Wuxi Baifude Biotechnology Co., Ltd.
← Technical Notes

Technical Note

Post-antibiotic recovery runs on more than one clock

"Supports the gut after antibiotics" is one of the most common positioning lines in probiotic and prebiotic products. Few spec sheets say what "recovery" means. A preprint posted last month argues that the usual answer, community composition returning to baseline, is the wrong clock: read at the composition level, the same antibiotic exposure looks resolved in weeks; read at the resistance level, it is still visible months later. This note covers what the argument rests on, where that evidence is thinner than the headline, and what it means for the question buyers actually ask us: which organism should go into a post-antibiotic product? The answer depends on which clock the product is meant to move.

Published 5 November 2026 · 10 min read

What the paper did

Design

A single-author perspective, posted to Preprints.org on 28 October 2026 and not yet peer reviewed. It reports no new data. It re-reads two older randomised trials and names the pattern it sees composition–resistance decoupling. The two trials it draws on:

  • 2015, mBio: 66 healthy adults in the UK and Sweden randomised to clindamycin, ciprofloxacin, amoxicillin, minocycline or placebo, with saliva and faeces sampled at baseline, immediately after treatment, and at 1, 2, 4 and 12 months.
  • 2007, Lancet: 224 healthy volunteers randomised to azithromycin (74), clarithromycin (74) or placebo (76), with pharyngeal swabs to day 180 tracking the proportion of oral streptococci resistant to macrolides.

Result

The composition clock is fast. On the author's reading of the 2015 trial, salivary composition recovered within about 30 days while faecal communities remained altered across follow-up of up to 12 months. The original trial also reported a marked loss of butyrate-producing species in faeces and enrichment of resistance-associated genes. The resistance clock is slow. In the 2007 trial, one course raised the proportion of macrolide-resistant pharyngeal streptococci above placebo by 53.4 percentage points at day 4 (azithromycin) and 50.0 points at day 8 (clarithromycin). The original paper reports both arms significantly above placebo at every time point, through day 180.

The author proposes a minimum standard: measure both clocks, report them with denominators, and keep measuring until resistance returns to baseline.

What it does and does not establish

The core caution holds. Composition returning to baseline is not the same as the antibiotic's effect being over. Two randomised trials in major journals point the same way.

The specific "weeks versus months" contrast is assembled, not measured in one cohort. Three things to keep in view:

  • The two clocks come from different trials, body sites and methods. The composition clock is from saliva and stool sequencing in 2015; the resistance clock is from culture of throat streptococci in 2007. Neither trial shows, in the gut of the same people on one timeline, composition recovering first and resistance later. The 2015 trial did find resistance-gene enrichment in stool, but shotgun metagenomics was run only on selected baseline/post-treatment pairs, not as a time course.
  • The day-180 denominator is small. The 2007 trial lost 20 participants by day 42 and 105 by day 180, close to half. The author's call for denominators applies to the day-180 figure the paper itself relies on.
  • These were healthy adults. In transplant, ICU or repeatedly treated populations both clocks may run slower, and not necessarily in the same ratio.

Two trials published this year happen to illustrate the critique.

  • A randomised subset in Gut Microbes (18 adults with overweight or obesity; 7 days of vancomycin, then 8 weeks of 2'-fucosyllactose or placebo) measured both microbiome and resistome. Resistome richness and diversity did not change with vancomycin, yet the presence of specific resistance genes did. Read only at the summary-metric level, the resistome looks unaffected.
  • A cirrhosis trial in the same journal (15 vs 15) found no significant resistance-gene enrichment with rifamycin SV MMX, over 30 days. By the preprint's logic, 30 days is about when the fast clock has reset and the slow one has barely been read.

Supportable: composition endpoints alone cannot establish post-antibiotic recovery; resistance needs its own measurement over a longer window. Not supportable: "composition recovers in weeks, resistance in months" as a measured, general rule.

"Which organism for post-antibiotic use?" Name the clock first

Buyers usually ask this assuming a single endpoint. There are at least three, and the evidence on each is very different.

Clock one: symptoms, in days. The deepest evidence, with named organisms.

  • This is preventing antibiotic-associated diarrhoea (AAD) during a course, not recovery afterwards, but it is where many products are actually used.
  • The 2019 Cochrane review in children pooled 33 randomised trials (6,352 participants): AAD in 8% of the probiotic arms versus 19% of controls, RR 0.45, moderate-certainty evidence. Dose mattered: studies at ≥ 5 billion CFU/day did better than lower doses.
  • ESPGHAN's 2016 recommendations: where risk factors apply, Lacticaseibacillus rhamnosus GG or Saccharomyces boulardii (both moderate-quality evidence, strong recommendation); for C. difficile-associated diarrhoea, S. boulardii (low quality, conditional).
  • Read the resolution carefully. ESPGHAN only rates organisms with at least two strain-specified RCTs. That is an admission rule, not a finding that other strains fail. And S. boulardii has a reason to work here that holds at the level of the organism: it is a yeast, so antibacterial drugs do not target it, and it is not killed alongside the resident bacteria. For how genus, species and subspecies (or variety) labels differ in what they guarantee, see Genus, species, subspecies: what each rank on a label actually guarantees.

Clock two: composition, in weeks. No organism has positive evidence to cite.

  • The 2'-FL trial above: no improvement in microbiome recovery versus placebo after 8 weeks (n = 18).
  • A 2018 Cell study sampled gut mucosa directly in mice and humans. Compared with spontaneous recovery, a multi-strain probiotic produced markedly delayed and persistently incomplete reconstitution of stool and mucosal communities; autologous faecal transplant restored them within days. It is small and used one specific blend, so it does not settle the category, but it does mean "probiotic = faster recovery" cannot be the default assumption.
  • On the diet side there is strong data that sugar intake during a course worsens disruption: see What people ate during antibiotics predicted how far their microbiome fell.

Clock three: resistance, in months. No product data in any study cited here.

  • No study cited here shows any organism or prebiotic bringing resistance back to baseline faster. On this clock, "which organism?" has no answer yet.
  • One thing is certain in the other direction: an organism dosed into a gut where resistance genes are being enriched should not itself carry transferable resistance genes. That is a strain-level safety question; the broiler synbiotic note covers the clearest case, Enterococcus faecium.

A note from an ingredient manufacturer

Baifude manufactures lactic acid bacteria, bifidobacteria, Bacillus, yeast and blended functional powders, and post-antibiotic products are a use case for many of our customers. Two things follow:

The 2018 Cell finding cuts against our category. We have reported it as published. It is a small study and not decisive, but we do not use the line "probiotics help recovery after antibiotics" ourselves.

This note reviews no specific product, and we do not rely on any study cited here to support claims for our own powders. The L. rhamnosus GG and S. boulardii evidence belongs to the strains and preparations used in those trials. Its specific numbers do not transfer to same-species material from other sources.

What this means for brands and formulators

1. A "recovery" claim should name its endpoint and its timepoint

"Recovery" can mean symptoms resolving, composition returning to baseline, or resistance returning to baseline, on timescales from days to months. A 4- or 8-week trial showing diversity rebounding speaks to the fast clock and says nothing about resistance. When substantiating a structure/function statement, ask your supplier which metric was measured, on which day, and how many participants were still in the analysis on that day.

2. Leave "reduces resistance genes" out of the copy

No study reviewed here shows a product returning resistance to baseline faster. Antimicrobial resistance is the most regulator- and media-sensitive part of the antibiotics conversation. Until there is direct data, keep it out of marketing materials.

3. Choose the clock before the organism

For AAD prevention during a course, the evidence centres on L. rhamnosus GG and S. boulardii, with a dose threshold (≥ 5 billion CFU/day did better in the paediatric review). For compositional recovery after a course, no organism has positive evidence to cite, and one study points the other way. These are two different products and should not share a claim.

4. If you fund a trial, build it to the preprint's minimum standard

Measure composition and resistome; report the analysed n at every timepoint; read the resistome by shotgun metagenomics at the gene level, not only as richness and diversity; follow up until resistance returns to baseline, not until composition does. It costs more, but it removes "wrong clock" as an objection to the result.

5. Document each strain's own resistance genes

Intrinsic resistance to some antibiotics is common in probiotic organisms (many lactobacilli and vancomycin, for example) and is not in itself a concern. The question is whether any resistance gene sits on a mobile element. Ask for strain-level genomic assessment, not a species-level statement, especially for a product meant for people who have just finished a course.

If you are assessing this direction

Where we can help is the strain and documentation layer: confirming what an organism is at species and strain level, providing strain-level antimicrobial susceptibility and genomic resistance assessments, and checking whether a cited study used the same strain, dose, population and timepoint. Blend and carrier feasibility can be discussed alongside. We will not use any study to support an efficacy claim for a product, ours or anyone else's.

Contact us

References

  1. Shen C. Composition–Resistance Decoupling: Why Microbiome Recovery After Antibiotics Is Measured on the Wrong Clock. Preprints.org, 28 October 2026 (preprint, not peer reviewed). doi:10.20944/preprints202610.2862.v1. Preprint
  2. Zaura E, Brandt BW, Teixeira de Mattos MJ, et al. Same Exposure but Two Radically Different Responses to Antibiotics: Resilience of the Salivary Microbiome versus Long-Term Microbial Shifts in Feces. mBio, 2015;6(6):e01693-15. doi:10.1128/mBio.01693-15. PubMed 26556275
  3. Malhotra-Kumar S, Lammens C, Coenen S, et al. 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)60235-9. PubMed 17292768
  4. Vliex LMM, Godinho Mimoso IF, Le GN, et al. Modulation of the gut microbiome and resistome by antibiotic use, but not by subsequent 2'-fucosyllactose intake, in adults with overweight or obesity. Gut Microbes, 2026;18(1):2746814. doi:10.1080/19490976.2026.2746814. PubMed 42878177
  5. Bajaj JS, Davis BC, Puri P, et al. Rifamycin use and gut microbial antibiotic resistance genes in cirrhosis: Results from a double-blind, placebo-controlled clinical trial. Gut Microbes, 2026;18(1):2757339. doi:10.1080/19490976.2026.2757339. PubMed 42975318
  6. Suez J, Zmora N, Zilberman-Schapira G, et al. Post-Antibiotic Gut Mucosal Microbiome Reconstitution Is Impaired by Probiotics and Improved by Autologous FMT. Cell, 2018;174(6):1406-1423.e16. doi:10.1016/j.cell.2018.08.047. PubMed 30193113
  7. Guo Q, Goldenberg JZ, Humphrey C, et al. Probiotics for the prevention of pediatric antibiotic-associated diarrhea. Cochrane Database Syst Rev, 2019;4:CD004827. doi:10.1002/14651858.CD004827.pub5. PubMed 31039287
  8. Szajewska H, Canani RB, Guarino A, et al. Probiotics for the Prevention of Antibiotic-Associated Diarrhea in Children. J Pediatr Gastroenterol Nutr, 2016;62(3):495-506. doi:10.1097/MPG.0000000000001081. PubMed 26756877

This note summarises and interprets published research and a preprint in order to explain what "post-antibiotic recovery" means on different endpoints and timescales, and how far each piece of evidence can be cited. Reference 1 is a preprint that has not been peer reviewed, and its arguments may change in review. This note is not a claim about the efficacy or safety of any product, is not directed at any company or brand, and is not medical or regulatory advice. Claim substantiation requirements differ between jurisdictions; the rules of the market concerned govern in every case. Findings apply only to the populations, strains, preparations, doses and timepoints described in the original sources. Antibiotics and probiotics should be used as directed by a clinician.