Association between Proton Pump Inhibitor and Clostridioides difficile Risk in ICU Patients: A Systematic Review and Meta-Analysis of Clinical Outcomes

Abstract

Introduction

Clostridioides difficile infection (CDI) is a leading cause of antibiotic-associated diarrhea in critically ill patients. This meta-analysis evaluated whether proton pump inhibitor-based stress ulcer prophylaxis increases CDI risk and examined associated outcomes, including mortality, gastrointestinal bleeding, and other clinical effects in ICU patients.

Methods

The study was prospectively registered on PROSPERO (CRD420251176540). Following PRISMA guidelines, PubMed, Scopus, Cochrane, and Google Scholar were searched (2007 – March 2025) for randomized controlled trials and observational studies comparing PPIs with H₂ receptor antagonists (H₂RAs) or placebo in ICU patients. Data were pooled using a random-effects model, and heterogeneity was assessed using the I² statistic.

Results

Eight studies (∼58,000 patients) met inclusion criteria. PPI use did not significantly increase CDI risk (OR 1.12; p = 0.49, I² = 73%) or mortality (RR 1.01; p = 0.70) but significantly reduced gastrointestinal bleeding risk (RR 0.60; p = 0.02), particularly versus H₂RAs (RR 0.72; p = 0.0008). No significant differences were found in ICU or hospital length of stay as well as for mechanical ventilation duration.

Conclusion

In ICU patients, PPI prophylaxis was associated with reduced gastrointestinal bleeding but showed no clear benefit for mortality, ICU or hospital length of stay, or duration of mechanical ventilation. The available evidence does not allow firm conclusions regarding CDI risk because of substantial heterogeneity and very low certainty of evidence. Compared with H2RAs, PPIs provide greater bleeding protection, although possible signals of longer ICU stay and slightly higher mortality require further investigation.

Introduction

Clostridioides difficile is the leading cause of antibiotic-associated diarrhea in hospitalized patients, with rising incidence, severity, and mortality. It has emerged as a significant pathogen in critically ill populations [1, 2]. While most infections follow antibiotic exposure, proton pump inhibitor (PPI) therapy commonly prescribed for stress ulcer prophylaxis (SUP) has also been implicated as a risk factor [3, 4]. Although PPIs reduce the risk of stress-related upper gastrointestinal (GI) bleeding from mucosal ischemia and reperfusion injury, the incidence of clinically significant bleeding in modern ICU practice has declined [5], raising concern that widespread PPI use may expose patients to unnecessary harm.

Previous systematic reviews have reported an association between PPI use and both initial and recurrent CDI, particularly when combined with antibiotics, whereas H₂ receptor antagonists appear to confer lower risk [6, 7]. Several biologically plausible mechanisms may explain the proposed association between PPIs and CDI. By increasing gastric pH, PPIs may reduce the gastric acid barrier, potentially allowing greater survival of ingested C. difficile spores and other enteric organisms. Acid suppression may also promote upper GI bacterial overgrowth and alter intestinal microbial composition, thereby reducing colonization resistance against C. difficile. In addition, PPIs have been reported to exert immunomodulatory effects, including effects on neutrophil and inflammatory function, which may further influence host susceptibility to infection [8, 9]. These observations underscore the clinical paradox: strategies to prevent one ICU complication may predispose patients to another.

Given the ongoing variability in PPI prescribing practices for SUP and uncertainty in balancing benefits against potential harms, we conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) and observational studies in critically ill patients. Our goal was to evaluate the association between PPI prophylaxis and major clinical outcomes, including GI bleeding, C. difficile infection, mortality, and resource utilization, to inform safer, evidence-based prescribing and guideline recommendations.

Methods

This systematic review and meta-analysis was conducted in compliance with the Cochrane [10] and PRISMA guidelines [11]. This study was registered with PROSPERO (CRD420251176540). PRISMA checklist is provided in the online supplementary materials (for all online suppl. material, see https://doi.org/10.1159/000553518).

Data Sources and Search Strategy

A comprehensive literature search was carried out using four major databases – PubMed, Scopus, Cochrane, and Google Scholar, from 2007 to March 2025. The search strategy included keywords such as “Proton Pump Inhibitors,” “Clostridioides difficile Infection,” “Mechanically ventilated patients,” and “ICU,” with full details provided in online supplementary File 1. No language filters were applied.

Two authors (Z.R., A.F.) independently reviewed and screened studies, with any disagreement resolved by a third author (M.I.) via mutual discussion. Duplicates were removed by manual screening. Full-text articles were then assessed against the inclusion criteria. The full study selection process is outlined in the PRISMA flowchart (online suppl. Fig. 1).

Eligibility Criteria

Eligible study designs were limited to RCTs and observational studies. Studies were deemed eligible if they involved ICU patients with PPIs as an intervention with a comparator group who used H₂ blockers or a placebo. The included studies were required to report at least one primary outcome of interest that included the rate of C. difficile infection, in-hospital mortality, and duration of ICU or hospital stay. Secondary outcomes included GI bleeding and duration of mechanical ventilation. Studies were excluded if they did not report outcomes of interest. In addition, conference abstracts, case reports series, as well as systematic reviews and meta-analysis were excluded.

Data Extraction and Quality Assessment

Data extraction was done independently via two reviewers (A.R.A., M.I.) using a standardized Excel-based template designed to apprehend key study characteristics, baseline demographic information, and outcome measures. Any inconsistencies were resolved by the third reviewer with consensus.

The risk of bias in RCTs was evaluated using the Cochrane Risk of Bias 2 (ROB 2) tool [12], which assesses five domains: bias arising from the randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. For non-randomized studies, risk of bias was assessed using the updated version of the Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I V2) tool [13], which evaluates seven domains: bias due to confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, and selection of the reported result. Risk of bias assessments were visually summarized using the robvis tool [14].

Statistical Analysis

Statistical analyses were conducted using Review Manager (RevMan, version 5.4; The Cochrane Collaboration, 2020) [15]. A random-effects model was applied to account for potential unmeasured confounding and between-study variability. For C. difficile infection, odds ratios (ORs) with corresponding 95% confidence intervals (CIs) were calculated. Risk ratios (RRs) were used for other dichotomous outcomes, while mean differences (MDs) were used for continuous outcomes. Statistical significance was defined as a p value <0.05. All results were displayed visually as forest plots.

Heterogeneity across studies was assessed using the I² statistic [16], with values >50% indicating substantial heterogeneity. To explore sources of heterogeneity, sensitivity analyses were conducted using a leave-one-out method to identify influential studies. Additionally, subgroup analyses were performed to examine the effects of age and comparator group type on outcome measures and heterogeneity. Publication bias was not assessed due to the presence of <10 studies included in our analysis.

Certainty of Evidence (GRADE)

We assessed certainty for each outcome using the GRADE framework, starting at high for randomized trials and low for observational studies, and downgrading for risk of bias, inconsistency (heterogeneity), indirectness, imprecision, and publication bias as applicable. Two reviewers judged domains independently with consensus resolution. For dichotomous outcomes, we translated relative effects into absolute risk differences per 1,000 using pooled control risks; continuous outcomes were judged on clinical importance of the MD and its 95% CI. Results are summarized in the summary-of-findings tables with final ratings (high/moderate/low/very low).

Results

A total of eight studies were included in this meta-analysis, comprising four RCTs [4, 17–19] and four observational studies [20–23], with an aggregate population of approximately 58,000 critically ill patients. Baseline characteristics of the RCT populations are summarized in Table 1, while those of the observational studies are presented in Table 2.

Table 1.

Baseline characteristics of the included RCTs

Author and year	Journal	Type of study	Total number of participants	Prophylaxis given, drug (n)		Males, n (%)		Age, years		Patients on mechanical ventilation, n (%)
				intervention	control	intervention	control	intervention	control	intervention	control
Abu El-Ella et al. [4] (2020)	Anales de Pediatria	RCT	144	Omeprazole (72)	Nothing (72)	34 (47.2%)	40 (55.6%)	24 months (7−106.5)a	24 months (6−72)a	28 (38.9%)	23 (31.9%)
Cook et al. [17] (2024)	The New England Journal of Medicine	RCT	4,821	IV pantoprazole (2,417)	0.9% sodium chloride (2,404)	1,534 (63.5%)	1,534 (63.8%)	58.2±16.4	58.3±16.4	2,417 (100.0%)	2,404 (100.0%)
Krag et al. [18] (2018)	The New England Journal of Medicine	RCT	3,298	IV pantoprazole (1,644)	0.9% sodium chloride (1,647)	1,039 (63%)	1,067 (65%)	67 (56–75)a	67 (55–75)a	1,273 (77%)	1,316 (80%)
Peptic Trial et al. [19] (2020)	JAMA	RCT	26,826	PPI (13,436)	H2 blockers (13,392)	8,577 (63.8%)	8,560 (63.9%)	58.6±17	58.2±17.2	13,436 (100%)	13,392 (100%)

^aMedian (IQR).

Table 2.

Baseline characteristics of included observational studies

Author	Journal	Type of study	Total number of participants	CDI	Male	Age	Patients on PPI, n (%)	Patients on mechanical ventilation, n (%)	Patients on antibiotics, n (%)
Faleck et al. [20] (2016)	American Journal of Gastroenterology	Retrospective (multicenter)	18,134	Yes	139 (51.3)	66 (54–76)a	155 (57.2)	196 (72.3)	239 (88.2)
				No	10,021 (56.1)	66 (54–76)a	11,075 (62)	8,595 (48.1)	12,000 (67.2)
Barletta et al. [21] (2014)	Critical Care BMC	Retrospective (case control)	408	Yes	113 (55)	68.6±15	170 (83)	151 (74)	191 (94)
				No	116 (57)	68.5±15	149 (73)	145(71)	177 (87)
Mi et al. [22] (2020)	Frontiers in Cellular and Infection Microbiology	Cross-sectional (multicenter)	531	Yes	55	–	31	–	–
				No	–	–	142	–	–
Ro et al. [23] (2016)	Gut and Liver	Retrospective	1,005	Yes	–	64.8±14.1	28 (73.7)	22 (57.9)	10 (26.3)
				No	–	60.7±16.1	416 (43.7)	448 (47.0)	231 (24.2)

^aMedian (IQR).

The four RCTs collectively enrolled 35,089 patients, with individual study sample sizes ranging from 144 to 26,826. PPIs were administered in 17,537 patients with 185 developing C. difficile infection. Across these trials, 63.8% of participants were male. Three RCTs focused on older adult populations, reporting mean ages ranging from 58 to 67 years. In contrast, the study by Abu El-Ella et al. [4] exclusively investigated pediatric patients, with a median age of 24 months. All RCTs included mechanically ventilated patients; two studies reported 100% of participants on mechanical ventilation, while the remaining two reported variable proportions of ventilated patients.

The four observational studies included a combined total of 20,078 critically ill ICU patients, of whom 11,305 (56.3%) were male. All studies focused on older adult populations, with a mean age greater than 60 years across all cohorts. Mechanical ventilation was required in 48.9% of patients. PPIs were administered to 12,166 patients (60.6%), and 614 patients developed C. difficile infection.

Quality Assessment

Among the RCTs, two studies were assessed as having a high risk of bias, while the remaining two were judged to have a low risk of bias. The high risk in both Abu El-Ella et al. [4] and the PEPTIC trial et al. [19] was primarily attributed to deviations from the intended intervention. Additionally, Abu El-Ella 2020 demonstrated high risk of bias in the measurement of outcomes (online suppl. Fig. 2 and 3).

For the observational studies, three out of four were determined to have a moderate risk of bias, with Faleck et al. [20] being the exception, rated as low risk. The most common sources of bias across the three studies were moderate risk due to confounding and selective reporting. Two studies exhibited bias related to missing data, and one study had concerns regarding outcome measurement (online suppl. Fig. 4 and 5).

Analysis

Clostridioides difficile Infection

Analysis revealed that the odds of developing C. difficile infection among ICU patients were higher in the PPI group compared to controls, although the difference was not statistically significant (n = 55,128; OR [95% CI]: 1.12 [0.81–1.56]; p = 0.49; I² = 73%, χ² = 22.54). Subgroup analysis by study design showed that RCTs demonstrated a nonsignificant reduction in odds with PPI use (OR [95% CI]: 0.95 [0.55–1.64]; p = 0.85), whereas observational studies showed a nonsignificant increase (OR [95% CI]: 1.29 [0.79–2.08]; p = 0.31). However, heterogeneity remained high in both subgroups (I² = 68% and 81%, χ² = 6.28 and 15.53, respectively), suggesting that study design was not a major source of heterogeneity (Fig. 1). Sensitivity analysis identified Ro et al. [23] as a key contributor to the overall heterogeneity. Removal of this study reduced heterogeneity from 73% to 51%. Subgroup analysis by age could not be conducted as the only pediatric study (Abu El-Ella et al. [4]) lacked estimable data (online suppl. Fig. 6).

Fig. 1.

Forest plot of ORs for Clostridioides difficile infection (CDI) in ICU patients receiving PPIs vs. controls. Pooled analysis shows a nonsignificant increase in CDI risk with PPI use.

Further subgroup analysis by type of comparator revealed nearly equivalent odds of infection between PPI and placebo groups (OR [95% CI]: 0.98 [0.71–1.34]; p = 0.89). However, patients receiving H₂RAs had nonsignificantly lower odds of C. difficile infection compared to those on PPIs (OR [95% CI]: 1.39 [0.64–3.02]; p = 0.41). Studies using H₂RA comparators also demonstrated substantial heterogeneity (I² = 87%, χ² = 15.17), indicating this subgroup may have contributed to between-study variability (online suppl. Fig. 7).

Mortality

Analysis of 34,966 ICU patients demonstrated no significant difference in mortality risk between those receiving PPIs and control groups (RR [95% CI]: 1.01 [0.96–1.07]; p = 0.70; I² = 31%, χ² = 4.35) (Fig. 2). Subgroup analysis by age revealed that the relative risk of mortality in adults was 1.01 (95% CI: 0.95–1.08; p = 0.77), while in children it was 0.90 (95% CI: 0.39–2.08; p = 0.81). Heterogeneity was moderate in adult studies (I² = 53%, χ² = 4.26), suggesting they were the primary contributors to between-study variability (online suppl. Fig. 8).

Fig. 2.

Forest plot of RRs for all-cause mortality in ICU patients on PPIs vs. controls. No statistically significant difference in mortality between the groups.

When stratified by comparator, mortality risk was slightly lower with PPIs compared to placebo (RR [95% CI]: 0.97 [0.91–1.04]; p = 0.45), although this difference was not statistically significant. Conversely, when compared to H₂ receptor antagonists, PPI use was associated with a marginally increased, yet still nonsignificant, mortality risk (RR [95% CI]: 1.05 [1.00–1.10]; p = 0.07) (online suppl. Fig. 9).

Gastrointestinal Bleeding

Analysis of GI bleeding among ICU patients showed that PPI use was associated with a significantly lower risk of bleeding compared to controls (RR [95% CI]: 0.60 [0.39–0.92]; p = 0.02; I² = 82%, χ² = 16.48). Sensitivity analysis identified Cook 2024 [17] as the primary source of heterogeneity; its exclusion reduced heterogeneity to 31% while maintaining statistical significance (p = 0.006) (Fig. 3).

Fig. 3.

Forest plot of RRs for gastrointestinal (GI) bleeding in ICU patients on PPIs vs. controls. PPI use significantly reduced the risk of GI bleeding.

Subgroup analysis by age demonstrated that PPIs significantly reduced the risk of GI bleeding in adults (RR [95% CI]: 0.52 [0.32–0.84]; p = 0.008), whereas no such benefit was observed in children (RR [95% CI]: 1.05 [0.62–1.80]; p = 0.85). The difference between these subgroups approached statistical significance (p = 0.06), suggesting a possible age-dependent effect (online suppl. Fig. 10). When stratified by type of comparator, PPIs nonsignificantly reduced GI bleeding risk compared to placebo (RR [95% CI]: 0.56 [0.29–1.11]; p = 0.10), though the high heterogeneity within this subgroup (I² = 85%, χ² = 13.56) limits interpretability. In contrast, compared to H₂ receptor antagonists, PPI use was associated with a statistically significant reduction in GI bleeding risk (RR [95% CI]: 0.72 [0.59–0.87]; p = 0.0008) (online suppl. Fig. 11).

ICU Stay

In ICU patients receiving PPIs, the mean ICU length of stay was slightly longer compared to controls; however, this difference did not reach statistical significance (MD [95% CI]: 0.27 [−0.02–0.56]; p = 0.07; I² = 35%) (Fig. 4).

Fig. 4.

Forest plot of MD in ICU length of stay in patients receiving PPIs vs. controls. PPI use associated with a nonsignificant increase in ICU stay duration.

Subgroup analysis by age showed a nonsignificant increase in ICU stay with PPIs in both adults (MD [95% CI]: 0.24 [−0.11–0.58]; p = 0.18) and children (MD [95% CI]: 1.00 [−0.89–2.89]) (online suppl. Fig. 12). Notably, heterogeneity was moderate in the adult subgroup (I² = 61%, χ² = 2.59), indicating variability in study effects and warranting cautious interpretation. Subgroup analysis by comparator type revealed no significant difference in ICU length of stay when PPIs were compared to placebo (MD [95% CI]: 0.06 [−0.40–0.52]; p = 0.80). However, when compared to H₂ receptor antagonists, PPI use was associated with a significantly longer ICU stay (MD [95% CI]: 0.37 [0.21–0.52]; p < 0.00001) (online suppl. Fig. 13).

Hospital Stay

The overall analysis showed that PPI use was associated with a nonsignificant reduction in total hospital length of stay compared to controls (MD [95% CI]: −0.21 [−1.45–1.03]; p = 0.74; I² = 73%, χ² = 7.29) (online suppl. Fig. 14). Sensitivity analysis identified Cook 2024 [17] as the major contributor to heterogeneity. Excluding this study reduced heterogeneity to 0% and altered the direction of the pooled effect, showing a nonsignificant increase in hospital stay among PPI users (MD [95% CI]: 0.33 [−0.25–0.90]; p = 0.26).

Subgroup analysis by age revealed a nonsignificant reduction in hospital stay among adults receiving PPIs (MD [95% CI]: −0.45 [−2.05–1.14]; p = 0.58); however, substantial heterogeneity was present (I² = 85%, χ² = 6.83), limiting the reliability of this finding. In pediatric patients, PPI use was associated with a nonsignificant increase in hospital stay (MD [95% CI]: 0.67 [−1.44–2.77]; p = 0.53) (online suppl. Fig. 15). When stratified by comparator type, PPIs nonsignificantly reduced hospital stay compared to placebo (MD [95% CI]: −0.55 [−2.46–1.37]; p = 0.58), while comparison with H₂ receptor antagonists showed a nonsignificant increase in hospital stay duration (MD [95% CI]: 0.30 [−0.30–0.90]; p = 0.32) (online suppl. Fig. 16).

Mechanical Ventilation

Across included studies, the use of PPIs in ICU patients was associated with no significant difference in the duration of mechanical ventilation compared to controls (MD [95% CI]: 0.01 days [−0.30–0.31]; p = 0.97; I² = 0%, χ² = 0.30) (online suppl. Fig. 17). The low heterogeneity indicates consistent findings across studies.

Subgroup analysis by age showed no meaningful difference in the duration of mechanical ventilation between patients receiving PPIs and controls. In adults, the MD was 0.01 days (95% CI: −0.30–0.31; p = 0.97), while in children, the MD was 0.00 days (95% CI: −0.74 to 0.74), indicating a neutral effect across age groups (online suppl. Fig. 18). Similarly, when stratified by comparator, no significant difference in ventilation duration was observed. The MD between PPIs and placebo was 0.00 days (95% CI: −0.31 to 0.31), and between PPIs and H₂ receptor antagonists was 1.26 days (95% CI: −3.26–5.78), both of which were statistically nonsignificant (online suppl. Fig. 19).

GRADE Certainty of Evidence across Outcomes

Mortality evidence is moderate certainty and indicates no important effect of PPIs (RR 1.01), with an absolute difference of ∼+2/1,000. GI bleeding shows a low-certainty reduction (RR 0.60) equating to ∼9 fewer events per 1,000, tempered by substantial heterogeneity. Evidence for C. difficile is very low certainty, with imprecision and inconsistency precluding firm conclusions. Continuous outcomes (ICU and hospital LOS, mechanical ventilation) are moderate to low certainty and collectively show no clinically meaningful change. Downgrading primarily reflected risk of bias in ≥1 trials, high I² for some outcomes, and CIs spanning clinically relevant effects (online suppl. Table 1).

Discussion

CDI imposes a substantial healthcare burden, particularly among patients with prolonged hospitalizations and complex care needs [24]. In this meta-analysis of approximately 58,000 patients from eight studies, PPI prophylaxis was associated with a significant reduction in GI bleeding risk without an accompanying increase in C. difficile infection. However, no significant differences were observed in mortality, ICU or hospital length of stay, or duration of mechanical ventilation. Although the prevalence of GI bleeding among critically ill ICU patients is relatively low, its prognostic significance remains uncertain, and stress SUP is still widely practiced [25]. The absolute bleeding risk is low in modern ICUs, raising questions about whether the relative benefit of PPIs translates into clinically meaningful outcomes. The observed benefit in bleeding prevention did not translate into a mortality advantage or increased CDI risk. However, interpretation requires caution. Although sensitivity analyses confirmed the robustness of the GI bleeding effect, heterogeneity remained high for CDI outcomes, with wide CIs that do not exclude a clinically important risk. This highlights ongoing uncertainty rather than definitive evidence of safety.

Differences between ICU and non-ICU populations may further explain discrepancies with prior meta-analyses. A recent review of 50 studies, including general ward patients, reported a significant increase in CDI risk among PPI users [26]. Our study, limited to ICU populations with a mean age of 58–67 years, did not reproduce this finding. This raises the possibility that patient age, comorbidity burden, and severity of illness modify the relationship between PPI use and CDI [27].

The mechanistic link between PPIs and CDI remains debated. Reduced gastric acidity may facilitate survival of C. difficile spores, while other risk factors including recent antibiotic use, advanced age, comorbidities, and prior hospitalization likely contribute more strongly to infection risk [6, 28–30]. Observational studies have yielded conflicting results, with some identifying PPIs as stronger risk factors than H₂RAs [31, 32]. Our findings suggest that in the ICU population, the impact of PPIs on CDI is less clear, and residual confounding remains a plausible explanation for the divergent results across study designs.

An important nuance is the potential difference between adult and pediatric populations. In subgroup analysis, PPIs significantly reduced GI bleeding in adults but not in children, with the interaction approaching statistical significance. However, these pediatric subgroup findings should be considered exploratory because they were derived from a single pediatric study and were limited by sparse or non-estimable outcome data. Therefore, no firm conclusions can be drawn regarding age-dependent effects, and further pediatric-specific studies are needed. Comparator choice also appeared to influence outcome estimates. Although PPIs were associated with a lower risk of GI bleeding compared with H2RAs, they were also associated with a significantly longer ICU length of stay in subgroup analysis. This finding warrants cautious interpretation. The absolute difference was small, and several alternative explanations should be considered, including greater baseline illness severity, confounding by indication, differences in ICU discharge practices, and unmeasured co-interventions that could not be fully accounted for across the included studies. Therefore, a direct causal relationship between PPI exposure and prolonged ICU stay cannot be inferred from these data. Nevertheless, this signal may be clinically relevant. Even modest increases in ICU length of stay can affect bed availability, resource utilization, and healthcare costs in critically ill populations. However, this finding was not accompanied by significant differences in hospital length of stay, duration of mechanical ventilation, or mortality, which tempers concern for a consistent drug-related harm signal. Accordingly, the observed increase in ICU length of stay should be interpreted as hypothesis generating and warrants further evaluation in future head-to-head studies comparing PPIs with H2RAs.

Study heterogeneity also warrants discussion. Cook et al. [17] contributed disproportionately to bleeding outcome variability, while Ro et al. [23] accounted for much of the CDI heterogeneity. Differences in outcome definitions, comparator arms, and patient populations likely explain these discrepancies. Importantly, all RCTs were published after 2018 and demonstrated more neutral safety signals compared with observational studies, underscoring the likelihood of confounding in non-randomized data.

Our findings have several clinical implications. The reduction in GI bleeding with PPIs remains clinically relevant for high-risk ICU patients, especially those receiving mechanical ventilation. However, the absolute incidence of bleeding is low, and the lack of survival benefit, coupled with possible signals of harm compared to H₂RAs, suggests that prophylaxis should be targeted rather than universal. In patients with lower bleeding risk, H₂RAs may represent a reasonable alternative. Our data highlight important gaps in knowledge. Pediatric data remain sparse, and future pediatric-specific studies are needed before any firm conclusions can be drawn regarding the balance of benefit and harm in this population. Uniform definitions for CDI and GI bleeding are needed to reduce heterogeneity in future analyses. Moreover, head-to-head RCTs comparing PPIs and H₂RAs powered for patient-centered outcomes mortality, functional recovery, and long-term complications are needed to clarify the optimal prophylactic strategy in critically ill patients.

Limitations

This meta-analysis has several limitations that should be considered when interpreting the findings. The number of eligible studies was limited, with only eight included, which restricts the statistical power of subgroup analyses and particularly limits conclusions in pediatric populations where data are sparse. Considerable heterogeneity was observed for several outcomes, most notably GI bleeding and CDI, which likely reflects differences in study design, patient populations, comparator groups, and outcome definitions. Although sensitivity analyses identified certain influential studies that contributed disproportionately to variability, substantial heterogeneity remained unexplained and lowers the certainty of pooled estimates. The inclusion of both RCTs and observational studies also introduces variability in study quality. Observational studies are inherently prone to confounding by indication and selective reporting, and effect estimates from these studies often diverged from those of randomized trials, suggesting residual confounding may have influenced the observed associations. Even among randomized trials, risk of bias was present; for example, the large PEPTIC trial and Abu El-Ella 2020 were judged to be at high risk of bias due to deviations from intended interventions and outcome measurement issues, raising concerns that protocol nonadherence or misclassification may have obscured true effects. Another limitation is the absence of standardized outcome definitions across studies. GI bleeding, CDI, and mortality were variably defined and inconsistently adjudicated, limiting comparability across trials and possibly contributing to heterogeneity. The inability to formally assess publication bias due to the small number of included studies further raises the possibility that unpublished negative trials might have influenced the direction of the pooled results. Generalizability is also restricted. Most included patients were older, mechanically ventilated adults in ICU settings, and evidence from pediatric populations was minimal. As a result, the applicability of these findings to general ward patients, younger populations, or less severely ill cohorts remains uncertain.

Despite these limitations, this study has several notable strengths. It was conducted in accordance with Cochrane and PRISMA guidelines, used a comprehensive search strategy across multiple databases, and applied rigorous independent screening, data extraction, and risk of bias assessments. The inclusion of both randomized trials and large observational cohorts allowed for a broad evaluation of outcomes in around 58,000 critically ill patients, making this one of the most comprehensive syntheses on the topic to date. By incorporating sensitivity and subgroup analyses, this review provides in-depth insights into the effects of PPIs in ICU populations and highlights areas where future research is most urgently needed.

Conclusion

ICU patients demonstrate that PPI prophylaxis significantly reduces GI bleeding, particularly in adults, but does not translate into lower mortality, shorter ICU or hospital stays, or reduced duration of mechanical ventilation. No statistically significant increase in C. difficile infection was identified; however, substantial heterogeneity and variability across study designs limit confidence in this finding. Comparisons with H₂ receptor antagonists revealed a potential trade-off, with fewer bleeding events but possible signals toward longer ICU stays and marginally higher mortality. Taken together, these results suggest that while PPIs remain effective for SUP in high-risk ICU patients, prophylaxis should not be applied universally and must be individualized to patient risk profiles. Future large-scale randomized trials, particularly in pediatric cohorts and direct PPI versus H₂RA comparisons, are needed to refine prophylaxis strategies and clarify the balance of benefit and harm.

Acknowledgments

Author Dr. Antonio Giuseppucci was not available to confirm co-authorship, but the corresponding author Dr. Puneet Kaur Bansal affirms that author Dr. Antonio Giuseppucci contributed to the paper, had the opportunity to review the final version to be published, and guarantees author Dr. Antonio Giuseppucci’s co-authorship status and the accuracy of the author contribution and conflict of interest statements.

Statement of Ethics

A Statement of Ethics is not applicable because this study is based exclusively on published literature.

Conflict of Interest Statement

The authors declare that they have no conflicts of interest to disclose.

Funding Sources

No funding was received for the design, conduct, analysis, interpretation, or publication of this study.

Author Contributions

The study was conceptualized and conducted under the direction of M.A.N. and A.B.S. Data collection and analysis were performed by A.R.A., M.I., M.A., A.F., P.K.B., M.N., and Z.R. Manuscript drafting was performed by A.R.A., A.M.Y., A.A., J.I., A.G., A.H.S., S.V., and A.U. Critical manuscript review was performed by all authors.

Funding Statement

No funding was received for the design, conduct, analysis, interpretation, or publication of this study.

Data Availability Statement

All data generated or analyzed during this study are included in this article and the online supplementary file. Further inquiries can be directed to the corresponding author.