Antibiotic Stewardship in Cirrhosis: Current Recommendations and Clinical Challenges

ABSTRACT

Bacterial infections are a major cause of mortality in cirrhosis, with antibiotic prophylaxis and empirical treatment embedded in international guidelines. However, the prevalence of multidrug‐resistant organisms and the epidemiology of infections have shifted considerably in recent decades. This review critically appraises the evidence underlying guideline recommendations on antibiotic prophylaxis for spontaneous bacterial peritonitis and acute variceal bleeding in cirrhosis, and antimicrobial management in acute liver failure. The evidence‐underlying current recommendations is limited and predates the current era of widespread antimicrobial resistance, challenging their applicability in current practice. Emerging evidence suggests that primary prophylaxis should be reconsidered, alternatives such as rifaximin warrant further evaluation against established fluoroquinolone regimens, and shorter antibiotic courses may be non‐inferior to standard durations. New double‐blind placebo‐controlled, non‐inferiority and Bayesian trials are needed to evaluate antibiotic regimens and microbiome‐ and barrier‐directed interventions, supported by improved diagnostics, while minimising treatment duration, toxicity and selection for antimicrobial resistance.

1. Introduction

Antibiotics are central to preventing and managing complications in cirrhosis, given the high incidence of bacterial infections and associated mortality [1, 2]. Susceptibility is driven by cirrhosis‐associated immune dysfunction, increased bacterial translocation and gut microbiome alterations [3, 4, 5]. In Europe, a 2025 meta‐analysis estimated that bacterial infections affect 38% of patients with cirrhosis, with a rising global trend over recent decades [6]. This trend is likely driven by improved chronic liver disease management leading to prolonged healthcare exposure, and increased use of invasive procedures [6, 7, 8].

In parallel, the epidemiology of infections has shifted, with an increasing prevalence of multidrug‐resistant organisms (MDROs) and changing infection sites, as infections other than spontaneous bacterial peritonitis (SBP) now account for most bacterial infections in cirrhosis [9, 10]. These include urinary tract infections (24.6%), respiratory tract infections (20%), bacteraemia (10.5%), and skin and soft‐tissue infections (7.2%) [10]. Causative organisms vary by infection site, with Enterococcus spp. commonly reported in urinary tract infections and Streptococcus spp. commonly reported in respiratory tract infections. Overall, Gram‐negative bacteria remain predominant, particularly Escherichia coli, whereas Gram‐positive organisms such as Enterococcus spp. are increasingly prevalent.

MDROs are defined as microorganisms with acquired non‐susceptibility to at least one agent in three or more antimicrobial categories [11]. Examples include Gram‐positive organisms such as methicillin‐resistant Staphylococcus aureus and vancomycin‐resistant Enterococcus faecium, and Gram‐negative organisms such as third‐generation cephalosporin‐resistant E. coli and carbapenem‐resistant Pseudomonas aeruginosa [12]. Worldwide, MDROs accounted for 34% of culture‐positive infections in patients with cirrhosis between 2015 and 2016, and were associated with increased mortality [9]. Repeated and prolonged antibiotic use drives MDRO emergence, underscoring the need for antimicrobial stewardship and reassessment of current prophylaxis and treatment strategies [13].

International guidelines consistently recommend antibiotic prophylaxis for primary and secondary prevention of SBP and acute variceal bleeding (AVB), although the latter represents pre‐emptive therapy [14, 15]. However, many recommendations are based on trials performed decades ago, before widespread antimicrobial resistance (AMR) and changing infection epidemiology, raising questions about their current applicability. Similar concerns apply to empirical antibiotic management in acute liver failure (ALF) [16]. Moreover, MDRO colonisation following antibiotic exposure is associated with increased mortality in liver transplant candidates and may complicate transplantation outcomes [17].

This review critically appraises the evidence underlying antibiotic prophylaxis in SBP and AVB in cirrhosis, examining whether the recent data support or challenge current recommendations. We also reassess the evidence for empirical antibiotic management in ALF.

2. Methods

Evidence underlying current guideline recommendations was identified from the latest international and national clinical practice guidelines, including the European Association for the Study of the Liver (EASL), American Association for the Study of Liver Diseases (AASLD), Baveno VI and VII consensus, British Society of Gastroenterology (BSG), National Institute for Health and Care Excellence (NICE), Asian Pacific Association for the Study of the Liver (APASL) and American College of Gastroenterology (ACG).

A targeted PubMed search identified articles up to April 2026, addressing antibiotic prophylaxis in SBP and AVB in cirrhosis, and empirical antibiotic management in ALF. Search terms included combinations of ‘cirrhosis’, ‘spontaneous bacterial peritonitis’, ‘variceal bleeding’, ‘acute liver failure’, ‘antibiotic’ and ‘prophylaxis’ alongside relevant MeSH terms. Priority was given to randomised controlled trials, meta‐analyses and large cohort studies. The search focussed on recent evidence that may support or challenge current strategies.

3. Results

3.1. Primary SBP Prophylaxis

Guidelines vary in recommendation strength regarding primary prophylaxis, reflecting uncertainty and limitations in the underlying evidence (Table 1) [14, 15, 21, 24, 26, 28]. Most guidelines recommend primary prophylaxis only for selected high‐risk patients, generally defined by low ascitic fluid protein (< 15 g/L [1.5 g/dL]) combined with advanced liver disease, renal dysfunction, or hyponatraemia (Supporting Information S1: Table S1) [14, 15, 26]. EASL and APASL recommend norfloxacin, while AASLD recommends ciprofloxacin or co‐trimoxazole, as norfloxacin is unavailable in the United States [14, 15, 26]. BSG recommends prophylaxis based on low ascitic fluid protein alone, without specifying a particular antibiotic regimen [24]. NICE takes the most restrictive approach, advising against routine prophylaxis while allowing consideration in severe liver disease, with antibiotic choice guided by local microbiological advice [28]. This approach reflects limited norfloxacin availability in the United Kingdom and fluoroquinolone restrictions related to adverse effects, AMR, and Clostridioides difficile infection [28, 31, 32]. Duration is rarely addressed, with only EASL and NICE explicitly recommending continuing prophylaxis until ascites resolves [14, 28].

TABLE 1.

Visual overview of guideline recommendations for antibiotic prophylaxis and empirical treatment.

Clinical setting and recommendation	EASL [14, 18]	AASLD* [15, 16, 19, 20]	Baveno VI, VII and VIII [21, 22, 23]	BSG [24, 25]	APASL [26, 27]	NICE [28, 29]	ACG** [30]
Primary SBP prophylaxis
Prophylaxis in selected high‐risk patients
Low ascitic fluid protein			Only in child‐pugh C
In combination with advanced liver disease, renal dysfunction, or hyponatraemia			Only child‐pugh C mentioned			Only severe liver disease mentioned
Norfloxacin recommended						Follow local microbiological advice
Ciprofloxacin or co‐trimoxazole listed as alternatives						Follow local microbiological advice
Continue until ascites resolves
Secondary SBP prophylaxis
Prophylaxis after previous SBP						Only considered in severe liver disease
Ciprofloxacin or co‐trimoxazole listed as alternatives						Follow local microbiological advice
Long‐term prophylaxis recommended			Discontinuation upon recompensation
AVB
Antibiotic prophylaxis recommended
Ceftriaxone preferred or recommended
Shorter course considered			Based on infection risk and cirrhosis severity
ALF
Microbiological surveillance recommended
Empirical antibiotics when infection is suspected or upon clinical deterioration
Routine antibiotic prophylaxis recommended

Note: Green indicates a recommendation by the guideline, orange indicates a conditional recommendation, red indicates a recommendation against, and grey indicates that the topic is not addressed. Dark shades indicate a strong recommendation, whereas lighter shades indicate a weak, ungraded* or conditional** recommendation.

Abbreviations: AASLD = American Association for the Study of Liver Diseases; ALF = Acute Liver Failure; APASL = Asian Pacific Association for the Study of the Liver; AVB = Acute Variceal Bleeding; BSG = British Society of Gastroenterology; EASL = European Association for the Study of the Liver; NICE = National Institute for Health and Care Excellence.

The key trial underlying primary prophylaxis recommendations is Fernandez et al. a placebo‐controlled trial with norfloxacin whose inclusion criteria largely shaped current prophylaxis criteria [33]. Among 68 patients, norfloxacin improved 3‐month survival (94% vs. 62%, p = 0.003), whereas the difference in 1‐year survival was borderline significant (60% vs. 48%, p = 0.05). However, mortality findings were inconsistent across the two meta‐analyses [34, 35]. Saab et al. found no significant effect (risk ratio [RR] 0.72, 95% confidence interval [CI] 0.48–1.06, p = 0.09), whereas Loomba et al. reported reduced mortality (odds ratio [OR] 0.60, 95% CI 0.37–0.97, p = 0.034), despite including the same four randomised trials. This discrepancy was attributed to Saab et al.’s use of additional unpublished data [36, 37]. Despite these inconsistencies, EASL provides a strong recommendation for primary prophylaxis in selected high‐risk patients [14].

AASLD also considered the NORFLOCIR trial by Moreau et al. (2018, n = 291), which compared norfloxacin with placebo in advanced cirrhosis, although < 5% had prior SBP [38]. This larger but underpowered trial did not confirm a survival benefit, with benefit suggested only in a post hoc subgroup with low ascitic fluid protein. A subsequent meta‐analysis by Facciorusso et al. suggested lower mortality with norfloxacin, although the effect remained borderline significant (RR 0.68, 95% CI 0.47–0.99, p = 0.04) [39]. Reflecting this uncertainty, AASLD states that primary prophylaxis may be considered in selected high‐risk patients [15]. APASL similarly recommends primary prophylaxis but grades the evidence as low based on the Cochrane meta‐analysis by Komolafe et al. [26, 40]. This meta‐analysis found no significant mortality benefit for any antibiotic regimen and graded the evidence as very low, without differences between primary and secondary prophylaxis.

Whether current guideline recommendations remain applicable is uncertain. Norfloxacin efficacy in preventing SBP has declined substantially over the past 3 decades, likely driven by rising MDROs [41, 42]. This prompted interest in alternatives to fluoroquinolones, including rifaximin and co‐trimoxazole [43, 44]. Rifaximin may reduce MDRO selection, given its local action in the gut with minimal systemic absorption [45]. It inhibits bacterial RNA polymerase, has broad‐spectrum activity against Gram‐positive and Gram‐negative bacteria, and modulates the gut microbiome [45, 46]. Established indications include traveller's diarrhoea and hepatic encephalopathy, with off‐label use for small intestinal bacterial overgrowth, and adverse events generally limited to mild nausea and respiratory or urinary tract infections [46, 47]. However, rifaximin may be selected for RNA polymerase mutations that mediate rifampicin resistance and cross‐resistance to daptomycin in vancomycin‐resistant E. faecium [48]. Although these mutations are not horizontally transferable, resistant strains may be selected rapidly and spread through clonal expansion [46, 48]. A large multinational cohort also found an almost two‐fold increased risk of AMR and multidrug resistance associated with rifaximin use [49].

Evidence for rifaximin in primary prophylaxis is mixed (Table 2) [43, 50, 51]. Assem et al. found alternating norfloxacin and rifaximin superior to norfloxacin alone, whereas Praharaj et al. found rifaximin and norfloxacin to be comparably effective in primary prophylaxis [43, 50]. However, Thévenot et al. found no 12‐month survival benefit with rifaximin over placebo or reduction in SBP or other liver‐related complications, although post hoc analysis suggested a survival benefit [51]. Similarly, the ASEPTIC trial (2025) found no benefit of co‐trimoxazole over placebo for survival or hospitalisation in 442 patients, even among patients with low ascitic fluid protein, leading the authors to argue against routine primary prophylaxis given AMR concerns [52, 58].

TABLE 2.

Overview of selected recent and ongoing trials of antibiotic prophylaxis.

Study	Design	Inclusion criteria	Intervention	Treatment duration	Infection outcome	Mortality
Primary SBP prophylaxis
Assem et al. [50]	RCT	Low ascitic protein level with advanced liver disease, renal impairment or hyponatraemia	Norfloxacin (n = 78) versus Rifaximin (n = 82) versus Monthly alternating rifaximin and norfloxacin (n = 79)	6 months	SBP a : 43.6% (NS) versus 31.7% (NS) versus 25.3% (p = 0.016 versus norfloxacin group)	Mortality a : 13.9% (NS) versus 9.8% (NS) versus 7.6% (NS)
Praharaj et al. [43]	RCT	Low ascitic protein level with advanced liver failure, renal impairment or hyponatraemia	Norfloxacin (n = 29) versus Rifaximin (n = 28)	6 months	SBP b : 24.1% versus 14.3% (NS)	Mortality b : 20.7% versus. 28.6% (NS)
Thévenot et al. [51]	RCT	Low ascitic protein level with severe liver failure, impaired renal function or hyponatraemia	Rifaximin (n = 72) versus Placebo (n = 82)	12 months	SBP*: 16.7% versus 22.5% (NS)	Mortality a : 38.9% versus 30% (NS)
O'Brien et al. [52]	RCT	Child‐pugh B and C with presence of ascites requiring diuretics or paracentesis	Co‐trimoxazole (n = 219) versus Placebo (n = 223)	18 months	—	Mortality: 46.1% versus 43.9% (NS)
Secondary SBP prophylaxis
Praharaj et al. [43]	RCT	Prior SBP	Norfloxacin (n = 33) versus Rifaximin (n = 26)	6 months	SBP b : 39.4% versus 7.7% (p = 0.004)	Mortality b : 36.4% versus 26.9% (NS)
NCT06234046 [53] (ongoing)	RCT	Recovered from an episode of SBP	Rifaximin (n = 40) versus Ciprofloxacin (n = 40)	6 months	—	—
NCT06199843 [54] (ongoing)	RCT	Prior SBP	Rifaximin (n = 140) versus Norfloxacin (n = 140)	12 months	—	—
AVB prophylaxis
Lee et al. [55]	RCT	Acute oesophageal variceal bleeding documented by endoscopic examination with variceal ligation	3 days (n = 38) versus 7 days (n = 33) of ceftriaxone	28 days	—	Mortality b : 0% versus 3% (NS)
Lo et al. [56]	Non‐inferiority trial	Presented with hematemesis and/or melena with confirmed acute gastroesophageal variceal haemorrhage	2 days (n = 52) versus 5 days (n = 48) of ceftriaxone and terlipressin	42 days	Bacterial infection b : 9.6% versus 14.5% (NS)	Mortality b : 1.9% versus 6.3% (NS)
Gupta et al. [57]	Non‐inferiority trial	Child‐pugh A with suspected variceal bleeding	5 days ceftriaxone (n = 86) versus No antibiotic treatment (n = 86)	6 weeks	Bacterial infection at day 5 a : 11.6% versus 7.0% (NS)	Mortality at day 5 a : 2.5% versus 0% (NS) Mortality at 6 weeks b : 2.5% versus 1.4% (NS)

Abbreviations: NS: not significant; RCT: Randomised Controlled Trial; SBP: Spontaneous Bacterial Peritonitis.

^a Based on (modified) Intention‐to‐Treat.

^b based on Per Protocol to Treat.

Overall, evidence supporting primary prophylaxis appears less robust than current guidelines suggest and predates the current era of widespread AMR. With ASEPTIC providing the first adequately powered trial in this setting, its negative findings warrant reconsideration of primary prophylaxis.

3.2. Secondary SBP Prophylaxis

Guidelines broadly recommend secondary prophylaxis after a prior SBP episode [14, 15, 21, 24, 26]. Norfloxacin is the standard regimen, with ciprofloxacin or co‐trimoxazole as alternatives if norfloxacin is unavailable (Table 1). AASLD and APASL recommend long‐term prophylaxis without specifying duration, whereas Baveno VIII recommends discontinuation upon recompensation [15, 22, 26].

Secondary prophylaxis is primarily supported by the 1990 trial by Ginés et al., which randomised 80 patients with prior SBP to norfloxacin or placebo and reduced 1‐year SBP recurrence from 68% to 20% (p = 0.0063) [59]. The superiority of norfloxacin over placebo was supported by the meta‐analysis by Mücke et al. which pooled predominantly older trials, although no clear superiority over alternative antibiotics was shown [42]. Ciprofloxacin and co‐trimoxazole are supported as alternatives by low‐quality evidence [24, 26].

Rifaximin appeared comparable with other antibiotics and in some meta‐analyses even more favourable, yet most guidelines await for further trials before recommending it [60, 61]. Praharaj et al. reported lower SBP recurrence with rifaximin than norfloxacin in the secondary prophylaxis subgroup (7.7% vs. 39.4%, p = 0.004) but no significant difference in mortality [43]. However, this subgroup analysis was limited by the small sample size and the trial was underpowered for mortality. While not directly addressing SBP prophylaxis, the LIVERHOPE trial (2025) found no reduction in the incidence of bacterial infections with rifaximin plus simvastatin compared with placebo in patients with decompensated cirrhosis (23.9% vs. 25.8%), although infection was a secondary outcome for which the trial was also not powered [62]. Further high‐quality trials are needed before rifaximin can be considered a valid alternative, with several registered trials currently evaluating its efficacy (Table 2) [53, 54].

More recent evidence on fluoroquinolone prophylaxis comes largely from cohort studies. Shetty et al. reported low SBP recurrence at 1 year, whereas Silvey et al. observed higher recurrence rates likely driven by fluoroquinolone resistance [63, 64]. Although studied in primary prophylaxis, Badal et al. (2023) reported fluoroquinolone resistance in 34% of patients receiving ciprofloxacin prophylaxis versus 14% in those without, with comparable findings for co‐trimoxazole (40% vs. 14%) [13]. Similarly, Mücke et al. reported secondary infections in 42.9% of patients during quinolone‐based prophylaxis in a mixed primary and secondary cohort, including SBP in 16.9% [41]. Quinolone‐resistant Gram‐negative bacteria were independently associated with prophylaxis failure. Regular MDRO screening has therefore been proposed to guide treatment and to re‐evaluate prophylaxis although evidence for alternative regimens remains limited.

While secondary prophylaxis is widely recommended as the standard of care, the evidence rests on a single placebo‐controlled trial from more than 30 years ago, with subsequent studies comparing different antibiotics rather than placebos. Uncertainties remain regarding optimal antibiotic choice, efficacy during widespread AMR, and prophylaxis duration.

3.3. AVB Prophylaxis

AVB mortality declined from 42.6% in 1980 to 14.5% by the 2000s, with studies in the 1990s showing improved survival with antibiotic prophylaxis by reducing infections and early rebleeding [65, 66, 67, 68]. Accordingly, guidelines uniformly recommend antibiotic prophylaxis (Table 1). Intravenous ceftriaxone remains the preferred agent due to its gram‐negative coverage, parenteral administration in critically ill patients, and increasing quinolone resistance [69]. However, recent European surveillance data show a 5.9% increase in third‐generation cephalosporin‐resistant E. coli bloodstream infections between 2019 and 2024, raising concerns about the long‐term efficacy of this strategy [12].

Baveno VIII recommends antibiotic prophylaxis in all patients presenting with AVB, while suggesting that duration and need may be individualised based on infection risk and cirrhosis severity [22]. Recommendations on optimal treatment duration have shifted from up to 7 days in older guidelines to shorter courses in recent AASLD and APASL guidelines [19, 27]. Since Baveno VI and the review by Fernandez et al. experts have called for trials omitting antibiotics in Child–Pugh A patients [22, 23, 70]. Crocombe proposed that patients with Child–Pugh score < 7 and successful endoscopic therapy may not require antibiotics, while shorter courses may suffice in higher‐risk patients without rebleeding [71]. Meta‐analyses by Prosty et al. and Bot et al. question the presumed mortality benefit of prolonged prophylaxis, even in patients with advanced liver disease (Child‐Pugh class B and C), with studies published after 2015 favouring shorter (2–3 days) or on‐demand strategies [55, 72, 73]. Supporting evidence includes a non‐inferiority study by Lo et al. a randomised controlled trial by Gupta et al. and several observational studies (Table 2) [56, 57, 74, 75, 76]. Together, these findings support a shift towards personalised, shorter antibiotic strategies, rendering older guideline recommendations on duration less applicable.

3.4. Management of ALF

Early recognition and treatment of infection are essential in ALF management, whereas antibiotic prophylaxis remains debated. All three guidelines recommend microbiological surveillance, including periodic cultures of sputum, urine and blood for bacterial and fungal organisms, although the optimal frequency remains unknown (Table 1) [18, 20, 30]. Recognition of bacterial infection in ALF is challenging, as clinical signs are frequently absent and C‐reactive protein and procalcitonin are often unhelpful [18, 20]. Hepatic encephalopathy at admission and a systemic inflammatory response syndrome score > 2 are predictors of bacteraemia [18]. Unexplained fever, leucocytosis, or deterioration in mental status should prompt investigation for infection. None of the guidelines provide a fixed antibiotic regimen or support routine prophylactic antibiotics in ALF.

Guidelines on empiric antibiotic treatment refer, directly or indirectly, to the prospective cohort study by Rolando et al. from the 1990s [77, 78]. This study evaluated 50 hospitalised patients with ALF, of whom 90% had microbiologically proven or clinical evidence of infection [77]. Despite antibiotic treatment, 30 patients died with proven infection documented in most cases. Although all guidelines acknowledge the limited and outdated evidence base, EASL strongly recommends empirical antibiotic therapy [18].

For antibiotic prophylaxis, EASL and AASLD refer to the 1993 trial by Rolando et al., with ACG referring to it indirectly [18, 30, 78, 79]. This trial randomised 104 patients with fulminant liver failure to a selective parenteral and enteral antimicrobial regimen (SPEAR), cefuroxime alone, or no initial antimicrobials [79]. Selective intestinal decontamination differs from systemic antibiotic prophylaxis, as it uses poorly absorbable oral antibiotics to reduce intestinal colonisation rather than systemic antibiotic exposure to prevent infection. SPEAR reduced infections compared with no prophylaxis (34% vs. 61%, p < 0.005) but offered no benefit over cefuroxime alone, suggesting this effect was driven by early systemic therapy rather than enteral decontamination. Be that as it may, overall survival did not differ between the groups. A retrospective study by Salmerón et al. found that selective intestinal decontamination reduced infections caused by enterobacteria, without improving survival [80]. EASL included a more recent retrospective cohort by Karvellas et al. of 1551 patients with ALF, in which antimicrobial prophylaxis did not reduce bloodstream infection (12.8% vs. 15.7%, p = 0.12) nor improve 21‐day survival (70% vs. 70%, p = 0.88) [81]. Given that the evidence is limited to a single trial and retrospective cohort data, all guidelines rated its quality as low.

Overall, evidence underlying ALF recommendations is limited, with recent data arguing against routine prophylaxis. In a related high‐risk population, a randomised controlled trial in patients with severe alcohol‐related hepatitis treated with prednisolone found that amoxicillin‐clavulanate did not improve 60‐day survival compared with placebo, despite reducing infections [82]. Moreover, multidrug‐resistant bacteria were more frequent in the antibiotic group. Similarly, current evidence is insufficient to recommend routine selective intestinal decontamination or systemic antibiotic prophylaxis in transplant candidates who may subsequently undergo immunosuppressive therapy [18, 20, 30]. Given the rarity and heterogeneity of ALF, and the risk of drug‐induced liver injury in severe liver dysfunction, high‐quality evidence supporting empirical antibiotic therapy remains lacking [83].

3.5. Future Perspectives

The evidence underlying current recommendations in cirrhosis is limited by methodological heterogeneity and small sample sizes across many studies, undermining the reliability of several meta‐analyses in this field. Many recommendations also rely on trials conducted 20–30 years ago, while rising MDRO prevalence challenges their effectiveness in current practice (Figure 1). This supports new double‐blind, placebo‐controlled randomised trials in contemporary cohorts to reassess established prophylactic strategies, such as norfloxacin‐based SBP prophylaxis, in the era of widespread AMR. Non‐inferiority trials may also be valuable, although they often require large sample sizes [84, 85]. A proposed approach uses the observed difference in efficacy rather than its confidence interval to reduce sample size, combined with predefined pharmacokinetic/pharmacodynamic parameters and quality of life [85]. Bayesian methods may further improve feasibility by incorporating previous trial data, to reduce sample size [86]. Future trials should also longitudinally assess the gut microbiome and resistome to determine how prophylactic regimens affect microbial composition and select for AMR genes [87, 88].

FIGURE 1.

Future perspectives for antibiotic prophylaxis and treatment in cirrhosis.

Beyond improved trial design, antimicrobial stewardship is essential to limit the rise of MDROs. Wider clinical adoption of rapid molecular diagnostics such as multiplex polymerase chain reaction syndromic tests, or whole‐genome sequencing, could identify pathogens and resistance genes simultaneously and enable more targeted antibiotic treatment [89, 90]. In ALF specifically, more reliable biomarkers are needed to distinguish bacterial infection from sterile inflammation [18, 30]. Presepsin, a soluble CD14 fragment released during innate immune activation, is a promising biomarker [91]. Higher presepsin concentrations have been reported in patients with sepsis compared with non‐infectious systemic inflammation [92]. Further research should determine its diagnostic value and optimal thresholds in ALF.

Improving vaccination coverage is another key strategy to reduce bacterial infections and antibiotic use in at‐risk populations, with pneumococcal vaccination rates reaching only 7.4% within the first year of chronic liver disease diagnosis [93]. The risk of invasive pneumococcal disease is 2–13‐fold higher in patients with chronic liver disease than in the general population [94]. Moreover, S. pneumoniae is a common cause of community‐acquired pneumonia and SBP in cirrhosis, supporting pneumococcal vaccination in this population [94, 95]. Beyond vaccination, shortening antibiotic treatment is another effective and low‐cost intervention, especially when improved trial designs as mentioned above are used. Collectively, these measures can strengthen antimicrobial stewardship in cirrhosis.

Even with optimal stewardship, alternatives like rifaximin may unintentionally drive cross‐resistance to daptomycin, underscoring the need for non‐antibiotic strategies [48]. Modulation of the gut microbiome and intestinal barrier may reduce AMR, bacterial translocation and infection risk. Novel strategies include faecal microbiota transplantation to reduce gut carriage of resistance genes, bacteriophage therapy to selectively target pathogenic strains, and microRNA‐based therapies to strengthen barrier function, potentially reducing infection burden and antibiotic use [96, 97].

4. Conclusion

Recommendations on antibiotic prophylaxis in cirrhosis and empirical treatment in acute liver failure rest on a limited evidence base that predates the current era of widespread antibiotic resistance, questioning their applicability in current practice. Emerging evidence suggests that primary prophylaxis in high‐risk patients should be reconsidered, alternative agents such as rifaximin warrant further evaluation against fluoroquinolone regimens, and shorter courses may be non‐inferior to standard durations. Robust evaluation will require new double‐blind, placebo‐controlled trials as well as non‐inferiority and Bayesian trial designs, supported by improved diagnostics, to evaluate antibiotic regimens and microbiome‐directed interventions that remain effective while minimising treatment duration, toxicity, and antibiotic resistance.

Funding

The authors have nothing to report.

Conflicts of Interest

S.E., K.K.I.M.B., and L.H.P.B. declare no competing interests. R.M.v.H. has received consultancy fees from Shionogi. A.O., M.A.v.A and J.M.v.H declare no competing interests. M.A.L. has received unrestricted research grants from the Jeroen Bosch Hospital (2019), the National Health Care Institute (Zorginstituut Nederland, 2020), the Dutch Digestive Foundation (MLDS, 2021), the Vaillant Fund (2022), the Dutch Gastroenterology Society (NVGE, 2023), the Amsterdam University Medical Centre (2024), and ZonMw (2025). M.A.L. has also received speaker fees from Gore (2024) and Takeda (2025), and an editorial board fee from Wiley (2026). All proceeds are paid to the institution.

Supporting information

Supporting Information S1

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.