ABSTRACT
-
Background/Aims
Infections with drug-resistant organisms (DROs) are associated with poor outcomes in cirrhosis. Rifaximin, widely used for hepatic encephalopathy (HE), could promote cross-resistance, but data regarding clinical impact are conflicting. Aim: Determine predictors of DROs in a global cirrhosis inpatient cohort focusing on preadmission rifaximin use.
-
Methods
From the global CLEARED consortium, we focused on cirrhosis inpatients with infections on/during admission. Clinical/demographic/medication, especially rifaximin details were recorded. The primary outcome was DRO development. Multivariable regression for DRO including clinical, medications, and country income was performed.
-
Results
2,949 infected inpatients (55.3 years, 62.9% male) were included. 12.2% of all and 24.4% of culture-positive infections developed DROs; these patients had higher HE (39 vs. 31%, P=0.003), hepatorenal syndrome (25 vs. 19%, P=0.006), lactulose (55 vs. 47%, P=0.008) and rifaximin use (34 vs. 27%, P=0.006) on crude comparisons but country-income distributions were similar. 29.7% were on pre-admission rifaximin, mostly HE-related; they had more advanced cirrhosis and from low/low-middle-income countries. Daptomycin was used in 1.5%, linked with DROs (5.0 vs. 1.0%, P<0.0001) without a difference in rifaximin use (1.4 vs.1.7%, P=0.61). On adjusted analysis, MELD-Na (1.03, 95% CI 1.02–1.04, P<0.001) increased, whereas male sex (0.73, 95% CI 0.58–0.92, P=0.008) and hepatitis-B (0.65, 95% CI 0.45–0.92, P=0.020) decreased DRO. Rifaximin was not associated with DROs overall (OR 1.07, 95% CI 0.80–1.43, P=0.65) or within income strata (high: 1.07, 95% CI 0.59–1.92, P=0.82, upper-middle: 1.18, 95% CI 0.74–1.85, P=0.49, low/low-middle: 1.04, 95% CI 0.60–1.81, P=0.90) despite sensitivity analyses.
-
Conclusions
In this large global cohort of hospitalized patients with cirrhosis and infections, 12% developed infections involving DROs. 30% had pre-admission rifaximin use which was not linked with daptomycin use or with DRO development on adjusted analysis overall or across country income groups.
-
Keywords: Antibiotic overuse; Disparities; CLEARED consortium; Infections; Daptomycin
Study Highlights
• Bullet point 1: Rifaximin is useful to prevent HE recurrence in cirrhosis. Recent pre-clinical analysis showed antimicrobial resistance emergence with rifaximin, but the clinical relevance is unclear.
• Bullet point 2: Analyzing 2949 cirrhosis patients hospitalized with infections in the global CLEARED Consortium across high/middle/low-income countries, DRO were found in 12.2% overall and in 24% of culture-positive infections, which increased mortality.
• Bullet point 3: 29% were on rifaximin, which did not affect the presence of DROs overall or within country-income strata on multi-variable analysis. Pre-clinical resistance emergence to rifaximin is likely a reflection of advanced cirrhosis and unlikely to be associated with DRO infections.
Graphical Abstract
INTRODUCTION
Patients with cirrhosis are at risk of developing infections, especially with drug-resistant organisms (DROs), which can lead to poor outcomes [
1–
3]. Acquisition and propagation of DROs is multifactorial and is worsened by the multiple contacts with health care systems, instrumentation, and exposure to antibiotics in patients with cirrhosis [
4]. These exposures and resistance patterns vary worldwide and the risk of DRO infections resulting from antibiotic prophylaxis for spontaneous bacterial peritonitis (SBP) has been described [
5–
7]. A recent study indicated that rifaximin used as treatment for hepatic encephalopathy (HE) causes cross-resistance to daptomycin in vancomycin-resistant
Enterococcus faecium [
8]. However, subsequent analyses have provided conflicting results, which is often complicated using focused database studies and short-term analyses [
9–
12]. A global cohort that provides perspective considering regional variations and resources would overcome the deficiencies of previous studies. We aimed to determine the predictors of DROs in a global cohort of patients with cirrhosis hospitalized with infections focusing on the relationship of resistant infections, to pre-admission rifaximin use and interaction with daptomycin.
MATERIALS AND METHODS
The CLEARED consortium consists of prospectively enrolled patients with cirrhosis admitted non-electively around the world [
13]. Cirrhosis details and history, demographics, initial inpatient course, and country income level using World Bank classifications: high-income countries (HICs), upper-middle-income countries (UMICs), and low- and lower-middle-income countries (L/LMICs) are collected. Inclusion criteria for CLEARED are confirmed cirrhosis, non-elective hospitalizations, and ability to obtain consent. For this manuscript, we only included patients from the CLEARED cohort with confirmed infection on or during admission using standard Infectious Diseases Society of America definitions [
14]. We excluded patients admitted electively, those with HIV, COVID-19 or prior transplants or those without any infections. The main outcome was the presence of a DRO, either on admission or during the hospital course; these were defined using standard criteria [
15]. We focused on fluoroquinolone resistance, carbapenemase expression, vancomycin resistant enterococci, and methicillin-resistant
Staphylococcus aureus (MRSA). The secondary outcome was the composite occurrence of either DRO or inpatient death. Other outcomes included death, liver transplant, nosocomial infection, and intensive care unit (ICU) transfer. Details of infections, rifaximin use, and inpatient and 30-day post-discharge course were collected.
Cohort characteristics around time of admission were summarized and compared between the those who were on rifaximin at admission and those who were not (
Supplementary Table 1). Similarly, we compared the admission characteristics between those who developed DROs during this admission. Multivariable logistic regression analysis was used to identify the effect of rifaximin on DRO development/composite inpatient death or DRO, adjusting for all other potential confounders that were significantly different between the outcome groups (at
P<0.05). Multicollinearity was measured using variance inflation factors (VIF) with a cutoff of 5 indicating high correlation between predictors and a cutoff of 2 indicating moderate correlation [
16]. Indication (HE, SBP, or neither) for rifaximin and treatment duration (number of months pre-admission that the patient was on rifaximin) were assessed for significance by adding these as predictors into the final multivariable model and assessing their effects. To address heterogeneity that may occur due to differences in country income level, we performed a stratified analysis of income distribution of countries separately [
17].
Multiple sensitivity analyses were conducted to address potential residual confounding that may have occurred when using this model-building approach or to address potential overlap between highly related predictors.
First, we performed a propensity score-matched (PSM) analysis of the cohort, where all variables shown in
Tables 1 and
2 (aside from outcomes and DRO-related data) were used to create balanced groups. The R package “MatchIt” [
18] was used, with optimal pair matching as the method of choice [
19]. The resulting cohort was then compared based on these variables (
Supplementary Table 2). We used logistic regression models in the matched cohort to assess the impact of rifaximin on the odds of both outcomes. Finally, we performed a double adjustment [
20], further including variables with
P<0.05 or standardized mean difference (SMD)>0.1 after matching into multivariable models (
Supplementary Table 3), which has been shown to remove further confounding that may occur even after PSM.
Secondly, we provided an alternative model-fitting approach that used backwards selection on the full predictor set (not simply those with P<0.05), returning parsimonious models for both outcomes. By including all potential predictors, this can identify combinations that may have been omitted when using statistical significance-based approaches alone.
Finally, we refitted the main multivariable models, which contained lactulose on admission & prior HE together, by first omitting lactulose (keeping prior HE in the model), and then the inverse of this (keeping lactulose while removing prior HE).
RESULTS
Of the 8,623 patients in the CLEARED consortium, 2,949 had an infection on or during admission, which was the cohort used for this analysis (
Supplementary Fig. 1 and
Supplementary Tables 4 and
5). Their mean age was 55.3±13.6 years, with the majority (62.9%) being men and alcohol being the most common cause (42%) of cirrhosis. Of these patients, the majority (54.9%) had an infection on admission, while the remainder had hospital-acquired or post-admission infections. In the six months prior to the index admission, 29.4% were admitted for an infection, and 8.8% had >2 prior infections. The in-hospital mortality rate was 20.4%, 30-days, mortality was 30.0%, 2.9% and 6.6% had liver transplants in-hospital and at 30-days respectively, and 26.6% were transferred to the ICU. The median hospital LOS was 12 days (interquartile range [IQR] 7–20). Almost a third (32.7%) were from high-income, 43% from upper middle and 24% were from low/low-middle income countries.
823 (29.7%) patients were on rifaximin treatment on admission. Most (81%) were on rifaximin for HE, while the remaining were on it for SBP prophylaxis. The majority (87%) were on the 550 mg BID dose, the rest were on th 400 mg TID dosage. The median duration of use (preadmission) was 2 months; 29% were started >6 months prior to admission, while 71% were started within 6 months. Patients on rifaximin were more likely to be younger, male, with alcohol- and MASH-related etiology of cirrhosis, and greater severity of disease, comorbidities and other admission medications suggestive of greater liver disease severity as measured by MELD-Na and cirrhosis history (
Table 1). Rifaximin use was highest in L/LMICs and lowest in UMICs.
In our analysis cohort, 361 (12.2%) had a DRO present and 837 (28.8%) had the composite outcome of DRO or death. Among the entire cohort, 1,478 had positive cultures (50.1%), of which 494 were gram-positive and 875 were gram-negative (the remainder were non-specified, fungal/viral were not included). This implies that the percentage of DROs represented 24.4% of the culture-positive infections. Specific DRO-related infection types included UTI (30.7%), spontaneous bacteremia (14.9%), respiratory tract infections (14.6%), SBP (13.6%), skin infections (4.9%), intra-abdominal infections (4.6%), and others (16.9%). Among these, women (39.0% vs. 24.7%; P=0.003) had significantly higher rates of UTI, men had higher rates of SBP (16.7% vs. 9.1%; P=0.04), and the remainder were not significantly different between genders or were too rare to compare. Of these DROs 13.0% were VRE related, 10% were MRSA related, 29.1% were fluroquinolone related, and 25.8% had carbapenemase resistance, the remainder were unspecified.
Patients with a DRO were and less likely to be male, less likely to have HBV or be on HBV antivirals and were more likely to have MASLD and hyperlipidemia (
Table 2). Patients with a DROs also had a worse MELD-Na on admission. These patients also had a higher proportion with history of prior AKI/HRS and HE with medications (lactulose, rifaximin) and a listing for LT. There was a trend towards higher prior infections and admission for infections and higher admission WBC count in those with infections who developed DROs.
Only 43 (1.5%) of patients were treated with daptomycin; their median (IQR) age was 59.00 [51.50, 67.50], 58.1% were men, and 46.5% had alcohol-related etiology of cirrhosis. 58.1% were from HIC
vs. UMI/L/LMICs which was significantly higher than the percentage of patients not on daptomycin (HIC 32.3%;
P<0.001). The rate of rifaximin use among these individuals was statistically similar to those not treated with daptomycin (1.4
vs. 1.7%;
P=0.61). Patients on daptomycin had significantly higher DRO emergence (5.0
vs. 1.0%;
P<0.0001), inpatient death (39.5%
vs. 20.1%;
P=0.003), and transplant (9.3%
vs. 2.8%;
P=0.036) rates, higher 30-day death (48.6%
vs. 29.7%;
P=0.021) and transplant (22.7%
vs. 6.4%;
P=0.009) rates, longer hospital LOS (median 16
vs. 12 days;
P=0.018), and were more likely to be transferred to the ICU (46.5%
vs. 26.3%;
P=0.005). While the number of patients receiving daptomycin was small, a post-hoc sensitivity analysis to determine the smallest difference that could be detected under our study design suggests that our study was powered to detect relatively small absolute differences in exposure rates, supporting the adequacy of the sample size (
Supplementary Results).
Variables affecting DRO development included male sex (adj. OR 0.73 [0.58–0.92];
P=0.008), HBV etiology (adj. OR 0.65 [0.45–0.92];
P=0.020), and higher MELD-Na (adj. OR 1.03 [1.02–1.04];
P<0.001). Rifaximin use was not significantly associated with DRO development (adj. OR 1.07 [0.80–1.43];
P=0.65) after adjustment for other covariates that significantly differed between the outcome groups (
Table 3). Indication and duration of rifaximin therapy did not impact DRO development. Variables affecting the composite outcome of DRO development or death were similar (
Table 3). Rifaximin use was not significantly associated with this outcome (adj. OR 1.24 [0.99–1.55],
P=0.07) after covariate adjustment. In all multivariable models, VIF were <2 in all scenarios, indicating an absence of multicollinearity.
In the propensity-score matched analysis, the matched cohort contained 679 patients per-group (1,358 total). Most variables were adequately balanced, but world bank income level, HBV etiology, prior ascites/VB/hyponatremia/AKI/transplant listing, lactulose, SBPPr, PPI, MELD-Na, and hemoglobin remained significant at
P<0.05 or had SMD>0.1 (
Supplementary Table 2). In the matched cohort, rifaximin usage at admission was not associated with DRO development (
P=0.758) which persisted after double adjustment (
P=0.918). The composite outcome of death/DRO was significantly higher in rifaximin patients in the matched cohort (
P=0.01), but after double adjustment this association was no longer significant (
P=0.253). A full multivariable model table can be seen in
Supplementary Table 3.
Re-fitting models with combinations of prior HE/lactulose did not affect the association between DRO development and rifaximin usage, but in the models that did not have both predictors included (
Supplementary Table 6), the composite outcome of inpatient death/DRO became significant.
Finally, using backwards selection to fit the multivariable models did not select rifaximin usage on admission as one of the final predictors included for both outcomes, indicating that this variable did not improve model fit. Specifically, the final multivariable model for DRO development contained only HBV etiology and MELD-Na, while the final multivariable model for the composite outcome contained age, income level, HBV etiology, prior ascites/HE/AKI, and MELD-Na.
Subgroup analysis of income level
On subgroup analysis of HICs, UMICs, and L/LMICs separately (
Supplementary Table 7), the lack of evidence of association between rifaximin and DRO development persisted. Rifaximin use was not associated with DRO isolation on multi-variable analysis in HIC (adj. OR 1.07 [0.59– 1.92];
P=0.82), UMIC (adj. OR 1.18 [0.74–1.85];
P=0.49), or L/LMICs (adj. OR 1.04 [0.60–1.81];
P=0.90). MELD-Na continued to be associated with DRO in UMIC (adj OR 1.05 [1.02–1.07];
P<0.001) and L/LMICs (adj OR 1.05 [1.02– 1.09];
P=0.002). In UMICs only HBV etiology (adj OR 0.50 [0.30–0.80];
P=0.005) and male sex (adj OR 0.67 [0.47– 0.95];
P=0.023) remained significantly associated with DRO emergence. Results were similar for the composite outcome.
DISCUSSION
In a globally representative cohort of prospectively enrolled hospitalized patients with cirrhosis with infections, 12% of total infections and 24% of culture-positive infections were due to DROs. Rifaximin use on admission was seen in almost 30% of patients, who tended to have worse cirrhosis characteristics. On adjusted analysis, rifaximin was not associated with DROs in the entire cohort and within country income levels.
The role of DROs and their prevention in cirrhosis is a critical issue since it can lead to poor outcomes [
6]. This was again demonstrated in our cohort where DRO isolation was linked with poor survival and a higher need for liver transplantation. Therefore, identifying endogenous and exogenous sources of DROs is critical to reduce this burden [
4]. Prior studies have shown that current/prior antibiotics, hospital exposures, and regional variations play a major role in DRO acquisition [
7,
21]. Our data show that higher disease severity, prior complications and associated medications such as rifaximin and lactulose, as well as a tendency towards higher infections and hospitalizations within 6 months were associated with DRO isolation. While these are expected in patients with cirrhosis, we aimed to study the impact of rifaximin on DRO development.
Rifaximin is an interesting non-absorbable antibiotic that can reduce ammonia generation even in the germ-free state [
22,
23]. The important role of rifaximin in HE recurrence prevention has been demonstrated multiple times without any real-world increase in either infections as a whole or DROs in particular [
24–
27]. Turner et al. [
8] performed preclinical and clinical assays showing daptomycin resistance in rifaximin users, but this was not extended towards clinical or DRO-related outcomes. Further studies using databases showed no short-term changes in infections, which were extended by in-depth patient-level analyses of antimicrobial resistance (AMR) genes [
9,
11,
12]. These data, combined with a prior trial of patients before and after rifaximin showed no major increase in AMR genes or resistant infections using rifaximin [
28]. A recent TriNetX study showed that patients with cirrhosis who were initiated on HE therapies showed a higher AMR rate and infections in those started on rifaximin versus not [
10]. However, databases often suffer from miscoding, matching issues, and uncertainty with outcomes. Additionally, subsequent large prospective trials with rifaximin in patients with advanced liver disease failed to show a signal related to infections or DROs [
29,
30].
Our data extends these prior analyses into a prospective global representative cohort which we isolated to only those who had infections. This reduces the risk of skewing by excluding non-infected individuals, in whom DROs would not be clinically sought. In our infected patient cohort, rifaximin was used in 30% of patients mostly for HE and majority of them started this within 6 months of the admission. As expected, rifaximin users were more advanced in their disease process compared to those who were not on it [
22]. And on crude comparisons, HE, lactulose and rifaximin were associated with DRO emergence. However, when controlled for all clinical variables, and country income levels, we did not find a significant association between rifaximin use on admission and isolation of DROs. This was consistent across country income levels. The results remained consistent even after multiple sensitivity analyses, such as inclusion of mortality, after propensity-score matching, and after analyzing HE diagnoses and HE therapies. Therefore, it is likely that the association of rifaximin with DROs is clinically a reflection of the patients in whom rifaximin is started rather than the rifaximin itself. Rifaximin has grade A evidence to reduce HE recurrence and a potentially small risk of increase in resistance that does not necessarily translate into infections or is clinically significant, and needs to be balanced against the higher risk of HE recurrence [
31–
33]. HE recurrence remains the most important preventable readmission among patients with cirrhosis, and the risk-benefit ratio of withholding rifaximin therapy should be carefully thought out [
34–
37].
To further investigate the findings of Turner et al. [
8], we focused on daptomycin and the potential for VRE, both of which are relatively rare in cirrhosis. In our current cohort 1.5% of patients were on daptomycin. Rifaximin use was equally distributed in patients with/without daptomycin use, and there was no statistically significant difference of VRE distribution with rifaximin. Daptomycin use was linked with higher DRO emergence and poor outcomes as expected, being a last-resort antibiotic. There was no consistent impact, and these daptomycin results mirroring previously published data in cirrhosis are likely not clinically significant [
11].
The DRO isolation rate depends on the awareness and availability of performing prompt cultures, which can vary worldwide [
7]. Approximately half of our infections were culture-positive, and a quarter had DROs. However, despite the limitations we found a similar rate of DROs across country income levels in this cohort. In addition, rifaximin use was higher in L/LMICs, which argues against potential lack of access to medications in these countries. Ultimately, we found no significant impact of rifaximin on DRO emergence regardless of country income levels. While higher MELD-Na is expected to be associated with more advanced liver disease and DROs, we found that male sex and HBV etiology were protective. The reasons behind these are unclear but these were driven by UMICs, especially China and Türkiye since they were also significant when isolated to these countries. The gender differences in sites of infection could explain this phenomenon partly as well. Since we restricted our cohort to only patients with infections, we did not find a major impact of SBP prophylaxis or PPIs on DROs. The specific impact of prior HE and AKI/HRS and associated medications on DROs is interesting since these have emerged as the major causes of admission in cirrhosis [
38]. Therefore, these are likely markers of hospital exposure rather than inherent risk factors [
15].
The representative nature of the data with a maximum of 100 subjects per site prevents skewing of the data from centers with higher resistance and is a strength. The granular nature of the data, uniform definitions, and central data capture with quality control across the CLEARED database is another strength. The limited number of patients per site (N=100) could reduce the generalizability with varying resources for infection management. Additionally, these were a risk of death occurring prior to DRO development, which could be addressed through the usage of competing risks regression. However, time-to-event data was not available in this cohort, which is a limitation.
In summary, in a prospectively enrolled inpatient cohort of infected patients with cirrhosis, DROs were isolated in 12% of infections, which did not vary significantly between country income levels. While prior HE and AKI, and lactulose and rifaximin use were significantly higher on crude analysis for DRO emergence, these were not significant on multivariable analysis. We conclude that rifaximin use was not associated with DRO emergence in this global cohort of inpatients with cirrhosis.
FOOTNOTES
-
Authors’ contribution
JSB conceptualized the study question, SS was involved in statistical analysis, BJB was responsible for data quality and curation, AC, PSK, FW, QX, RI, MT, WKS, AT, PH, JG, and MAS are part of the Steering committee for CLEARED who were involved in critical revisions and study conduct. All other investigators were involved in study conduct.
-
Acknowledgements
Partly supported by VA Merit Review I01CX002472, NIH NCATS UM1TR004360, and an investigator-initiated grant from Bausch. None of the funders had any role in the research design, conduct, or decision to publish. AI was used to create the graphical abstract using Notebook LLM. There are no other uses of AI in the manuscript.
-
Conflicts of Interest
JSB’s institution receives grant support from Bausch. None for any other collaborator.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Table 5.
CLEARED investigators (other than those on the cover) with emails for those with 20 or more subjects per site who need to be on the PubMed listing
cmh-2026-0228-Supplementary-Table-5.pdf
Table 1Cohort characteristics split by rifaximin on admission
Table 1
|
Characteristic |
Overall (n=2,949) |
No rifaximin (n=2,126) |
On rifaximin (n=823) |
P-value |
|
Age (yr) |
56.00 [46.00, 65.00] |
56.00 [46.00, 66.00] |
55.00 [45.00, 64.00] |
0.026 |
|
Male sex |
1,854 (62.9) |
1,293 (60.8) |
561 (68.2) |
<0.001 |
|
World Bank income group classification |
|
|
|
<0.001 |
|
High |
964 (32.7) |
704 (33.1) |
260 (31.6) |
|
|
Upper middle |
1,278 (43.3) |
1,062 (50.0) |
216 (26.2) |
|
|
Low/low-middle |
707 (24.0) |
360 (16.9) |
347 (42.2) |
|
|
Etiology |
|
Alcohol |
1,240 (42.0) |
856 (40.3) |
384 (46.7) |
0.002 |
|
MASLD |
542 (18.4) |
358 (16.8) |
184 (22.4) |
0.001 |
|
Hepatitis B |
516 (17.5) |
433 (20.4) |
83 (10.1) |
<0.001 |
|
Hepatitis C |
318 (10.8) |
224 (10.5) |
94 (11.4) |
0.529 |
|
Auto-immune liver diseases |
174 (5.9) |
122 (5.7) |
52 (6.3) |
0.608 |
|
Cryptogenic |
220 (7.5) |
164 (7.7) |
56 (6.8) |
0.444 |
|
Other |
112 (3.8) |
84 (4.0) |
28 (3.4) |
0.554 |
|
Comorbidities |
|
Diabetes |
854 (29.0) |
583 (27.4) |
271 (32.9) |
0.004 |
|
Hypertension |
721 (24.4) |
533 (25.1) |
188 (22.8) |
0.225 |
|
Hyperlipidemia |
350 (11.9) |
252 (11.9) |
98 (11.9) |
>0.999 |
|
Cirrhosis related history |
|
Prior ascites |
2,024 (68.6) |
1,352 (63.6) |
672 (81.7) |
<0.001 |
|
Prior variceal bleed |
774 (26.2) |
492 (23.1) |
282 (34.3) |
<0.001 |
|
Prior hepatic encephalopathy |
938 (31.8) |
414 (19.5) |
524 (63.7) |
<0.001 |
|
Prior hyponatremia |
576 (19.5) |
306 (14.4) |
270 (32.8) |
<0.001 |
|
Prior acute kidney injury/HRS |
573 (19.4) |
302 (14.2) |
271 (32.9) |
<0.001 |
|
Hospitalized in past 6 mo |
1,458 (49.4) |
959 (45.1) |
499 (60.6) |
<0.001 |
|
Infections in past 6 mo |
867 (29.4) |
556 (26.2) |
311 (37.8) |
<0.001 |
|
Prior listing for liver transplant |
363 (12.3) |
191 (9.0) |
172 (20.9) |
<0.001 |
|
Medications on admission |
|
Β-blockers |
958 (32.5) |
622 (29.3) |
336 (40.8) |
<0.001 |
|
Diuretics |
1,626 (55.1) |
1,074 (50.5) |
552 (67.1) |
<0.001 |
|
Lactulose |
1,429 (48.5) |
698 (32.8) |
731 (88.8) |
<0.001 |
|
SBP prophylaxis |
412 (14.0) |
199 (9.4) |
213 (25.9) |
<0.001 |
|
Statins |
267 (9.1) |
191 (9.0) |
76 (9.2) |
0.888 |
|
Proton-pump inhibitors |
1,293 (43.8) |
814 (38.3) |
479 (58.2) |
<0.001 |
|
HBV antivirals |
472 (16.0) |
365 (17.2) |
107 (13.0) |
0.007 |
|
Admission reasons |
|
Infection |
1,618 (54.9) |
1,125 (52.9) |
493 (59.9) |
0.001 |
|
Liver related |
2,553 (86.6) |
1,808 (85.0) |
745 (90.5) |
<0.001 |
|
Non-liver related |
92 (3.1) |
76 (3.6) |
16 (1.9) |
0.03 |
|
Admission labs |
|
Hemoglobin |
10.00 [8.30, 11.71] |
10.20 [8.40, 12.00] |
9.40 [8.00, 10.97] |
<0.001 |
|
WBC |
7.50 [4.57, 11.80] |
7.30 [4.50, 11.60] |
8.00 [4.89, 12.50] |
0.028 |
|
INR |
1.61 [1.31, 2.10] |
1.55 [1.30, 2.00] |
1.82 [1.44, 2.40] |
<0.001 |
|
Sodium |
134.0 [129.0, 137.9] |
134.0 [130.0, 138.0] |
132.0 [128.0, 137.0] |
<0.001 |
|
Creatinine |
1.02 [0.71, 1.70] |
0.98 [0.70, 1.57] |
1.20 [0.80, 1.99] |
<0.001 |
|
Bilirubin |
3.81 [1.58, 10.70] |
3.40 [1.41, 9.86] |
4.94 [1.95, 12.73] |
<0.001 |
|
Albumin |
2.70 [2.30, 3.20] |
2.70 [2.30, 3.20] |
2.70 [2.30, 3.10] |
0.009 |
|
MELD-Na |
23.00 [17.00, 29.00] |
22.00 [16.00, 28.00] |
26.00 [20.00, 32.00] |
<0.001 |
|
DRO details |
|
Any DRO |
361 (12.2) |
238 (11.2) |
123 (14.9) |
0.006 |
|
VRE-related DRO |
47 (1.6) |
28 (1.3) |
19 (2.3) |
0.078 |
|
MRSA-related DRO |
36 (1.2) |
26 (1.2) |
10 (1.2) |
>0.999 |
|
Fluoroquinolone DRO |
105 (3.6) |
76 (3.6) |
29 (3.5) |
>0.999 |
|
Carbapenemase DRO |
93 (3.2) |
51 (2.4) |
42 (5.1) |
<0.001 |
|
Outcomes |
|
In-hospital mortality |
591 (20.4) |
354 (16.9) |
237 (29.5) |
<0.001 |
|
Mortality or hospice discharge |
641 (22.1) |
399 (19.1) |
242 (30.1) |
<0.001 |
|
In-hospital transplant |
82 (2.9) |
53 (2.5) |
29 (3.7) |
0.141 |
|
ICU transfer |
784 (26.6) |
476 (22.4) |
308 (37.4) |
<0.001 |
|
Hospital LOS |
12.00 [7.00, 20.00] |
12.00 [7.00, 21.00] |
11.00 [6.00, 19.50] |
0.01 |
|
30-Day readmission |
562 (29.4) |
412 (28.9) |
150 (31.1) |
0.375 |
|
30-Day mortality |
745 (30.0) |
475 (26.9) |
270 (37.7) |
<0.001 |
|
30-Day transplant |
126 (6.6) |
78 (5.5) |
48 (10.0) |
0.001 |
|
Composite, death or DRO |
837 (28.8) |
492 (23.8) |
316 (37.7) |
<0.001 |
Table 2Cohort characteristics split by DRO development
Table 2
|
Characteristic |
No DRO (n=2,588) |
DRO (n=361) |
P-value |
|
Age (yr) |
56.00 [46.00, 65.00] |
55.00 [47.00, 64.00] |
0.922 |
|
Male sex |
1,650 (63.8) |
204 (56.5) |
0.009 |
|
World Bank income group classification |
|
|
0.418 |
|
High |
857 (33.1) |
107 (29.6) |
|
|
Upper middle |
1,114 (43.0) |
164 (45.4) |
|
|
Low/low middle income |
617 (23.8) |
90 (24.9) |
|
|
Etiology |
|
Alcohol use |
1,093 (42.2) |
147 (40.7) |
0.625 |
|
MASLD |
463 (17.9) |
79 (21.9) |
0.078 |
|
Hepatitis B |
477 (18.4) |
39 (10.8) |
<0.001 |
|
Hepatitis C |
289 (11.2) |
29 (8.0) |
0.088 |
|
Auto-immune liver diseases |
145 (5.6) |
29 (8.0) |
0.086 |
|
Cryptogenic |
190 (7.3) |
30 (8.3) |
0.583 |
|
Other |
96 (3.7) |
16 (4.4) |
0.599 |
|
Comorbidities |
|
Diabetes |
737 (28.5) |
117 (32.4) |
0.139 |
|
Hypertension |
630 (24.3) |
91 (25.2) |
0.77 |
|
Hyperlipidemia |
289 (11.2) |
61 (16.9) |
0.002 |
|
Cirrhosis related history |
|
Prior ascites |
1,776 (68.6) |
248 (68.7) |
>0.999 |
|
Prior variceal bleed |
683 (26.4) |
91 (25.2) |
0.678 |
|
Prior overt hepatic encephalopathy |
798 (30.8) |
140 (38.8) |
0.003 |
|
Prior hyponatremia |
494 (19.1) |
82 (22.7) |
0.119 |
|
Prior acute kidney injury/HRS |
483 (18.7) |
90 (24.9) |
0.006 |
|
Hospitalized in past 6 mo |
1,266 (48.9) |
192 (53.2) |
0.143 |
|
Infections in past 6 mo |
747 (28.9) |
120 (33.2) |
0.099 |
|
Prior listing for liver transplant |
305 (11.8) |
58 (16.1) |
0.025 |
|
Medications on admission |
|
Β-blockers |
848 (32.8) |
110 (30.5) |
0.416 |
|
Diuretics |
1,423 (55.0) |
203 (56.2) |
0.696 |
|
Lactulose |
1,230 (47.5) |
199 (55.1) |
0.008 |
|
Rifaximin |
700 (27.0) |
123 (34.1) |
0.006 |
|
SBP prophylaxis |
366 (14.1) |
46 (12.7) |
0.524 |
|
Statins |
229 (8.8) |
38 (10.5) |
0.346 |
|
Proton-pump inhibitors |
1,143 (44.2) |
150 (41.6) |
0.378 |
|
HBV antivirals |
442 (17.1) |
30 (8.3) |
<0.001 |
|
Admission reasons |
|
Infection admission |
1,437 (55.5) |
181 (50.1) |
0.061 |
|
Liver related admission |
2,243 (86.7) |
310 (85.9) |
0.739 |
|
Non-liver related admission |
75 (2.9) |
17 (4.7) |
0.091 |
|
Admission labs |
|
Hemoglobin |
10.00 [8.30, 11.80] |
9.60 [8.00, 11.30] |
0.012 |
|
WBC |
7.33 [4.50, 11.62] |
8.42 [4.90, 12.90] |
0.067 |
|
INR |
1.60 [1.30, 2.10] |
1.80 [1.42, 2.30] |
<0.001 |
|
Sodium |
134.0 [129.0, 138.0] |
134.0 [129.0, 137.0] |
0.542 |
|
Creatinine |
1.00 [0.71, 1.66] |
1.18 [0.76, 2.00] |
0.001 |
|
Bilirubin |
7.59 (9.08) |
9.36 (11.59) |
0.001 |
|
Albumin |
2.75 (0.67) |
2.71 (0.69) |
0.288 |
|
MELD-Na |
22.98 (8.27) |
25.10 (7.79) |
<0.001 |
|
Outcomes |
|
In-hospital mortality |
476 (18.7) |
115 (32.4) |
<0.001 |
|
Mortality or hospice discharge |
517 (20.3) |
124 (34.9) |
<0.001 |
|
In-hospital transplant |
56 (2.2) |
26 (7.4) |
<0.001 |
|
ICU transfer |
635 (24.5) |
149 (41.3) |
<0.001 |
|
Hospital LOS |
11.00 [7.00, 19.00] |
16.00 [9.00, 28.00] |
<0.001 |
|
30-Day readmission |
495 (29.1) |
67 (32.4) |
0.366 |
|
30-Day mortality |
606 (28.0) |
139 (44.0) |
<0.001 |
|
30-Day transplant |
93 (5.5) |
33 (15.6) |
<0.001 |
Table 3Multivariable logistic regression model for association with DRO development and composite risk of death/DRO
Table 3
|
Variable*
|
OR (95% CI) |
P-value |
|
Outcome: DRO development |
|
Rifaximin on admission |
1.07 (0.80–1.43) |
0.649 |
|
Male sex |
0.73 (0.58–0.92) |
0.008 |
|
HBV etiology |
0.65 (0.45–0.92) |
0.020 |
|
MELD-Na |
1.03 (1.01–1.04) |
<0.001 |
|
Outcome: Composite death or DRO |
|
Rifaximin on admission |
1.24 (0.99–1.55) |
0.07 |
|
Male sex |
0.81 (0.68–0.98) |
0.028 |
|
HBV etiology |
0.59 (0.45–0.76) |
<0.001 |
|
MELD-Na |
1.08 (1.07–1.10) |
<0.001 |
Abbreviations
low/low-middle income countries
methicillin-resistant Staphylococcus aureus; PSM, propensity score-matched
spontaneous bacterial peritonitis
standardized mean difference
upper-middle income country
variance inflation factors
REFERENCES
- 1. Piano S, Brocca A, Mareso S, Angeli P. Infections complicating cirrhosis. Liver Int 2018;38:Suppl 1. 126-133.
- 2. Cai J, Yip TC, Wong GL, Wong VW, Lui GC, Sun J, et al. Microbiological profile of bacterial infections in patients with cirrhosis over time: a 20-year territory-wide study. Clin Gastroenterol Hepatol 2026;24:150-160.
- 3. Jeong JY, Cho JY, Kang D, Kim JH, Park SJ, Joo EJ, et al. Impact of serious infection on mortality of patients with chronic liver disease: a nationwide population-based cohort study. Clin Gastroenterol Hepatol 2026;24:443-452e6.
- 4. Bajaj JS, Kamath PS, Reddy KR. The evolving challenge of infections in cirrhosis. N Engl J Med 2021;384:2317-2330.
- 5. Silvey S, Patel N, Tsai SY, Nadeem M, Sterling RK, Markley JD, et al. Higher rate of spontaneous bacterial peritonitis recurrence with secondary spontaneous bacterial peritonitis prophylaxis compared with no prophylaxis in 2 national cirrhosis cohorts. Am J Gastroenterol 2025;120:1066-1075.
- 6. Piano S, Singh V, Caraceni P, Maiwall R, Alessandria C, Fernandez J, et al. Epidemiology and effects of bacterial infections in patients with cirrhosis worldwide. Gastroenterology 2019;156:1368-1380e10.
- 7. Cao Z, Wong F, Choudhury AK, Kamath PS, Topazian M, Torre A, et al. Global prevalence and characteristics of infections and clinical outcomes in hospitalised patients with cirrhosis: a prospective cohort study for the CLEARED Consortium. Lancet Gastroenterol Hepatol 2024;9:997-1009.
- 8. Turner AM, Li L, Monk IR, Lee JYH, Ingle DJ, Portelli S, et al. Rifaximin prophylaxis causes resistance to the last-resort antibiotic daptomycin. Nature 2024;635:969-977.
- 9. Deng ZL, Safaei N, Schütte SL, Ohlendorf V, Maasoumy B, McHardy AC. High prevalence and local dissemination of daptomycin-resistance mutations for Enterococcus faecium in cirrhotic patients. Gastroenterology 2026;170:228-230.
- 10. Kuo CH, Carter GP, Howden BP, Huang TW, Cao Z, Yeo YH, et al. Increased risk of antimicrobial resistance in patients with cirrhosis and hepatic encephalopathy using rifaximin. Nat Commun 2025;17:674.
- 11. Silvey S, Patel N, Khoruts A, Bajaj JS. Rifaximin does not increase the rate of 30-day mortality in patients with cirrhosis and daptomycin in two national US-based cohorts. Gastroenterology 2025;169:349-351e3.
- 12. Yu X, Jin Y, Zhou W, Xiao T, Wu Z, Su J, et al. Rifaximin modulates the gut microbiota to prevent hepatic encephalopathy in liver cirrhosis without impacting the resistome. Front Cell Infect Microbiol 2022;11:761192.
- 13. Bajaj JS, Choudhury AK, Xie Q, Kamath PS, Topazian M, Hayes PC, et al. Global disparities in mortality and liver transplantation in hospitalised patients with cirrhosis: a prospective cohort study for the CLEARED Consortium. Lancet Gastroenterol Hepatol 2023;8:611-622.
- 14. Bajaj JS, O’Leary JG, Reddy KR, Wong F, Olson JC, Subramanian RM, et al. Second infections independently increase mortality in hospitalized patients with cirrhosis: the North American consortium for the study of end-stage liver disease (NACSELD) experience. Hepatology 2012;56:2328-2335.
- 15. Centers for Disease Control and Prevention (CDC). Background of the Management of Multidrug-Resistant Organisms in Healthcare Settings 2006 guideline. CDC web site. <https://www.cdc.gov/infection-control/hcp/mdro-management/background.html>. Accessed 1 January 2026.
- 16. Marcoulides KM, Raykov T. Evaluation of variance inflation factors in regression models using latent variable modeling methods. Educ Psychol Meas 2019;79:874-882.
- 17. Metreau E, Young KE, Eapen SG. World Bank country classifications by income level for 2024–2025. World Bank Group web site. <https://blogs.worldbank.org/en/opendata/world-bank-country-classifications-by-income-level-for-2024-2025>. Accessed 1 January 2026.
- 18. Ho DE, Imai K, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal 2007;15:199-236.
- 19. Hansen BB, Klopfer SO. Optimal full matching and related designs via network flows. J Comput Graph Stat 2006;15:609-627.
- 20. Nguyen TL, Collins GS, Spence J, Daurès JP, Devereaux PJ, Landais P, et al. Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance. BMC Med Res Methodol 2017;17:78.
- 21. Markley JD, Bajaj JS. Rethinking antibiotic prophylaxis for spontaneous bacterial peritonitis in patients with cirrhosis: first, do no harm. Clin Infect Dis 2025;80:710-714.
- 22. Caraceni P, Vargas V, Solà E, Alessandria C, de Wit K, Trebicka J, et al. The use of rifaximin in patients with cirrhosis. Hepatology 2021;74:1660-1673.
- 23. Kang DJ, Kakiyama G, Betrapally NS, Herzog J, Nittono H, Hylemon PB, et al. Rifaximin exerts beneficial effects independent of its ability to alter microbiota composition. Clin Transl Gastroenterol 2016;7:e187.
- 24. Mullen KD, Sanyal AJ, Bass NM, Poordad FF, Sheikh MY, Frederick RT, et al. Rifaximin is safe and well tolerated for long-term maintenance of remission from overt hepatic encephalopathy. Clin Gastroenterol Hepatol 2014;12:1390-1397e2.
- 25. Chang C, Huang CH, Tseng HJ, Yang FC, Chien RN. Realworld experience of the one-year efficacy of rifaximin add-on to lactulose is superior to lactulose alone in patients with cirrhosis complicated with recurrent hepatic encephalopathy in Taiwan. J Pers Med 2021;11:478.
- 26. Kawaratani H, Namisaki T, Kondo Y, Tatsumi R, Kawabe N, Tanabe N, et al. Real-world setting of efficacy and safety of 3 years of rifaximin administration in Japanese patients with hepatic encephalopathy: a multicenter retrospective study. J Clin Med 2025;14:1358.
- 27. Jesudian AB, Gagnon-Sanschagrin P, Bungay R, Easson K, Bellefleur R, Guérin A, et al. Rifaximin reduces rehospitalization risk in patients with cirrhosis and overt hepatic encephalopathy. Hepatol Commun 2026;10:e0874.
- 28. Shamsaddini A, Gillevet PM, Acharya C, Fagan A, Gavis E, Sikaroodi M, et al. Impact of antibiotic resistance genes in gut microbiome of patients with cirrhosis. Gastroenterology 2021;161:508-521e7.
- 29. Thévenot T, Elkrief L, Bureau C, Bardou-Jacquet E, Rosa I, Nguyen-Khac E, et al. Effect of rifaximin in patients with severe cirrhosis and ascites: a randomized double-blind placebo-controlled trial. J Hepatol 2025;83:1320-1327.
- 30. Pose E, Jiménez C, Zaccherini G, Campion D, Piano S, Uschner FE, et al. Simvastatin and rifaximin in decompensated cirrhosis: a randomized clinical trial. JAMA 2025;333:864-874.
- 31. Doern GV, Brecher SM. The clinical predictive value (or lack thereof) of the results of in vitro antimicrobial susceptibility tests. J Clin Microbiol 2011;49:S11-S14.
- 32. European Association for the Study of the Liver. EASL Clinical Practice Guidelines on the management of hepatic encephalopathy. J Hepatol 2022;77:807-824.
- 33. Vilstrup H, Amodio P, Bajaj J, Cordoba J, Ferenci P, Mullen KD, et al. Hepatic encephalopathy in chronic liver disease: 2014 Practice Guideline by the American Association for the Study of Liver Diseases and the European Association for the Study of the Liver. Hepatology 2014;60:715-735.
- 34. Tapper EB. Rifaximin for the prevention of readmissions for patients with hepatic encephalopathy - the price is right. Liver Int 2016;36:1252-1254.
- 35. Tapper EB, Aberasturi D, Zhao Z, Hsu CY, Parikh ND. Outcomes after hepatic encephalopathy in population-based cohorts of patients with cirrhosis. Aliment Pharmacol Ther 2020;51:1397-1405.
- 36. Tapper EB, Halbert B, Mellinger J. Rates of and reasons for hospital readmissions in patients with cirrhosis: a multistate population-based cohort study. Clin Gastroenterol Hepatol 2016;14:1181-1188e2.
- 37. Bajaj JS, O’Leary JG, Tandon P, Wong F, Kamath PS, Biggins SW, et al. Targets to improve quality of care for patients with hepatic encephalopathy: data from a multi-centre cohort. Aliment Pharmacol Ther 2019;49:1518-1527.
- 38. Gu W, Hortlik H, Erasmus HP, Schaaf L, Zeleke Y, Uschner FE, et al. Trends and the course of liver cirrhosis and its complications in Germany: nationwide population-based study (2005 to 2018). Lancet Reg Health Eur 2021;12:100240.
, Patrick S. Kamath2, Florence Wong3, Qing Xie4, Ramazan Idilman5, Mark Topazian6, Wai-Kay Seto7, Aldo Torre8, Peter Hayes9, Jacob George10, Mario Alvares-da-Silva11, Brian J. Bush12, Scott Silvey12, Ashok Choudhury13 on behalf of CLEARED Investigators
