Aliment Pharmacol Ther 2005; 22: 365–372. doi: 10.1111/j.1365-2036.2005.02624.

Meta-analysis: non-pathogenic yeast Saccharomyces boulardii in

the prevention of antibiotic-associated diarrhoea
H. SZAJEWSKA* & J. M RUKOWICZ 
*Department of Pediatric Gastroenterology and Nutrition, The Medical University of Warsaw;  Polish Institute for Evidence-
based Medicine and Medycyna Praktyczna, Cracow, Poland
Accepted for publication 7 July 2005

Results: Of 16 potentially relevant clinical trials identified,

SUMMARY five randomized-controlled trials (1076 participants) met
Background: Antibiotic-associated diarrhoea occurs in the inclusion criteria for this systematic review. Treat-
up to 30% of patients who receive antibiotics but can be ment with S. boulardii compared with placebo reduced the
prevented with probiotics. risk of antibiotic-associated diarrhoea from 17.2% to
Aim: To systematically evaluate the effectiveness of 6.7% (RR: 0.43; 95% CI: 0.23–0.78; random effect
Saccharomyces boulardii in preventing antibiotic-associ- model). The number needed to treat to prevent one case of
ated diarrhoea in children and adults. antibiotic-associated diarrhoea was 10 (95% CI: 7–16).
Methods: Using medical subject headings and free- No side-effects were reported.
language terms, the following electronic databases were Conclusions: A meta-analysis of data from five random-
searched for studies relevant to antibiotic-associated ized-controlled trials showed that S. boulardii is moder-
diarrhoea and S. boulardii: MEDLINE, EMBASE, CINAHL ately effective in preventing antibiotic-associated
and The Cochrane Library. Additional sources were diarrhoea in children and adults treated with antibiotics
obtained from references in reviewed articles. Only for any reason (mainly respiratory tract infections).
randomized-controlled trials were considered for study For every 10 patients receiving daily S. boulardii with
inclusion. antibiotics, one fewer will develop antibiotic-associated
diarrhoea.

and up to 2 months after cessation of treatment.2–4 The

INTRODUCTION
incidence of diarrhoea in children receiving broad-
Antibiotic-associated diarrhoea (AAD) is defined as spectrum antibiotics ranges from 11% to 40%.5, 6 Almost
otherwise unexplained diarrhoea that occurs in associ- all antibiotics, particularly those active against anaer-
ation with the administration of antibiotics.1 Although obes, can cause diarrhoea, but the risk seems to be higher
no infectious agent is found in most cases, the bacterial with aminopenicillins, a combination of aminopenicillins
agent commonly associated with AAD, particularly in and clavulanate, cephalosporins and clindamycin.7, 8
most severe episodes (pseudomembranous colitis), is Preventive measures include the use of probiotics,
Clostridium difficile.2 AAD occurs in approximately which are live microbial food ingredients that are
5–30% of patients between the initiation of therapy beneficial to health.9 The most commonly used probi-
otics are lactic acid bacteria, such as lactobacilli or
Correspondence to: Dr H. Szajewska, Department of Paediatric Gastroen- bifidobacteria, but non-bacterial organisms, such as a
terology and Nutrition, The Medical University of Warsaw, 01-184
Warsaw, Dzialdowska 1, Poland. non-pathogenic yeast Saccharomyces boulardii, also have
E-mail: [email protected] been used. The rationale for the use of probiotics in AAD

2005 Blackwell Publishing Ltd 365

366 H. SZAJEWSKA & J. MRUKOWICZ

is based on the assumption that the use of antibiotics Types of outcome measures. The ‘primary’ outcome
leads to a disturbance in the normal intestinal micro- measure was the incidence of diarrhoea or AAD (as
flora and that this is a key factor in the pathogenesis of defined by the investigators). The ‘secondary’ outcome
AAD.10 In a randomized, double-blind, placebo- measures were as follows: the incidence of C. difficile
controlled study, to which one of us (H.S.) contributed, diarrhoea; mean frequency of bowel movements; mean
it was demonstrated that S. boulardii (250 mg po b.d.) is duration of diarrhoea; the need for discontinuation of
effective in preventing AAD in children 6 months to the antibiotic treatment, hospitalization to manage the
14 years of age treated with antibiotics for otitis media diarrhoea (in out-patients) or intravenous rehydration
and/or respiratory tract infections.11 in any of the study groups and adverse events.
Two systematic reviews of the use of different strains of
probiotics in the prevention of AAD have been pub-
Search strategy
lished.12, 13 A problem with both of these reviews is a
lack of power, which likely reflects more restrictive The following electronic databases were systematically
search criteria (e.g. language restriction). In addition, searched for relevant studies: MEDLINE (1966 – March
the literature search in these reviews was restricted to 2005), EMBASE (1980 – March 2005), Cumulative
only two databases, and authors did not report whether Index to Nursing and Allied Health (CINAHL, 1982 –
they reviewed references in retrieved articles. Results March 2005), The Cochrane Database of Systematic
from both systematic reviews suggest that probiotics, Reviews (Issue 1, 2005) and The Cochrane Controlled
including S. boulardii, have the potential to be useful in Trials Register (Issue 1, 2005). The search strategy
this situation but that further data are needed. Critics of included use of a validated filter for identifying
using a meta-analytical approach to assess the efficacy controlled trials,14 which was combined with a topic-
of probiotics argue that beneficial effects of probiotics specific strategy. The search terms were diarrhoea/
seem to be strain-specific, thus, pooling data on different diarrhoea, antibiotic-associated/antibiotic associated, C.
strains may result in misleading conclusions. difficile, probiotics and S. boulardii. Furthermore, refer-
Given these concerns the present review was under- ence lists from the original studies and review articles
taken to update data on the effectiveness and safety of identified were screened, and key experts in the field
only one probiotic micro-organism – S. boulardii – in the were approached for unpublished material. No limit was
prevention of AAD. imposed regarding the language of publication, but
certain publication types (i.e. letters to the editor,
abstracts, proceedings from scientific meetings) were
METHODS excluded.
Inclusion criteria
Methods of review
Electronic databases (see Search strategy) were system-
atically searched to identify studies appropriate for Included and excluded studies. Two reviewers independ-
inclusion in this systematic review. Inclusion criteria ently screened titles and abstracts identified according to
were as follows. the above-described search strategy. All potentially
relevant articles were retained and the full text of these
Types of studies. Randomized-controlled trials (RCTs) studies examined to determine which studies satisfied
that compared S. boulardii with placebo or no additional the inclusion criteria. The same reviewers independ-
intervention. ently carried out data extraction, using standard data
extraction forms. Studies reported in languages other
Types of participants. Adults and children who received than those familiar to the authors were translated.
antibiotics for any reason. Discrepancies between the reviewers were resolved by
discussion.
Types of interventions. Patients in the experimental
groups received S. boulardii as an adjunct to antibiotics. Study quality. Two reviewers independently, but with-
Patients in the control group received placebo or no out being blinded to the authors or journal, assessed the
additional intervention. quality of studies that met the inclusion criteria. Use of

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372

 META-ANALYSIS: S. BOULARDII IN THE PREVENTION OF DIARRHOEA 367

the following strategies associated with good quality (i) definition of diarrhoea; (ii) age of participants –
studies was assessed: (i) allocation concealment, (ii) adults vs. children; (iii) dose of S. boulardii and (iv)
blinding of investigators, participants, outcome asses- duration of follow-up.
sors and data analysts (yes/no/not reported), (iii) To test for publication bias, we used a test for
intention-to-treat (ITT) analysis (yes/no) and (iv) com- asymmetry of the funnel plot proposed by Egger
prehensive follow-up. et al.15 This test detects funnel plot asymmetry by
Allocation concealment was considered adequate determining whether the intercept deviates significantly
when the randomization method used did not allow from zero in a regression of the normalized effect
the investigator or the participant to identify or estimate (estimate divided by its standard error) against
influence the intervention group before enrolment of precision (reciprocal of the standard error of the
eligible participants in the study. However, the quality estimate) weighted by the reciprocal of the variance of
of the allocation concealment was considered unclear the estimate.
when randomization was used but no information
about the method was available and inadequate, when
RESULTS
inappropriate methods of randomization (e.g. alternate
medical record numbers, unsealed envelopes, tossing The meta-analyses reported here are presented
the coin) were used. according to the standards set out in the 1999
In regard to the ITT analysis, an answer of ‘yes’ meant Quality of Reporting of Meta-analyses (QUOROM)
that the authors had specifically reported undertaking statement.16
this type of analysis and/or that our own study
confirmed this finding. Conversely, a ‘no’ meant that
Description of studies
authors did not report use of ITT analysis and/or that
we could not confirm its use on study assessment. To The search yielded 16 citations. Five RCTs met the
evaluate the completeness of patient follow-up, we inclusion criteria for this systematic review11, 17–20 (see
determined the percentage of participants excluded or Table 1). The remaining 11 studies were excluded.21–31
lost to follow-up. Table 2 summarizes characteristics of the excluded
trials, including the reasons for exclusion.
The five selected studies recruited a total of 1076
Statistical methods
participants (564 in the experimental group and 512 in
The data were analysed using statsdirect software the control group). All studies were placebo controlled.
(2, 3, 8; 2004-04-17). The binary measure (prevalence There was considerable clinical heterogeneity among
of diarrhoea or AAD) for individual studies and pooled the trials in settings (in-patients or out-patients), age of
statistics is reported as the risk ratio (RR) between the participants (except one, all were performed in adults),
experimental and control groups with 95% confidence daily dose of the study product (200 mg–1 g) and the
intervals (CI). We calculated the number needed to treat type of antibiotics administered. There were also wide
(NNT) as the inverse of pooled absolute risk differences differences in the duration of follow-up, which varied
and 95% CI. The weighted mean difference between the from 211, 18 to 7 weeks20 after the cessation of antibi-
treatment and control groups was selected to represent otic treatment or was not specified.17, 19 Furthermore,
the difference in continuous outcomes. The weights there was variability in definitions of outcome meas-
given to each study are based on the inverse of the ures. The most commonly used definition of the
variance. diarrhoea was the presence of three or more loose (or
We used the Q test (chi-square statistics) with an a of watery) stools per 24 h, but criteria for its duration
0.1 to test heterogeneity among pooled estimates. varied from 24 h (two studies)17, 19 to at least 48 h
When there was statistically significant heterogeneity (three studies).11, 18, 20
in outcomes across studies, we conducted sensitivity The methodological quality of the trials also varied
analyses according to each of the four parameters of (Table 1). Allocation concealment was unclear in all of
trial methodological quality. A priori subgroup analysis the trials. Although all were double-blind studies, it
was planned based on factors that could potentially often was not stated who was blinded. Completeness of
influence the magnitude of the treatment effect: follow-up was unclear in one study19 and inadequate in

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372

 Table 1. Characteristics of the included RCTs
368

Number of pati-
ents with AAD
Allocation Age Dose Duration of Follow- Antibiotic Definition of
Trial concealment1 Blinding ITT2 FU3 N Participants (years) (per day) intervention up studied diarrhoea or AAD SB Placebo

Adam Unclear DB No No 388 Out-patients >15 200 For the duration No data b-Lactam ‡2 stools/day, 9/199 33/189
et al.17 mg of antibiotic TX antibiotics or liquid consistency
(minimum 5 days; tetracycline (oral) (as estimated
experimental group: from the table)
6.8 ± 0.24 days;
control group:
6.84 ± 0.25 days)
Surawicz Unclear DB No No 180 Hospitalized Adults 1g Within 48 h of 2 weeks (?) Various ‡3 loose or watery 11/116 14/64
et al.18 antibiotic initiation (penicillin, stools/day for
and 2 weeks after clindamycin, at least 2 days
(exact data not given) cephalosporins)
McFarland Unclear DB Yes Yes 193 Hospitalized 18–86 1g Within 72 h of 7 weeks b-lactam ‡3 loose stools/ 7/97 14/96
H. SZAJEWSKA & J. MRUKOWICZ

et al.20 antibiotic initiation antibiotics day for at least 2

and 3 days after the (oral or consecutive days;
antibiotic was intravenous) AAD – diarrhoea
discontinued associated with
(exact data not given) at least one
b-lactam antibiotic
with no other
aetiology of
diarrhoea identified

Lewis Unclear DB No Yes 72 Hospitalized >65 226 For the duration of No data Various ‡3 loose stools 7/33 5/36
et al.19 mg antibiotic treatment within a 24-h period
(exact data not given)
Kotowska Unclear DB Yes Yes 269 Hospitalized 6 months– 500 For the duration 2 weeks Various Diarrhoea: ‡3 loose or 4/119 22/127
et al.11 and 14 years mg of antibiotic treatment watery stools per day
out-patients (experimental for a minimum of 48 h
group: 7.8 ± 1 days; during and/or up to
control group: 2 weeks after the end
8.1 ± 1 days) of antibiotic treatment
AAD: as above, caused
by Clostridium difficile or
for otherwise unexplained
diarrhoea

1
Adequate: randomization method described that would not allow investigator/caregivers to identify or influence the intervention group before eligible participants entered the study; Unclear:
randomization stated but no information about method used was provided; Inadequate: use of an inappropriate method of randomization (e.g. alternate medical record numbers or unsealed envelopes)
and/or any information in the study indicating that investigators or participants could influence the intervention group.
2
Yes: specifically reported by authors that ITT analysis was undertaken and this was confirmed by our study assessment; Yes: not stated but confirmed by our study assessment; No: not reported and
lack of ITT analysis confirmed by our study assessment (patients who were randomized were not included in the analysis because they did not receive the study intervention, they withdrew from the
study or were not included because of protocol violation); No: stated but not confirmed by our study assessment.
3
Completeness of follow-up: trials with >80% follow-up of participants.
AAD, antibiotic-associated diarrhoea; DB, double blinding; SB, Saccharomyces boulardii; TX, treatment; RCT, randomized-controlled trial; ITT, intention-to-treat; FU, follow-up.

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372

 META-ANALYSIS: S. BOULARDII IN THE PREVENTION OF DIARRHOEA 369

Table 2. Characteristics of the excluded studies

Study Reason(s) for exclusion

Buts et al.30 Saccharomyces boulardii for Clostridium difficile-associated enteropathies in infants

Donat24 Non-randomized, controlled clinical trial
Elmer and McFarland25 RCT; designed to clarify the effectiveness of S. boulardii in recurrent C. difficile disease.
Not prevention study
Erdeve et al.26 Non-randomized, prospective controlled trial
Hotz21 Non-randomized, clinical trial
McFarland et al.29 RCT; explored the effect of S. boulardii in combination with standard antibiotics
for C. difficile disease. Not prevention study
Potts et al.23 RCT (abstract). Later reported as a full paper19
Schellenberg et al.31 Open trial. Treatment of C. difficile with S. cerevisiae
Surawicz et al.22 RCT; explored the effect of S. boulardii in the treatment of C. difficile colitis (not prevention study)
Surawicz et al.28 RCT; explored the effect of S. boulardii in the treatment of C. difficile disease

RCT, randomized-controlled trial.

another.18 ITT analysis was performed in only two S. boulardii in the prevention of C. difficile diarrhoea
trials.11, 20
Only one study11 (performed in children) evaluated the
effect of S. boulardii in the prevention of C. difficile
S. boulardii in the prevention of diarrhoea diarrhoea. The risk of documented C. difficile diarrhoea
was lower in the S. boulardii group compared with the
Treatment with S. boulardii compared with placebo
placebo group, but the difference was of borderline
reduced the risk of diarrhoea in patients treated with
statistical significance (RR: 0.3; 95% CI: 0.1–1.04).
antibiotics (as defined by the study investigators) from
17.2% to 6.7% (RR: 0.43; 95% CI: 0.23–0.78, random
effect model; Figure 1). For every 10 patients receiving Mean frequency of bowel movements and mean duration of
daily S. boulardii with antibiotics, one fewer would diarrhoea
develop diarrhoea (NNT: 10; 95% CI: 7–16).
These outcome measures were either not reported in the
The pooled effect size of three trials11, 18, 20 that used a
studies included in this systematic review11, 17, 19 or
more conservative definition of diarrhoea was 0.38
were reported in a manner that does not allow meta-
(95% CI: 0.22–0.63, random effect model; NNT: 10;
analysis.18, 20
95% CI: 7–18).

Relative risk meta-analysis plot (random effects)

Adam 0.26 (0.13. 0.52)

Surawicz 0.43 (0.21. 0.89)

McFarland 0.49 (0.21. 1.14)

Lewis 1.53 (0.56. 4.20)

Kotowska 0.19 (0.07. 0.52)

Figure 1. Plot of relative risk of antibiotic- Combined [random] 0.43 (0.23. 0.78)
associated diarrhoea in patients treated
with Saccharomyces boulardii compared with 0.01 0.1 0.2 0.5 1 2 5
placebo. Relative risk (95% confidence interval)

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372

370 H. SZAJEWSKA & J. MRUKOWICZ

Need for discontinuation of the antibiotic treatment, DISCUSSION

hospitalization to manage the diarrhoea (in out-patients)
Meta-analysis of the various studies assessing the
and/or intravenous rehydration in any of the study groups
effectiveness of S. boulardii in preventing AAD is difficult,
Only one trial11 (performed in children) addressed these as these studies are heterogeneous and the definition of
outcomes. There was no need for discontinuation of diarrhoea is not consistent. Consequently, this assess-
antibiotic treatment, hospital treatment because of ment is imperfect. However, consistent with previous
diarrhoea in the out-patients, or intravenous rehydra- findings, this meta-analysis of data from five RCTs
tion in any of the study groups. showed that S. boulardii is effective in preventing AAD in
adults and children treated with antibiotics for any
reason (mainly for respiratory tract infections). For
Adverse events
every 10 patients receiving daily S. boulardii with
The S. boulardii was well-tolerated, and no adverse antibiotics, one fewer will develop AAD. One charac-
events associated with this therapy were reported. teristic that makes our meta-analysis distinct from
previous reviews is that it focuses exclusively on only
one probiotic strain. Misuse or unsubstantiated use of
Homogeneity test
probiotics would be limited if more emphasis was given
The statistical test of homogeneity yielded a significant to analysis by probiotic agent, as in the case of our
result (v2 ¼ 10; P < 0.04). Significant heterogeneity meta-analysis on S. boulardii.
was attributable to the inclusion of elderly participants We made every effort to avoid publication bias,
in one of the trials (all participants were >65 years).19 including performing a comprehensive search using a
Exclusion of this trial by Lewis et al.19 resulted in a range of databases. We also reviewed references from
homogenous group of four studies involving 1007 retrieved articles. As only a limited number of experts in
patients (v2 ¼ 2.9; P ¼ 0.4). The significance of the the field were contacted for unpublished material,
pooled effect of S. boulardii remained when calculations publication bias cannot be fully ruled out. However,
were made with four homogenous trials only (RR: 0.33; based on the results of bias detection analysis (funnel
95% CI: 0.22–0.5, random effect model). plot and statistical test for its asymmetry), we believe
that the risk of not having identified important,
randomized, methodologically relevant trials is rather
Publication bias
low. Nevertheless, it is important to stress that the
A funnel plot (Figure 2) and Egger et al.’s15 regression power of the statistical methods that investigate evi-
asymmetry test (P ¼ 0.5 and 95% CI included 0) did not dence for funnel plot asymmetry is limited, particularly
show any publication bias or other small sample bias. for meta-analyses based on a small number of studies.
Thus, the results from our meta-analysis should be
Bias assessment plot viewed with caution. Similarly, given the small number
0.35 of studies, statistical conclusions on determinants of
heterogeneity might be flawed.
The mechanism by which S. boulardii, a non-patho-
0.40
genic yeast, exerts its action in preventing AAD is
Standard error

unclear. Possible mechanisms, which have been dem-

0.45 onstrated in animals, include the production of a
protease that inactivates a receptor for toxin A of
C. difficile, secretion of increased levels of secretory
0.50 immunoglobulin A (IgA) and IgA antitoxin A, and
competition for attachment sites.32–34 Saccharomyces
boulardii has also been shown to block C. difficile
0.55
-2 0 -1.5 -1.0 -0.5 0.0 0.5 adherence to cells in vitro.35
Log (relative risk) The results of this meta-analysis show that S. boulardii
significantly reduces the risk of diarrhoea in patients
Figure 2. Plot of publication bias.

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372

 META-ANALYSIS: S. BOULARDII IN THE PREVENTION OF DIARRHOEA 371

treated with antibiotics for various purposes (but mainly seems that the judicious use of antibiotics remains the
respiratory tract infections). However, they do not allow best method of preventing AAD.
conclusions about the efficacy of S. boulardii in prevent-
ing diarrhoea attributable to any single antibiotic class.
ACKNOWLEDGEMENT
Results from one recent RCT suggest that S. boulardii
effectively prevents diarrhoea caused by amoxicillin in This study was funded in part by a grant from The
combination with clavulanate as well as intravenously Medical University of Warsaw.
administered cefuroxime.11 Larger trials will be neces-
sary to further address these issues.
REFERENCES
The duration of follow-up after antibiotic treatment in
the trials included in this meta-analysis varied. As 1 Bartlett JG. Antibiotic-associated diarrhoea. N Engl J Med
diarrhoea may occur up to 2 months following cessa- 2002; 346: 334–9.
2 Bartlett JG, Chang TW, Gurwith M, Gorbach SL, Onderdonk
tion of such treatment, some cases of AAD may have
AB. Antibiotic-associated pseudomembranous colitis due to
been missed. In addition, only one of these trials toxin producing clostridia. N Engl J Med 1978; 298: 531–4.
(performed in children) assessed the prevalence of 3 Wiström J, Norrby SR, Myhre EB, et al. Frequency of
serious complications of AAD, such as the need to antibiotic-associated diarrhoea in 2462 antibiotic-treated
terminate antibiotic treatment or hospitalize the patient hospitalized patients: a prospective study. J Antimicrob
Chemother 2001; 47: 43–50.
for management of diarrhoea and/or intravenous
4 McFarland LV. Epidemiology, risk factors and treatments for
rehydration.11 These authors observed no such cases antibiotic-associated diarrhoea. Dig Dis 1998; 16: 292–307.
in children, thus, the baseline risk in this age group 5 Turck D, Bernet JP, Marx J, et al. Incidence and risk factors of
appears to be very low. Unfortunately, these important oral antibiotic-associated diarrhoea in an outpatient pediatric
outcomes have not been investigated in trials performed population. J Pediatr Gastroenterol Nutr 2003; 37: 22–6.
in adults. 6 Elstner CL, Lindsay AN, Book LS, Matsen JM. Lack of
relationship of Clostridium difficile to antibiotic-associated
Whereas no adverse effects were observed in any of the
diarrhoea in children. Pediatr Infect Dis 1983; 2: 364–6.
included trials, the administration of S. boulardii is not 7 Barbut F, Meynard JL, Guiguet M, et al. Clostridium difficile-
without risk. One caveat about S. boulardii is that it can associated diarrhoea in HIV infected patients: epidemiology and
cause fungemia.36–39 Most complications have occurred risk factors. J Acquir Immune Defic Syndr 1997; 16: 176–81.
in immunocompromised subjects or in patients with 8 McFarland LV, Surawicz CM, Stamm WE. Risk factors for
Clostridium difficile carriage and C. difficile-associated diarrhoea
other life-threatening illnesses managed in intensive
in a cohort of hospitalized patients. J Infect Dis 1990; 162:
care units. Although other probiotics (e.g. lactobacilli, 678–84.
bifidobacteria) so far used in clinical trials have 9 Diplock AT, Aggett PJ, Ashwell M, et al. Scientific concepts of
generally been described as safe and well-tolerated, it functional foods in Europe: consensus document. Br J Nutr
is not clear whether they could be used in the 1999; 81 (Suppl. 1): S1–S27.
prevention of AAD in immunocompromised patients; 10 Surawicz CM. Probiotics, antibiotic-associated diarrhoea and
Clostridium difficile diarrhoea in humans. Best Pract Res Clin
as there have been some concerns expressed, this
Gastroenterol 2003; 17: 775–83.
requires further evaluation. 11 Kotowska M, Albrecht P, Szajewska H. Saccharomyces boulardii
In conclusion, the results from this meta-analysis of in the prevention of antibiotic-associated diarrhoea in
five RCTs support the value of S. boulardii as adjunctive children: a randomized double-blind placebo-controlled trial.
treatment in patients on antibiotic therapy. In addition Aliment Pharmacol Ther 2005; 21: 583–90.
12 D’Souza AL, Rajkumar C, Cooke J, Bulpitt CJ. Probiotics in
to ease of administration, coadministration of S. boular-
prevention of antibiotic associated diarrhoea: meta-analysis.
dii with antibiotics has potential cost–benefits. However, Br Med J 2002; 324: 1361–4.
some issues need to be resolved before widespread use of 13 Cremonini F, di Caro S, Nista EC, et al. Meta-analysis: The
probiotics is advocated, e.g. by means of identifying effect of probiotic administration on antibiotic-associated
populations at high risk for AAD. There is a need for diarrhoea. Aliment Pharmacol Ther 2002; 16: 1461–7.
well designed RCTs to evaluate the efficacy of S. boulardii 14 Robinson KA, Dickersin K. Development of a highly sensitive
search strategy for the retreival of reports of controlled trials
in preventing AAD caused specifically by C. difficile or
using PubMed. Int J Epidemiol 2002; 31: 150–3.
those antibiotics that are most likely to cause diarrhoea, 15 Egger M, Smith GD, Schneider M, Minder C. Bias in meta-
as well as the effectiveness of S. boulardii in treating analysis detected by a simple, graphical test. Br Med J 1997;
AAD. Although available data are encouraging, it 315: 629–34.

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372

372 H. SZAJEWSKA & J. MRUKOWICZ

16 Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF, 28 Surawicz CM, McFarland LV, Greenberg RN, et al. The search
for the QUOROM Group. Improving the Quality of Reports of for a better treatment for recurrent Clostridium difficile disease:
Meta-analyses of randomised controlled trials: the QUOROM use of high-dose vancomycin combined with Saccharomyces
statement. Lancet 1999; 354: 1896–900. boulardii. Clin Inf Dis 2000; 31: 1012–7.
17 Adam J, Barret A, Barret-Bellet C. Essais cliniques contrôlés en 29 McFarland LV, Surawicz CM, Greenberg RN, et al. A
double insu de l’ultra-levure lyophilisée: étude multicentrique randomized placebo-controlled trial of Saccharomyces
par 25 médicins de 388 cas. Gaz Med Fr 1977; 84: 2072–8. boulardii in combination with standard antibiotics for
18 Surawicz CM, Elmer GW, Speelman P, McFarland LV, Chinn J, Clostridium difficile disease. JAMA 1994; 271: 1913–8.
van Belle G. Prevention of antibiotic-associated diarrhoea by Erratum in: JAMA 1994; 272: 518.
Saccharomyces boulardii: a prospective study. Gastroenterology 30 Buts JP, Corthier G, Delmee M. Saccharomyces boulardii
1989; 96: 981–8. for Clostridium difficile-associated enteropathies in infants.
19 Lewis SJ, Potts LF, Barry RE. The lack of therapeutic effect of J Pediatr Gastroenterol Nutr 1993; 16: 419–25.
Saccharomyces boulardii in the prevention of antibiotic-related 31 Schellenberg D, Bonningotn A, Champion CM, Lancaster R,
diarrhoea in elderly patients. J Infect 1998; 36: 171–4. Webb S, Main J. Treatment of Clostridium difficile diarrhoea
20 McFarland LV, Surawicz CM, Greenberg RN, et al. Prevention with brewer’s yeast. Lancet 1994; 343: 171–2.
of beta-lactam-associated diarrhoea by Saccharomyces boulardii 32 Pothoulakis C, Kelly CP, Joshi MA, et al. Saccharomyces
compared with placebo. Am J Gastroenterol 1995; 90: 439– boulardii inhibits Clostridium difficile toxin A binding and
48. enterotoxicity in rat ileum. Gastroenterology 1993; 104:
21 Hotz J. Prevention of antibiotic-associated diarrhoea by 1108–15.
Saccharomyces boulardii: a prospective study. Z Gastroenterol 33 Wilson KH, Perini I. Role of competition for nutrients in
1990; 28: 259–60 (German). suppression of Clostidium difficile by the colonic microflora.
22 Surawicz CM, McFarland LV, Elmer G, Chinn J. Treatment of Infect Immun 1988; 56: 2610–4.
recurrent Clostridium difficile colitis with vancomycin and 34 Qamar A, Aboudola A, Warny M, et al. Saccharomyces
Saccharomyces boulardii. Am J Gastroenterol 1989; 84: boulardii stimulates intestinal immunoglobulin A immune
1285–7. response to Clostridium difficile toxin A in mice. Infect Immun
23 Potts L, Lewis SJ, Barry R. Randomised double blind placebo 2001; 69: 2762–5.
controlled study of the ability of Saccharomyces boulardii to 35 Tasteyre A, Barc MC, Karjalainen T, et al. Inhibition of in vitro
prevent antibiotic related diarrhoea [abstract]. Gut 1996; 38 cell adherence of Clostridium difficile by Saccharomyces
(Suppl. 1): A61. boulardii. Microbiol Pathogens 2002; 32: 219–25.
24 Donat F. Prevention of intestinal side-effects of antibiotics with 36 Bassetti S, Frei R, Zimmerli W. Fungemia with Saccharomyces
perenterol. Munch Med Wochenschr 1970; 112: 74–7 cerevisiae after treatment with S. boulardii. Am J Med 1998;
(German). 105: 71–2.
25 Elmer GW, McFarland LV. Suppression by Saccharomyces 37 Rijnders BJA, Van Wijngaerden E, Verwaest C, Peetermans
boulardii of toxigenic Clostridium difficile overgrowth after WE. Saccharomyces fungemia complicating S. boulardii
vancomycin treatment in hamsters. Antimicrob Agents treatment in a non-immunocompromised host. Intensive
Chemother 1987; 31: 129–31. Care Med 2000; 26: 825.
26 Erdeve O, Tiras U, Dallar Y. The probiotic effect of 38 Zunic P, Lacotte J, Pegoix M, et al. S. boulardii fungemia.
Saccharomyces boulardii in a pediatric age group. J Trop Apropos of a case. Therapie 1991; 46: 498–9.
Pediatr 2004; 50: 234–6. 39 Cassone M, Serra P, Mondello F, et al. Outbreak of Saccharomyces
27 Schellenberg D, Bonington A, Champion CM, Lancaster R, cerevisiae subtype boulardii fungemia in patients neighboring
Webb S, Main J. Treatment of Clostridium difficile diarrhoea those treated with a probiotic preparation of the organism. J Clin
with brewer’s yeast. Lancet 1994; 343: 171–2. Microbiol 2003; 41: 5340–3.

2005 Blackwell Publishing Ltd, Aliment Pharmacol Ther 22, 365–372