FormalPara Key Summary Points

Antibiotic-associated diarrhoea is a short-term sign of gut dysbiosis (i.e. the disruption of a previously stable, functionally complete gut microbiota) in adults and children.

Antibiotic-associated gut dysbiosis might have clinically significant long-term impacts on many organ systems, although the exact mechanisms have yet to be characterized.

The timely, evidence-based, appropriately dosed co-administration of specific probiotics can help to prevent or resolve antibiotic-associated diarrhoea and dysbiosis.

Further research on risk factors for antibiotic-associated diarrhoea is required.

Introduction

Gut dysbiosis corresponds to the disruption of a previously stable, functionally complete microbiota and is associated with harmful short- and long-term effects on health [1]. A concrete example of gut dysbiosis—notably because it results in diarrhoea as a visible sign in between 10% and 40% of patients [2,3,4,5]—is that induced by (or associated with) antibiotic treatments; this has been referred to variously as “antibiotic-related dysbiosis” [6], “antibiotic-induced dysbiosis” [7] and “antibiotic-associated dysbiosis” [8]; we shall adopt the last term, hereafter abbreviated as AAdys.

In our experience, clinicians are not sufficiently aware of (i) the incidence of antibiotic-associated diarrhoea, (ii) the potential short- and long-term health consequences of AAdys, and (iii) the ability of probiotics to mitigate antibiotic-associated diarrhoea. Hence, in the present commentary, we summarize our understanding of the significance of AAdys and indicate how it can be managed through the evidence-based administration of specific probiotics. Although we did not perform a formal literature search for this experience-based commentary, most of the publications cited herein will be identified by a PubMed search with logical combinations of the following keywords: gut, intestine*, dysbiosis, antibio*, microbiot*, microb*, diarrhea*, diarrhoea*, Clostrid*, AAdys, probiotic, Saccharomyces boulardii CNCM I-745, Lactobacillus rhamnosus, resilience.

Antibiotics and Gut Microbiota

Despite legitimate concerns about the growing threat posed by microbial resistance [9], today’s antibiotics are still among the miracles of modern medicine [10]. Indeed, antibiotics are highly effective and very safe drugs [11]. However, the potential rapid, intense, negative effects of antibiotics—even narrow-spectrum ones—on the composition of intestinal microbiota should not be underestimated: widespread reductions in diversity across many phyla, the potential loss of whole microbial communities, overgrowth by pathogenic species (such as Clostridioides difficile), and an increase in the spread of antibiotic-resistance genes [12,13,14,15]. Very generally, antibiotic treatments tend to increase levels of Enterobacteriaceae, Bacteroidaceae, enterococci, and resistant Escherichia coli and decrease levels of bifidobacteria, lactobacteria, actinobacteria, Lachnospiraceae, and streptococci [14, 15]; however, there are marked differences between and within antibiotic classes in this respect [15].

Short-Term Negative Health Impact of Antibiotic Treatment: Antibiotic-Associated Diarrhoea

Antibiotic-associated diarrhoea is a common, serious, clinical problem. By way of an example, a study of internal medicine wards in four Belgian hospitals found that 743 (29.2%) of the 2543 hospitalized adult patients had been treated with antibiotics, and that 9.6% of these antibiotic users developed diarrhoea (starting between the 1st and 16th days of treatment) [16]. The risk factors for antibiotic-associated diarrhoea were found to be age greater than 70 years, treatment with a combination of antibiotics, the use of proton pump inhibitors (PPIs), endoscopy, diabetes, kidney problems, the decubitus position, and hospitalization in a nephrology ward [16]. It must be borne in mind that the 90% of people who do not experience patent antibiotic-associated diarrhoea are still likely to experience AAdys to some extent—at least in terms of compositional changes that may persist over time [17, 18].

In our experience, antibiotic-associated diarrhoea has particularly significant effects on frail patients, such as those requiring intensive care [19, 20]. In an intensive care setting, the antibiotics’ impact on the microbiota is amplified by the effects of vasopressor use, highly processed enteral nutrition products, gastrointestinal tract surgery, age, PPI use, and long-term opiate use [19,20,21]. Interestingly, in a subset of healthy volunteers treated with antibiotics, the reduction in microbial diversity was similar to that seen in patients in the intensive care unit [17, 18].

Furthermore, antibiotic-associated diarrhoea can reveal or lead to a C. difficile infection of the gut, the signs of which can range from mild watery diarrhoea to life-threatening pseudomembranous colitis. C. difficile is present in up to 50% of asymptomatic infants, up to 10% of hospital patients, and 1–5% of healthy adults [22,23,24,25,26]. C. difficile can expand in the event of gut microbiota dysbiosis, and the standard treatment of a C. difficile infection with an antibiotic can result in further perturbation of the microbiota. Hence, a resilient gut microbiota is essential for recovery from C. difficile infection [27, 28]. Other treatment options include probiotics (see below), live biotherapeutic products, and faecal microbiota transplantation [29,30,31].

Hence, in older and/or fragile patients, the short-term consequences of AAdys can be serious and even life-threatening. In infants, the physician should also be concerned about the life-long health consequences of AAdys, as described below.

Dysbiosis in Infancy and the Long-Term Health Impact of Antibiotic Treatments

It is known that an “adult” microbiota is attained at around the age of 3–5 years; schematically, the first 1000 postnatal days constitute a critical, relatively unstable period in the microbiota’s development [32, 33]. Controversially, it has been suggested that the future infant’s microbiota starts to develop before birth: bacterial DNA has been detected in placental biopsies and amniotic fluid samples collected under supposedly sterile conditions, as well as in the meconium [34,35,36,37]. However, contamination during sample collection or preprocessing is thought to be a major problem and calls the foetal microbiome into question [38].

At birth, caesarean section prevents the transfer of maternal Bifidobacterium and Lactobacillus species and increases the risk of colonization by opportunistic pathogens present in the hospital environment, such as Enterococcus, Enterobacter, and Klebsiella [39]. This caesarean-linked delay in establishment of the microbiota is thought to be associated with an increased risk of obesity and other noncommunicable diseases in later life [40].

During the first years of life, the gut microbiota is sensitive to dysbiosis induced or accentuated by a variety of environmental factors, such as the mode of delivery, the diet (e.g. mother’s milk vs. formula milk), the ingestion or overgrowth of pathogenic microorganisms, and exogenous compounds (including antibiotics) [41]. There is evidence from studies of cohorts in Finland and in Germany that neonatal exposure to antibiotics impairs growth (resulting in a low body mass index (BMI) for the first 6 years of life) [42]. This effect was particularly marked in boys. However, the literature data on neonatal antibiotic exposure and subsequent BMI are subject to debate [43]. Trasande et al. found an association between antibiotic exposure in the first 6 months of life and a high BMI at the age of 10 to 38 months but not at the age of 7 years [44]. Murphy et al. observed an association between antibiotic exposure in infancy and the risk of an elevated BMI but only in boys [45]. However, other studies failed to find a relationship [46, 47]. Furthermore, it has been suggested that an elevated BMI is more strongly linked to broad-spectrum antibiotics (such as macrolides) than to narrow-spectrum antibiotics [48, 49].

It appears that gut dysbiosis in infancy might contribute—along with other factors—to an elevated lifetime incidence of noncommunicable diseases and disorders, such as atopic disorders (including asthma), immune/inflammatory disorders (including inflammatory bowel disease), neurological disease, diabetes, kidney disorders, obesity/overweight, necrotising enterocolitis, infantile colic, stress, anxiety, and social behavioural difficulties [50,51,52,53,54,55,56,57]. However, the underlying mechanisms—possibly metabolic in nature—have yet to be identified and/or fully characterized.

Resilience and a “Healthy” Microbiota

As discussed above, antibiotic treatment results in dysbiosis and therefore an abnormal, “unhealthy” microbiota in the gut. The problem is then one of defining a “normal” or “healthy” gut microbiota, the re-establishment of which should be a clinical goal. However, compositional definitions are not necessarily adequate for defining a “healthy” microbiota, and researchers have focused on functional profiles. Turnbaugh et al.’s study of adult monozygotic and dizygotic twin pairs showed that the phylogenetic beta-diversity of the bacterial lineages present in the gut microbial community varied markedly from one individual to another but that the broad functional profiles (as judged from categories of genes and types of metabolic enzymes) were remarkably similar [58]. Arumugam et al.’s Sanger sequencing study identified three robust clusters (enterotypes) that were not nation- or continent-specific from among 39 individuals (22 metagenomes from Danish, French, Italian and Spanish individuals, plus other published Japanese and American datasets) [59]. These enterotypes are also found in non-human primates [60]. Arumugam et al. pointed out that abundant molecular functions are not necessarily provided by abundant species [59]. Hence, we suggest that a more explanatory, scientifically justified but admittedly longer term for “dysbiosis” is “disruption of a previously stable, functionally complete microbiota”.

One key feature of a “healthy” microbiota is resilience [61,62,63]. A useful analogy for resilience is that of a “ball” (the microbiota) at rest in a minimum (a trough) in a functional landscape (Fig. 1). In the event of a small disruption, the ball might move up the side of the minimum at first but then will roll back to its initial position—thanks to resilience. However, in the event of a large disruption (such as that caused by antibiotics), the ball leaves its baseline position and falls into a less functional, metastable, local minimum: this is dysbiosis. Functional stability appears to be a feature of a healthy microbiota. The Human Microbiome Project Consortium studied study (PMID: 22699609) of 4788 specimens from 242 screened and phenotyped adults (129 male, 113 female) [64]. To assess function, 749 samples were sequenced using 101-bp paired-end Illumina shotgun metagenomic reads and a functional (metabolic) sequence database that included the Kyoto Encyclopedia of Genes and Genomes Orthology. The researchers reported that the within-subject variation in metabolic function over time was consistently lower than the between-subject variation, and suggested that the stability of an individual’s microbial community is a feature associated with health [64]. After the first few years of life, each human microbial reaches a homeostatic climax composition [65]. However, one practical problem with assessing the functional stability approach is that clinical studies of ill patients rarely include any data on the baseline microbiota prior to hospital admission or the onset of serious illness.

Fig. 1
figure 1

A diagrammatic representation of the concepts of antibiotic-associated dysbiosis and resilience. Eubiosis corresponds to a functionally stable state, indicated here as a trough. When the microbiota is perturbed by a variety of factors, collective resilience (a combination of microbiotic and host factors that have yet to be fully characterized) maintains the stable eubiotic state [61,62,63]. However, this resilience may not be sufficient to counter the major perturbation caused by the administration of antibiotics—a perturbation that affects not only the microbiota but also the gut’s barrier functions and immune functions. The administration of antibiotics triggers a critical transition from eubiosis to dysbiosis—a less functional but meta-stable state [1]

Full size image

Although resilience appears to be linked mainly to the functional diversity of species present, there are constant metabolic, immunological, and even neurological interactions between the microbiota and the host tissues; the host’s roles in the onset and prevention of dysbiosis should not be ignored [66,67,68,69,70]. For example, antibiotic treatment in mice is associated with disruption of the intestinal tight junction barrier, decreased tight junction protein expressions, disrupted ZO-1 morphology, activation of the NLRP3 inflammasome and autophagy, and a macrophage-dependent increase in inflammatory T helper 1 responses [71, 72]. AAdys and changes in species diversity can alter epithelial homeostasis by altering Toll-like receptor signalling and immunoregulation [73]. Studies in animal models show that antibiotics have both short- and long-term effects on gastrointestinal motility, some of which are due to changes in host gene expression and metabolism [74, 75]. All these host factors may contribute to antibiotic-associated diarrhoea.

Preventing, Mitigating and Reversing AAdys: Role of Probiotics

In the context of AAdys, re-establishment of a normal (functional) gut microbiota requires completion of the clinically indicated course of antibiotics, time (i.e. recovery through the microbiota’s natural resilience), and appropriate support, including probiotics.

Probiotics have been defined as “live microorganisms which, when administered in adequate amounts, confer a health benefit on the host” [76]. Systematic reviews and meta-analyses of clinical trials have confirmed the strain-specific efficacy of some probiotics in the treatment and/or prevention of antibiotic-associated diarrhoea in children and in adults [3, 77,78,79,80]. This is particularly the case for the yeast S. boulardii CNCM I-745 and the bacterium L. rhamnosus GG [3, 77,78,79,80]. On the basis of these findings, the European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) recommended the administration of sufficient doses (at least 5 × 109 colony-forming units (CFU) per day) of specific strains of probiotics in children presenting risk factors (as judged by the prescribing physician) for antibiotic-associated diarrhoea, upon initiation of the antibiotic treatment [78]. The World Gastroenterology Organisation’s Global Guidelines include the same recommendation [81]. In its 2020 guidelines, the American Gastroenterology Association suggested the use of certain strains and strain combinations of probiotics in the prevention of C. difficile infections as a conditional recommendation but emphasized the unclear or high risk of bias in most of the trials considered [82]. Lastly, the importance of strain specificity is likely to be influenced by the regulatory regimen applied to probiotics; considering probiotics as “live biotherapeutic products” (rather than food supplements) requires more detailed documentation and closer scrutiny of efficacy and safety [83, 84].

Unusually for a probiotic, S. boulardii CNCM I-745 is a eukaryote and is not therefore directly affected by antibiotics—making it especially suitable for administration in an indication of antibiotic-associated diarrhoea [85, 86]. S. boulardii is naturally absent from the human intestinal microbiota, achieves stable population levels after 3 days of administration as a probiotic, and is no longer detected in the stools 2–5 days after the discontinuation of administration [87]. This probiotic appears to have several modes of action: direct binding to pathogens, a reduction in local inflammation, antitoxin effects, effects on digestive enzymes, the stimulation of short-chain fatty acid production by Lachnospiraceae and Ruminococcaceae, and the formation of a protective environment for beneficial species at the intestine’s mucus layer (Fig. 2) [87,88,89,90].

Fig. 2
figure 2

Mechanisms of action of the probiotic Saccharomyces boulardii CNCM I-745 in the context of antibiotic-associated dysbiosis. Antibiotics perturb the microbiota and the gut’s barrier functions and immune functions (left panel) [33, 37, 68, 70, 104]. The probiotic S. boulardii CNCM I-745 notably counters these perturbations and promotes a return to eubiosis by increasing microbial diversity, stimulating the microbial production of short-chain fatty acids (SCFAs), reinforcing the gut’s barrier functions and diminishing local inflammation (right panel) [87,88,89,90]

Full size image

Szajewska and Kolodziej’s meta-analysis showed that relative to a placebo or the absence of probiotic treatment, the administration of S. boulardii reduced the risk of antibiotic-associated diarrhoea in adults and children (in the context of Helicobacter pylori eradication) by 8.5–18.7% (risk ratio (RR) [95% confidence interval (CI)] = 0.47 [0.38–0.57]) [79]. It should be noted that Szajewska and Kolodziej’s meta-analysis of 21 selected studied covered four antibiotic treatment settings: children treated for infections, children treated for H. pylori eradication, adults treated for infections, and adults treated for H. pylori eradication. Although the overall heterogeneity was not statistically significant (v2 = 28.44, P = 0.10, I2 = 30%), the heterogeneity differed from one treatment setting to another; this raises the level of uncertainty about the absolute effect size. Furthermore, the meta-analysis covered trials in which S. boulardii was compared with a placebo and trials in which it was compared with the absence of treatment; this mixture makes it more difficult to determine an absolute effect size. Lastly, treatment with S. boulardii was associated with a significantly lower risk of C. difficile-associated diarrhoea in children (RR [95%CI] 0.25 [0.08–0.73]), although it should be noted that only two studies in this treatment setting were included in the meta-analysis [79, 80].

Similarly, another meta-analysis of studies in adults and children by Szajewska and Kolodziej found that the administration of L. rhamnosus GG reduced the risk of antibiotic-associated diarrhoea (RR [95%CI] = 0.48 [0.26–0.89]) [80]. However, this overall significant effect was driven by the effect observed in five studies in children; the effect was not significant in adults alone. Lastly, in a recent systematic review of 29 publications, Fernández-Alonso et al. found that despite (i) the lack of standardized protocols for analysing the gut microbiota and (ii) the wide range of products studied, the co-administration of probiotics with antibiotics prevents some of the changes in gut microbial diversity and composition induced by antibiotics—including the restoration of health-promoting bacteria and the recovery of alpha diversity [91]. In most of the studies reviewed by Fernández-Alonso et al., the probiotic and antibiotic treatments were initiated at the same time, and the probiotic was terminated at the same time as (or a week or two later than) the antibiotic.

Despite the proven benefits of probiotics on the prevention of antibiotic-associated diarrhoea in adults and children (especially those having already experienced this problem), only 10–60% of antibiotic prescriptions are accompanied by a probiotic [91, 92]. Physicians’ attitudes to the co-prescription of antibiotics and probiotics appear to vary greatly from one country to another [93]. On the basis of our experience and the results of Fernández-Alonso et al.’s systematic review, and with a view to preventing or mitigating antibiotic-associated diarrhoea, we recommend the initiation of adequately dosed probiotic treatment at the same time as antibiotic initiation or (in view of the speed of action of antibiotics) within the following 48 h. The typically recommended dose of S. boulardii CNCM I-745 for the prevention and treatment of antibiotic-associated diarrhoea is 250 mg once daily in children and 500 mg once daily (or 250 mg twice daily) in adults [78, 94]. Underdosing of S. boulardii CNCM I-745 is a potential issue in research and in clinical practice: Kelesidis and Pothoulakis pointed out that the dose of S. boulardii is not reported consistently in clinical trials (complicating meta-analyses) and/or is quoted in different units (organisms per 100 ml, organisms per day, colony forming units per day, or grams per day) [95]. The dose of L. rhamnosus GG recommended by the ESPGHAN for preventing nosocomial diarrhoea is at least 109 CFU per day for the duration of the hospital stay [96, 97]. The probiotic S. boulardii can be taken with food but not with hot liquids (e.g. tea or coffee). Although S. boulardii CNCM I-745 is generally well tolerated, some people may experience side effects like gas, bloating, and constipation. Lastly, S. boulardii-based products are contraindicated for seriously ill or immunosuppressed patients because of the risk of Saccharomyces cerevisiae fungemia—a rare but potentially life-threatening problem [98].

Perspectives

Even though the evidence for probiotics’ efficacy in antibiotic-associated diarrhoea is robust, there are many open research questions. Firstly, there is a need for additional long-term, non-interventional, epidemiological studies of the effects of antibiotic treatment on the growth and development of exposed infants. Secondly, research could usefully focus on the effects of probiotics in patients taking a dysbiosis-inducing drug other than an antibiotic (such as PPIs, cancer drugs, and NSAIDs) alone or in combination with an antibiotic [99]. Thirdly, further investigation of risk factors for antibiotic-associated diarrhoea (age, comorbidities, the antibiotic class, the pattern of previous episodes, etc.) is required so that the individuals who would most benefit from preventive or therapeutic administration of probiotics can be identified. Fourthly, there is a clear need to identify one or more “resilience biomarkers”, i.e. reliable markers of the microbiota’s status under perturbed and non-perturbed conditions. These might be functional markers, rather than compositional diversity markers. Indeed, the definition of dysbiosis as a change in composition has been criticized, and it may not be clear whether changes in composition are causes or consequences of concomitant health problems [100]. Fifthly, valuable scientific and clinical insights might be gained through comparative studies of strains within a given probiotic species. Lastly, it would also be useful to know whether the presence of S. boulardii CNCM I-745 limits the spread of antibiotic resistance genes. Animal models are likely to be of value when addressing some of the above mechanistic questions, since (for example) S. boulardii CNCM I-745 appears to have much the same effects in mice as it does in humans [101]. In a recent study of a faecal microbiota transfer model in the mouse, 16S and ITS2 rRNA sequencing revealed that the administration of S. boulardii CNCM I-745 was associated with a more rapid recovery of bacterial populations after a course of amoxicillin-clavulanate antibiotics [102].

With regard to primary care, we suggest that efforts should be focused on preventing AAdys and promoting resilience through lifestyle, nutritional and biological measures. Currently, microbiota analyses (whether based on 16S RNA sequencing, shotgun sequencing, or other techniques) are not useful in clinical practice; the physician’s skill resides in finding and interpreting clinical clues that indicate the presence of AAdys. In line with ongoing efforts to promote evidence-based antibiotic stewardship, we think that prescribers need more information on the known short- and long-term effects of antibiotic classes and other compounds on the microbiota [103]. Lastly, a focus on the prescription of evidence-based probiotics should help to limit unjustified, underdosed and/or potentially dangerous treatments with non-prescription products—especially in children.

Conclusion

Our key messages and recommendations are as follows: (i) the microbiota is important for short- and long-term health; (ii) almost everyone has a “normal” functional microbiota at some point, and this microbiota is resilient; (iii) antibiotics have overwhelming positive effects on health but also have negative short-term (and perhaps long-term) effects that require further investigation; (iv) antibiotic-associated diarrhoea is not only a clinical problem but also a marker of dysbiosis; (v) there is a growing body of evidence to show that the timely administration (upon initiation of antibiotic treatment or within 48 h) of an adequately dosed, clinically validated, evidence-based probiotic (such as S. boulardii CNCM I-745 and L. rhamnosus GG) can help to prevent or resolve antibiotic-associated diarrhoea and dysbiosis and promote a return to a resilient, pre-antibiotic state, and (vi) there is a need for additional research on the variables that influence susceptibility to antibiotic-associated diarrhoea and the effectiveness of probiotic treatments.