Appendix D: Appraising Evidence for a Clinical Question about Probiotics for Antibiotic Associated Diarrhea

Transcription

1 Appendix D: Appraising Evidence for a Clinical Question about Probiotics for Antibiotic Associated Diarrhea Prepared by Miranda So, BSc, BScPhm, PharmD Clinical scenario: A physician on your adult general medicine unit wants to prescribe probiotics as routine prevention for antibiotic associated diarrhea, including diarrhea caused by Clostridium difficile, for all of his patients. He thinks that new information, published in the past few years, has become available and requests your opinion on the issue. Clinical question: What is the role of probiotics in preventing antibiotic associated diarrhea (including C. difficile diarrhea) in adult patients who have been admitted to hospital? Search results: A search of MEDLINE (refer to MEDLINE search example) yielded 15 articles, of which 1 was a meta analysis of randomized controlled trials (RCTs), 2 were RCTs, and the rest were review articles, editorials, or letters to the editor. Only the meta analysis and the 2 RCTs are evaluated here. The RCTs are summarized according to the patients intervention comparator outcome (PICO) system. Critical appraisal of the meta analysis follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta Analyses ) 1 checklist, published in See also [insert link] Appendix C ( EBM Workshop: Appraising Systematic Reviews ). An additional meta analysis was identified using keyword searching, and it is summarized as the last item in this document. Note: the symbol is used in the following tables should be interpreted as not applicable. Lönnermark E, Friman V, Lappas G, Sandberg T, Berggren A, Adlerberth I. Intake of Lactobacillus plantarum reduces certain gastrointestinal symptoms during treatment with antibiotics. J Clin Gastroenterol 2010;44(2): Patients Intervention Comparator Outcome Swedish hospital (single centre study) Inclusion criteria: Aged 16 or above, started on antibiotics no more than 48 h before entry into study. Expected duration of treatment 7 14 d. Must be able to keep diary and provide stool samples. Exclusion criteria: Chronic gastrointestinalgi disease, 200 ml PO once daily of a drink made from blueberries and 5% oats gruel containing L. plantarum 5 x 10 7 colonyforming units (CFU)/mL given within 48 h of antibiotic initiation, 200 ml PO once daily of the same drinks without L. plantarum (n = 83) Same prohibition Primary outcomes: (1) Proportion of patients in each group with development of diarrhea ( 3 loose or watery stools per day for 2 d): 7.5% in treatment group vs. 6% in placebo group; odds ratio (OR) 1.4, 95% confidence interval (CI) , p = (2) Proportion of patients in each group with 1

2 immumosuppression (because of treatment or disease), acute diarrheal disease, use of laxatives within the week before enrolment, hypersensitivity to ingredients in test or placebo drinks, heart valve disease or surgery, antibiotic treatment 2 wk before enrolment, or treatment with metronidazole or vancomycin on enrolment. Total n = 239 patients: 137 inpatients outpatients; 76 withdrew. Reasons for antibiotic therapy on enrolment: respiratory infection (40%), skin and soft tissue infection (32%), urinary tract infection (18%), septicaemia/meningitis (6%), and other diagnoses (4%). Is the study valid? Were patients randomly assigned to treatment groups? Yes continued until 1 wk after end of antibiotic treatment (n = 80) Consumption of foods or products containing live bacteria prohibited during study period of consumption of live bacteria from other sources positive test result for C. difficile toxin in feces after treatment with antibiotics: 3 patients in each group (no significant difference). Secondary outcome: Risk of experiencing loose or watery stools not meeting criteria of diarrhea significantly lower in treatment group: 57% vs. 71% (during antibiotic treatment); OR 0.69, 95% CI , p = Conclusion: L. plantarum could have a preventive effect on milder GI symptoms during treatment with antibiotics. Randomization by computer generated list performed by investigator not involved in enrolment of patients or analysis of data Was everyone blinded to treatment? Yes Described as double blind, but blinding process not specified Was the study controlled? Yes Placebo drink made of same ingredients except L. plantarum Were treatment and control patients similar at the beginning of study? Yes More women in treatment group; more patients in treatment group had positive results for C. difficile toxin on enrolment, but difference was not significant Were all patients accounted for? Yes Number of patients who withdrew stated, and reasons were described Were data analyzed according to the intention totreat No Patients who withdrew or dropped out not included in analysis principle? Were patients treated similarly during the study except for the study treatment? Yes Same prohibition on consuming foods that might contain live bacteria, but significantly more patients in treatment group were taking a proton pump inhibitor or H 2 receptor antagonist; treatment with antibiotics not significantly different How was the study funded? Partly funded by industry (company producing probiotic); several investigators owned stocks of company What were the study s results? What were the primary and secondary end points? What was the difference in outcomes between treatment and control groups? Primary end points: Proportion of patients with development of diarrhea by a priori definition; proportion of patients with positive test results for C. difficile toxin in feces Secondary end point: Proportion of patients with development of loose or watery stools that did not meet criteria for diarrhea Diarrhea: 6 patients (7.5%) in treatment group vs. 5 (6%) in placebo group; OR 1.4, 95% CI , p = 0.86 C. difficile toxin in feces: 3 patients in each group (no statistical data provided) Loose stools: 57% vs. 71% during treatment with antibiotics 2

3 Were the differences statistically significant? Clinically significant? Primary end points: proportion of patients developing diarrhea were not statistically significant between treatment and control groups; proportion of patients tested positive for C. difficile toxin were also not statistically significant between treatment and control groups. Secondary end point: Risk of loose stool significantly lower in treatment group Was absolute risk reduction or relative risk reduction reported? Was the number needed to treat calculated? If so, what was the value? Does this study matter to my patients? Does this study look at outcomes my patients care about? Were the patients in the study similar to my patients? Clinical significance: Although the risk of having loose stools was lower, whether the duration of loose stools was shortened was unclear. There was no quality of life assessment, and it is unknown how clinically significant the results were for participants. The incidence of diarrhea (as per definition) and C. difficile diarrhea were lower than what has been reported in the literature (most trials have reported diarrhea in 15% 20% of placebo treated patients and 25% 50% reduction in diarrhea with probiotic therapy relative to placebo 1 ). In this study, a lowerthan usual baseline incidence of diarrhea may have contributed to the nonsignificanct results. No No Yes Diarrhea, C. difficile associated diarrhea, and other GI adverse effects associated with antibiotic treatment are common concerns for patients. Yes Patients were treated with antibiotics for commonly encountered infections that require admission to hospital (and 55% of patients in the study were inpatients). Choices of antibiotics were representative of standard agents used for such infections. Do the benefits of treatment outweigh the risks No No adverse effects were recorded, but neither of the primary end points were statistically and costs? significant. How does this study change my practice? Should I change my practice? No Treatment with L. plantarum did not result in reduction in diarrhea (as per study definition) or C. difficile diarrhea. The risk of loose stools was reduced, but there was no clear indication of how the reduced loose stools risk affected participants, or whether duration of diarrhea was shorter than in the placebo group. Costs of L. plantarum were not reported. Many commercially available probiotics contain a mixture of Lactobacilli spp. The practicality of acquiring a single species of probiotic (i.e., L. plantarum) in Canada with the specific dose used in the study is unclear. Can I change my practice? How will I change my practice? 3

4 Reference 1. Katz JA. Probiotics for the prevention of antibiotic associated diarrhea and Clostridium difficile diarrhea. J Clin Gastroenterol 2006;40: Hickson M, D Souza AL, Muthu N, Rogers TR, Want S, Rajkumar C, et al. Use of probiotic Lactobacillus preparation to prevent diarrhoea associated with antibiotics: randomised double blind placebo controlled trial. BMJ 2007 doi: /bmj Patients Intervention Comparator Outcome Recruited from 3 hospitals in London, UK, mainly from medical, orthopedic, geriatric wards Inclusion criteria: Age over 50. Antibiotics prescribed (single or multiple; IV or PO). Able to eat and drink. Exclusion criteria: Diarrhea on admission or within preceding week, reported recurrent Probiotic yogourt drink (Actimel, Danone, France) containing L. casei DN (L. casei imunitass) 1.0 x 10 8 CFU/mL, Streptococcus thermophilus 1.0 x 10 8 Longlife, sterile milkshake (Yazoo, Campina, Netherlands) Dose = 100 g (97 ml) BID 0.5 h AC or 2 h PC n = 56 diarrhea, bowel disease, intake of high risk CFU/mL, and L. antibiotics (clindamycin, cephalosporins, bulgaricus 1.0 x 10 7 aminopenicilins), more than 2 course of CFU/mL started within antibiotics in past 4 weeks, severe lifethreatening illness, immunosupression, GI 48 h of antibiotic surgery, cardiac disease, regular probiotic before initiation admission, lactose intolerance. Dose = 100 g (97 ml) BID 0.5 h AC or 2 hr PC n = 57 Total n = 113: 1760 patients screened, 135 met study criteria, 22 lost to follow up. Primary outcome: Occurrence of diarrhea (>2 liquid stools a day for 3 d in quantities in excess of normal for each patient), as recorded by nurses and confirmed by investigators: 12% in probiotic group vs. 34% in placebo group, p = Absolute risk reduction (ARR) 22%, 95% CI 7% 37%; number needed to treat (NNT)= 5. Secondary outcome: Occurrence of C. difficile infection (episode of diarrhea combined with detection of toxins A or B or both in a stool sample) (data from 109 patients): 0% in probiotic group vs. 17% in placebo group, p = 0.001; ARR 17%, 95% CI 7% 27%, NNT = 6. Conclusions: Probiotic drink in this study reduced the incidence of antibiotic associated diarrhea and C. difficile diarrhea. The authors also stated this has potential to decrease morbidity, healthcare costs. and mortality if used routinely in patients over age 50. Is the study valid? Were patients randomly assigned to treatment Yes By an independent statistician groups? Was everyone blinded to treatment? Yes Patients, researchers, nursing staff, but not pharmacy staff. Pharmacy staff replaced commercial labels on original bottles of both treatment and placebo yoghurt with a generic label. Nursing staff poured out 100 ml of either treatment or placebo yoghurt from the bottles for administration to patients. Was the study controlled? Yes Placebo was 100 ml of a sterile drink, described as identical in colour and consistency with the 4

5 treatment formulation Were treatment and control patients similar at the Yes Clinical characteristics of patients comparable at baseline, but statistical analyses not reported beginning of study? Were all patients accounted for? Yes 12 in treatment group and 10 in placebo group lost to follow up for various reasons Were data analyzed according to the intention totreat principle? No Although specified as intention to treat analysis, patients assigned to groups but lost to follow up were not included because of a lack of data on incidence of diarrhea Were patients treated similarly during the study except for the study treatment? Yes Antibiotic treatments were assigned risk levels (low, medium, high) on the basis of their tendency to cause diarrhea. Distribution of antibiotic treatment among the 3 risk levels appeared to be similar between probiotic and placebo patients, but numerically more patients in control group were given high risk antibiotics (aminopenicillin, cephalosporins), whereas more patients in the probiotic group were given medium risk antibiotics (tetracyclines, sulphonamides, quinolones, macrolides). No statistical analyses were provided in describing these differences. How was the study funded? What were the study s results? What were the primary and secondary end points? What was the difference in outcomes between treatment and control groups? Were the differences statistically significant? Clinically significant? Was absolute risk reduction or relative risk reduction reported? Was the number needed to treat calculated? If so, what was the value? Does this study matter to my patients? Does this study look at outcomes my patients care about? Were the patients in the study similar to my patients? Do the benefits of treatment outweigh the risks and costs? How does this study change my practice? Use of high risk antibiotics appeared to violate exclusion criterion of intake of high risk antibiotics (clindamycin, cephalosporins, aminopenicillins), but no explanation was provided. Partly by a manufacturer of commercially available yoghurt; investigators received funding from this manufacturer, and one of them was a member of the company s advisory group Primary end point: Incidence of antibiotic associated diarrhea Secondary end point: Incidence of C. difficile diarrhea Primary end point: 12% in probiotic group vs. 34% in placebo group, p = Secondary end point: 0% in probiotic group vs. 17% in placebo group, p = Yes Statistically and clinically significant for both end points Yes Primary end point: ARR 22%, 95% CI 7% 37% Secondary end point: ARR 17%, 95% CI 7% 27% Yes Primary end point: 5 Secondary end point: 6 Yes Y/N No Antibiotic associated and C. difficile diarrhea are potential adverse effects that could result in morbidity and even mortality in hospital inpatients. In terms of age and reasons for antibiotic treatment, yes (commonly seen on a general medicine unit); in terms of exclusion criteria, no. 5

6 Should I change my practice? No Only 7% of screened patients (potential target population) participated in the study. As such, generalization of probiotics to routine use is not justified 2 (see Conclusions, above). Importantly, healthcare costs and mortality were not prespecified outcome measures, and therefore it cannot be concluded that these parameters would be positively affected. Exclusion criteria were highly selective (e.g., patients taking high risk antibioitics, the very people in whom it is particularly important to prevent such diarrhea, were excluded), which further limits applicability, 3 yet this critierion appears to have been violated without explanation (see also comments above). Can I change my practice? How will I change my practice? References 2. Wilcox MH, Sandoe JA. Probiotics and diarrhea: data are not widely applicable [letter]. BMJ 2007;335: Billyard T. Probiotics and diarrhea: no high risk antibiotics? [letter]. BMJ 2007;335:171. McFarland LV. Meta analysis of probiotics for the prevention of antibiotic associated diarrhea and the treatment of Clostridium difficile disease. Am J Gastroenterol 2006;101: *Only the results pertaining to prevention of antibiotic associated diarrhea are presented below (refer to clinical scenario and clinical question above). Patient populations and study design Sources: PubMed, MEDLINE, Google Scholar ; no language restrictions Inclusion criteria: RCTs in humans published in peer reviewed journals Exclusion: Preclinical studies, case reports/series, phase I safety studies, reviews, duplicate reports, trials of unspecified probiotics, prebiotics, inconsistent outcomes Intervention Comparator Outcome Single strains (Saccharomyces boulardii, Lactobacillus rhamnosus GG, Bacillus clausii, Bifidobacterium longum, Clostridium butyricum miyairir, Lactobacillus acidophilus, Enterococus faecium SF 68), mixtures of 2 types of probiotics, or single probiotic in combination with a prebiotic Types of controls used in the studies and their method of delivery or presentation not specified in the metaanalysis Primary outcome (antibiotic associated diarrhea [ 3 loose stools/d for 2 d or 5 loose stools/48 h] within 2 mo of antibiotic exposure): Adults (16 RCTs): 7 studies (44%) showed significant efficacy Pediatric patients (9 RCTs): 6 studies (67%) showed significant efficacy Significant heterogeneity among studies (χ 2 for heterogeneity 82.5, p < 0.001) Overall relative risk of antibiotic associated diarrhea was 0.43, 95% CI , p < 0.001, based on random effects model, favouring probiotic Funnel plot indicated modest publication bias, but Begg and Mazumdar rank correlation test did not show significant evidence of publication bias Subgroup analyses by probiotic type: S. boulardii (6 RCTs): relative risk of antibiotic associated diarrhea 0.37, 95% CI 6

7 For indication of antibiotic associated diarrhea prophylaxis: 104 publications identified, but only 25 RCTs included in analysis: 16 in adults and 9 in pediatric patients; total n = 2810 Daily doses from 1 x 10 7 to 1 x CFU (mean daily dose 3 x 10 9 CFU) Duration of treatment: 5 d to 8 wk (median 2 wk) , p < , based on fixed effects model L. rhamnosus GG (6 RCTs): relative risk 0.31, 95% CI , p = 0.006, based on random effects model mixtures of Lactobacillus spp. and Bifidobacterium spp. (7 RCTs): relative risk 0.51, 95% CI , p < , based on fixed effects model Adverse effects: No bacteremia, fungemia, or other serious adverse events reported As this was a meta analysis, the PRISMA checklist was used to critically appraise the study. Section/topic # Checklist item Reported on page # / Comments Title 1 Identify the report as a systematic review, meta analysis, or both. Yes (p. 812) Structured summary TITLE ABSTRACT 2 Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number. INTRODUCTION Yes (p. 812) Rationale 3 Describe the rationale for the review in the context of what is already known. Yes (p. 813: Previous meta analyses in 2002 included only 11 trials and did not discuss possible reasons for conflicting findings, and also no detailed study information. A recent Cochrane review for C. difficile diarrhea did not include probiotics. ) Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, outcomes, and study design (PICOS). Yes (p. 813: to assess efficacy and safety of probiotics for prevention of antibioticassociated diarrhea and treatment of C. difficile diarrhea ); PICOS not specified in the objectives 7

8 Section/topic # Checklist item Reported on page # / Comments Protocol and registration Eligibility criteria Information sources METHODS 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide registration information including registration number. 6 Specify study characteristics (e.g., PICOS, length of follow up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale. 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched. Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated. Study selection Data collection process 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta analysis). 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators. Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made. No; not described in the text Yes (p. 813; see also inclusion and exclusion criteria described in table summarizing study results, above), but finer study characteristics such as duration of follow up and PICOS requirements were not specified in study selection criteria Yes (p. 813; databases, years searched, secondary and hand searches, specific search terms, and language restrictions described), but date of last search not provided; study authors contacted for data not reported in original publication (under Data Extraction ) No Yes (p. 814: Quality of Reporting of Meta Analyses (QUOROM) flow diagram describing identification of studies and reasons for exclusion, e.g., review articles or lacking controls) Yes (p. 814: parameters extracted from publications and use of a standardized table are described; contact with original authors stated) Yes (p. 813: PICOS stated, but not funding sources; specifications such as handling of trials with multiple probiotics described; statement included that author of meta analysis was sole searcher of trials, with search results verified 8

9 Section/topic # Checklist item Reported on page # / Comments Risk of bias in individual studies Summary measures Synthesis of results Risk of bias across studies Additional analyses Study selection Study characteristics 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis. using historic searches [under Data Extraction ]) Yes (p. 813: quality assessment done at study level using a scale published by US Preventive Services Task Force, based on randomization, study design, sample size, generalizability, study biases, outcome assessment; study quality not integrated with weighting, which was based only on sample size) 13 State the principal summary measures (e.g., risk ratio, difference in means). Yes (p : relative risk with 95% confidence interval to be used to assess efficacy; other statistical analyses also described) 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I 2 ) for each meta analysis. 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies). 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, metaregression), if done, indicating which were pre specified. RESULTS 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram. 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow up period) and provide the citations. Yes (p. 814: tests of heterogeneity and use of random effects vs. fixed effects model described) Yes (p. 814: use of funnel plots and adjusted rank correlation test described, along with Begg and Mazumdar method to assess publication bias) Yes (p. 814 subgroup analysis based on probiotic type, dose, and indication specified a priori) Yes (p. 814: study selection process and rationale for exclusion described in flow diagram and in text) Yes (p. 815: 25 trials with characteristics described in Table 1, along with citations) Risk of bias 19 Present data on risk of bias of each study and, if available, any outcome level Yes (p. 815: ranking of quality with US Preventive Services Task Force scale [ranging 9

10 Section/topic # Checklist item Reported on page # / Comments within studies assessment (see item 12). from 1 to 3, for poor to good ] included in Table 1; the 25 RCTs included for antibioticassociated diarrhea scored 2 or 3) Results of individual studies Synthesis of results Risk of bias across studies Additional analysis Summary of evidence 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group; (b) effect estimates and confidence intervals, ideally with a forest plot. 21 Present results of each meta analysis done, including confidence intervals and measures of consistency. Yes (p. 815: efficacy data for each trial provided in Table 1; forest plot provided in Figure 2 on p. 818; effect estimate and confidence interval stated in text on p. 816) Yes (p. 816: see comments above) 22 Present results of any assessment of risk of bias across studies (see Item 15). Yes (p. 819: risk of publication bias [ modest presence ] shown in Figure 3 funnel plot; Begg and Mazumdar rank correlation test indicated no bias) 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta regression [see Item 16]). DISCUSSION 24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers). Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review level (e.g., incomplete retrieval of identified research, reporting bias). Yes (p. 816: subgroup analyses based on type of probiotic described in Table 2) Yes (p : estimates of effects for each outcome stated; comparisons with previous meta analyses or pooled analyses to validate present findings reported; relevance to healthcare provider described) Yes (p : small sample sizes, heterogeneity mentioned; acknowledged: limitation of combining adult and pediatric studies in antibiotic associated diarrhea analyses with justification; different types of probiotic; different doses of probiotic; publication bias noted; safety data with prolonged use lacking) Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and Yes (p. 820: The present meta analyses suggest that probiotics can significantly reduce the 10

11 Section/topic # Checklist item Reported on page # / Comments implications for future research. FUNDING incidence of AAD [antibiotic associated diarrhea]... Future studies should expand the types of probiotics tested and pay careful attention to proper study design and sample size considerations ) Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review. Should I change my practice? Not described in text This meta analysis by and large followed the reporting transparency guidelines in the PRISMA checklist, and its results are convincing. Nonetheless, it had significant flaws in methodology that affect the validity of the results: (1) Combining results from studies of different populations (pediatric and adult) assumed that the efficacy and magnitude of effect for the probiotic (i.e., protective effect) were independent of patients age, which may not be correct. 4 A systematic review might have been more appropriate. (2) Probiotics vary in efficacy, and it may be inappropriate to combine them in estimating effects, although this potential limitation was overcome by the subgroup analyses. (3) Five of the 25 RCTs examining antibiotic associated diarrhea included patients receiving treatment for Helicobacter pylori, who differed from hospital inpatients receiving broad spectrum antibiotics. 5 (4) Extraction of data by a single investigator meant that there was no validation of the assessment of study quality. 5 (5) In 5 studies, the definition of diarrhea differed from that of the a priori definition for inclusion provided by the author of the meta analysis, and they should therefore have been excluded. 5 However, in response to this criticism, the author has stated that excluding those studies did not significantly affect the pooled estimate of risk. 6 Although the overall effects of probiotics appear to be protective, the appropriate delivery method, appropriate dose, and appropriate patient population all remain unclear from this meta analysis. References: 4. Dendukuri N, Brophy J. Inappropriate use of meta analysis to estimate efficacy of probiotics [letter]. Am J Gastroenterol 2007;102: Lewis S. Response to the article: McFarland LV. Meta analysis of probiotics for the prevention of antibiotic associated diarrhea and the treatment of Clostridium difficile disease [letter]. Am J Gastroenterol 2007;102: McFarland LV. Response to Drs. Ramnarace and Dendukuri [Letters to the Editor] Am J Gastroenterol 2007;102: Kale Pradhan PB, Jassal HK, Wilhelm SM. Role of Lactobacillus in the prevention of antibiotic associated diarrhea: a meta analysis. Pharmacotherapy 2010;30: Patient populations and study design Intervention Comparator Outcome 11

12 Sources: MEDLINE, EMBASE, Cochrane registry from inception to May 2008; limited to English and humans Inclusion: RCTs that reported occurrence of antibiotic associated diarrhea in patients receiving singleagent Lactobacillus therapy or a placebo; adults or pediatric patients Excluded: non Lactobacillus therapy or Lactobacillus in combination with another probioticor trial not meeting all inclusion criteria 238 publications screened, and 10 RCTs included in analysis: 6 in adults and 9 in pediatric patients; total n = 1862 Single agent Lactobacillus spp. in form of sachets, capsules, fermented milk Duration of probiotic treatment ranged from 3 d, to throughout the entire duration of antibiotic treatment, to 30 d Dosage: 2 x 10 9 to 4 x CFU/d Frequency: OD to QID Follow up period: 2 d to 3 mo Types of controls used in the studies and their method of delivery or presentation not specified in the meta analysis Primary outcome: Occurrence of antibiotic associated diarrhea (not defined): risk ratio(rr) 0.35, 95% CI Subgroup analyses: Adults: RR 0.24, 95% CI (significant) Pediatric patients: RR 0.44, 95% CI (not significant) Adverse effects: No significant difference between treatment and placebo Publication bias present: Funnel plot asymmetric, with one clear outlier Statistical heterogeneity present: I 2 = 79% (greater than 50% requires use of random effects model in estimating overall effects) Conclusion: Prophylactic single agent Lactobacillus therapy given during antibiotic treatment reduces risk of antibiotic associated diarrhea relative to placebo; therapy well tolerated Section/topic # Checklist item Reported on page # / Comments Title 1 Identify the report as a systematic review, meta analysis, or both. Yes (p. 119) Structured summary TITLE ABSTRACT 2 Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number. Yes (p. 119) INTRODUCTION Rationale 3 Describe the rationale for the review in the context of what is already known. Yes (p : incidence of antibioticassociated diarrhea, pathophysiology, 12

13 Section/topic # Checklist item Reported on page # / Comments Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, outcomes, and study design (PICOS). Protocol and registration Eligibility criteria Information sources METHODS 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide registration information including registration number. 6 Specify study characteristics (e.g., PICOS, length of follow up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale. 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched. Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated. Study selection Data collection process 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta analysis). 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators. Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made. complications, risk factors and associated high healthcare costs, mechanism of action explaining protective effects of probiotics) Yes (p. 120: evaluation of efficacy of singleagent Lactobacillus regimen in prevention of antibiotic associated diarrhea) No Yes (p. 121: randomization to either singleagent Lactobacillus therapy or placebo; databases and timeframe specified; language limitation [English only]), but publication status not specified Yes (p. 121: database with dates of coverage), but not stated if there was contact with original study investigators; also secondary searches of reference lists in review articles not mentioned No, but limits applied were stated Yes (p. 121: QUORUM flow diagram [Figure 1]) Yes (p. 121: two independent reviewers extracted data from all eligible studies with a standardized form; consensus reached in cases of discrepancy); did not state whether original investigators were contacted Yes (p. 122), but funding source not stated 13

14 Section/topic # Checklist item Reported on page # / Comments Risk of bias in individual studies Summary measures Synthesis of results Risk of bias across studies Additional analyses 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis. Yes (p.121: assessment of quality of studies with Jadad scoring system [validated and published] at study level described), but cutoff of quality assessment score for exclusion not specified; however text states (p. 122) that included studies scored 2 5 (range 0 5, where 0 = poorest quality) 13 State the principal summary measures (e.g., risk ratio, difference in means). Yes (p. 121: RR with 95% CI) 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I 2 ) for each meta analysis. 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies). 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, metaregression), if done, indicating which were pre specified. Yes (p. 121: use of I 2 test to assess heterogeneity) Yes (p.121: funnel plot to assess publication bias) Yes (p : by age [over or under 18), but subgroup analysis not prespecified Study selection Study characteristics Risk of bias within studies Results of individual studies RESULTS 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram. 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow up period) and provide the citations. 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12). 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group; (b) effect estimates and confidence intervals, ideally with a forest plot. Yes (p : Figure 1; reasons for exclusion also stated in text) Yes (p. 122, Table 1), but follow up period not provided; Jadad score not provided for individual studies, except for statement in text that quality scores of included studies ranged from 2 to 5 Yes (p. 122), but description not extensive, except for range of Jadad score, from 2 to 5 (see above) Yes (p. 123, Figure 2) 14

15 Section/topic # Checklist item Reported on page # / Comments Synthesis of results Risk of bias across studies Additional analysis Summary of evidence 21 Present results of each meta analysis done, including confidence intervals and measures of consistency. Yes (p : text and forest plot for each age group) 22 Present results of any assessment of risk of bias across studies (see Item 15). Yes (p. 122: I 2 statistic for heterogeneity stated; p Figure 3 funnel plot to assess publication bias) 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta regression [see Item 16]). DISCUSSION 24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers). Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review level (e.g., incomplete retrieval of identified research, reporting bias). Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications for future research. FUNDING Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review. Should I change my practice? Overall this was a persuasive, well done meta analysis, with a focused clinical question regarding probiotic type (single agent Lactobacillus) and indication (prevention of antibiotic associated diarrhea). It may have been inappropriate to combine studies of adults and children in the overall estimates, but subgroup analyses based on age were conducted to address this limitation (subgroup analyses were not defined a priori). Proposed explanation of difference in benefits between adult and pediatric populations was given. For example, the paediatric studies included a broad range of age (2 weeks to 14 years), but those under 6 might be at greater risk of developing antibiotic associated diarrhea. Also, paediatric patients may have other risk factors for developing diarrhea, such as rotavirus. Limitations: small sample size (as small as 38 in one pediatric study and as small as 60 in one adult study); combination of outpatients and inpatients (number in each group not clearly 15 No Yes (p ), but did not specify relevance to key groups Yes (p : small sample size of individual studies; heterogeneity [including adults vs. children; dosage forms and doses, with only 2 trials using recommended dose of >10 10 CFU/d; duration of treatment]; publication bias) Yes (p.125: single agent Lactobacillus therapy effective as prophylaxis for antibiotic associated diarrhea, but evaluation for dosage forms and regimens warranted) No

16 Section/topic # Checklist item Reported on page # / Comments specified); no specification of antimicrobial agents, duration of antibiotic therapy, and indications (types of infections being treated). All of these limitations affect the generalizability of the results. There was a discrepancy between the text and Table 1 regarding duration of treatment with probiotic (maximum 30 d in Table 1 and in text on p. 124, but 14 d in text on p. 122). Overall Summary of the Evidence for this Clinical Question: The data presented in these 4 studies are conflicting, and key questions regarding dosage forms, delivery methods, regimens, and duration of prophylaxis remain unanswered. In terms of type of probiotic, the 2 meta analyses summarized above indicate that Lactobacillus and Saccharomyces appear to be the most promising. In terms of risks, none of the meta analyses or RCTs reported serious adverse events; however, complications such as S. boulardii septicemia and Lactobacillus endocarditis have been reported in the literature. Concerns about translocation of probiotic organisms from the intestine to other organs have also been raised (McFarland. Am J Gastroenterol 2006;101: ). In particular, caution should be exercised for severely ill patients who receive nutrition or antibiotics through a potentially open portal (catheter or nasogastric tube) (McFarland. Am J Gastroenterol 2006;101: ). In conclusion, on the basis of current data, routine use of probiotics as prophylaxis for antibiotic associated diarrhea, including C. difficile diarrhea, may not be warranted, because of the important questions that remain unanswered. LITERATURE CITED 1. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta analyses: the PRISMA statement. BMJ 2009;339:b2535. Available from: 16