Technical Note

What people ate during antibiotics predicted how far their microbiome fell

A study published in Nature tracked 9,419 individual meals eaten by 173 hospitalised patients and matched them against what happened to their gut microbiome afterwards. Intake of sweets and sugars during antibiotic exposure predicted more severe dysbiosis: lower α-diversity and greater expansion of the pathobiont Enterococcus. In mice, sucrose supplementation made antibiotic-induced Enterococcus expansion both larger and longer-lasting.

Published 15 October 2026 · 6 min read

What the study did

Diet and the microbiome have been studied for years, but most of that work asks what a long-term dietary pattern does. This study asked a narrower and more useful question: during the few days when the microbiome is already being violently disturbed, does the specific food group eaten change how far it falls?

Diet was recorded meal by meal, not by questionnaire. 9,419 meals eaten by 173 patients hospitalised for haematopoietic cell transplantation were recorded precisely and broken down by food group. The inpatient setting is what made that level of precision possible.
The microbiome side was longitudinal and paired. 158 of those patients had paired longitudinal microbiome samples. Bayesian inference was used to link what was eaten to what changed afterwards, in the right temporal order.
The finding. Sweets and sugars consumed during antibiotic exposure predicted worse dysbiosis: a larger drop in α-diversity and greater Enterococcus expansion.
The mouse work supplied a direction of causality. Sucrose supplementation in mice increased antibiotic-induced Enterococcus expansion and prolonged it.

The authors' own wording is careful: avoiding a diet rich in simple sugars during antibiotic treatment may mitigate microbiota disruption, and they state explicitly that independent cohorts are needed to generalise the finding and to evaluate microbiota-sparing interventions.

How far this evidence reaches

This is a Nature paper with unusually good dietary data, and it still has firm limits. Stating them is more useful than overselling the result.

Three limits worth naming

The human arm is observational. Meal-by-meal records are far better than retrospective recall, and Bayesian inference handles the sequence of events, but patients were not randomised to eat sugar or avoid it. This is a predictive relationship, not an RCT result.
The cohort is transplant patients. High-dose antibiotics, mucosal barrier injury, immune reconstitution. The fragility of these patients' microbiomes is not on the same scale as that of a healthy adult taking a one-week course for a respiratory infection. The direction may transfer; the magnitude should not be assumed to.
The endpoints are microbiome measures, not clinical outcomes. What was measured is α-diversity and relative Enterococcus abundance — not infection rate, length of stay or survival. The mouse experiment establishes causality for one link in the chain (sucrose → Enterococcus expansion), not for "less sugar → a healthier patient".

So what the study supports is this: during antibiotic exposure, sugar is a dietary variable now backed by high-quality data and worth controlling. What it does not support is the claim that cutting sugar prevents antibiotic damage to the microbiome.

What this means for brand owners and formulators

1. Dietary guidance on the pack is not a product claim

"Post-antibiotic recovery" is a common positioning. What this study offers is advice about how the user eats during those days, not a conclusion about what the product does. Putting it in the directions for use or an accompanying dietary note speaks to eating behaviour and makes no assertion about product efficacy — regulatorily a different category of statement altogether.

2. Carriers and sweeteners are worth revisiting — with the dose gap stated honestly

Probiotic sachets and powder sticks commonly use sucrose, glucose and maltodextrin as carriers and sweeteners. To be clear: this study measured dietary intake of sweets and sugars, not excipients. A few grams of sugar in one sachet is not on the same scale as a day's confectionery intake, and the study never tested excipients. Using this paper to argue that a sugar-free product is therefore better is stretching the evidence past what it says.

That said, if the product is used precisely during or after antibiotic therapy, moving to a sweetener system that is not metabolised in the upper gut is a low-cost and defensible formulation choice. The justification is mechanistic plausibility, not a direct finding of this study.

3. Here Enterococcus appears as a pathobiont — and strain-level evidence does not erase that

The same genus has supplied working probiotic strains for decades, and in this study it is the organism that expands after antibiotics. It is worth being precise about why both are true, because “safety is strain-specific” is easily misused to wave away a real species-level risk. Some species in this genus carry opportunistic-pathogen traits in their own right — E. faecalis most clearly, with clinical isolates frequently carrying vanA, esp, hyl and gelE. The species level does carry information, and the information it carries is a constraint.

That is exactly why EFSA issued a dedicated safety guidance for E. faecium in animal nutrition: a strain with an ampicillin MIC above 2 mg/L, or carrying any one of the marker genes IS16, esp and hyl_Efm, should not be used as a feed additive. A species that needs its own screening threshold is a species whose risk at that level is real.

So the correct relationship is this: the species decides eligibility; the strain decides difference. First confirm the species is permitted for the intended application and passes the screens specific to that species; only then discuss what this strain does that others do not. Strain-level evidence narrows within the species-level gate — it is not a way past it.

And it is why species identity has to be confirmed genomically rather than inherited from historical naming: if you are not certain which species you are holding, you cannot know which safety gate applies. See: the species on the label and the species in the genome are not always the same.

If you are formulating in this space

For post-antibiotic recovery formats, three things are worth discussing together: strain selection, the carrier and sweetener system, and viability of the live organisms in the finished blend. We can supply strain-level documentation, stability data and a view on formulation feasibility.

We will not use this study to support an efficacy claim for any product. It shows that one dietary variable during antibiotic exposure is worth controlling; between that and "this product works" sits a trial nobody has run.

References

Dai A, Ballweg A, Jogia W, et al. Sugar-rich foods exacerbate antibiotic-induced microbiome disruption. Nature, 30 September 2026. doi:10.1038/s41586-026-11077-3. PubMed 42816625
EFSA FEEDAP Panel. Guidance on the safety assessment of Enterococcus faecium in animal nutrition. EFSA Journal. Full text

This note summarises and interprets published research on diet and microbiome disturbance. It is not a claim about the efficacy of any product and is not medical advice. The findings apply to the population, microbiome endpoints and experimental conditions described in the original paper. Antibiotic use should follow medical direction.