Deciphering meta-analytic results: a mini-review of probiotics for the prevention of paediatric antibiotic-associated diarrhoea and Clostridium difficile infections

Abstract

Meta-analyses are used to evaluate pooled effects of a wide variety of investigational agents, but the interpretation of the results into clinical practices may be difficult. This mini-review offers a three-step process to enable healthcare providers to decipher pooled meta-analysis estimates into results that are useful for therapeutic decisions. As an example of how meta-analyses should be interpreted, a recent meta-analysis of probiotics for the prevention of paediatric antibiotic-associated diarrhoea (AAD) and the prevention of Clostridium difficile infections (CDI) will be used. First, the pooled results of this meta-analysis indicates a significant protective efficacy for AAD is found when the 16 different types of probiotics are combined (pooled relative risk (RR) = 0.43, 95% confidence interval (CI)=0.33-0.56) and also a significant reduction of paediatric CDI (pooled RR=0.34, 95%CI=0.16-0.74) was found pooling four different types of probiotics. Secondly, because the efficacy of probiotics is strain-specific, it is necessary to do a sensitivity analysis, restricting the meta-analysis to one specific strain. Two strains, Saccharomyces boulardii lyo and Lactobacillus rhamnosus GG showed significant efficacy for paediatric AAD when pooled (pooled RR for S. boulardii = 0.43, 95%CI=0.21-0.86 and pooled RR for L. rhamnosus GG = 0.44, 95%CI=0.20-0.95). Thirdly, if studies within probiotic types have different results, it is prudent to examine these studies individually to determine the reasons why non-significant differences in efficacy were found. By drilling down through these three analytic layers, physicians will be confident in recommending the correct probiotic strain to their patients.

Keywords: Clostridium difficile; antibiotic-associated diarrhoea; meta-analysis; probiotics; randomised clinical trials; safety.