Criteria for considering studies for this review
Types of studies
Parallel-group controlled trials (randomised or quasi-randomised) which have compared anti-H.pylori therapy with or without placebo versus anti-H.pylori therapy with probiotic supplementation, and have reported rates of successful eradication any time after completion of therapy. We will include only those studies available as full papers, irrespective of language, or blinding. In cross-over trials, we will include data from the first time period only.
Types of participants
H.pylori-infected persons (adults or children), irrespective of presence or absence of symptoms, receiving anti-H.pylori therapy. We will include studies irrespective of timing of probiotic administration (before, simultaneously, after, or any combination of these) in relation to antibiotics.
Types of interventions
The intervention will be anti-H.pylori therapy, defined as a combination of one proton pump inhibitor medication plus at least two antibiotics, which are supplemented either with a probiotic (treatment group) or a placebo/no intervention (control group). We will include studies irrespective of type, dose, or duration of antibiotics or probiotics.
Types of outcome measures
We will assess the following outcomes:
Helicobacter pylori eradication rate;
Overall rate of adverse events, defined as any untoward medical occurrence not necessarily having a causal relationship with the treatment but resulting in either dose reduction or discontinuation of therapy (ICH-GCP 1997).
Rate of treatment completion, defined as intake of anti-H.pylori drugs for at least 80% of the planned duration;
Rate of severe adverse events, defined as an event during therapy which would satisfy one of the following criteria:
which requires hospitalisation or prolongs present hospitalisation;
which requires intervention for management;
which is life-threatening;
which leads to persistent or significant disability for the participant;
which causes a congenital anomaly or birth defect;
which causes death (ICH-GCP 1997).
Rate of occurrence of diarrhoea;
Rate of occurrence of metallic taste;
Rate of occurrence of epigastric pain/discomfort.
Search methods for identification of studies
We will limit our search to literature published after the discovery of H.pylori in 1984 (Marshall 1984).
We will search the following databases electronically:
Searching other resources
We will search for further studies by browsing the reference lists of the identified studies. We will also contact study authors for any clarification or supplementary information, if needed.
Data collection and analysis
Selection of studies
Both review authors (AG and RA) will independently scan all titles identified by the literature searches. We will go through abstracts of identified studies to assess their design and relevance for this review. The two review authors will then go through the full text of selected articles independently, to look for inclusion criteria at initial assessment. While reviewing the full text of a study, we will check eligibility criteria using a printed checklist. We will resolve any discrepancy between the two review authors by discussion. While reviewing the titles, abstracts, or full text articles, the review authors will not be blinded to the journal of the original publication, or to the names and affiliations of the authors. For any studies that are excluded, we will provide the reason for exclusion.
Data extraction and management
The two unblinded review authors will independently extract the data, which they will enter into a predefined electronic data extraction sheet, designed for this review. Both review authors will cross-check their extracted data, resolving any discrepancies between their findings by discussion.
We will extract the following data for each study:
General information: Study ID, year of publication, language, whether available as abstract or as full text.
Trial characteristics: Study design, randomised or quasi-randomised, method of randomisation, sequence generation, allocation concealment, blinding, number of participants with missing data, follow-up duration, first-line therapy or therapy for treatment failed previously;
Study population: Children or adults, number of participants in each arm, median (or mean) age, sex, symptomatic or asymptomatic participants, method of H.pylori detection before and after therapy;
Intervention: Antibiotic regimen and duration of therapy; type of probiotic (bacteria or Sachharomycetes), dose and duration of probiotics supplementation; placebo;
Outcome measures: Eradication rate, overall adverse events, metallic taste, diarrhoea, epigastric pain/discomfort, compliance rate, treatment discontinuation due to severe adverse events.
Assessment of risk of bias in included studies
Risk of bias will be assessed according to guidance in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The two review authors will independently assess risks of bias for each included study, using The Cochrane Collaboration’s 'Risk of bias' tool. This tool assesses each included study for risk of bias under domains of sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other issues. We will grade each study separately as being at low, high, or unclear risk of bias. We will present the risks of bias in two ways, namely 'Risk of bias graph' and 'Risk of bias summary'. 'Risk of bias summary' will give an information about different biases in individual studies included in analysis where as 'Risk of bias graph' gives an overview of overall proportions of different biases in included studies. The review authors will resolve any disagreement by discussion, and will contact the trial authors for clarification as and when required.
Measures of treatment effect
All our planned primary and secondary outcomes are dichotomous variables. We will report treatment effects as risk ratios (RRs), with 95% confidence intervals (CIs). We will calculate the number needed to treat for an additional beneficial outcome (NNTB) or harmful outcome (NNTH), using an ‘overall baseline risk', and will conduct sensitivity analyses using 'a range of different baseline risks'.
Unit of analysis issues
We will use studies with a parallel-group design where participants are individually assigned to one of the treatment groups with a single measurement for each outcome from each participant. The unit of analysis will therefore be individuals assigned to each group.
Dealing with missing data
On encountering missing data, we will contact authors of the original study by email to obtain missing data and relevant details. If we fail to gather any fruitful information after contacting study authors, we will impute data for missing values, assuming that all missing participants in the intervention and control groups had poor outcomes (failed to eradicate H.pylori). We will assess the impact of missing data on results by conducting sensitivity analyses based on 'best-case' and 'worst-case' scenarios. The 'best-case' scenario is that in which all participants with missing outcomes in the control group are considered to have had an unfavourable outcome, whereas those in the intervention group are considered as having a favourable outcome. The 'worst-case' scenario is that in which all participants with missing outcomes in the control group are taken to have favourable outcomes whereas those in the intervention group are taken as having unfavourable outcomes. For each trial, we will report whether the analysis was performed on an intention-to-treat (ITT) or on a per-protocol basis. Whenever possible, we will conduct an ITT analysis which aims to include all participants randomised into a trial, irrespective of what happened subsequently in terms of losses to follow-up, deviation from the protocol, dropouts, etc.
Assessment of heterogeneity
We will check for heterogeneity between the included trials by visual examination of the forest plots to check for overlapping confidence intervals. We will assess statistical heterogeneity among different studies using the Chi² test of heterogeneity with a 10% level of significance (P < 0.10). The Chi² test assesses whether observed differences between the study groups can be reasonably expected to have occurred by chance. The Chi² test has limited power in meta-analysis; thus, whereas a statistically significant result in this test may indicate the presence of heterogeneity, a non-significant result cannot be taken to exclude heterogeneity. We will also measure heterogeneity by applying the I² statistic (Higgins 2003). We will consider heterogeneity to be substantial if I² is greater than 50% or if the P value is less than 0.10. If we detect substantial heterogeneity, we will recheck the data, to exclude mistakes in data extraction or data entry. We will explore heterogeneity using sub-group analysis and sensitivity analysis.
Assessment of reporting biases
Apart from assessing the risk of selective outcome bias in included studies, as mentioned above, we will look for potential publication bias by generating a funnel plot, provided that at least 10 studies are available.
We will carry out all statistical analyses using Review Manager 5 software (RevMan 2012). We will conduct meta-analysis using both random-effects and fixed-effect models to ensure the robustness of our findings. If we find a difference between the two results then both will be presented; otherwise only the results of the fixed-effect model will be presented.
Subgroup analysis and investigation of heterogeneity
In case of substantial heterogeneity, we will perform subgroup analyses to explore the heterogeneity. If possible, we will do the following subgroup analyses in our review.
Clinical status of subjects: symptomatic or asymptomatic;
Age group: children or adults;
Therapy status: initial first-line therapy or second-line therapy for treatment failures;
Type of therapy: triple drug or quadruple drug therapy;
Probiotic composition: Saccharomyces species- or bacterial species-based probiotics. Studies using combinations of both will be excluded from this subgroup analysis;
Probiotics with placebo or no placebo;
Probiotics used in form of proprietary brands or non-proprietary edibles such as yogurt, curd, fermented milk products, etc.
For comparison of subgroups, we will use the Chi² test with a P value set at 0.05.
We will conduct sensitivity analyses for studies with a low risk of bias.