lung condition sometimes called chronic lung disease of prematurity, or bronchopulmonary dysplasia. Their lungs are unfinished, they usually need help to breathe, and immaturity, inflammation and the need for breathing support are together thought to leave lasting damage, so that some babies still need extra oxygen weeks later. Doctors have long looked for something to interrupt that process early.

One idea centred on a bacterium called Ureaplasma, which is found in the lungs of some premature babies. A review combining 39 studies found that babies with Ureaplasma in their lungs were substantially more likely to be diagnosed with chronic lung disease [1]. Azithromycin — a common antibiotic from the family known as macrolides — kills Ureaplasma and also calms inflammation, so it offered two possible ways to help at once. Earlier trials were too small to give a clear answer [2]; one US trial included 121 babies, found no meaningful difference in how many died or had serious breathing problems by around two years of age, and argued a much larger trial was needed [3].

That larger trial was carried out. Called the AZithromycin ThErapy for Chronic Lung Disease of Prematurity (AZTEC) trial, it ran in 28 specialist neonatal intensive care units across the United Kingdom, enrolling 799 babies born before 30 weeks of pregnancy who needed breathing support in their first three days and giving each either azithromycin through a drip for ten days or an identical-looking placebo (a dummy treatment). Neither families nor staff knew which a baby had received — the design that gives the most trustworthy answer. The result was clear and disappointing: 166 of 394 babies given azithromycin (42%) survived without moderate or severe chronic lung disease, compared with 179 of 402 (45%) given placebo — a difference small enough to be chance — and the researchers concluded azithromycin should not be used for this purpose [4]. A later analysis pooling six trials and 1,360 babies agreed [5]; we covered that part of the story in a companion article, Azithromycin does not prevent bronchopulmonary dysplasia.

Why resistance was the question everyone wanted answered

Antibiotic resistance happens when bacteria carry genes that let them survive a particular antibiotic; the more that antibiotic is used, the more such bacteria are favoured. Worldwide, about 1.27 million deaths in 2019 were directly attributable to antibiotic-resistant bacteria [6]. In neonatal intensive care, antibiotics are the most commonly prescribed medicines of all, and overuse encourages resistant bacteria that then circulate in the unit [7].

Azithromycin had already been shown to drive resistance elsewhere. In a carefully controlled study in healthy volunteers, it clearly increased resistant bacteria in the throat within days [8]. In The Gambia, after whole communities were given azithromycin to control the eye infection trachoma, carriage of resistant Staphylococcus aureus in the nose jumped from about 9% to about 34% within a month, then fell back to about 7% by six months [9]. In a study in Niger, run alongside a larger trial called MORDOR I, in which 30 communities were randomly assigned to give young children azithromycin twice a year or a placebo, throat and gut resistance were both substantially higher after two years in the azithromycin communities [10], and a review of ten trials in young children in lower-income countries found the same pattern [11].

What nobody knew was what happens in babies born very early. Their gut bacteria are only just establishing themselves, they are in hospital rather than at home, and almost all of them receive antibiotics in their first days because infection is hard to rule out and dangerous to miss. Earlier work had shown that this environment leaves premature babies' guts dominated by hardy, resistant bacteria [12], and the research group behind AZTEC had already found resistance to azithromycin to be common in premature babies' guts even before any trial medicine — probably picked up from their mothers [13]. Because AZTEC was placebo-controlled, it could separate what the antibiotic did from what ordinary care did anyway, and checking for resistance was planned from the start [2].

What the two studies did

The two studies looked at the same trial from two ends of the body. The first examined the breathing passages: 1,108 samples of fluid suctioned from the breathing tube or the back of the nose, taken from 348 babies before treatment started and then around days 5, 10 and 14 — the first babies recruited, the group closing at 348 rather than the planned 350 when recruitment paused for the COVID-19 pandemic in March 2020. The researchers searched each sample for six genes known to make bacteria resistant to macrolide antibiotics, and grew bacteria from 150 samples to see which organisms were there and which other antibiotics they could withstand [14].

The second study examined the gut, using 1,261 stool samples from 693 babies over the first three weeks — babies born at a median of 27 weeks of pregnancy, weighing a median of 903 grams (about two pounds) at birth. The researchers read the genetic fingerprints of all the bacteria present, measuring how varied the community was and which types dominated, and tested for the same six resistance genes. Because some hospitals routinely gave probiotics (live "good" bacteria given as a supplement), they could also look at how probiotics and the antibiotic interacted, although probiotic use depended on the hospital rather than on chance [15].

What they found

In the breathing passages, the overall amount of resistance barely moved, and it was already high in both groups. Just under two-thirds of samples carried at least one macrolide-resistance gene: 361 of 567 (63.7%) from placebo babies and 346 of 541 (63.9%) from azithromycin babies. Only one of the six genes, erm(C), became more common in treated babies, consistently from day 5 onwards. The background level is high most likely because of ordinary care: roughly 86–88% of babies in each group were on antibiotics on the day they were born and the day after, and 28% were still on them at two weeks [14].

Growing the bacteria revealed what that resistance looks like. Out of 150 samples, 68 grew bacteria, and azithromycin-resistant bacteria grew far more often from treated babies. Of the 77 bacterial colonies identified, about half were Staphylococcus epidermidis and just over a third Staphylococcus haemolyticus — ordinary skin bacteria that nevertheless cause a large share of serious infections in premature babies [16]. Nearly all (75 of 77) carried at least one macrolide-resistance gene, and they resisted many other antibiotics too: all 68 staphylococcal colonies resisted ampicillin, 67 of 68 flucloxacillin and 62 of 68 gentamicin, while only one resisted vancomycin, an antibiotic commonly used in neonatal care [14]. These organisms are already difficult, whether or not a baby received azithromycin.

In the gut, the antibiotic's effect was clearer. Babies given azithromycin had a less varied mixture of gut bacteria, measurable from day 5 on one measure of variety and from day 10 on another. They had less Bifidobacterium — one of the bacteria most associated with healthy early gut development — which fell from the third most common type in placebo babies to the sixth in treated babies, and by three weeks fewer treated babies had any detectable Bifidobacterium. Bacteria from the Enterobacter and Klebsiella group, and staphylococci, became relatively more common, and the total number of macrolide-resistance genes was higher in treated babies, with erm(C) again standing out [15]. This is close to what other early antibiotics do: in 74 very premature babies, standard antibiotics increased Enterobacter and reduced bacterial variety in weeks two and three [17], and a trial in 147 babies born near or at term also found less Bifidobacterium and more Klebsiella [18].

Probiotics were helpful but not protective. Babies given probiotics had more varied gut bacteria and more Bifidobacterium, in line with research showing probiotics are one of the strongest influences on the premature gut [19], but they did not cancel out the antibiotic: babies given both still had less variety and less Bifidobacterium than babies given probiotics with placebo. Babies exposed to probiotics were also somewhat more likely to carry the erm(C) gene — an observation rather than a proven effect, since probiotic use followed hospital policy rather than chance, and earlier studies point in different directions, one suggesting probiotics might reduce antibiotics' effects [20] and another finding no lasting effect [21].

What this means for families

The first practical point is that azithromycin is not recommended for preventing chronic lung disease [4], and these studies give no reason to revisit that: the drug did not help, and it did have a cost.

The second is more reassuring than it may first sound. Resistant bacteria were already present in most samples from these babies, in both groups — they come mainly from the mother, the hospital environment, and the antibiotics nearly every very premature baby needs in the first days when infection cannot be ruled out [13][14]. Azithromycin added to that picture, but did not create it. And "resistant bacteria in a sample" is not the same as "infection": these studies counted genes and grew bacteria, and did not examine whether any baby became ill because of resistance.

The third point is the one the researchers themselves emphasise. The airway team concluded that because resistance is already common whatever a baby receives, every antibiotic should be used carefully and stopped as early as it safely can be, with clear routines for stopping antibiotics in babies showing no sign of infection [14]. The gut team concluded that, given the effects on gut bacteria and resistance genes, azithromycin may not be a good choice for premature babies [15]. One fair and welcome ward-round question is: "Is my baby still on antibiotics, and what is the plan for stopping them?" Related reading is in our article on neonatal antibiotic resistance and stewardship.

What researchers are working on next

The biggest open question is time. Both studies stopped at around three weeks, and the gut changes had not reverted by then — far too short to know whether they eventually do. Other research has found that very premature babies given early antibiotics can carry extra resistance genes and hardy resistant bacteria well after going home [22], while babies born at or near term in the trial mentioned earlier largely returned to normal within a year [18]; finding out which path premature babies follow needs sampling that continues after discharge. Two further questions follow: whether carrying these resistant bacteria changes the infections babies get later, and whether probiotics or newer approaches such as bacteriocins and postbiotics can protect the developing gut when antibiotics are genuinely necessary. For now, the clearest lesson from a trial that did not work is a practical one: in babies for whom antibiotics are nearly unavoidable, the decision worth getting right is not which extra antibiotic to add, but how soon the ones already running can safely stop.

References

  1. Lowe J, Watkins WJ, Edwards MO, et al. Association between pulmonary ureaplasma colonization and bronchopulmonary dysplasia in preterm infants: updated systematic review and meta-analysis. Pediatr Infect Dis J. 2014;33(7):697-702. doi:10.1097/INF.0000000000000239 ↩
  2. Lowe J, Gillespie D, Hubbard M, et al. Study protocol: azithromycin therapy for chronic lung disease of prematurity (AZTEC) - a randomised, placebo-controlled trial of azithromycin for the prevention of chronic lung disease of prematurity in preterm infants. BMJ Open. 2020;10(10):e041528. doi:10.1136/bmjopen-2020-041528 ↩
  3. Viscardi RM, Terrin ML, Magder LS, et al. Randomized trial of azithromycin to eradicate Ureaplasma respiratory colonization in preterm infants: 2-year outcomes. Pediatr Res. 2022;91(1):178-187. doi:10.1038/s41390-021-01437-2 ↩
  4. Lowe J, Gillespie D, Aboklaish A, et al. Azithromycin therapy for prevention of chronic lung disease of prematurity (AZTEC): a multicentre, double-blind, randomised, placebo-controlled trial. Lancet Respir Med. 2024;12(8):608-618. doi:10.1016/S2213-2600(24)00079-100079-1) ↩
  5. Joseph M, Murali Krishna M, Karlinski Vizentin V, Provinciatto H, Ezenna C. Azithromycin for Prevention of Bronchopulmonary Dysplasia and Other Neonatal Adverse Outcomes in Preterm Infants: An Updated Systematic Review and Meta-Analysis. Neonatology. 2026;123(1):74-83. doi:10.1159/000547537 ↩
  6. Antimicrobial Resistance Collaborators. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis. Lancet. 2022;399(10325):629-655. doi:10.1016/S0140-6736(21)02724-002724-0) ↩
  7. Rallis D, Giapros V, Serbis A, Kosmeri C, Baltogianni M. Fighting Antimicrobial Resistance in Neonatal Intensive Care Units: Rational Use of Antibiotics in Neonatal Sepsis. Antibiotics (Basel). 2023;12(3):508. doi:10.3390/antibiotics12030508 ↩
  8. Malhotra-Kumar S, Lammens C, Coenen S, Van Herck K, Goossens H. Effect of azithromycin and clarithromycin therapy on pharyngeal carriage of macrolide-resistant streptococci in healthy volunteers: a randomised, double-blind, placebo-controlled study. Lancet. 2007;369(9560):482-490. doi:10.1016/S0140-6736(07)60235-960235-9) ↩
  9. Bojang E, Jafali J, Perreten V, et al. Short-term increase in prevalence of nasopharyngeal carriage of macrolide-resistant Staphylococcus aureus following mass drug administration with azithromycin for trachoma control. BMC Microbiol. 2017;17(1):75. doi:10.1186/s12866-017-0982-x ↩
  10. Doan T, Arzika AM, Hinterwirth A, et al. Macrolide Resistance in MORDOR I - A Cluster-Randomized Trial in Niger. N Engl J Med. 2019;380(23):2271-2273. doi:10.1056/NEJMc1901535 ↩
  11. Luchen CC, Chibuye M, Spijker R, et al. Impact of antibiotics on gut microbiome composition and resistome in the first years of life in low- to middle-income countries: A systematic review. PLoS Med. 2023;20(6):e1004235. doi:10.1371/journal.pmed.1004235 ↩
  12. Gibson MK, Wang B, Ahmadi S, et al. Developmental dynamics of the preterm infant gut microbiota and antibiotic resistome. Nat Microbiol. 2016;1:16024. doi:10.1038/nmicrobiol.2016.24 ↩
  13. Gallacher DJ, Zhang L, Aboklaish AF, et al. Baseline azithromycin resistance in the gut microbiota of preterm born infants. Pediatr Res. 2024;95(1):205-212. doi:10.1038/s41390-023-02743-7 ↩
  14. Aboklaish AF, Watkins WJ, Gallacher D, et al. Assessment of macrolide resistance in the respiratory tract of preterm-born infants during treatment with azithromycin in the AZTEC clinical trial. Thorax. 2026;81(9):838-844. doi:10.1136/thorax-2025-224094 ↩
  15. Ahearn-Ford S, Aboklaish A, Young GR, et al. The impact of prophylactic azithromycin on the developing preterm gut microbiome and resistome. Gut Microbes. 2026;18(1):2748147. doi:10.1080/19490976.2026.2748147 ↩
  16. Dong Y, Speer CP, Glaser K. Beyond sepsis: Staphylococcus epidermidis is an underestimated but significant contributor to neonatal morbidity. Virulence. 2018;9(1):621-633. doi:10.1080/21505594.2017.1419117 ↩
  17. Greenwood C, Morrow AL, Lagomarcino AJ, et al. Early empiric antibiotic use in preterm infants is associated with lower bacterial diversity and higher relative abundance of Enterobacter. J Pediatr. 2014;165(1):23-29. doi:10.1016/j.jpeds.2014.01.010 ↩
  18. Reyman M, van Houten MA, Watson RL, et al. Effects of early-life antibiotics on the developing infant gut microbiome and resistome: a randomized trial. Nat Commun. 2022;13(1):893. doi:10.1038/s41467-022-28525-z ↩
  19. Beck LC, Masi AC, Young GR, et al. Strain-specific impacts of probiotics are a significant driver of gut microbiome development in very preterm infants. Nat Microbiol. 2022;7(10):1525-1535. doi:10.1038/s41564-022-01213-w ↩
  20. Esaiassen E, Hjerde E, Cavanagh JP, et al. Effects of Probiotic Supplementation on the Gut Microbiota and Antibiotic Resistome Development in Preterm Infants. Front Pediatr. 2018;6:347. doi:10.3389/fped.2018.00347 ↩
  21. Yuu EY, Bührer C, Eckmanns T, et al. The gut microbiome, resistome, and mycobiome in preterm newborn infants and mouse pups: lack of lasting effects by antimicrobial therapy or probiotic prophylaxis. Gut Pathog. 2024;16(1):27. doi:10.1186/s13099-024-00616-w ↩
  22. Gasparrini AJ, Wang B, Sun X, et al. Persistent metagenomic signatures of early-life hospitalization and antibiotic treatment in the infant gut microbiota and resistome. Nat Microbiol. 2019;4(12):2285-2297. doi:10.1038/s41564-019-0550-2 ↩