Skip to main content

Foundational Trials

AntibioCor

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Prophylactic Antibiotics on Mortality in Severe Alcohol-Related Hepatitis: A Randomized Clinical Trial
  • Acronym: AntibioCOR-HAA (AntibioCor)
  • Year: 2023
  • Journal published in: JAMA
  • Citation: Louvet A, Labreuche J, Dao T, Thévenot T, Oberti F, Bureau C, et al. Effect of prophylactic antibiotics on mortality in severe alcohol-related hepatitis: a randomized clinical trial. JAMA. 2023;329(18):1558-1566.

Context & Rationale

  • Background
    • Severe alcohol-related hepatitis (AH) has high short-term mortality, with many deaths occurring within the first 2 months.
    • Prednisolone has been used to improve short-term outcomes in selected patients with severe AH, but benefit is modest and time-limited.34
    • Bacterial infection is common in severe AH (both at presentation and after corticosteroid initiation) and is strongly associated with mortality.12
    • Corticosteroid exposure appears to increase susceptibility to infection in severe AH, raising concern that infection prevention could be a key mediator of any survival benefit.2
    • Antibiotics might plausibly improve outcomes by preventing early infections and modulating gut-derived inflammatory signalling, but prolonged broad-spectrum therapy risks antimicrobial resistance and antibiotic-associated harms.
  • Research Question/Hypothesis
    • In adults with biopsy-proven severe AH (MELD ≥21) treated with prednisolone, does prophylactic amoxicillin/clavulanate for 30 days reduce 60-day all-cause mortality compared with placebo?
    • Mechanistic hypothesis: reducing early infections would translate into fewer complications (including hepatorenal syndrome) and improved survival.
  • Why This Matters
    • Severe AH frequently requires high-dependency/critical care support; infection-related decompensation is a major driver of deterioration.
    • Empiric “prophylactic” antibiotics are sometimes used in immunosuppressed cirrhosis/AH populations despite uncertain net benefit.
    • A definitive RCT addressing survival and harms is essential to balance any infection reduction against antimicrobial stewardship risks.

Design & Methods

  • Research Question: Among hospitalised adults with biopsy-proven severe alcohol-related hepatitis receiving prednisolone, does adding prophylactic amoxicillin/clavulanate for 30 days reduce 60-day all-cause mortality compared with placebo?
  • Study Type: Multicentre, randomised, double-blind, placebo-controlled, investigator-initiated trial (25 centres in France and Belgium; enrolment June 2015 to May 2019).
  • Population:
    • Setting: inpatient hepatology services (multicentre; international within Europe).
    • Key inclusion: age 18–75 years; jaundice onset <3 months; heavy alcohol consumption (>40 g/day women; >50 g/day men); transjugular liver biopsy confirming AH; Maddrey discriminant function >32; MELD score ≥21.
    • Key exclusions (examples): sepsis at randomisation; untreated infection/antibiotic cessation <7 days; type 1 hepatorenal syndrome before treatment; severe renal insufficiency; uncontrolled gastrointestinal bleeding; contraindication to prednisolone or beta-lactams.
  • Intervention:
    • Prednisolone 40 mg orally once daily for 30 days.
    • Amoxicillin 1 g + clavulanate 125 mg oral suspension three times daily for 30 days (total amoxicillin 3 g + clavulanate 375 mg per day).
  • Comparison:
    • Prednisolone 40 mg orally once daily for 30 days.
    • Matching placebo oral suspension three times daily for 30 days.
  • Blinding: Double-blind (participants, clinicians, investigators, and outcome assessors); indistinguishable placebo; central allocation.
  • Statistics: A total sample size of 280 was planned to detect an absolute 14% increase in 60-day survival (from 73% to 87%) with 80% power at a two-sided α of 0.05; inflated to 292 to allow ~5% loss to follow-up; primary analysis used an intention-to-treat approach (with exclusion of participants without consent) and time-to-event methods (Cox models for mortality; competing-risk methods for infection and hepatorenal syndrome).
  • Follow-Up Period: 180 days (scheduled assessments at days 7, 14, 21, 30, 45, 60, 90, and 180).

Key Results

This trial was not stopped early. No interim analyses were planned.

Outcome Amoxicillin/clavulanate + prednisolone Placebo + prednisolone Effect p value / 95% CI Notes
All-cause mortality at 60 days (primary) 24/142 (17.3%) 31/142 (21.9%) HR 0.77 95% CI 0.45 to 1.31; P=0.33 Between-group difference −4.7%; 95% CI −14.0 to 4.7
All-cause mortality at 90 days 29/142 (21.2%) 33/142 (23.9%) HR 0.87 95% CI 0.53 to 1.43; P=0.58 Between-group difference −2.8%; 95% CI −12.7 to 7.2
All-cause mortality at 180 days 43/142 (32.6%) 48/142 (35.5%) HR 0.89 95% CI 0.57 to 1.39; P=0.62 Between-group difference −2.9%; 95% CI −14.3 to 8.4
Incidence of infection at 60 days 42/142 (29.7%) 59/142 (41.5%) SHR 0.62 95% CI 0.41 to 0.91; P=0.02 Competing-risk analysis; between-group difference −11.8%; 95% CI −23.0 to −0.7
Incidence of infection at 0–30 days (post hoc) 19/142 (13.4%) 41/142 (28.9%) SHR 0.41 95% CI 0.24 to 0.70; P=0.001 On-treatment window; between-group difference −15.5%; 95% CI −24.9 to −6.1
Incidence of infection at 31–60 days (post hoc) 23/116 (20.0%) 18/95 (18.9%) SHR 1.06 95% CI 0.57 to 1.96; P=0.86 Off-treatment window; between-group difference 1.0%; 95% CI −9.8 to 11.9
Incidence of hepatorenal syndrome at 60 days 11/142 (7.7%) 12/142 (8.5%) SHR 0.91 95% CI 0.40 to 2.05; P=0.81 Between-group difference −0.7%; 95% CI −5.0 to 3.7
Response to corticosteroids at day 7 (Lille score <0.45) 80/141 (56.7%) 76/138 (55.1%) OR 1.03 95% CI 0.83 to 1.27; P=0.78 Between-group difference 1.7%; 95% CI −10.0 to 13.4
MELD score at 60 days 16.9 (4.8) 17.3 (6.2) Mean diff −0.41 95% CI −1.98 to 1.17; P=0.61 Mean (SD); n=100 vs n=91
Multidrug-resistant bacteria colonisation 29/91 (31.9%) 8/94 (8.5%) Not reported P=0.004 Safety surveillance sampling; denominators reflect available swabs
Serious adverse events 51/141 (36.2%) 57/143 (39.9%) Not reported P=0.52 As-treated safety population
  • Prophylactic amoxicillin/clavulanate reduced infections at 60 days (29.7% vs 41.5%; SHR 0.62; 95% CI 0.41 to 0.91; P=0.02), with the effect concentrated during the first 30 days (13.4% vs 28.9%; SHR 0.41; 95% CI 0.24 to 0.70; P=0.001).
  • No statistically significant reduction in 60-day mortality (17.3% vs 21.9%; HR 0.77; 95% CI 0.45 to 1.31; P=0.33), and no signal for benefit at 90 or 180 days.
  • Despite infection reduction, there was no improvement in hepatorenal syndrome, prednisolone response (Lille <0.45), or MELD at day 60, and multidrug-resistant colonisation was higher in the antibiotic group (31.9% vs 8.5%; P=0.004).

Internal Validity

  • Randomisation and Allocation: Centralised computer-generated randomisation (stratified by centre; block size of 4) with allocation concealment; double-blinding and matching placebo minimise selection and performance bias.
  • Drop out or exclusions: 292 randomised; 8 excluded from analysis due to absence/withdrawal of consent; analysed population 284 (142 per group); 15 did not receive allocated intervention (8 vs 7), creating potential dilution but limited impact on follow-up completeness.
  • Performance/Detection Bias: Mortality is objective and robust to ascertainment bias; infection ascertainment is more clinically mediated, but blinding reduces differential diagnosis and treatment escalation based on allocation.
  • Protocol Adherence: Completion of 30 days of study drug was 101/142 (71.1%) vs 86/142 (60.6%); completion of 30 days of prednisolone was 95/142 (66.9%) vs 99/142 (69.7%); clinician-assessed “good” adherence was 61.8% vs 63.3%, implying meaningful non-adherence in both arms that could attenuate effect size.
  • Baseline Characteristics: Groups were broadly comparable in severity: baseline MELD median (IQR) 24.9 (22.9–27.7) vs 25.8 (23.0–28.4); Maddrey score mean (SD) 69.0 (23.9) vs 71.2 (24.7); Glasgow Alcoholic Hepatitis Score median (IQR) 8.0 (8.0–9.0) vs 8.0 (8.0–9.0).
  • Heterogeneity: Multicentre recruitment with variable enrolment across centres; stratified randomisation mitigates (but does not eliminate) centre-level practice heterogeneity (e.g., infection screening intensity, antimicrobial prescribing, escalation thresholds).
  • Timing: Treatment required biopsy confirmation and randomisation in hospitalised patients; patients with recent antibiotic exposure required a washout, potentially shifting enrolment away from the earliest, highest-risk infectious window.
  • Dose: Amoxicillin/clavulanate was given at a high-dose oral regimen for 30 days (3 g/375 mg per day), long enough to influence early infection risk but also long enough to select antimicrobial resistance; prednisolone was delivered at standard dosing (40 mg daily for 30 days).
  • Separation of the Variable of Interest: Clear biological separation in early infection incidence (0–30 days: 13.4% vs 28.9%) that disappeared after antibiotic cessation (31–60 days: 20.0% vs 18.9%), supporting a time-limited prophylactic effect.
  • Key Delivery Aspects: Both arms received prednisolone, so the comparison isolates incremental prophylactic antibiotics; incomplete course completion (particularly for study drug) likely biases towards the null.
  • Crossover: Major deviations included receipt of non-allocated intervention (7 vs 6), introducing potential contamination; primary results were consistent in sensitivity analyses excluding major deviations.
  • Adjunctive therapy use: Not reported.
  • Outcome Assessment: Death was unambiguous; secondary outcomes (infection, hepatorenal syndrome) depended on clinical diagnosis and may vary by centre, but were analysed with competing-risk methods and were directionally coherent with the intervention window.
  • Statistical Rigor: Prespecified sample size was achieved and the primary endpoint analysis followed the superiority framework; secondary endpoints were treated as exploratory without multiplicity correction, increasing risk of false-positive secondary findings.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong, supported by rigorous randomisation, allocation concealment, and blinding with near-complete follow-up; limitations include incomplete adherence, some contamination, and exploratory handling of multiple secondary outcomes.

External Validity

  • Population Representativeness: Participants represent a high-risk subset of severe AH (MELD ≥21) confirmed by transjugular biopsy; this may be more selective than real-world practice where biopsy is not always performed.
  • Important exclusions: Patients with sepsis at randomisation, untreated infection, recent antibiotic exposure without washout, major renal dysfunction, or uncontrolled bleeding were excluded, limiting applicability to ICU patients with active infection and multi-organ failure.
  • Healthcare system context: Conducted in France/Belgium specialist centres; antimicrobial resistance ecology, infection screening protocols, and thresholds for ICU escalation may differ internationally.
  • Applicability: Findings are most applicable to prednisolone-eligible severe AH patients without active sepsis at the point of steroid initiation; extrapolation to other antibiotics, shorter/longer durations, or different stewardship environments is uncertain.

Conclusion on External Validity: Generalisability is moderate for prednisolone-treated, high-risk AH populations in specialist centres, but is limited for critically ill patients with uncontrolled infection, for settings without routine biopsy confirmation, and for regions with different resistance patterns.

Strengths & Limitations

  • Strengths:
    • Randomised, double-blind, placebo-controlled design with concealed allocation.
    • Clinically meaningful primary endpoint (60-day all-cause mortality) with extended follow-up to 180 days.
    • Focused on a biologically and prognostically enriched high-risk population (MELD ≥21), improving event rates and relevance.
    • Demonstrated on-treatment biological separation in infection rates, strengthening mechanistic inference.
  • Limitations:
    • Assumed a large survival benefit (absolute 14%); observed mortality was lower than expected, reducing power to detect smaller clinically relevant effects.
    • Incomplete adherence and some contamination (receipt of non-allocated intervention) could dilute true effects.
    • Infection outcomes are clinically adjudicated and may vary across centres despite blinding.
    • Secondary outcomes were exploratory without multiplicity correction; interpretation should prioritise effect sizes and consistency.
    • Higher multidrug-resistant colonisation in the antibiotic group raises important stewardship concerns.

Interpretation & Why It Matters

  • Clinical practice
    Routine prophylactic amoxicillin/clavulanate added to prednisolone in severe AH reduced early infections but did not improve survival, arguing against blanket prophylaxis as a mortality-modifying strategy in this population.
  • Mechanistic signal
    The dissociation between infection reduction (especially days 0–30) and unchanged mortality, hepatorenal syndrome, steroid response, and MELD suggests that preventing early bacterial infection alone is insufficient to alter the dominant trajectory of severe AH in many patients (e.g., liver failure, bleeding, systemic inflammation).
  • Antimicrobial stewardship
    The marked increase in multidrug-resistant colonisation (31.9% vs 8.5%) indicates potential downstream harm that may offset any short-term infection reduction and is highly relevant to ICU ecology and hospital-wide resistance pressure.
  • Trial design implications
    Future strategies may need precision targeting (e.g., biomarker- or risk-enriched prophylaxis), shorter antibiotic exposure, or alternative approaches (e.g., improved screening, early de-escalation algorithms, gut-directed therapies) to avoid resistance while addressing infection risk.

Controversies & Subsequent Evidence

  • Power and clinical plausibility: the planned detectable effect (absolute 14% improvement in 60-day survival) was large; contemporary commentary emphasised that smaller, more plausible benefits would require substantially larger trials, leaving uncertainty around modest benefit or harm despite a “negative” primary result.5
  • Infection reduction without survival benefit: observational data link infection tightly to mortality in severe AH and suggest corticosteroids increase infection susceptibility, yet AntibioCor showed that substantially fewer early infections did not translate into improved survival, challenging a simple causal model where infection prevention alone is sufficient to improve outcomes.12
  • Antibiotic choice and duration: the editorial highlighted the tension between preventing community-acquired infections and the risks of 30 days of broad-spectrum antibiotics (selection of multidrug resistance and collateral microbiome effects), which is reinforced by the observed colonisation signal in the trial.5
  • Hypothesis-generating subgroup: among patients with infection treated before randomisation, mortality was numerically lower with prophylaxis (HR 0.40; 95% CI 0.13 to 1.16), but the wide CI and lack of interaction significance mean this should not be used for practice decisions.
  • Subsequent synthesis: a 2024 systematic review and meta-analysis addressing prophylactic antibiotics in steroid-treated alcohol-associated hepatitis concluded that the evidence remains uncertain and does not establish a clear survival benefit, despite signals for infection reduction in some analyses.6
  • Broader evidence base: large trials and meta-analyses of corticosteroids in severe AH show limited, short-term benefit and emphasise infection as a key competing risk, contextualising why an infection-focused co-intervention may not yield mortality gains when liver failure dominates trajectories.34

Summary

  • In severe, biopsy-proven alcohol-related hepatitis with MELD ≥21 treated with prednisolone, 30 days of prophylactic amoxicillin/clavulanate did not reduce 60-day mortality (17.3% vs 21.9%; HR 0.77; 95% CI 0.45 to 1.31).
  • Prophylaxis reduced infection over 60 days (29.7% vs 41.5%; SHR 0.62; 95% CI 0.41 to 0.91), with benefit concentrated during the first 30 days of treatment.
  • There was no improvement in hepatorenal syndrome, prednisolone response (Lille <0.45), or MELD at day 60.
  • Multidrug-resistant colonisation was significantly higher in the antibiotic group (31.9% vs 8.5%).
  • The trial provides high-quality evidence that “blanket” prolonged antibiotic prophylaxis in steroid-treated severe AH is not a survival-modifying strategy and carries stewardship-relevant harms.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • AntibioCor is best interpreted as a negative mortality trial with a positive, time-limited infection signal and a clinically important antimicrobial resistance signal.

Overall Takeaway

AntibioCor provides high-quality evidence that routine 30-day prophylactic amoxicillin/clavulanate added to prednisolone in severe alcohol-related hepatitis reduces early infections but does not improve short-term or 6-month survival. Given the increased multidrug-resistant colonisation, the findings strongly caution against blanket antibiotic prophylaxis and reinforce prioritising targeted infection screening and treatment alongside judicious antimicrobial stewardship.

Overall Summary

  • Prophylactic amoxicillin/clavulanate reduced early infections but did not reduce mortality and increased multidrug-resistant colonisation in prednisolone-treated severe alcohol-related hepatitis.

Bibliography