Background

Antibiotic-associated diarrhoea (AAD), defined as diarrhoea in association with antibiotic treatment and without an alternative cause, occurs in 5–39% people within 12 weeks of exposure to antibiotics.^1,2 The predominant mechanism underlying AAD is disruption of the commensal gut flora. This impairs colonisation resistance and facilitates the emergence of a range of gut pathogens.^2,3 The major diarrhoeal pathogen associated with antibiotic treatment is Clostridium difficile, which accounts for 15–39% of AAD cases.⁴ Altered commensal flora may also result in diarrhoea through changes in the metabolism of carbohydrates, short-chain fatty acids and bile acids, and some antibiotics cause diarrhoea through direct effects such as increased gut motility.^2,5

Clostridium difficile is an anaerobic bacterium that produces resistant spores that persist long term in the environment. Transmission is faecal–oral and 4–21% patients may acquire the infection during admission to hospital through contact with colonised patients, contaminated fomites and the hands of health-care staff. C. difficile diarrhoea (CDD) occurs in both endemic and outbreak scenarios.⁶ Since 2003, an increase in the frequency and severity of CDD in North America and Europe has been attributable to the emergence of the hypervirulent 027 strain, which may produce higher amounts of toxin.^7–9

Antibiotic-associated diarrhoea occurs typically in older people admitted to hospital.^1,2 ADD complicates treatment with many antibiotic classes but especially broad-spectrum penicillins, cephalosporins, clindamycin and long-duration antibiotic treatments.¹⁰ Additional risk factors include prolonged hospital stay, previous hospital admission, previous gastrointestinal surgery and use of a nasogastric tube (NGT).³ In a retrospective study of European hospitals, risk factors for CDD included age ≥ 65 years, severe comorbidity and recent treatment with cephalosporins and aminopenillin-beta (β)-lactamase inhibitor combinations.¹¹ In a prospective study of people aged > 18 years admitted to Canadian hospitals, age, exposure to antibiotics, treatment with proton pump inhibitors (PPIs) and previous hospital admission within the last 2 months predicted CDD.¹²

The severity of AAD varies greatly. Although usually a mild, self-limiting illness, it is associated with prolonged hospital stay and increased health-care costs. C. difficile infection may remain asymptomatic, but intestinal pathology results from secretion of toxins A and B causing increased mucosal fluid secretion and inflammation. Symptoms range from mild, self-limiting diarrhoea to severe diarrhoea complicated by pseudomembranous colitis, toxic megacolon and death.^6,8

Estimates of the financial burden of C. difficile infection (CDI) to the health-care service have varied between $2454 and $16,464 for every health-care-acquired case in the USA,^13–15 £4107 in the UK¹⁶ and €7147 in Germany.¹⁷ The annual cost of health care for CDD in the USA has been estimated to be $3B.^18,19 Nosocomial CDI increases hospital stay by between 7 and 26 days,^16,17,20 and prolonged hospital admission was identified as the main cost driver in most studies.^13,15,16 Furthermore, an increase in length of hospital stay due to more severe disease in recent years has resulted in a rise of the incremental cost of CDI.²¹

Treatment

Uncomplicated AAD usually responds to withdrawal of the offending antibiotics. CDD usually responds to treatment with metronidazole or vancomycin, but 20–25% patients go on to suffer from a recurrent form of the disease. Novel modes of treatment include probiotics, immunoglobulin infusion and faecal transplant from healthy donors.^3,6

Prevention of antibiotic-associated and Clostridium difficile diarrhoea

The frequency and severity of CDD in hospitals in the industrialised world have led to comprehensive strategies for prevention. These include decontamination of the environment, hand washing and isolation of patients with diarrhoea.⁶ Antibiotic stewardship programmes have also been effective in reducing infection rates.^6,22,23

Probiotics

Probiotics are defined as live microbial organisms which, when administered in adequate numbers, are beneficial to health.²⁴ Although clinical trials are undertaken to determine whether or not a microbial preparation has a health benefit in a specific population, the term ‘probiotic’ is commonly used to refer to the preparation being evaluated and it is used in this sense here. Bacteria used as probiotics are among the organisms ‘generally regarded as safe’ by the Food and Agriculture Organization of the United Nations.²⁵ Live bacteria from several genera and the yeast Saccharomyces boulardii have been administered to vulnerable groups such as preterm infants and people with human immunodeficiency virus (HIV) infection without adverse effects.²⁶ Adverse events occurring in clinical trials evaluating the prevention of AAD have not been ascribed to probiotic intake.²⁶ Administration of lactic acid bacteria has been associated in rare cases with septicaemia in immunocompromised people and endocarditis in people with artificial heart valves.²⁷ Despite the apparent low risk of adverse events, careful assessment of safety in clinical trials is recommended.²⁶

Rationale

Specific probiotic strains have been identified in vitro and in vivo to possess several mechanisms that may prevent or ameliorate AAD and CDD through enhanced barrier function of the gut epithelium.²⁸ First, potential pathogens are killed or inhibited by the secretion of antimicrobial peptides and probiotics compete for attachment sites on intestinal mucus and epithelium. Acidification of the gut contents through the production of lactic acid also inhibits pathogen growth. Second, mucosal immunity is enhanced through increased secretory immunoglobulin A production and stimulation of antimicrobial peptide secretion by host cells. Finally, through direct effects on the epithelium, probiotics increase the secretion of mucus, enhance tight junction integrity and decrease epithelial cell apoptosis.

At the time that probiotic lactobacilli and bifidobacteria in antibiotic-associated diarrhoea and Clostridium difficile diarrhoea in the elderly (PLACIDE) was developed, meta-analyses had suggested that probiotics may be effective in the prevention of AAD. McFarland pooled data from 25 randomised controlled trials (RCTs), which included a total of 2810 adults and children.²⁹ There was significant heterogeneity (χ² = 82.5; p < 0.001) in a fixed-effects model. In random-effects analysis, the relative risk (RR) for AAD was 0.43 [95% confidence interval (CI) 0.31 to 0.58] in participants receiving a probiotic. This meta-analysis included all of the trials included in previous systematic reviews, which had broadly similar findings.^30–33 Although meta-analysis has provided some evidence that probiotics may be effective in the prevention of AAD, the marked differences between trials in the microorganisms evaluated (single strains and mixtures of bacteria, the yeast S. boulardii), administration regimens (mode of delivery, number of organisms, probiotics combined with prebiotics), patient populations (age and exposure to antibiotics) and period of follow-up for AAD probably underlie the statistical heterogeneity in the results and weaken the evidence for probiotic effectiveness.

Much less evidence from clinical trials was available for probiotics in the prevention of CDD. A pilot study in elderly hospitalised patients reported that 30 out of 138 (21.7%) patients developed diarrhoea and 5 out of 69 (7.2%) in the placebo group compared with 2 out of 69 (2.9%) who had received a combination of Lactobacillus acidophilus and Bifidobacterium bifidum tested positive for C. difficile toxin. Stool culture suggested that the main effect of the probiotic was neutralisation of toxin rather than prevention of colonisation.³⁴ Thomas et al. had assessed Lactobacillus rhamnosus GG for the prevention of AAD in 267 adults who were monitored for an average of 21 days, but the number of patients from whom stool samples had been collected and in which C. difficile was detected was too small to assess probiotic effect.³⁵ In randomised trials of 193³⁶ and 180 adult patients,³⁷ the occurrence of CDD was similar in those receiving S. boulardii and those receiving the placebo. Although there was trial evidence for probiotics in the treatment of established CDD or the prevention of recurrence,²⁹ we are not aware of any other studies that had assessed probiotics in the prevention of CDD in adults.

We are not aware of any studies that have assessed the effect of probiotic on quality of life (QoL) in patients at risk of AAD.

Selection of the probiotic preparation

Although several mechanisms whereby probiotic organisms enhance gut barrier function have been identified,²⁸ it remains unclear which of these are most relevant to the prevention or amelioration of AAD and to what extent these mechanisms are common to many different probiotic organisms or are strain specific. Therefore, the scientific evidence to inform the selection of a specific probiotic intervention for the prevention of AAD is limited. The meta-analyses included trials that had evaluated many different bacterial preparations and the yeast S. boulardii.²⁹ The bacterial interventions included single strains, mixtures of different organisms and mixtures of probiotics with prebiotics. Dosages (number of organisms) and modes of administration also varied markedly between studies. Factors associated with greater efficacy in preventing AAD in meta-analysis included the use of S. boulardii or L. rhamnosus GG, mixtures of probiotics and preparations with high numbers of organisms.²⁹ Efficacy was similar for bacterial preparations [five trials conducted on 384 adults and children; odds ratio (OR) 0.34; 95% CI 0.19 to 0.61] and yeast preparations (four trials conducted on 830 participants; OR 0.39; 95% CI 0.25 to 0.62).³¹

In an attempt to maximise gut colonisation and, therefore, colonisation resistance, we used a multistrain preparation of Lactobacillus sp. and Bifidobacterium sp., bacterial species that had been evaluated extensively in clinical trials, with a high number of viable bacteria (total 6 × 10¹⁰ organisms per day). We intended to undertake a pragmatic trial to assess the clinical effectiveness and cost-effectiveness of the probiotic preparation in older people receiving antibiotics in secondary health-care settings representative of those in industrialised counties and with the causes of diarrhoea determined by routine laboratory practice.