This meta-analysis of 34 clinical trials supports the efficacy of probiotics in the prevention of AAD, expanding information from earlier meta-analyses. Our analysis shows that the preventive effect of probiotic administration is present across different probiotic species, is observed equally in children and adults, and appears to be independent of the concomitant antibiotics used and of the indication for the antibiotic treatment. The overall effect remains statistically significant even when restricting the analysis to studies with low risk of bias.
This meta-analysis tackles a clinically relevant issue. The incidence of AAD is variable yet remarkable, estimated between 5% and 39% of subjects undergoing any antibiotic treatment. Antibiotic-associated diarrhoea results in patient discomfort, potential loss of adherence to treatment, increased costs and can lengthen hospital stays in the in-patient setting.
The pathophysiological determinants of AAD are probably multiple. For instance, certain antibiotics, such as erythromycin and other macrolides, possess gastroduodenal prokinetic effects through agonistic activity on the motilin receptor. Most antibiotics may also significantly affect the composition and balance of the normal gut flora, in turn altering the bacterial breakdown of carbohydrates and the concentration of conjugated and unconjugated bile acids in the gut, which may result in alteration in the bowel habits from osmotic and motor mechanisms. Alteration of the gut microecology may favour excess proliferation of organisms that possess the potential for inducing greater water and electrolyte secretion, or that may act as true pathogens even in individuals with preserved mucosal immunity, such as C. difficile, Klebsiella ocitoca, staphylococci and candida spp.[66,67] In these settings, supplementation with probiotics during antibiotic treatment has been proposed to enhance the gut mucosal barrier function of the host, to favour competitive exclusion of potentially proliferating pathogen species among commensal bacteria, and to positively affect the host's immune response. In addition, certain probiotics may have direct anti-microbial effects against specific pathogens. For example, a bacteriocin produced by L. Salivarius was shown to be protective against Listeria monocytogenes, and S. Boulardii can directly bind to C. difficile-produced toxin A.
The supplementation of probiotics during antibiotic treatment to treat or prevent AAD is currently not a standard of care. Prior meta-analyses of probiotics for the prevention of AAD have shown similar risk reductions. Two meta-analyses, including 9 and 22 studies reported RRs (95% CI) of 0.37 (0.26, 0.53) and 0.40 (0.27, 0.57), respectively.[5,70] A more recent meta-analysis, which included 25 studies, reported a risk reduction of 0.43 (95% CI: 0.31–0.58). The present meta-analysis includes 20 of the 25 studies analysed previously, excluding the remaining studies for various inadequate design and reporting issues,[19,25,31,71,72] and includes 14 new studies published since the previous meta-analysis. Our estimates suggest a similar reduction in the risk of developing AAD while on a probiotic, compared to placebo.
A greater risk reduction was seen in the pooled analysis of studies of H. pylori eradication regimens. Possible explanations for this difference are both biological and methodological. The beneficial effect of probiotics on gastrointestinal (GI) symptoms, such as diarrhoea, may be enhanced when the concomitant antibiotic treatment is used to treat a GI condition. In addition, the antibiotic regimens for H. pylori often include multiple agents and commonly are associated with diarrhoea. It is also possible that H. pylori infection results in a more prominent bacterial flora imbalance, which may be the basis for a greater perceived effect of probiotic administration. Clinical studies of H. pylori eradication also used defined antibiotic regimens, and all but one study were conducted in asymptomatic volunteers, resulting in more homogeneity in study protocol and patient population within and between studies.
Strengths of the present meta-analysis in addition to the larger number of studies included are the adherence to updated consensus guidelines (PRISMA) including an assessment of risk of bias rather than quality or adequate reporting of methods, and the numerous sensitivity analyses performed.
Our results should also be interpreted in the context of several limitations. Our inclusion criteria and search strategy may have missed clinical trials with non-diarrhoea primary outcomes but in which the incidence of diarrhoea was explicitly measured. For example, all studies on H. pylori with the primary outcome of H. pylori eradication may not have been retrieved; however, this strategy also limits the search to studies published by investigators with a specific interest in the outcome of diarrhoea and potentially non-retrieved studies could conceivably had less rigorous collection of data on the outcome of diarrhoea (for example, patient self-report). The pooled estimate of probiotic efficacy resulted from studies that varied in the populations studied (adult vs. paediatric, in-patient vs. out-patient), in nature and modalities of intervention (probiotic strains and their doses, antibiotics used, duration of treatment) and in the outcomes considered (definitions of diarrhoea and its incidence, duration of follow-up). We attempted to ascertain the impact of these factors on our estimates by means of meta-regression using the log odds ratio of diarrhoea during antibiotic therapy in the placebo group as the dependent variable, and by subgroup analyses. While the presence of risk factors for AAD has not been studied and therefore cannot be accounted for, the patient populations in the included studies may have differed in their underlying risk for AAD.
Subgroup analyses by type of probiotic showed similar effects for the most commonly studied probiotic strains. These sensitivity analyses suggest an efficacy shared by most probiotics in the prevention of AAD.
With the exception of studies in H. pylori eradication, there was considerable between-study and within-study variability of the antibiotics regimens administered, thereby preventing a reliable subgroup analysis of effect by antibiotic administered. Notably, there was also substantial variability in the underlying overall health conditions of the study participants. Such variability, however, could also suggest that the effects of probiotics are generalisable to wider patient populations. An additional limitation is that the review was limited to studies published in English and to full text articles. A high-quality meta-analysis including non-English language papers published until 2006, however, points to effect sizes similar to those observed in the present analysis.
Probiotic administration is usually purported as devoid of side effects. In patients who are in critical conditions and in those with severe immune compromise, potential harmful effects of probiotics have been reported.[73,74] We did not systematically extract data related to adverse events and thus number needed to harm was not calculated; however, there were no serious adverse effects related to probiotics reported in any of the studies. Finally, while we report a NNT to aid in the interpretation of the findings, we recognise the limitations of this summary statistic when applied to pooled results especially when there is significant heterogeneity across studies.
In conclusion, this meta-analysis shows the preventive effect of probiotic supplementation on the incidence of AAD to be relatively consistent across different probiotic species used, various antibiotic regimens and indications, including H. pylori eradication, and in adult and paediatric populations.
As part of AP&T's peer-review process, a technical check of this meta-analysis was performed by Dr Y. Yuan.
Aliment Pharmacol Ther. 2012;35(12):1355-1369. © 2012 Blackwell Publishing