Ehvidence
Favorable Efficacy (RCT)

S. boulardii for pediatric acute diarrhea (China)

Saccharomyces boulardii CNCM I-745 vs. Standard rehydration (open control, no probiotic) · for Pediatric acute gastroenteritis (PAGE) · real-time analysis of 10 studies · updated 2026-06-06

Across 10 randomized trials of Chinese children with acute diarrhea, adding S. boulardii CNCM I-745 to standard rehydration raised the proportion of children who improved and who were cured, and was well tolerated.

Efficacy (RCT): Efficacy evidence on clinical outcomes. Effects are risk ratios with number-needed-to-treat where a baseline risk is available. Glossary →

Interpretation & tips

All 10 trials were open-label with no placebo, single-country (China), and mostly moderate quality, and publication bias was detected — so the effect size is probably optimistic. Diarrhea duration was also markedly shorter with S. boulardii (see the Duration tab), and adverse events were fewer (RR 0.19). The study was funded by the manufacturer (Biocodex).

20% higher rate of total effectiveness
Pooled RR 1.20 (95% CI 1.15–1.26) across 10 studies, 1,062 patients (random-effects, I² 0%).
1.20
Pooled RR
1.15–1.26
95% CI
10
Studies
1,062
Patients
7
NNT

Forest plot—Total effectiveness

Wu 2021 1.16 [1.03–1.32] Chen QJ 2020 1.24 [1.05–1.45] Qiu 2018 1.29 [1.07–1.55] Yao 2018 1.32 [1.06–1.65] Cao 2017 1.16 [1.04–1.29] Tan 2015 1.17 [1.04–1.33] Yang 2015 1.21 [1.04–1.42] Chen LL 2014 1.21 [1.02–1.44] Lv 2014 1.27 [1.03–1.57] Qu 2012 1.25 [1.04–1.50] Pooled (RE) 1.20 [1.15–1.26] 0.1 0.25 0.5 1 2 4 ← favors control favors treatment →

Study results—Total effectiveness

Study Design Dose / regimen Treatment Control RR [95% CI] Improvement NNT Weight
Wu 2021 RCT S. boulardii CNCM I-745 500 mg/d 50/51 43/51 1.16 [1.03–1.32] 16% 8 15%
Chen QJ 2020 RCT S. boulardii CNCM I-745 sachet 47/49 38/49 1.24 [1.05–1.45] 24% 6 9%
Qiu 2018 RCT S. boulardii CNCM I-745 500 mg/d 45/48 35/48 1.29 [1.07–1.55] 29% 5 7%
Yao 2018 RCT S. boulardii CNCM I-745 sachet 61/63 22/30 1.32 [1.06–1.65] 32% 5 5%
Cao 2017 RCT S. boulardii CNCM I-745 sachet 89/94 77/94 1.16 [1.04–1.29] 16% 8 20%
Tan 2015 note RCT S. boulardii CNCM I-745 sachet 54/55 46/55 1.17 [1.04–1.33] 17% 7 15%
Yang 2015 RCT S. boulardii CNCM I-745 sachet 46/48 38/48 1.21 [1.04–1.42] 21% 6 9%
Chen LL 2014 RCT S. boulardii CNCM I-745 sachet 40/42 33/42 1.21 [1.02–1.44] 21% 7 8%
Lv 2014 RCT S. boulardii CNCM I-745 sachet 39/43 30/42 1.27 [1.03–1.57] 27% 6 5%
Qu 2012 RCT S. boulardii CNCM I-745 sachet 50/55 40/55 1.25 [1.04–1.50] 25% 6 7%

RR < 1 favors treatment for outcomes where lower is better. Rows in gray have a confidence interval crossing 1 (individually inconclusive). “~” marks effects reported as OR/HR and treated as RR-approximations. “excl” = excluded from pooling (e.g. reviews). For this outcome higher is better, so RR > 1 favors treatment.

Notes & interpretation

  • Tan 2015—Rated high risk of bias in the review.

Background

A systematic review and meta-analysis (McFarland & Li 2025) of 10 RCTs (1,125 children) conducted in China, none previously included in English-language reviews. All trials gave conventional rehydration to both arms and used open (non-placebo) controls.

Topic methodology & caveats

Effectiveness and cure use the per-trial 2×2 counts from Table 1 of McFarland & Li 2025 (number rated effective, and number cured, per arm). These are higher-is-better outcomes, so a risk ratio above 1 favors S. boulardii; our random-effects estimates (effectiveness RR ≈ 1.20, cure RR ≈ 1.48) are close to the review's Mantel-Haenszel 1.22 and 1.47. Duration of diarrhea is a continuous outcome: each trial's standardized mean difference (Hedges' g) is computed live from its reported mean ± SD, and pooled by random-effects, exactly like every other outcome — no trial excluded and nothing hardcoded. (For reference, the source review reported a pooled SMD of −1.63 using a Bayesian model; ours differs because of the model and because Yao 2018's very small SDs carry more weight here.) The pooled adverse-event result (RR 0.19) is described but not encoded.

Studies