Probiotic in the Prevention of Ventilator-Associated Pneumonia in Critically Ill Patients

Evidence From Meta-Analysis and Trial Sequential Analysis of Randomized Clinical Trials

Yue-chen Sun; Chen-yi Wang; Hai-li Wang; Yao Yuan; Jian-hong Lu; Lei Zhong


BMC Pulm Med. 2022;22(168) 

In This Article


This study was written following the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines (Additional file 1: Appendix 1).[29]

Search Strategy and Selection Criteria

The clinical questions were specified using the PICO framework listed in Additional file 2: Appendix 2. Two writers (ZL and SYC) independently searched the PubMed, Embase and the Cochrane library databases to identify randomized controlled trials (RCTs) or quasi-RCTs that addressed the efficacy of probiotic, prebiotic or symbiotic supplementation in preventing VAP among critically ill patients from the inception to 10 October 2021, without language restriction. The keywords were as follows: "probiotic", "prebiotic", "synbiotic", "ventilator-associated pneumonia", "Randomized Controlled Trial", etc. The Additional file 2: Appendix 2 provided a full description of the search strategy. Moreover, the reference lists of relevant papers were selectively hand-searched to capture any additional studies.

We excluded studies if they were duplicate publications, case reports, letters, reviews, case–control studies, cohort studies or non-human studies. Trials eligibility were carried out by the two independent authors (ZL and SYC) through screening titles, abstracts and even reading the full text.

The primary outcome was as follows: the incidence of VAP; Secondary endpoints included: ICU/hospital/28-/90-day mortality, bacteremia, catheter-related bloodstream infection (CRBSI), diarrhea, ICU-acquired infections, infectious complications, pneumonia, urinary tract infection (UTI) and wound infection.

Data Extraction

The relevant data of included articles were extracted by two separate authors (ZL and SYC) and were summarized in Table 1. We contacted original authors to ask for any relevant missing information whenever possible, for example, the Mahmoodpoor study.[30]

Assessment of Study Quality

We evaluated the quality of each eligible studies in adherence to The Cochrane Collaboration's tool,[31] including selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Simultaneously, the strength of evidence for all outcomes in adults studies was assessed using GRADE (Grading of Recommendations Assessment, Development and Evaluation) methodology.

Sensitivity Analysis and Publication Bias

We did a sensitivity analysis so as to appraise the stability of the pooled effect estimates. The publication bias was also examined by the two authors via the funnel plot and statistical tests (Begg's Test and Egger's Test).[31] In addition, we conducted a trim and fill analysis.

Statistical Analysis

For each trial, the dichotomous outcomes were reported as relative risk (RR) along with 95% confidence interval (CI). The median and range/interquartile range were converted to mean and standard deviation using the formulas described by one previous study.[32] The between-study heterogeneity was determined in accordance with the Chi-square test, P values and the I2 index. In view of the conservative of random-effects model, we used this model to pool all data.[33] In order to determine whether the accumulated evidence was sufficient and conclusive, a trial sequential analysis (TSA) was performed in our study. The TSA version beta (, Stata 12.0 (StataCorp, College Station, TX, USA) and Review Manager Version 5.3.5 software ( were implemented to analyze data. A two-tailed P-value < 0.05 was considered statistically significant.

Probiotics, prebiotics and synbiotics were equal for analysis in our meta-analysis.

Considering the difference in neonates/children and adults, we analyzed the data separately. In addition, we conducted subgroup analyses based on the strain types (prebiotic vs. synbiotic vs. probiotic), the risk of bias (low risk vs high risk) and the center (multi-center vs. single-center). We also applied a cumulative meta-analysis by publication year.