Probiotics and synbiotics as adjunctive therapies for heart failure: a systematic review with network and pairwise...

This study aimed to evaluate the use of probiotics and synbiotics as microbiome-directed adjunctive therapies for heart failure (HF), while explicitly determining whether any eligible prebiotic-only trials existed.

We searched eight English- and Chinese-language databases from inception to 1 May 2026 and conducted expanded sensitivity searches in PubMed, Embase, and Web of Science on 15 July 2026. Random-effects network meta-analysis (NMA) was used only for connected multi-node evidence; two-node evidence was analyzed pairwise. As placebo, usual care, and food comparators were not clinically exchangeable, comparator-stratified sensitivity analyses of direct evidence were conducted using restricted maximum likelihood (REML) and Hartung-Knapp (HK) confidence intervals (CIs).

A total of 12 randomized studies were included in the systematic review (14 reports); 11 contributed to at least one primary or sensitivity quantitative analysis, and 10 contributed to the primary analyses. The revision-stage search audit identified no additional eligible randomized trials and no eligible prebiotic-only trials. Across all the trial populations, 1,021 participants were randomized, 973 of whom were allocated to review-eligible arms. The left ventricular ejection fraction (LVEF) network included 7 studies and 539 participants in the analysis set; however, this network was star-shaped, had no closed loops, and depended on a combined comparator node. The exploratory NMA estimate for multi-strain bacterial probiotics versus the combined comparator was a mean difference (MD) of 2.13 percentage points (95% confidence interval [CI] -0.10 to 4.35). The comparator-stratified sensitivity analysis for usual care yielded an MD of 1.83 percentage points (95% CI -2.33 to 5.98). All synthesized results were assessed as having either some concern or a high risk of bias. The effect estimates were generally imprecise, and the certainty of evidence was low or very low. Adverse event definitions and denominators were inconsistent.

The available evidence is sparse and affected by significant non-transitivity, risk of bias, and imprecision. Network comparisons and rankings should be considered hypothesis-generating; therefore, no probiotic or synbiotic adjunct can be recommended for routine HF management.

CRD420261413595. Public URL: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261413595.