Discussion

In this systematic review and network meta-analysis comparing the effects of prebiotics, probiotics, and synbiotics on gastrointestinal symptoms and clinical outcomes in critically ill patients, synbiotics and prebiotics were associated with a lower risk of gastrointestinal symptoms than control, whereas probiotics showed a directionally similar but statistically nonsignificant effect. No statistically significant differences were observed among the active interventions. On note, analyses on individual GI symptom suggested different effects between types of biotics, while these results should not be interpreted as demonstrating comparative superiority among them.

Although no significant differences were observed among the active interventions, the association between synbiotic use and a lower risk of gastrointestinal symptoms compared with control is biologically plausible. Synbiotics combine live microorganisms with substrates that promote microbial growth and metabolic activity [[7], [8], [9]], potentially helping to counteract dysbiosis and the depletion of short-chain fatty acids in critically ill patients [20,21]. Supporting this biological rationale, an RCT in patients with sepsis reported increased fecal counts of Bifidobacterium and Lactobacillus and higher short-chain fatty acid concentrations following synbiotic administration [22].

Differential effects of biotic interventions on clinical outcomes were observed. Although synbiotics and prebiotics were associated with a lower risk of gastrointestinal symptoms than control, probiotics and synbiotics were associated with a lower risk of VAP. Notably, this discrepancy may reflect distinct mechanisms of action. Specifically, prebiotics primarily modulate gut microbiome composition and SCFA production, which may more directly influence gastrointestinal function by enhancing intestinal barrier integrity and motility [23,24]. Conversely, the associations of probiotics and synbiotics with a lower VAP risk may be partly explained by the gut–lung axis. Through this pathway, critical illness–related dysbiosis and disruption of the intestinal barrier may promote bacterial translocation, systemic inflammation, and impaired pulmonary host defense [[25], [26], [27]]. A randomized controlled trial in patients with sepsis demonstrated that synbiotic administration maintained beneficial gut flora and significantly increased SCFA levels, along with reduced VAP incidence [22]. This finding is also consistent with prior pairwise meta-analytic evidence showing that synbiotics, but not probiotics alone, were associated with a significant reduction in VAP incidence [28].

Further, subgroup analyses indicated heterogeneous treatment effects across patient populations and intervention characteristics. Probiotics and synbiotics were associated with a lower incidence of gastrointestinal symptoms in surgical ICU patients, whereas significant benefits in medical ICU patients were observed only with prebiotics. Importantly, these differences may reflect variation in baseline dysbiosis or underlying pathophysiology, although the optimal biotic approach for each population remains uncertain and requires further investigation [4,29,30]. Most included trials assessed 7–13 days of treatment, during which synbiotics remained beneficial. Nonetheless, the optimal biotic therapy duration warrants additional study. In trials incorporating oligosaccharide supplementation, both prebiotics and synbiotics significantly reduced the incidence of gastrointestinal symptoms, suggesting that specific prebiotic components may be key determinants of efficacy.

Data on gastrointestinal symptoms suitable for the primary NMA were available in only 33 of the 72 included RCTs, likely reflecting the historical focus of ICU biotics trials and previous meta-analyses on infectious outcomes, particularly VAP, and the lack of standardized gastrointestinal outcome definitions [31,32]. Compared with the previous Bayesian NMA, the present study used a frequentist framework, compared individual biotic classes with placebo or usual care rather than broader nutritional strategies, focused on gastrointestinal symptoms rather than VAP prevention, and included 72 rather than 31 RCTs [12]. The two studies therefore address distinct clinical questions through different network structures, with the present work offering a larger evidence base focused on gastrointestinal function in critically ill patients. Strain- and fiber-specific effects could not be assessed because of heterogeneity in the formulations used across trials. Given that probiotic effects are strain-specific and that prebiotic fibers may differ in their effects on the gut microbiota and SCFA production, pooling diverse formulations may obscure clinically relevant effects [33]. The lower risk observed with oligosaccharide-containing regimens further suggests that intervention composition may influence efficacy, supporting future trials of well-characterized biotic formulations using standardized gastrointestinal outcomes.

This study has some limitations. Definitions and assessment methods for gastrointestinal symptoms, probiotic formulations, and prebiotic fiber types varied across studies, contributing to clinical heterogeneity and limiting recommendations regarding optimal regimens [34,35]. Direct comparisons between active interventions were sparse, and most estimates relied on indirect evidence through the control node [36]. Several studies had some concerns or a high risk of bias, and sensitivity analyses excluding high-risk studies did not fully support the primary findings, particularly for prebiotics. Significant inconsistency was also detected for infectious complications. In addition, the search excluded several databases, trial registries, and non-English full-text articles, potentially introducing publication and language bias. Estimates for ICU length of stay and mechanical ventilation duration should be interpreted cautiously because these outcomes are typically right-skewed, despite the use of validated methods to convert medians to means. Finally, certainty of evidence was not formally assessed using the GRADE framework. Given the risk of bias, small-study effects, sparse direct evidence, clinical heterogeneity, and absence of significant differences between active interventions, the findings and treatment rankings should be interpreted cautiously and should not independently inform clinical guidelines.