Abstract
What is the evidence of uncertainty?We conducted a review of meta-analyses, updated with subsequent randomised controlled trials. We searched PubMed, EMBASE, the Cochrane Library, and Clinical Evidence in October 2011 for meta-analyses published in the past five years in English and trials published after their search dates on probiotics to prevent antibiotic associated diarrhoea using the search terms “probiotic” and “antibiotic associated diarrhoea [or diarrhea]”. We excluded prevention studies, small pilot studies, studies that were not placebo controlled, studies published only in abstract form, studies focusing on antibiotic associated diarrhoea caused by a single organism (such as Clostridium difficile), and systematic reviews without meta-analysis.The commonest outcome measure was diarrhoea, defined as three loose stools in a 24 hour or 48 hour period.⇓ The type of probiotic tested, study populations, and effect sizes varied widely between studies, with both statistically significant2 and non-significant13 14 findings for the primary outcome and widely differing rates of antibiotic associated diarrhoea. Many of the trials identified in the systematic reviews were of poor quality.6 9 18 Reasons included poor allocation concealment, inadequate power, possible publication bias, variation in outcome measures, lack of intention to treat analyses, variation in follow-up duration, lack of cost-benefit data, variation in illness severity, and the small proportion of eligible patients enrolled. We found no head to head comparisons of probiotic strains.View this table: In this window In a new window Systematic reviews of randomised placebo controlled trials (RCTs) and subsequent individual trials of probiotics to prevent antibiotic associated diarrhoea