Highlights
– A 2025 systematic review and bayesian meta-analysis of 14 randomized trials (1,322 participants) found a 97.3% probability that shorter durations of prophylactic antibiotics (including none) are noninferior to 5–7 days for all-cause mortality after upper gastrointestinal bleeding in cirrhosis (RD 0.9%; 95% CrI −2.6 to 4.9).
– Shorter durations had a lower probability of noninferiority for early rebleeding (73.8%) and were associated with more trial-defined bacterial infections (RD 15.2%; 95% CrI 5.0 to 25.9).
– Studies published after 2004 showed higher probabilities of noninferiority across outcomes, but trials were generally low-to-moderate quality with heterogeneous infection definitions and no adverse-event reporting.
Background
Patients with cirrhosis who present with upper gastrointestinal (GI) bleeding—most commonly from esophageal or gastric varices—are at increased risk of bacterial infections, rebleeding, and death. International guidance developed over the past decade (including Baveno consensus statements and EASL/AASLD recommendations) has recommended antibiotic prophylaxis, typically for 5–7 days, for hospitalized cirrhotic patients with variceal bleeding. The rationale is twofold: to prevent bacterial infections that can precipitate rebleeding and organ failure, and to reduce mortality observed in earlier trials.
However, practice and supportive care have evolved substantially—vasoactive agents, urgent endoscopic band ligation, broader use of proton pump inhibitors for nonvariceal bleeding, earlier rescue transjugular intrahepatic portosystemic shunt (TIPS), and improved intensive care—raising the question whether older trial evidence still supports routine 5–7 day antibiotic courses for all patients with cirrhosis and upper GI bleeding.
Study design
Prosty and colleagues performed a systematic review and bayesian random-effects meta-analysis (JAMA Intern Med. 2025) to reassess the evidence supporting current prophylaxis recommendations. Key methods and inclusion criteria:
- Data sources: Embase, MEDLINE, and CENTRAL searched to September 25, 2024, for randomized clinical trials (RCTs) of adult patients with cirrhosis and upper GI bleeding.
- Study selection: RCTs comparing longer versus shorter durations of prophylactic systemic antibiotics (or any prophylaxis versus none). Exclusions included observational and pediatric studies, non-systemic antibiotics, and trials not reporting mortality or early rebleed.
- Outcomes: Primary outcome was all-cause mortality with a prespecified 5% noninferiority margin (risk difference scale). Secondary outcomes included early rebleeding and study-defined bacterial infections.
- Analysis: Random-effects bayesian meta-analysis using a noninformative prior for effect and a weakly informative prior for heterogeneity. A post hoc subgroup analysis focused on trials published after 2004 to account for contemporaneous advances in care.
- Quality assessment: Risk of bias was assessed with the Cochrane Risk of Bias 2 tool.
Key findings
Fourteen RCTs enrolling 1,322 participants were included; 90.9% of patients had variceal bleeding. Trial characteristics were heterogeneous: antibiotics used, durations (1–10 days), comparator arms (no prophylaxis or shorter courses), and definitions of infection varied. Overall study quality ranged from low to moderate; none reported adverse events attributable to antibiotics.
Primary outcome—All-cause mortality
– Shorter antibiotic durations (including omission of prophylaxis) had a 97.3% probability of being noninferior to longer (5–7 day) prophylaxis for all-cause mortality.
– Estimated risk difference (RD) for mortality was 0.9% favoring longer prophylaxis, but the 95% credible interval (95% CrI) crossed both harm and benefit (RD −2.6 to 4.9), consistent with no clear treatment effect.
Secondary outcomes—Early rebleeding and bacterial infections
– Early rebleeding: Shorter durations had a 73.8% probability of noninferiority for early rebleeding. Point estimate RD was 2.9% in favor of longer prophylaxis, but with a wide CrI (−4.2 to 10.0), limiting certainty.
– Bacterial infections: Trials reported more study-defined bacterial infections with shorter or absent prophylaxis (RD 15.2%; 95% CrI 5.0 to 25.9), a result that was statistically robust within the included trials.
Temporal subgroup (post-2004) findings
– In trials conducted or published after 2004, the probabilities that shorter durations were noninferior increased across outcomes (mortality, early rebleeding, infection), suggesting that contemporary care may attenuate the benefits historically ascribed to routine prolonged prophylaxis.
Interpretation of effect sizes
Although the pooled point estimates suggested small absolute differences in mortality and rebleeding favoring longer prophylaxis, credible intervals were wide and compatible with clinically unimportant effects. The increased rate of trial-defined infections with shorter prophylaxis was more consistent—but heterogeneity in how infections were defined, diagnosed, and adjudicated across older trials raises concern that detection bias or inconsistent ascertainment may have inflated this signal.
Study strengths and limitations
Strengths:
- Focus on randomized evidence with explicit noninferiority framing and prespecified clinical margin.
- Use of bayesian meta-analytic methods which provide direct probabilities of noninferiority and handle uncertainty in small or heterogeneous evidence sets.
- Temporal subgroup analysis acknowledging changes in standard care and supportive interventions.
Limitations:
- Most included trials were small, older, and at low-to-moderate quality by contemporary standards; many lacked blinding and had heterogeneous outcome definitions, particularly for bacterial infections.
- Antibiotic agents, local resistance patterns, and supportive care have changed since many trials were conducted; older trials often predate routine endoscopic band ligation, vasoactive drugs, and early TIPS strategies.
- No trials reported antibiotic adverse events, and none were large enough to reliably assess rare but important harms (e.g., Clostridioides difficile infection, selection for resistant organisms).
- The meta-analysis included trials comparing both shorter versus 5–7 day regimens and any prophylaxis versus none; pooling these comparisons increases clinical heterogeneity.
Expert commentary and clinical implications
Why might mortality not improve despite fewer infections with prophylaxis? If modern hemostatic and critical care reduce rebleeding and organ failure, then the incremental mortality benefit from preventing bacterial infections may be smaller than in earlier eras. The observed increase in reported infections with shorter prophylaxis is biologically plausible, but heterogeneity in detection and potential biases (open-label designs, varied diagnostic thresholds) limit confidence that these infections were clinically consequential.
Practical implications for clinicians:
- Interpret the results as provocative but not definitive. The meta-analysis challenges the evidence base underlying routine universal 5–7 day prophylaxis, but it does not by itself mandate immediate guideline changes.
- Individualize prophylaxis decisions. Consider routine short-course or selective prophylaxis for lower-risk patients (e.g., Child-Pugh A–B without active infection, prompt endoscopic control) while continuing standard 5–7 day regimens for higher-risk patients (e.g., severe cirrhosis, ongoing shock, uncontrolled bleeding, renal dysfunction, or documented infection at presentation).
- When antibiotics are used, choose agents in line with local resistance patterns. Many guidelines favor third-generation cephalosporins (e.g., ceftriaxone) over oral fluoroquinolones where quinolone resistance is common.
- Antimicrobial stewardship and monitoring: Be vigilant for antibiotic adverse events and resistant organisms. Document indications and planned duration clearly and reassess daily.
Policy and research implications:
- High-quality, contemporary RCTs are needed that randomize patients with cirrhosis and upper GI bleeding to well-defined antibiotic strategies (e.g., none, single-dose or 48–72 hour course, vs 5–7 days), stratified by bleeding source and severity, and incorporating modern endoscopic and rescue therapies.
- Trials should prespecify robust, objective infection definitions and collect data on antimicrobial resistance and harms (including C. difficile), and consider health-system outcomes such as length of stay and antibiotic exposure.
Conclusion
The 2025 bayesian meta-analysis by Prosty et al. questions the historic mortality benefit attributed to routine 5–7 day antibiotic prophylaxis after upper GI bleeding in patients with cirrhosis. While prophylaxis reduced the incidence of trial-defined infections, methodological concerns and heterogeneity limit confidence that this translates into mortality benefit in modern practice. Clinicians should balance the older guideline recommendations with evolving evidence, local microbiology, patient risk factors, and principles of antimicrobial stewardship. Definitive practice change will require contemporary, adequately powered randomized trials.
Funding and clinicaltrials.gov
See the original publication for detailed funding and disclosures: Prosty C, Noutsios D, Dubé LR, et al. Prophylactic Antibiotics for Upper Gastrointestinal Bleeding in Patients With Cirrhosis: A Systematic Review and Bayesian Meta-Analysis. JAMA Intern Med. 2025;185(10):1194-1203. doi:10.1001/jamainternmed.2025.3832.
References
1. Prosty C, Noutsios D, Dubé LR, Baden R, Davar K, Freling S, Bhuket T, Yee HF Jr, Spellberg B, McDonald EG, Lee TC. Prophylactic Antibiotics for Upper Gastrointestinal Bleeding in Patients With Cirrhosis: A Systematic Review and Bayesian Meta-Analysis. JAMA Intern Med. 2025 Oct 1;185(10):1194-1203. doi:10.1001/jamainternmed.2025.3832.
2. De Franchis R; Baveno VI Faculty. Expanding consensus in portal hypertension: Report of the Baveno VI Consensus Workshop: Stratifying risk and individualizing care for portal hypertension. Journal of Hepatology. 2015;63(3):743-752. doi:10.1016/j.jhep.2015.05.022.
3. European Association for the Study of the Liver (EASL). EASL Clinical Practice Guidelines: Management of decompensated cirrhosis. Journal of Hepatology. 2018;69(2):406-460. doi:10.1016/j.jhep.2018.03.024.
Note: Additional guideline documents from regional societies (AASLD, national guidelines) and local antimicrobial resistance data should be consulted when applying these findings in practice.
