[Link to free full-text BMJ article PDF | BMJ extract | PubMed® abstract]
Christopher C Butler, professor of primary care medicine,
Donna Duncan, senior project manager,
Kerenza Hood, professor of statistics
Correspondence to: C C Butler firstname.lastname@example.org
Probiotics are thought to combat antibiotic associated diarrhoea through restoring resistance to colonisation by pathogenic bacteria after the normal colonic microflora have been damaged by antibiotics, by breaking down non-absorbable compounds into absorbable products, by interfering with pathogenic toxins, and by enhancing immunity.What is the evidence of uncertainty?
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 significant and non-significant 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.
Systematic reviews of randomised placebo controlled trials (RCTs) and subsequent individual trials of probiotics to prevent antibiotic associated diarrhoea
|Number of studies and/or total number of participants, care setting|
Intervention: organism in probiotic and daily dose (colony forming units)
|Outcome (risk ratio) for antibiotic associated diarrhoea|
|McFarland 2010 (1976-2009)||10 RCTs, 1858 adults, 4 trials in hospitalised patients, 1 outpatient, 3 in patients receiving antibiotic treatment for H pylori infection||S boulardii, ranging from 4×109 to 2×1010||0.47 (95% CI 0.35 to 0.63)|
|Avadhani 2010 (unclear)||8 RCTs, 1220 adults, inpatients||3 trials of S boulardii, 1 of L rhamnosus, 4 of mixed strains that included L casei, L acidophilus, L bulgaricus, S thermophilus, B bifidum, and L rhamnosus, range of doses||0.56 (95% CI 0.44 to 0.71)|
|McFarland 2006 (1977-2005)||25 RCTs, 2810 children and adults, inpatients and outpatients including H pylori treatment||6 trials S boulardii, 6 trials L rhamnosus, 6 other single strains, and 7 mixed strains, ranging from 1×107 to 1×1011(mean dose 3×109)||Combined 0.43 (95% CI 0.31 to 0.58);|
S boulardii 0.37 (95% CI 0.26 to 0.52);
L rhamnosus 0.31 (95% CI 0.13 to 0.72)
|Kale-Pradham 2010 (inception -May 2008)||10 RCTs, 1862 children and adults, inpatients and outpatients||Single agent lactobacillus, ranging from 2×10 to 4×1010||Combined 0.35 (95% CI 0.19 to 0.67);|
adults 0.24 (95% CI 0.08 to 0.75);
children 0.44 (95% CI 0.18 to 1.08)
|Sazawal 2006 (inception -February 2006)||19 RCTs, children and adults, inpatients and outpatients||Single (5 L rhamnosus) and mixed, ranging from 1×710 to 1×1010||0.48 (95% C1 0.35 to 0.65)|
|Szajewska 2006 (1966-December 2005)||6 RCTs, 766 children, inpatients and outpatients||2 RCTs lactobacillus GG, one S boulardii, 3 mixed, dose range unclear||0.44 (95% CI 0.25 to 0.77)|
|Johnston 2007 (inception to August 2006)||9 RCTs, 1946 children, inpatients and outpatients||6 single, 3 mixed containing (alone or in combination) Lactobacillus spp, Bifidobacterium spp, Streptococcus spp, S boulardii, ranging from 8.25×106 to 4×1010||Per-protocol analysis 0.49 (95% CI 0.32 to 0.74); intention to treat analysis 0.90 (95% CI 0.50 to 1.63); 5 studies of higher dose (5 to 40 × 109day) 0.35 (95% CI 0.25 to 47); 3 studies of low dose (<5 sup="sup">95>|
744 of 1120 (66.5%) eligible participants were not recruited
Low dose 0.64 (95% CI 0.42 to 0.97); 15.5% low dose, 28.2% high dose intervention, and 44.1% placebo treated patients developed diarrhoea
78 children aged 5 months to 16 years with respiratory tract infections, inpatients and outpatients
|B longum, L rhamnosus, and L plantarum, twice daily at 108||0.50 (95% CI 0.06 to 3.50); 2.5% intervention and 5.3% placebo treated children developed diarrhoea|
Six placebo controlled trials are in progress examining the effect of probiotics in preventing antibiotic associated diarrhoea in hospitalised patients. These studies will provide information on probiotics to prevent antibiotic associated diarrhoea in a wider range of hospitalised patients and may be large enough to provide information on which subgroups of patients are at greatest risk and are most likely to benefit.
There is an absence or insufficiency of high quality evidence to support routine use of probiotics to prevent antibiotic associated diarrhoea in all people, regardless of age, comorbidity, and care setting. For example, few trials have been done in primary care, and we found none from intermediate and social care settings. We found no pragmatic, open implementation studies.
What should we do in the light of uncertainty?
Good evidence exists to support using probiotics with S boulardii and Lactococcus rhamnosus GG (ATCC 53103) to prevent antibiotic associated diarrhoea, with emerging evidence for certain mixed strains that include L casei or L acidophilus. Probiotics also seem to be more effective at higher doses. However, because insufficient evidence exists to support routinely using probiotics for this purpose, and because of the low incidence and generally mild severity of antibiotic associated diarrhoea in otherwise healthy people, we recommend against routine use of probiotics in all people taking antibiotics to prevent antibiotic associated diarrhoea. Nevertheless, probiotics are cheap and safe, so routine use with antibiotics is justified in frail patients in hospital and possibly in children. Those who have previously had antibiotic associated diarrhoea should be offered probiotics when they are treated with antibiotics, regardless of setting, but probiotics should be avoided in people who are seriously immunocompromised. As probiotics seem more effective at higher doses, doses of at least 50 billion colony forming units should be used; probiotics should be taken for the duration of antibiotic treatment and continued for a week thereafter.
Evidence about the effectiveness of many strains is absent or insufficient. Head to head studies of probiotic strains are needed, as well as more studies to identify groups of patients at greatest risk and most likely to benefit, especially in the community and in intermediate care.
[Link to free BMJ article PDF for full text, images, and references]
© 2012 BMJ Publishing Group Ltd