Epidemiology, Predictors of Antimicrobial Resistance and Empiric Treatment Strategies for Spontaneous Bacterial Peritonitis in Cirrhosis: A Multicentre SBP‐INDIA Study

ABSTRACT

Background and Aims

The rising burden of multidrug‐resistant (MDRO) and carbapenem‐resistant organisms (CRO) among patients with cirrhosis and spontaneous bacterial peritonitis (SBP) is compromising the effectiveness of standard empiric therapy. This multicentre study evaluated resistance patterns, independent predictors of MDRO and CRO infection, and clinical outcomes to inform risk‐based empiric therapy and antimicrobial stewardship in ascitic infections.

Methods

We analysed clinical characteristics and outcomes of hospitalized patients with cirrhosis and culture‐positive ascitic fluid infections from 13 centres across India. SBP was defined as ascitic neutrophil count (ANC) ≥ 250/mm³; culture‐positive cases with ANC < 250/mm³ were classified as non‐neutrocytic bacterascites (NNBA). Predictors of MDRO, CRO and mortality were assessed through regression models.

Results

Among 319 patients (median age 49 years; 79.9% male; 65.8% with acute‐on‐chronic liver failure), 74.6% of isolates were gram‐negative. MDRO and CRO were identified in 53.6% and 24.1% of infections, respectively. Nosocomial/healthcare‐associated acquisition, circulatory failure, elevated leukocyte count and SBP prophylaxis were independently associated with MDRO infection, while the nosocomial setting, leukocyte count and circulatory failure predicted CRO infection. MDRO infection was associated with poorer survival. Respiratory failure, liver failure, systemic inflammation, presence of MDRO and higher ACLF severity independently predicted mortality. Nearly one‐third of culture‐positive infections had ANC < 250/mm³ yet showed similar mortality to SBP. Based on clinical setting and risk profile, an empiric treatment framework was proposed.

Conclusions

There is a high burden of MDRO and CRO in ascitic infections among hospitalized patients with cirrhosis, with clinically relevant impact on outcomes. Empiric antibiotic selection should be guided by acquisition setting, illness severity and resistance risk factors, with stewardship oversight. Similar outcomes between NNBA and SBP suggest that microbiological evidence, rather than neutrophil count alone, should guide decisions.

Epidemiology and microbiological profile of ascitic fluid infections was studied in a multicentre retrospective study comprising 319 patients of cirrhosis with acute decompensation. Infection by MDRO was seen in 53.6% of infections. Patients with shock had high mortality, irrespective of mode of acquisition. Comparison of spontaneous bacterial peritonitis (SBP) with non‐neutrocytic bacterascites (NNBA) revealed similar mortality.

1. Introduction

Spontaneous bacterial peritonitis (SBP) remains the most common infection in cirrhosis and is a key event in decompensated disease, conferring substantial short‐term mortality and high recurrence [1]. International guidelines recommend immediate empiric antibiotics guided by acquisition setting and local resistance patterns, with reassessment at 48 h where response is suboptimal [2]. Yet, two major challenges threaten the effectiveness of guideline‐based treatment of SBP patients: (a) rapidly rising antimicrobial resistance among patients with cirrhosis, and (b) limited guidance to empiric and stewardship decisions due to low ascitic culture positivity in routine clinical practice.

The global epidemiology of infections in cirrhosis demonstrates marked geographic heterogeneity in multidrug resistance. In a global study, the prevalence of multidrug‐resistant (MDR/MDRO) bacteria among culture‐proven infections was ~34% overall, with the highest rates being reported from India (≈73%) [1]. A recent meta‐analysis [3] and a large 20‐year study from Hong Kong [4] have also demonstrated similar results, with the burden of MDR infections increasing over time. This context is clinically important, especially where extended‐spectrum beta lactamase (ESBL) and carbapenem resistance are common; third‐generation cephalosporins may lead to failure in a substantial proportion of hospitalized patients, especially those with healthcare‐associated or nosocomial acquisition, acute‐on‐chronic liver failure (ACLF) or shock. Although randomized evidence in nosocomial SBP supports broader empiric therapy (meropenem plus daptomycin) over cephalosporin monotherapy [5], such data have limited representation from high burden settings, and generalization requires similarity of local resistance patterns.

A second unresolved issue is the clinical significance of culture‐positive ascitic infections without neutrophilic ascites. SBP definitions operationally rely on ascitic polymorphonuclear leukocyte count (ANC ≥ 250 cells/mm³), because cultures are often negative [2]. However, the prognostic relevance of culture positivity with ANC < 250 (non‐neutrocytic bacterascites, NNBA phenotype) is less well defined, particularly in the context of rising antimicrobial resistance. If such patients experience outcomes comparable to classic SBP, then reliance on ANC thresholds alone could delay treatment and obscure stewardship decisions. Moreover, the ability of AF‐ANC levels ≥ 250 to pick SBP has been reported to be suboptimal in some settings; below 50% in settings with manual counting, use of empiric antimicrobials, sample handling and poor reporting [6].

Against this background, we conducted a multicentric study focusing on culture‐proven ascitic fluid infections among hospitalized adults with cirrhosis of the liver. We aimed to: (a) define the epidemiology and resistance profile (including MDRO and CRO burden) across acquisition settings; (b) identify predictors of MDRO infection; (c) quantify short‐term outcomes and predictors of 30‐day mortality; and (d) compare clinical features and outcomes of SBP versus culture‐positive ANC < 250 (NNBA phenotype), thereby clarifying whether ANC‐based triage remains reliable in an era of rising resistance.

2. Methods

2.1. Study Design and Setting

This was a multicentre study involving 13 medical centres with dedicated gastroenterology and hepatology services across India. A retrospective analysis of prospectively maintained hospital databases was conducted. Data were available from 2013 to 2024 at the coordinating centre (PGIMER, Chandigarh) and from 2022 to 2024 at the collaborating centres. Centre‐wise contributions are provided in Table S1. Ethics approval was obtained from the individual institutional ethics committees (IEC/2022/SPL‐1350). The study adhered to the Declaration of Helsinki (1975), Indian Council of Medical Research (ICMR) guidelines on biomedical research, STROBE statements and Good Clinical Practice standards.

2.2. Participants and Variables

Adult hospitalized patients (≥ 18 years) with cirrhosis and culture‐positive spontaneous ascitic fluid infection were included. Exclusion criteria were: secondary bacterial peritonitis; hepatic or extrahepatic malignancy (including solid tumours and hematologic malignancies); use of immunosuppressive therapy other than corticosteroids for alcohol‐associated hepatitis or autoimmune hepatitis; prior organ transplantation; human immunodeficiency virus infection; tuberculosis; pregnancy or lactation; and missing key clinical or outcome data. A retrospective cohort of patients with culture‐negative neutrocytic ascites (CNNA) was identified from the coordinating centre (PGIMER, Chandigarh) during 2013–2024. This was compared with patients who had ANC ≥ 250 with positive ascitic fluid culture using propensity score matching (PSM) to minimize the effect of baseline confounding variables. Since this analysis included patients of CNNA from only one centre, this was not combined with the analysis of SBP and NNBA.

Ascitic fluid samples were obtained by diagnostic paracentesis, and 10 mL was inoculated at the bedside into blood culture bottles (BACTEC system). Clinical and laboratory parameters at the time of infection diagnosis were extracted from in‐hospital medical records. Outcomes at 30 and 90 days were assessed from follow‐up records. Collected variables included demographic characteristics, aetiology of cirrhosis, acute precipitating events, vital parameters, laboratory values (liver function, renal function, blood counts, inflammatory markers), type of infection according to acquisition setting (community‐acquired, healthcare‐associated or nosocomial), risk factors for infection, previous healthcare exposure, medication history and procalcitonin levels where available. Microbiological data included organism identification, antimicrobial susceptibility profiles and resistance patterns. Antimicrobial susceptibility testing at all centres was performed according to the latest Clinical and Laboratory Standards Institute (CLSI) guidelines [7]. All variables were recorded at the time of ascitic fluid collection yielding a positive culture.

2.3. Definitions

SBP was defined as ascitic fluid absolute neutrophil count (ANC) ≥ 250 cells/μL in the presence of a positive monomicrobial ascitic fluid culture, without any surgically treatable intra‐abdominal source of infection [8]. Culture‐negative neutrocytic ascites (CNNA) was defined as ascitic fluid ANC ≥ 250 cells/μL in the absence of a positive monomicrobial ascitic fluid culture, without any surgically treatable intra‐abdominal source of infection [8]. Ascitic fluid neutrophil count < 250 cells/μL in the presence of a positive monomicrobial ascitic fluid culture, without any surgically treatable intra‐abdominal source of infection, was defined as non‐neutrocytic bacterascites (NNBA) [8]. Acute decompensation (AD) was defined in hospitalized patients of cirrhosis with the acute development of grade 2–3 ascites, gastrointestinal bleeding, hepatic encephalopathy and/or bacterial infection [9]. ACLF was defined according to EASL‐CLIF criteria as acute deterioration of cirrhosis with AD and organ failures [9]. Infections were classified by acquisition setting as: Nosocomial: diagnosed ≥ 48 h after hospital admission and not incubating at admission. Healthcare‐associated: diagnosis in patients with healthcare contact in the preceding 3 months or hospitalization within the past 30 days [2, 10]. Community‐acquired: all remaining cases. Antimicrobial use was assessed by comparing prescribed agents with isolate susceptibility profiles; therapy was considered appropriate when the initially administered empiric antibiotic matched the susceptibility profile of the identified organism. Detailed data on timing, dose and duration of therapy were not available for more granular analysis.

Multidrug‐resistant (MDR) infection was defined as non‐susceptibility to at least one agent in at least three antimicrobial categories [11]. Extensively drug‐resistant (XDR) bacteria were defined as non‐susceptibility to at least one agent in all but two or fewer antimicrobial categories [11]. Pan‐drug‐resistant (PDR) bacteria were defined as non‐susceptibility to all tested antimicrobial agents [11]. Specific resistance phenotypes were recorded according to published criteria, including: ESBL‐producing Enterobacterales, CRE (carbapenem‐resistant Enterobacterales), CRAB (carbapenem‐resistant Acinetobacter baumannii ), CRPA (carbapenem‐resistant Pseudomonas aeruginosa ), MDR Enterococcus faecium , MDR Staphylococcus aureus , VRE (vancomycin‐resistant Enterococcus) [11]. CROs were considered a subset of MDR infection. Organ failures were defined using EASL‐CLIF criteria: Liver—bilirubin ≥ 12 mg/dL, coagulation—INR ≥ 2.5, renal—creatinine ≥ 2 mg/dL, cerebral—grade 3–4 hepatic encephalopathy, respiratory—SpO₂/FiO₂ < 214 or PaO₂/FiO₂ < 200, circulatory—mean arterial pressure < 65 mmHg or vasopressor requirement. ACLF severity was assessed using organ failure count, EASL‐CLIF grades and CLIF‐C ACLF score. Severity of acute decompensation was assessed using Child‐Turcotte‐Pugh (CTP), MELD, MELD‐Na and CLIF‐C OF scores.

2.4. Statistical Analysis

Normality of continuous variables was assessed using the Shapiro–Wilk test. Normally distributed variables were reported as mean ± standard deviation, and non‐normally distributed variables as median (interquartile range). Categorical variables were expressed as counts and percentages. Between‐group comparisons were performed using Student's t‐test or Mann–Whitney U‐test for continuous variables, and chi‐square test or Fisher's exact test for categorical variables, as appropriate. Survival analysis was performed using Kaplan–Meier curves, with comparisons by log‐rank test. Univariable and multivariable logistic regression analyses were used to identify independent predictors of MDRO and CRO infection. To minimize multicollinearity, variables with variance inflation factor > 5 were excluded. Variables significant at p < 0.05 in univariable analysis were considered for multivariable modelling and further refined based on clinical relevance and backward elimination. Univariable and multivariable Cox proportional hazards regression was performed to identify predictors of 30‐day mortality. Variables were selected based on clinical plausibility and incremental model building and avoiding collinearity, with final model selection guided by performance metrics (Harrell's C‐index). Three multivariable cox regression models were subsequently developed. Since one of the aims was to determine whether presence of MDRO is an independent predictor of mortality, it was included as a variable in all three models. In addition, Model 1 included clinical parameters of severity of acute decompensation. Model 2 included systemic inflammatory markers and MELD‐Na. Model 3 included individual organ failures. Propensity score matching (PSM) was performed using the MatchIt (v4.7.2) and cobalt (v4.6.2) packages in R. The initial cohort comprised 258 patients, including 125 with CNNA and 133 with SBP (culture positive with ANC ≥ 250 cells/mm³). Propensity scores, representing the probability of belonging to the culture‐positive group based on baseline characteristics, were estimated using logistic regression with the following covariates: age, sex, diabetes, hypertension (HTN), healthcare contact, chronic aetiology and MELD‐Na score. Patients were matched using 1:1 nearest‐neighbour matching without replacement, with a calliper width of 0.2 of the standard deviation of the logit of the propensity score. Covariate balance was assessed using standardized mean differences (SMD), with an SMD < 0.1 considered indicative of adequate balance. A two‐sided p‐value < 0.05 was considered statistically significant. Statistical analyses were performed using Stata Now/SE version 19.5 (Stata Corp).

3. Results

3.1. Baseline Characteristics

From a total of 1535 patients that were screened, 319 patients were analysed from 13 centres (Figure 1 and Table S1). The median age was 49 years, with male predominance (n = 255; 79.9%). Median Child‐Turcotte‐Pugh (CTP) score was 11 and median MELD‐Na was 26 (Figure 2 and Table 1). Acute‐on‐chronic liver failure (ACLF) was present in 210 patients (65.8%). At enrolment, 109 patients (34.2%) required vasopressors and 70 (21.9%) required mechanical ventilation. Thirty‐ and ninety‐day survival were 62.7% and 60.5%, respectively.

FIGURE 1.

Study flow chart for participants included in the study (culture‐positive ascitic fluid infections). MDRO, multidrug‐resistant organism.

FIGURE 2.

(A) Geographic distribution of participating centres across India. (B) Antimicrobial resistance patterns among multidrug‐resistant organism infections in cirrhosis patients with spontaneous ascitic fluid infections (including SBP and NNBA). (C) Microbiological profile of ascitic fluid infections (SBP and NNBA) stratified by presence or absence of shock. CRAB, carbapenem‐resistant Acinetobacter baumannii ; CRE, carbapenem‐resistant Enterobacterales; CRO, carbapenem‐resistant organism; CRPA, carbapenem‐resistant Pseudomonas aeruginosa ; ESBL, extended‐spectrum β‐lactamase–producing organisms; HCA, healthcare‐associated; MDR‐EF, multidrug‐resistant Enterococcus faecium ; MDRO, multidrug‐resistant organism; MDR‐S, multidrug‐resistant Staphylococcus spp.; NNBA, non‐neutrocytic bacterascites; PDR, pan‐drug‐resistant; SBP, spontaneous bacterial peritonitis; VRE, vancomycin‐resistant Enterococci; XDR, extensively drug‐resistant.

TABLE 1.

Baseline characteristics of patients with and without multidrug‐resistant organism (MDRO) infection.

Characteristics	All patients (n = 319)	No MDRO (n = 148)	MDRO bacteria (n = 171)	p
Age (years) a	49.0 (39.0–58.0)	49.5 (39.0–57.0)	49.0 (40.0–59.0)	0.637
Male gender b	255 (79.9)	113 (76.4)	142 (83.0)	0.178
Comorbidities b
Diabetes mellitus	105 (32.9%)	53 (35.8)	52 (30.4)	0.366
Hypertension	60 (18.8%)	29 (19.6)	31 (18.1)	0.849
Dyslipidaemia	18 (5.6%)	10 (7.9)	8 (5.5)	0.582
Public hospital b	218 (68.3%)	91 (61.5)	127 (74.3)	0.020
Systolic blood pressure (mm Hg) a	110 (100–120)	112.0 (100.0–124.0)	110.0 (100.0–120.0)	0.085
Diastolic blood pressure (mm Hg) a	70 (60–80)	70.0 (60.0–80.0)	70.0 (60.0–80.0)	0.959
Mean arterial pressure (mm Hg) a	83.3 (76.7–92.7)	84.7 (76.7–93.3)	83.3 (76.0–91.0)	0.343
Heart rate (beats per minute) a	100 (86–110)	96.0 (85.8–110.0)	107.0 (88.0–111.0)	0.037
Number of SIRS criteria b	2 (1–3)	1.0 (0.0–3.0)	2.0 (1.0–3.0)	0.003
Presence of ≥ 2 SIRS b	160 (50.2%)	61 (41.2)	99 (57.9)	0.004
SpO2/FiO2 ratio a	461.9 (447.6–466.7)	466.7 (457.1–466.7)	461.9 (253.5–466.7)	0.126
Use of mechanical ventilation b	70 (21.9%)	20 (13.5)	50 (29.2)	0.001
Prior decompensations b
Jaundice	111 (36.2%)	49 (34.5)	62 (37.6)	0.661
Ascites	186 (62.4%)	85 (62)	101 (62.7)	0.998
HE	71 (24.1%)	32 (23.7)	39 (24.4)	1.000
Gastrointestinal bleed	80 (25.6%)	35 (24.5)	45 (26.6)	0.761
No. of past decompensations a	1.0 (0.0–2.0)	1.0 (0.0–2.0)	1.0 (0.0–2.0)	0.472
Time since first decompensation (months) a	5.0 (2.0–12.0)	5.0 (1.4–12.0)	6.0 (2.0–12.0)	0.693
Acute decompensation by EASL b	109 (34.2)	58 (39.2)	51 (29.8)	0.101
Presence of ACLF—EASL b	210 (65.8)	90 (60.8)	120 (70.2)
Acute insult b
Alcohol	46 (14.4)	27 (18.2)	19 (11.1)	0.110
Hepatitis B virus	25 (7.8)	7 (4.7)	18 (10.5)
Hepatitis C virus	3 (0.9)	1 (0.7)	2 (1.2)
Gastrointestinal bleed	4 (1.3)	3 (2)	1 (0.6)
Infection	225 (70.5)	99 (66.9)	126 (73.7)
Alcohol + infection	7 (2.2)	4 (2.7)	3 (1.8)
Acute viral hepatitis	1 (0.3)	1 (0.7)	0 (0)
Autoimmune hepatitis	2 (0.6)	1 (0.7)	1 (0.6)
Gastrointestinal bleed + infection	2 (0.6)	1 (0.7)	1 (0.6)
Unknown	4 (1.3)	4 (2.7)	0 (0)
Chronic insult b
Viral	62 (19.4)	21 (14.2)	41 (24.0)	0.017
Alcohol	202 (63.3)	91 (61.5)	111 (64.9)
Autoimmune hepatitis	14 (4.4)	9 (6.1)	5 (2.9)
MASLD	15 (4.7)	10 (6.8)	5 (2.9)
Others	26 (8.2)	17 (11.5)	9 (5.3)
Grade of ascites b
Grade 2	181 (56.7%)	93 (62.8)	88 (51.5)	0.053
Grade 3	138 (43.3%)	55 (37.2)	83 (48.5)
Grade of HE b
Grade 0	13 (4.1)	8 (5.4)	5 (2.9)	0.009
Grade 1	112 (35.1%)	65 (43.9)	47 (27.5)
Grade 2	138 (43.3%)	54 (36.5)	84 (49.1)
Grade 3	55 (17.2%)	20 (13.5)	35 (20.5)
Grade 4	1 (0.3%)	1 (0.7)	0 (0)
Gastrointestinal bleed b	35 (11.0)	13 (8.8)	22 (12.9)	0.325
Acute kidney injury b	128 (40.1)	55 (37.2)	73 (42.7)	0.373
Haemoglobin (g/dL) a	8.6 (7.4–10.1)	8.6 (7.4–10.2)	8.5 (7.5–9.9)	0.700
TLC (per cu. mm.) a	8.6 (5.83–13.5)	7.7 (5.2–11.6)	10.0 (6.8–15.8)	0.001
Neutrophil percentage a	78 (68–85.5)	78.0 (70.0–84.0)	78.2 (66.0–86.0)	0.931
Lymphocyte percentage a	13 (7–21.9)	14.0 (8.0–21.1)	12.2 (7.1–21.8)	0.788
Neutrophil lymphocyte ratio a	5.9 (3–11.3)	5.7 (3.1–11.1)	6.4 (2.9–11.3)	0.788
Platelet count (per cu. mm.) a	92,000 (60,000–140,000)	91,000.0 (62,000.0–133,500.0)	95,000.0 (59,745.0–142,500.0)	0.596
Sodium (mEq/L) a	133 (129–138)	133.0 (129.0–136.0)	133.0 (128.1–139.0)	0.358
Potassium (mEq/L) a	4 (3.6–4.6)	4.0 (3.6–4.6)	4.1 (3.7–4.6)	0.502
Urea (mg/dL) a	44.8 (25–78)	37.5 (23.6–69.7)	45.6 (28.2–84.8)	0.139
Creatinine (mg/dL) a	1.2 (0.8–2.01)	1.2 (0.8–2.0)	1.2 (0.8–2.0)	0.855
Bilirubin (mg/dL) a	4.5 (2.2–12)	4.6 (1.9–11.7)	4.3 (2.5–12.1)	0.695
AST (IU/L) a	55.9 (39–109)	61.0 (39.1–108.8)	54.5 (38.8–108.8)	0.973
ALT (IU/L) a	35 (24–58)	37.0 (26.7–53.7)	33.3 (22.7–59.2)	0.516
ALP (IU/L) a	133 (96–219)	138.0 (97.5–204.0)	127.5 (93.0–230.0)	0.948
Protein (g/dL) a	6 (5–6.7)	6.1 (5.3–6.7)	5.7 (4.8–6.5)	0.066
Albumin (g/dL) a	2.6 (2.2–3.1)	2.6 (2.1–3.0)	2.7 (2.3–3.1)	0.083
Prothrombin time (seconds) a	22.05 (17.4–30)	21.0 (17.0–28.7)	23.7 (18.0–31.0)	0.042
INR a	1.8 (1.42–2.4)	1.7 (1.4–2.3)	1.9 (1.5–2.5)	0.198
CRP (mg/dL) a	43.6 (13–79)	44.4 (17.1–64.2)	38.0 (12.0–110.4)	0.779
Procalcitonin (ng/mL) a	1.02 (0.47–3)	1.1 (0.5–2.7)	1.0 (0.4–3.8)	0.335
Type of SBP b
Community acquired	137 (42.9)	80 (54.1)	57 (33.3)	< 0.001
Healthcare associated	61 (19.1%)	15 (10.1)	46 (26.9)
Nosocomial	121 (37.9%)	53 (35.8)	68 (39.8)
Ascitic fluid cell count (per cu. mm.) a	574 (230–1800)	574.0 (320.0–1675.0)	514.0 (200.0–1800.0)	0.626
Ascitic fluid neutrophil a	70 (60–80)	70.0 (60.0–80.0)	70.0 (54.2–80.0)	0.453
Ascitic fluid neutrophil count (per cu. mm) a	320 (38–1176)	320.0 (43.5–1125.0)	327.5 (35.6–1153.2)	0.790
Ascitic fluid neutrophil count < 250 b	96 (30%)	48 (32.43)	48 (28.07)	0.397
Ascitic fluid protein (g/dL) a	1.2 (0.81–1.8)	1.2 (0.8–1.8)	1.2 (0.9–1.8)	0.753
Ascitic fluid albumin (g/dL) a	0.52 (0.3–0.9)	0.5 (0.3–0.9)	0.5 (0.3–0.8)	0.799
Ascitic fluid sugar (mg/dL) a	109 (82–140)	110.0 (94.0–140.0)	108.0 (67.2–137.2)	0.278
Gram‐negative b	238 (74.6)	102 (68.9)	136 (79.5)	0.041
Gram‐positive b	89 (27.9)	47 (31.8)	42 (24.6)	0.192
Polymicrobial infection b	21 (6.58)	4 (2.7)	17 (9.94)	0.009
Drug treatment b
Rifaximin	141 (44.2%)	62 (41.9)	79 (46.2)	0.510
Albumin	226 (70.8%)	107 (72.3)	119 (69.6)
NSBB	54 (47.4%)	30 (50)	24 (44.4)	0.685
Statins	33 (19.4%)	16 (21.9)	17 (17.5)	0.603
SBP prophylaxis	65 (20.4%)	26 (17.6)	39 (22.8)	0.308
Liver failure b	82 (25.7)	37 (25.0)	45 (26.3)	0.889
Renal failure b	83 (26)	39 (26.4)	44 (25.7)	1.000
Circulatory failure b	102 (34.2)	34 (23.0)	75 (43.9)	< 0.001
Coagulation failure b	75 (23.5)	32 (21.6)	43 (25.1)	0.543
Cerebral failure b	58 (18.2)	22 (14.9)	36 (21.1)	0.199
Respiratory failure b	68 (21.3)	18 (12.2)	50 (29.2)	< 0.001
SOFC a	1.0 (0.0–2.0)	1.0 (0.0–2.0)	2.0 (0.0–3.0)	0.002
ACLF grade b
Grade 0	109 (34.2)	58 (39.2)	51 (29.8)	0.001
Grade 1	82 (25.7%)	48 (32.4)	34 (19.9)
Grade 2	55 (17.2%)	16 (10.8)	39 (22.8)
Grade 3	73 (22.9%)	26 (17.6)	47 (27.5)
Grade of ACLF a	1 (0–2)	1 (0–2)	2 (0–3)	0.002
MELD a	22.4 (16.0–29.0)	22.0 (16.0–27.6)	22.8 (16.1–30.1)	0.356
MELD‐Na a	26.0 (19.6–30.9)	24.6 (20.1–29.7)	27.0 (19.3–31.9)	0.287
CTP a	11.0 (10.0–13.0)	11.0 (10.0–12.0)	12.0 (10.0–13.0)	0.062
CLIF C OF a	9.0 (8.0–12.0)	9.0 (7.0–11.0)	10.0 (8.0–12.0)	0.001
30‐day survival b	200 (62.7%)	109 (73.6)	91 (53.2)	< 0.001
90‐day survival b	193 (60.5%)	105 (70.9)	88 (51.5)	0.001

Note: Values in bold italics denote significant p < 0.05.

Abbreviations: ACLF, acute‐on‐chronic liver failure; ALP, alkaline phosphatase; ALT, alanine aminotransferase; APASL, Asia Pacific Association for Study of Liver; AST, aspartate aminotransferase; CLIF, chronic liver failure; CRP, C‐reactive protein; CTP, Child‐Turcotte‐Pugh; EASL, European Association for Study of Liver; FiO₂, fraction of inspired oxygen; HE, hepatic encephalopathy; INR, international normalized ratio; MASLD, metabolic dysfunction associated steatohepatitis; MELD, model for end‐stage liver disease; NSBB, non‐specific beta blocker; OF, organ failure; SBP, spontaneous bacterial peritonitis; SIRS, systemic inflammatory response syndrome; SOFC, simple organ failure count; SpO₂, oxygen saturation; TLC, total leukocyte count.

^a Median [IQR].

^b n (%).

All patients had acute decompensation (AD) of cirrhosis with or without ACLF. Infection was the most common acute precipitating event (70%) of AD, and alcohol‐associated liver disease was the most common underlying aetiology (63%) of cirrhosis. Overall, 74.6% of patients were infected with gram‐negative isolates, and 27.9% with gram‐positive. Multidrug‐resistant organisms (MDRO) and carbapenem‐resistant organisms (CRO) were identified in 171 (53.6%) and 77 (24.1%) patients, respectively.

3.2. Comparison of MDRO vs. Non‐MDRO Infections

A greater proportion of patients with MDRO infection were admitted to public sector hospitals compared to private hospitals (74.3% vs. 25.7%; p = 0.020) (Table 1). MDRO infection was significantly more frequent in nosocomial and healthcare‐associated infections than in community‐acquired infections. Patients with MDRO infection had higher heart rate (107 [88–111] vs. 96 [85.8–110]; p = 0.037) and met more SIRS criteria (2 [1–3] vs. 1 [0–3]; p = 0.004). Mechanical ventilation was required more often in the MDRO group (50 [29.2%] vs. 20 [13.5%]; p = 0.001). Although CTP, MELD, and MELD‐Na scores were similar between groups, patients with MDRO infection had more severe organ dysfunction, reflected by higher number of organ failures (2 [0–3] vs. 1 [0–2]; p = 0.002), higher ACLF grade (2 [0–3] vs. 1 [0–2]; p = 0.002), higher CLIF‐C OF score (10 [8–12] vs. 9 [7–11]; p = 0.001), higher CLIF‐C ACLF score (48.3 [40.5–56.1] vs. 43.5 [35.7–50.4]; p < 0.001). Past decompensation history did not differ between groups. Grade 2 and 3 hepatic encephalopathy was more frequent in the MDRO group (49.1% vs. 36.5% for grade 2 HE and 20.5% vs. 13.5% for grade 3 HE; p = 0.009). Patients with MDRO infection had higher total leukocyte count (10 [6.8–15.8] vs. 7.7 [5.2–11.6]; p = 0.001) and slightly prolonged prothrombin time (23.7 [18–31] vs. 21 [17–28.7]; p = 0.042). Neutrophil percentage, neutrophil‐to‐lymphocyte ratio, ascitic fluid cell count, and ascitic neutrophil count did not differ between groups.

On multivariable logistic regression (Table S2, Figure S2), independent predictors of MDRO infection were type of hospital admission, presence of circulatory failure, nosocomial/healthcare‐associated infection, higher leukocyte count, use of SBP prophylaxis. In separate multivariable models for carbapenem‐resistant organisms (Tables S3 and S4), independent predictors were circulatory failure, nosocomial/healthcare‐associated acquisition and higher leukocyte count.

3.3. Microbiological Profile

Gram‐negative infections were more frequent among MDRO cases than non‐MDRO cases (79.5% vs. 68.9%; p = 0.041), while gram‐positive infections were similar (24.6% vs. 31.8%; p = 0.192). Among MDRO infections, XDR and PDR phenotypes were observed in 29 (17.0%) and 13 (4.1%) patients, respectively.

In the MDRO group, the most frequently isolated organisms were E. coli (n = 54; 31.6%), K. pneumoniae (n = 35; 20.5%), Enterococcus spp. (n = 25; 14.6%), Staphylococcus spp. (n = 15; 8.8%), A. baumannii (n = 14; 8.2%), Pseudomonas spp. (n = 12; 7.0%), Burkholderia spp. (n = 11; 6.4%). Other organisms were uncommon. In the non‐MDRO group, the most frequent isolates were: E. coli (n = 29; 19.6%), Staphylococcus spp. (n = 21; 14.8%), A. baumannii (n = 19; 12.8%), Streptococcus spp. (n = 17; 11.5%), K. pneumoniae (n = 15; 10.1%), Pseudomonas spp. (n = 15; 10.1%), Burkholderia spp. (n = 13; 8.8%) (Table S5).

3.4. Type of SBP by Acquisition Setting

Community‐acquired SBP was observed in 137 patients (42.9%), nosocomial in 121 (37.9%) and healthcare‐associated in 61 (19.1%). Because nosocomial and healthcare‐associated infections had similar outcomes and microbiological profiles, they were analysed together (Figure 2 and Figure S1).

Compared with community‐acquired SBP, patients with nosocomial/healthcare‐associated infection had: higher heart rate (107 [88–112] vs. 94 [80–110]; p < 0.001), more SIRS criteria (2 [1–3] vs. 1 [0–2]; p = 0.002), higher prevalence of ACLF (71.0% vs. 59.0%; p = 0.028), higher prevalence of MDRO infection (62.6% vs. 41.6%; p < 0.001), higher prevalence of CRO infection (30.2% vs. 16.1%; p = 0.003), more frequent circulatory failure (39.0% vs. 27.7%; p = 0.036), higher CLIF‐C OF score (10 [8–12] vs. 9 [7–12]; p = 0.021), higher procalcitonin levels (1.36 [0.51–3.48] vs. 0.75 [0.37–2.26]; p = 0.024) (Table S6). Use of rifaximin was more frequent among patients with nosocomial/healthcare‐associated infection (50.5% vs. 35.7%; p = 0.008). Thirty‐ and ninety‐day survival did not differ significantly by acquisition setting (Table S6). However, across all acquisition categories, patients presenting with shock had significantly higher 90‐day mortality (Figure 2C, Figure S1).

Microbiologically, nosocomial/healthcare‐associated infections had higher rates of: Enterobacter spp. (5.5% vs. 0.7%; p = 0.027), ESBL‐producing organisms (46.7% vs. 32.1%; p = 0.011), MDR Staphylococcus (7.2% vs. 1.5%; p = 0.014), CRE (26.4% vs. 14.6%; p = 0.013), while community‐acquired SBP had a higher proportion of Pseudomonas infection (12.4% vs. 5.5%; p = 0.028) (Tables S7 and S8).

3.5. Survival Rates and Predictors of Mortality

Thirty‐day survival rate was 62.7% (n = 200) and 90‐day survival was 60.5% (n = 193). Survival was significantly worse among patients with MDRO infection at both 30 days (53.2% vs. 73.6%; p < 0.001) and 90 days (51.5% vs. 70.9%). Among patients who died within 30 days, 67.2% had MDRO infection. Patients who died by 30 days were clinically sicker, with lower blood pressure, higher heart rate, higher oxygen requirement, lower SpO₂/FiO₂ ratio and greater need for mechanical ventilation. They had more severe hepatic encephalopathy, higher prevalence of acute kidney injury and more frequently presented with ACLF (Table 2). Over half of patients with grade 3 ACLF died within 30 days. Patients who died had higher disease severity scores (CTP, MELD, MELD‐Na, CLIF‐C OF and CLIF‐C ACLF). There was no difference in mortality based on the type of SBP, gastrointestinal bleeding, medication exposure, SBP prophylaxis or common comorbidities. Laboratory parameters associated with mortality included higher leukocyte count, urea, creatinine, bilirubin, transaminases, prothrombin time, INR and procalcitonin (Table 2). Ascitic neutrophil count did not differ between survivors and non‐survivors. Appropriate antimicrobials as per culture sensitivity were received in 204/319 patients and were linked to lower 30‐day mortality (OR: 0.59; 95% CI: 0.37–0.94, p = 0.029) and 90‐day mortality (OR: 0.61; 95% CI: 0.39–0.98, p = 0.041).

TABLE 2.

Clinical and laboratory characteristics stratified by 30‐day outcome (survivors vs. non‐survivors).

Variable	Died (n = 119)	Alive (n = 200)	p
Age a	47 [39–57]	50 [40–60]	0.49
Gender (male) b	99 (83.2)	156 (78)	0.263
Public hospital b	81 (68.06)	137 (68.5)	0.936
Systolic blood pressure (mm Hg) c	108.9 ± 18.32	114.4 ± 15.9	0.006
Diastolic blood pressure (mm Hg) c	66.4 ± 11.8	72.14 ± 10.4	< 0.001
Mean arterial pressure (mm Hg) c	80.89 ± 13.35	86.23 ± 10.83	< 0.001
Heart rate (beats per minute) c	106.15 ± 18.41	96.13 ± 15.92	< 0.001
Presence of ≥ 2 SIRS b	100 (84.03)	60 (30)	< 0.001
SpO2 c	96.92 ± 2.73	97.28 ± 1.78	0.232
FiO2 c	38.23 ± 24.11	23.55 ± 9.1	< 0.001
SpO2/FiO2 c	330.31 ± 150.21	439.07 ± 76.86	< 0.001
Use of ventilator b	53 (44.5)	17 (8.5)	< 0.001
No. of past decompensations a	1 [0–2]	1 [0–2]	0.130
Time since first decompensation (months) a	6 [2–12]	5 [1.5–12]	0.350
Past jaundice b	47 (39.5)	64 (32)	0.155
Past ascites b	74 (62.2)	112 (56)	0.050
Past HE b	29 (24.4)	42 (21)	0.229
Past gastrointestinal bleed b	35 (29.4)	45 (22.5)	0.205
Comorbidities b
Diabetes mellitus	35 (29.4)	70 (35)	0.304
Hypertension	16 (13.4)	44 (22)	0.059
Dyslipidaemia	4 (3.4)	14 (7)	0.347
Acute decompensation b	18 (15.1)	91 (45.5)	< 0.001
ACLF b	101 (84.9)	109 (54.5)
Alcohol‐associated ACLF b	28 (23.5)	25 (12.5)	0.010
Ascites b
Grade 2	70 (58.8)	111 (55.5)	0.562
Grade 3	49 (41.2)	89 (44.5)
Grade of HE a	2 [1–3]	2 [1–2]	< 0.001
Gastrointestinal bleed b	17 (14.3)	18 (9)	0.144
Acute kidney injury b	64 (53.7)	64 (32)	< 0.001
Type of SBP b
Community acquired	46 (38.6)	91 (45.5)	0.260
Healthcare associated	21 (17.6)	40 (20)
Nosocomial	52 (43.7)	69 (34.5)
Drug use b
Rifaximin	60 (50.4)	81 (40.5)	0.084
NSBB	14/35 (40)	40/79 (50.6)	0.294
Statins	10/55 (18.2)	23/115 (20)	0.779
Albumin	89 (74.7)	137 (68.5)	0.232
SBP prophylaxis b	22 (18.5)	43 (21.5)	0.518
Haemoglobin (g/dL) c	8.89 ± 2.03	8.63 ± 1.89	0.257
TLC (per cu. mm.) a	10,360 [6900–17,590]	7900 [5400–12,150]	0.002
Neutrophil percentage a	79 [70–86]	77.55 [67.9–85.25]	0.276
Lymphocyte percentage a	11 [6–20]	14 [9–22.8]	0.089
Neutrophil lymphocyte ratio a	7.18 [3.47–13.5]	5.65 [2.9–9.5]	0.097
Platelet count (per cu. mm.) a	96,000 [55,000–135,000]	92,000 [64,000–145,000]	0.524
Sodium (mEq/L) a	133 [128–138]	133 [129–137.5]	0.662
Potassium (mEq/L) a	4.2 [3.56–4.7]	4 [3.7–4.5]	0.295
Urea (mg/dL) a	61 [35–101]	36 [22.5–58.2]	< 0.001
Creatinine (mg/dL) a	1.7 [0.97–2.6]	0.99 [0.74–1.56]	< 0.001
Bilirubin (mg/dL) a	8 [3.7–19]	3.2 [1.6–7.45]	< 0.001
AST (IU/L) a	78 [42.15–146]	49 [36.5–94]	< 0.001
ALT (IU/L) a	41 [28–62.7]	49 [36.5–94]	0.007
ALP (IU/L) a	116 [85–220]	138 [99.1–215]	0.111
Protein (g/dL) a	5.65 [4.57–6.7]	6.1 [5.27–6.7]	0.100
Albumin (g/dL) a	2.6 [2.1–3.1]	2.62 [2.2–3.095]	0.341
Prothrombin time (seconds) a	28 [21–40]	20 [16.5–25.8]	< 0.001
INR a	2.2 [1.7–3]	1.6 [1.325–2.1]	< 0.001
CRP (mg/dL) a	45 [18.67–112.07]	43.45 [12–64]	0.064
Procalcitonin (ng/mL) a	1.61 [0.52–3.82]	0.8 [0.43–2.15]	0.014
Ascitic fluid cell count (per cu. mm.) a	485 [216.5–2802.5]	574 [250–1510]	0.937
Ascitic fluid neutrophil a	67.5 [52–80]	70 [60–80]	0.135
Ascitic fluid neutrophil count (per cu. mm) a	350 [48–1800]	314 [36.4–1029.65]	0.321
Ascitic fluid neutrophil < 250 (n) b	33 (27.7)	63 (31.5)	0.478
Ascitic fluid protein (g/dL) a	1.2 [0.8–1.77]	1.2 [0.82–1.8]	0.669
Ascitic fluid albumin (g/dL) a	0.5 [0.31–0.83]	0.52 [0.3–0.9]	0.863
Ascitic fluid sugar (mg/dL) a	106 [61–144]	109 [91–138]	0.240
Ascitic fluid ADA a	10 [4–17.7]	9 [5–13]	0.468
Presence of MDR organism b	80 (67.2)	91 (45.5)	< 0.001
Presence of XDR organism b	13 (10.92)	16 (20)	0.380
Presence of PDR organism b	6 (5.0)	7 (3.5)	0.703
Presence of CRO b	37 (31.1)	40 (20)	0.025
Polymicrobial infection b	10 (8.4)	11 (5.5)	0.312
SOFC a	2 [1–4]	1 [0–1]	< 0.001
ACLF grade a	2 [1–3]	1 [0–1]	< 0.001
Grade 3 ACLF b	58 (56.3)	15 (12.6)	< 0.001
Liver failure b	47 (39.5)	35 (17.5)	< 0.001
Renal failure b	44 (36.9)	39 (19.5)	0.001
Coagulation failure b	45 (37.8)	30 (15)	< 0.001
Cerebral failure b	40 (33.6)	18 (9)	< 0.001
Respiratory failure b	53 (44.5)	15 (7.5)	< 0.001
Circulatory failure b	66 (55.4)	43 (21.5)	< 0.001
MELD a	28.54 [21.49–36.26]	19.92 [14.5–25]	< 0.001
MELD‐Na a	29.93 [24.47–37.49]	23.25 [18.45–29.07]	< 0.001
CTP a	12 [11–13]	11 [10–12]	< 0.001
CLIF‐C of score a	12 [10–14]	9 [7–10]	< 0.001

Note: Values in bold italics denote significant p < 0.05.

Abbreviations: ACLF, acute‐on‐chronic liver failure; ALP, alkaline phosphatase; ALT, alanine aminotransferase; APASL, Asia Pacific Association for Study of Liver; AST, aspartate aminotransferase; CLIF, chronic liver failure; CRO, carbapenem‐resistant organism; CRP, C‐reactive protein; CTP, Child‐Turcotte‐Pugh; EASL, European Association for Study of Liver; FiO₂, fraction of inspired oxygen; HE, hepatic encephalopathy; INR, international normalized ratio; MASLD, metabolic dysfunction associated steatohepatitis; MDR, multidrug resistant; MELD, Model for end‐stage liver disease; NSBB, non‐specific beta blocker; OF, organ failure; PDR, pan‐drug resistant; SBP, spontaneous bacterial peritonitis; SIRS, systemic inflammatory response syndrome; SOFC, simple organ failure count; SpO₂, oxygen saturation; TLC, total leukocyte count; XDR, extensively drug resistant.

^a Median [IQR].

^b n (%).

^c Mean ± standard deviation.

Univariate cox regression analysis identified various predictors of 30‐day mortality (Table S9). In multivariate cox regression Model 1, presence of MDRO (HR 1.5; 95% CI 1.01–2.22; p = 0.043) and CLIF‐C OF score (HR 1.34; 95% CI 1.26–1.43; p < 0.001) independently predicted 30‐day mortality (Table S10). In Model 2, high MELD‐Na score (HR 1.06; 95% CI 1.04–1.08; p < 0.001) and presence of MDRO (HR 1.59; 95% CI 1.08–2.34; p = 0.018) were associated with increased mortality risk, while absence of SIRS (HR 0.18; 95% CI 0.11–0.29; p < 0.001) protected against mortality (Table S10). In Model 3, presence of respiratory failure (HR 2.37; 95% CI 1.47–3.81; p < 0.001) and liver failure (HR 1.61; 95% CI 1.05–2.46; p = 0.027) independently predicted mortality (Table S10). Kaplan–Meier analysis showed worse survival in patients with MDRO infection (p < 0.001), CRO infection (p = 0.026) and with increasing ACLF grade (p < 0.001). However, survival did not differ by acquisition setting (p = 0.345) (Figure 3).

FIGURE 3.

Kaplan–Meier survival curves stratified by (A) presence of MDRO, (B) presence of CRO, (C) grade of ACLF and (D) type of SBP (community‐acquired vs. nosocomial/healthcare‐associated). ACLF, acute‐on‐chronic liver failure; CA, community‐acquired; CRO, carbapenem‐resistant organism; MDRO, multidrug‐resistant organism; SBP, spontaneous.

3.6. Patients With Culture‐Positive Infection and ANC < 250 Cells/μL

Ninety‐six patients had culture‐positive ascitic infection with ANC < 250 cells/μL. Median age was 48 years [39–55], and 82.3% were male. Median ascitic fluid total leukocyte count was 225 [105–466] cells/μL, and median ANC was 10.9 [0–98.5]. Compared with patients with ANC ≥ 250, those with ANC < 250 had lower heart rate (95.9 ± 18.3 vs. 101.6 ± 17.0; p = 0.007), more frequent presentation with acute decompensation (44.8% vs. 24.2%; p < 0.001), less frequent ACLF (51.0% vs. 72.0%; p < 0.001), less frequent albumin use (61.5% vs. 74.9%; p = 0.015) (Table 3). Patients with ANC ≥ 250 had higher rates of acute kidney injury, grade 3 ACLF, circulatory failure, and higher CLIF‐C OF and CLIF‐C ACLF scores. Despite these differences, mortality was similar between groups (NNBA vs. SBP): 30‐day mortality: 38.5% vs. 34.4% (p = 0.478) and 90‐day mortality: 39.9% vs. 38.5% (p = 0.819) (Table 3).

TABLE 3.

Clinical characteristics stratified by ascitic fluid neutrophil count < 250 cells/μL versus ≥ 250 cells/μL.

Variable	ANC ≥ 250/μL (n = 223)	ANC < 250/μL (n = 96)	p
Age a	50 [39–60]	48 [39–55]	0.187
Gender (male) b	176 (78.9)	79 (82.3)	0.491
Public hospital b	147 (65.9)	71 (73.9)	0.157
Systolic blood pressure (mm Hg) c	111.76 ± 16.92	113.59 ± 17.47	0.402
Diastolic blood pressure (mm Hg) c	69.52 ± 10.62	70.89 ± 13.06	0.826
Mean arterial pressure (mm Hg) c	83.78 ± 11.76	85.12 ± 13.02	0.389
Heart rate (beats per minute) c	101.57 ± 16.97	95.89 ± 18.31	0.007
Presence of ≥ 2 SIRS b	117 (52.4)	43 (44.8)	0.209
SpO2 c	97.13 ± 2.25	97.2 ± 1.99	0.829
FiO2 c	29.35 ± 17.06	28.01 ± 18.77	0.585
SpO2/FiO2 c	395.57 ± 122.11	412.25 ± 114.87	0.332
Use of ventilator b	54 (24.22)	16 (16.67)	0.135
No. of past decompensations a	1 [0–2]	1 [0–1.5]	0.02
Time since first decompensation (months) a	6 [2–12]	2 [1–11]	0.004
Past jaundice b	82 (38.14)	29 (31.52)	0.269
Past ascites b	135 (66.18)	51 (54.26)	0.048
Past HE b	53 (26.24)	18 (19.35)	0.199
Past gastrointestinal bleed b	66 (30)	14 (15.22)	0.006
Comorbidities b
Diabetes mellitus	88 (39.46)	17 (17.71)	< 0.001
Hypertension	45 (20.2)	15 (15.6)	0.340
Dyslipidaemia	12 (6.45)	6 (6.9)	0.890
Acute decompensation b	62 (28)	47 (49)	< 0.001
ACLF b	161 (72)	49 (51)
Alcohol‐associated ACLF b	28 (12.56)	25 (26.04)	0.003
Ascites b
Grade 2	126 (56.5)	55 (57.29)	0.896
Grade 3	97 (43.5)	41 (42.7)
Grade of HE a	2 [1–2]	2 [1–2]	0.07
Gastrointestinal bleed b	23 (10.3)	12 (12.5)	0.567
Acute kidney injury b	105 (47.1)	23 (23.9)	< 0.001
Type of SBP b
Community acquired	95 (42.6)	42 (43.7)	0.982
Healthcare associated	43 (19.28)	18 (18.75)
Nosocomial	85 (38.12)	36 (37.5)
Drug use b
Rifaximin	100 (44.84)	41 (42.71)	0.725
NSBB	45 (51.14)	9 (34.6)	0.138
Statins	29 (21.5)	4 (11.43)	0.180
Albumin	167 (74.89)	59 (61.46)	0.015
SBP prophylaxis b	50 (22.42)	15 (15.6)	0.167
Empiric antibiotics used b
Third‐generation cephalosporin	26 (11.6%)	11 (11.5%)	0.959
BL‐BLI combination	65 (29.1%)	28 (29.2%)	0.997
Carbapenems	94 (42.6%)	33 (34.4%)	0.194
Polymyxin B or Colistin	13 (5.8%)	6 (6.25%)	0.885
Anti‐MDR Staphylococcus or Enterococcus	57 (25.5%)	20 (22.2%)	0.372
Haemoglobin (g/dL) c	8.86 ± 2.07	8.43 ± 1.57	0.069
TLC (per cu. mm.) a	9200 [5680–13,750]	7900 [6213–12,620]	0.475
Neutrophil percentage a	78.3 [70–86]	77 [65.45–80.7]	0.089
Lymphocyte percentage a	13.2 [7–21.9]	12 [8.75–22.05]	0.632
Neutrophil lymphocyte ratio a	5.85 [3.04–12.28]	6.14 [2.86–9.28]	0.514
Platelet count (per cu. mm) a	92,000 [58,000–136,000]	92,000 [71,500–153,000]	0.256
Sodium (mEq/L) a	132.5 [128.8–138]	133 [129–139]	0.229
Potassium (mEq/L) a	4.01 [3.6–4.6]	4 [3.7–4.5]	0.699
Urea (mg/dL) a	43.6 [24–78]	45 [27.6–80]	0.487
Creatinine (mg/dL) a	1.22 [0.8–2.1]	1.015 [0.8–1.94]	0.406
Bilirubin (mg/dL) a	4.8 [2.2–12.6]	3.83 [2.52–8.38]	0.326
AST (IU/L) a	62 [40–129]	49 [37–85]	0.043
ALT (IU/L) a	38.65 [25–62]	32 [22–51]	0.062
ALP (IU/L) a	135 [98–219.5]	122 [84–194]	0.174
Protein (g/dL) a	5.9 [5–6.7]	6.1 [5.1–6.8]	0.900
Albumin (g/dL) a	2.6 [2.1–3]	2.62 [2.39–3.125]	0.058
Prothrombin time (seconds) a	23 [17.5–31]	21 [17.3–27.3]	0.217
INR a	1.86 [1.4–2.52]	1.73 [1.46–2.2]	0.416
CRP (mg/dL) a	38 [12.01–80.16]	59.05 [34.16–73.83]	0.141
Procalcitonin (ng/mL) a	1.075 [0.5–3.81]	0.78 [0.4–2]	0.118
Ascitic fluid cell count (per cu. mm) a	800 [347.5–2800]	225 [105–466]	< 0.001
Ascitic fluid neutrophil a	78 [65–85]	60 [20–70]	< 0.001
Ascitic fluid neutrophil count (per cu. mm) a	928 [405–2712]	10.89 [0–98.5]	< 0.001
Ascitic fluid protein (g/dL) a	1.2 [0.88–1.81]	1.2 [0.8–1.7]	0.800
Ascitic fluid albumin (g/dL) a	0.55 [0.3–0.9]	0.52 [0.38–0.7]	0.716
Ascitic fluid sugar (mg/dL) a	102 [65–144]	116 [107–135]	0.005
Ascitic fluid ADA a	8.5 [5–15]	9.5 [9–15]	0.216
Presence of MDR organism b	123 (55.16)	48 (50)	0.397
Presence of XDR organism b	16 (7.17)	13 (13.54)	0.070
Presence of PDR organism b	6 (2.7)	7 (3.3)	0.110
Presence of CRO b	54 (24.2)	23 (23.9)	0.961
Polymicrobial infection b	17 (7.62)	4 (4.17)	0.254
SOFC a	1 [0–2]	1 [0–2]	0.003
ACLF grade a	1 [0–2]	1 [0–2]	0.003
Grade 3 ACLF b	55 (24.66)	18 (18.75)	0.004
Liver failure b	61 (27.35)	21 (21.8)	0.304
Renal failure b	60 (26.9)	23 (23.9)	0.582
Coagulation failure b	57 (25.5)	18 (18.75)	0.188
Cerebral failure b	45 (20.2)	13 (13.5)	0.159
Respiratory failure b	52 (23.3)	16 (16.7)	0.183
Circulatory failure b	86 (38.5)	23 (23.9)	0.012
MELD a	23.6 [16.3–29.8]	20.8 [15.9–27.8]	0.095
MELD‐Na a	27.1 [20.84–31.6]	23.4 [19.1–29.9]	0.052
CTP a	12 [10–13]	11 [10–12]	0.060
CLIF‐C OF SCORE a	10 [8–12]	9 [7–11]	0.001
Appropriate empiric antibiotic b	146 (65.5%)	58 (60.4%)	0.523
30‐day mortality b	86 (38.5)	33 (34.4)	0.478
90‐day mortality b	89 (39.9)	37 (38.5)	0.819

Note: Values in bold italics denote significant p < 0.05.

Abbreviations: ACLF, acute‐on‐chronic liver failure; ALP, alkaline phosphatase; ALT, alanine aminotransferase; APASL, Asia Pacific Association for Study of Liver; AST, aspartate aminotransferase; BL‐BLI, Beta‐lactam/beta‐lactamase inhibitor; CLIF, chronic liver failure; CRO, carbapenem‐resistant organism; CRP, C‐reactive protein; CTP, Child‐Turcotte‐Pugh; EASL, European Association for Study of Liver; FiO₂, fraction of inspired oxygen; HE, hepatic encephalopathy; INR, international normalized ratio; MASLD, metabolic dysfunction associated steatohepatitis; MDR, multidrug resistant; MELD, model for end‐stage liver disease; NSBB, non‐specific beta blocker; OF, organ failure; PDR, pan‐drug resistant; SBP, spontaneous bacterial peritonitis; SIRS, systemic inflammatory response syndrome; SOFC, simple organ failure count; SpO₂, oxygen saturation; TLC, total leukocyte count; XDR, extensively drug resistant.

^a Median [IQR].

^b n (%).

^c Mean ± Standard deviation.

3.7. Comparison of CNNA With SBP

After propensity score matching, a total of 180 patients were retained in the final cohort, comprising 90 patients in each group (Table S11). Covariate balance before and after matching was assessed using standardized mean differences (SMD) and visualized using a Love plot (Figure S3). Prior to matching, several baseline variables demonstrated substantial imbalance, including age (SMD 0.279), diabetes (SMD 0.210), healthcare contact (SMD −0.279), chronic aetiology (SMD 0.376) and MELD‐Na (SMD −0.403). After matching, there was a marked improvement in covariate balance across all variables. Age (SMD 0.047), sex (SMD 0.056), diabetes (SMD 0.056), hypertension (SMD 0.011), healthcare contact (SMD −0.044), chronic aetiology (SMD −0.076) and MELD‐Na (SMD 0.0005) were all well balanced, with absolute SMD values below the predefined threshold of 0.1 (Table S12). The propensity score distance also showed a substantial reduction in imbalance (from SMD 0.820 to 0.122), indicating improved overlap between groups. Overall, these findings demonstrate that propensity score matching effectively minimized baseline differences between the groups, resulting in a well‐balanced matched cohort with minimal residual imbalance.

More patients with SBP had history of previous ascites (70.5% vs. 43.3%; p < 0.001) or hepatic encephalopathy (29.9% vs. 14.4%; p = 0.022) when compared to patients with CNNA. Alcohol was the most common aetiology of cirrhosis in patients with CNNA (63.3% vs. 10%; p < 0.001) while chronic viral hepatitis was the most common aetiology of cirrhosis in patients with SBP (68.9% vs. 3.3%; p < 0.001). Mean arterial pressure was higher in patients with CNNA (85.7 [75.4–92.5] vs. 81.7 [74.2–88]; p = 0.03). Grade 2 (62 [68.9%] vs. 32 [35.6%]; p < 0.001) and 3 HE (23 [25.6%] vs. 17 [18.9%]; p < 0.001) along with grade 3 ascites (60 [66.7%] vs. 39 [43.3%]; p = 0.003) were more common in patients with CNNA than with SBP. Among laboratory parameters, patients with SBP had lower sodium (131 [128–136] vs. 134.5 [131–139]; p = 0.003) and serum protein (5.5 [3.3–6.3] vs. 6.2 [5.6–6.9]; p < 0.001), but higher ALP (136 [102.2–233.2] vs. 116 [86–162]; p = 0.011) and procalcitonin levels (1.2 [0.7–3.8] vs. 0.7 [0.3–2]; p = 0.006). ACLF grade, organ failure count, type of organ failures, CLIF‐C OF score, MELD score and MELD‐Na were similar in both the groups (Table S11). Patients with SBP had higher ascitic fluid neutrophil percentage (79.5 [70–85] vs. 65 [52.2–79.8]; p < 0.001) and lower ascitic fluid sugar (101.5 [69.5–126.2] mg/dL vs. 110 [87.5–146] mg/dL; p = 0.04) than patients with CNNA. More patients with CNNA were on rifaximin (76.7% vs. 52.2%; p < 0.001), while more patients with SBP had received prior SBP prophylaxis (20% vs. 2.2%; p < 0.001). Thirty‐day survival was similar (52.2% vs. 63.3%; p = 0.174); however, 90‐day survival was better in patients with SBP than those with CNNA (61.1% vs. 38.9%; p = 0.003).

3.8. Empiric Antibiotic Framework

Based on the microbiological profile and stratified risk analysis (Figure 2C), an empiric antibiotic framework was developed (Table 4). Infections with shock were analysed as a separate category because of substantial differences in outcomes and epidemiology compared with infections without shock (Figure 2). Among patients without shock, nosocomial and healthcare‐associated infections were grouped together due to their similar epidemiological characteristics (Figure 2). Differences in microbiological patterns were primarily driven by MDRO prevalence rather than acquisition setting alone. Among patients presenting with shock, MDRO prevalence was high across all settings, with higher rates in healthcare‐associated/nosocomial infections. Among hemodynamically stable patients, healthcare‐associated/nosocomial infection was associated with higher rates of MDRO, CRE and MDR gram‐positive organisms. The proposed framework emphasizes stratification by illness severity and MDRO risk factors rather than acquisition setting alone to support rational empiric therapy and stewardship.

TABLE 4.

Proposed empiric antibiotic regimens for ascitic fluid infections in Indian patients with cirrhosis.

Condition	Epidemiology	Empiric antimicrobial choice (hospitalized patients)	Quality of evidence; strength of recommendation	Stewardship justification
Community‐acquired SBP without shock	Gram‐negative—71.7% Gram‐positive—29.3% MDR—36.4% XDR—7.0% PDR—3.0% ESBL—28.3% CRO—12.1% MDR GP—2.0%	Low MDR risk a : Cefotaxime 2 g IV q8h or ceftriaxone 2 g IV q24h (5–7 days)	High; Strong	3rd‐gen cephalosporin have RCT‐driven efficacy and remain standard for low‐risk CA SBP [12, 13, 14].
		Strong MDR risk a : Piperacillin–tazobactam (TZP) 4.5 g IV q6h (extended infusion preferred) OR meropenem 1 g IV q8h	Moderate; Strong	However, CA SBP (without shock) shows MDR ~36.4 and ESBL ~28.3 and CRO 12.1, which is higher than many Western cohorts, so ceftriaxone should be restricted to low‐risk CA presentations. Meropenem is preferred in those with high severity; CLIF‐SOFA > 7 based on study demonstrating lower mortality in carbapenem treated individuals than TZP [15].
Healthcare‐associated/Nosocomial SBP without shock	Gram‐negative—73.8% Gram‐positive—27.0% MDR—54.0% XDR—10.8% PDR—5.4% ESBL—39.6% CRO—25.2% MDR GP—12.6%	Low CRO risk b : Meropenem 1 g IV q8h (extended infusion) add anti‐gram‐positive MDRO only if indicated (see remarks): vancomycin/teicoplanin	Moderate; Strong	For nosocomial SBP, meropenem ± glycopeptide/lipopeptide is suggested by EASL [2], and TZP is only a primary approach in areas with low MDRO prevalence. Indian HCA/nosocomial SBP has MDR ~54%, ESBL ~39.6%, CRO ~25.2%, making cephalosporin‐first strategies high risk for failure. An RCT shows meropenem + daptomycin more effective than ceftazidime in nosocomial SBP; efficacy predicted survival [5]. Anti‐GP cover should be conditional, especially when (a) prior MRSA/VRE colonization, (b) severe sepsis, (c) indwelling lines/skin focus, (d) high local VRE/MRSA burden.
		High CRO risk b : Add targeted CR pathway depending on likely mechanism in patients with high CRO risk b : (A) Suspected CRE (KPC/OXA‐48‐like): consider ceftazidime–avibactam; (B) Suspected MBL‐producer: ceftazidime–avibactam + aztreonam; (C) Suspected CRAB: high‐dose ampicillin–sulbactam ± second agent (polymyxin or institutional policy)	Low; Conditional	Polymyxin/tigecycline as default empiric CR coverage is increasingly toxic. Contemporary AMR guidance emphasizes mechanism‐directed therapy for CRE/CRAB and explicitly supports CAZ‐AVI + aztreonam approaches for MBL settings and CRO management frameworks [16]. Use as empiric only when probability is high. CRO is noted up to ~25% of cases in nosocomial SBP without shock.
SBP presenting with shock (septic shock)	Gram‐negative—77.9% Gram‐positive—27.5% MDR—68.8% XDR—9.1% PDR—3.7% ESBL—52.3% CRO—33.9% MDR GP—11.0%	Meropenem + agent active against gram‐positive MDROs: e.g., vancomycin/teicoplanin If high risk for CROs b : add mechanism‐directed CR agent (above) and involve infectious disease expert early	Moderate; Strong	EASL emphasizes immediate empiric therapy [2]; severe infection/sepsis warrants broader initial therapy, with modification if poor early PMN response. Indian SBP patients with shock shows MDR ~68.8% and CRO ~33.9%, supporting broad initial cover and early reassessment or de‐escalation every 48–72 h.

Note: Do not default to MDR Gram‐positive cover for all SBP. Use it selectively (device‐related infection, known colonization, severe sepsis with high local MRSA/VRE, concurrent SSTI/pneumonia where GP pathogens are plausible). Early response check (48 h): failure of ascitic PMN to fall < 25% at 48 h suggests resistant pathogens/secondary peritonitis and should trigger therapy modification and evaluation. Aminoglycosides are discouraged as empiric therapy in SBP because of nephrotoxicity risk. Dosing note (general): Prefer extended infusion for TZP/meropenem in sepsis/ACLF when feasible; adjust for renal function. Re‐evaluate every 48–72 h. De‐escalate the antimicrobials as per the microbiological reports to narrower spectrum and sensitive antibiotics.

Abbreviations: CA, community‐acquired; CAZ‐AVI, ceftazidime + avibactum; CRAB, carbapenem‐resistant Acinetobacter baumanii; CRE, carbapenem‐resistant Enterobacterales; CRO/CR‐bacteria, carbapenem‐resistant organisms; ESBL, ESBL‐producing Enterobacterales; HCA, healthcare‐associated; MDRO, multidrug‐resistant organisms; MRSA/VRE, methicillin‐resistant S. aureus /vancomycin‐resistant Enterococci; TZP, tazobactum + piperacillin.

^a Risk factors for MDROs (to upgrade CA SBP to broader therapy): recent hospitalization in 3 months, receipt of broad‐spectrum antibiotics in last 3 months, quinolone prophylaxis, invasive procedures or devices, second infection, multisite or polymicrobial infection, high organ failure scores, severity of ACLF, circulatory failure, ventilator support, and known colonization with MDROs.

^b Risk factors for CROs (to trigger CR‐pathway consideration): circulatory failure, quinolone prophylaxis, multisite infection, ventilator support, prior broad‐spectrum antibiotics, infection as acute precipitant, known colonization with CRO, and likely mechanism may be derived from genotypic expression of colonizing bacteria or local epidemiological patterns.

4. Discussion

Spontaneous bacterial peritonitis (SBP) remains a prevalent complication in cirrhosis and is associated with high short‐term mortality of 50% in 6 months [17, 18]. In routine practice, empiric therapy is often necessary because ascitic culture positivity is limited, yet rising antimicrobial resistance is reducing the reliability of standard regimens. In this multicentre Indian cohort of hospitalized cirrhosis patients with culture‐positive spontaneous ascitic infections (n = 319), we observed a high burden of MDRO (53.6%) and CRO (24.1%), with important implications for early empiric antibiotic choice, escalation strategy and stewardship.

More than half of culture‐positive ascitic infections were caused by MDROs, and one‐quarter were carbapenem‐resistant. Gram‐negative organisms predominated (74.6% overall), and gram‐negative infection was more frequent among MDRO cases. This is in line with previous studies from India that reported a prevalence of MDR infections in patients with cirrhosis from 49% to 60% [19, 20]. However, the global estimates vary across income categories and settings with average estimate of 34% [1, 3]. A global epidemiological study reported infections in cirrhosis from India with high prevalence of 73% MDRO with ESBL, Methicillin‐resistant Staphylococcus aureus (MRSA), and VRE to be the most common organisms [1]. Majority patients had a gram‐negative organism causing SBP, seen in 74% patients. Among MDROs, E. coli (31.6%) was the most isolated organism followed by K. pneumoniae (20.5%), Enterococcus (14.6%), Staphylococcus spp. (8.8%), A. baumanii (8.2%), Pseudomonas (7%) and Burkholderia (6.4%). Among patients without MDROs, the most isolated organism was E. coli (19.6%), followed by Staphylococcus spp. (14.8%), A. baumanii (12.8%), Streptococcus (11.5%), K. pneumoniae (10.1%), Pseudomonas spp. (10.1%), Burkholderia (8.8%), Enterococcus (5.4%) and Enterobacter (4.7%). Our results are in line and additionally demonstrate SBP specific estimates and baseline probabilities to guide empiric decisions in cirrhosis [1].

In our cohort, short‐term outcomes were poor overall survival, reflecting a high acuity inpatient population, with ACLF in 65.8% and substantial vasopressor (34.2%) and ventilator (21.9%) requirements. Importantly, MDRO infection was associated with worse outcomes, including lower 30‐day survival (53.2% vs. 73.6%) and lower 90‐day survival (51.5% vs. 70.9%) compared with non‐MDRO infections. These findings confirm previous observations [21] and underscore that resistance in ascitic infections is not a microbiological detail; it is a clinically relevant determinant of prognosis in hospitalized cirrhosis. Clinically severe disease, as determined by MELD‐Na score and CLIF‐C OF score and presence of organ failures like liver failure and respiratory failure, independently predicted mortality along with presence of MDRO infection. Absence of systemic inflammation had a protective effect on short‐term mortality. These findings reinforce a clinically important point that while resistance influences outcomes, early mortality is also strongly driven by organ failure and systemic inflammation. Therefore, antibiotic strategy must be integrated with timely recognition of sepsis physiology and aggressive supportive care, particularly respiratory and circulatory support.

A key contribution of this study is the identification of pragmatic clinical predictors that can support risk‐based empiric therapy. In multivariable analysis, MDRO infection was independently associated with nosocomial/healthcare‐associated acquisition, circulatory failure, higher leukocyte count, SBP prophylaxis and type of hospital admission. CRO infection showed a similar pattern, with nosocomial/healthcare‐associated acquisition, circulatory failure and higher leukocyte count as independent predictors. These predictors are readily available at presentation and can be used to define ‘high resistance‐risk’ phenotypes where broader initial empiric coverage may be appropriate while avoiding unnecessary escalation in lower‐risk patients. Emerging data also links SBP prophylaxis with the rising burden of MDRO infections [22]. In our analysis, MDRO risk was associated with quinolone prophylaxis rather than rifaximin use, consistent with findings from previous studies [23]. Rifaximin was not independently linked to MDRO infections. Although it was more frequently used in patients with nosocomial/healthcare‐associated SBP, likely reflecting greater disease severity and use for hepatic encephalopathy management rather than a direct effect on SBP prevention.

Community‐acquired infection was the most common acquisition category (42.9%), yet MDRO and CRO rates were substantially higher in nosocomial/healthcare‐associated infections (MDRO 62.6% vs. 41.6%; CRO 30.2% vs. 16.1%). This supports current guideline principles that acquisition setting matters. However, survival did not differ by acquisition category in this cohort. Instead, MDRO, CRO, shock/organ failure and ACLF severity appeared to be stronger determinants of outcome. Likewise, CRO infections are associated with mortality in cirrhosis [12]. This reinforces the need to move beyond acquisition setting alone and adopt a combined risk stratification approach based on both resistance risk and illness severity.

A major and clinically actionable result is that 96 patients (approximately one‐third) had culture‐positive ascitic infection with ANC < 250 cells/μL, yet 30‐ and 90‐day mortality were similar to patients meeting the standard ANC threshold for SBP. While patients with ANC ≥ 250 were more unwell by several severity measures, outcomes remained comparable. Our findings align with previously published literature demonstrating comparable 6‐month mortality rates between patients with SBP and NNBA [24]. Several mechanisms can explain this similarity despite the apparent lower systemic inflammatory burden in patients with NNBA. First, a proportion of patients with NNBA would have received intravenous antibiotics prior to hospitalization or diagnostic paracentesis, partially improving the ANC to levels below 250 cells/mm³, while residual infection persists. Second, NNBA can represent an early stage of bacterial translocation that can progress to SBP on repeat ascitic fluid analysis in a subset of patients leading to similar long‐term outcomes. Third, although SBP was associated with a higher incidence of circulatory failure in our cohort, MELD score was comparable in both groups, and a similar proportion of patients in each group received appropriate empiric antibiotics. These factors likely mitigated the expected excess mortality in patients with SBP, shifting the primary determinants of prognosis towards underlying cirrhosis severity, rather than the degree of local ascitic fluid inflammation alone. Our results suggest that in hospitalized patients with cirrhosis, culture‐positive ascitic infection should not be dismissed as ‘low‐risk’ solely because ANC is below 250. In an era of high resistance burden, restricting treatment decisions to ANC thresholds risks delayed or inadequate therapy, and may weaken stewardship decisions by under‐recognizing clinically meaningful infection. These data support recognizing culture‐positive infection with ANC < 250 as a clinically important phenotype requiring prompt risk factor identification and context‐appropriate antibiotics in the presence of signs and symptoms of infection.

We observed that a higher proportion of MDRO cases came from public sector hospitals. This likely reflects differences in referral patterns, acuity, crowding and healthcare exposure rather than a single causal factor. The correct inference is not that ‘public hospitals cause MDRO’, but that the resistance burden may cluster in settings managing higher‐acuity patients and higher healthcare exposure‐exactly the environments where stewardship, infection prevention and surveillance are most essential. This finding should be framed as a systems‐level signal requiring targeted stewardship and IPC strengthening rather than as an attribution of fault.

Strengths include the multicentre design including 13 centres, a large culture‐positive cohort, detailed organism and resistance characterization, and linkage to clinically relevant outcomes including short‐term survival. The comparison between SBP and culture‐positive ANC < 250 infections is a particular strength because it addresses a common clinical dilemma. Limitations include the tertiary referral setting with hospitalized setting, which may inflate the observed resistance burden compared to outpatient or lower‐acuity settings. Some potential confounders were not captured (e.g., specific prior antibiotic exposures, proton pump inhibitor use, sarcopenia, renal replacement therapy), and SIRS may be imperfect in cirrhosis. Despite these limitations, the consistency of the resistance signal, the clear association of MDRO/CRO with survival, and the clinically actionable predictors support the broader relevance of the findings for inpatient cirrhosis care in similar settings.

In conclusion, among hospitalized Indian patients with culture‐positive spontaneous ascitic infections, MDRO (53.6%) and CRO (24.1%) were frequent and associated with worse survival. Nosocomial/healthcare‐associated acquisition, circulatory failure and leukocyte count consistently predicted MDRO/CRO infection, supporting risk‐based early stratification and targeted escalation. Early mortality was associated with MELD‐Na, MDRO status, appropriateness of antibiotics, organ failures and systemic inflammation, highlighting the need to integrate right antibiotics with organ support and immune modulating therapies. Nearly one‐third of culture‐positive infections had ANC < 250 yet had similar 30‐ and 90‐day mortality to SBP, supporting treatment decisions that give substantial weight to microbiological evidence and clinical context rather than ANC threshold alone. Lastly, empiric regimens should be individualized based on acquisition setting, resistance risk, and severity phenotype, with early reassessment and de‐escalation guided by cultures and clinical response.

Author Contributions

Nipun Verma: conceptualization, data curation, investigation, methodology, project administration, supervision, validation, writing – original draft, writing – review and editing. Rohit Mehtani: writing – original draft, visualization, methodology, formal analysis, writing – review and editing. Arun Valsan: methodology, investigation, writing – review and editing. Pratibha Garg: data curation, investigation, writing – review and editing. Rohit Kumar Nadda: investigation, writing – review and editing, methodology, formal analysis. Shilpa Prasad: investigation, writing – review and editing. Anand Sharma: investigation, writing – review and editing. Rohit Gupta: investigation, writing – review and editing. Balram Ji Omar: investigation, writing – review and editing. Naga Srikanth Darla: investigation, writing – review and editing. Roshan Agarwala: investigation, writing – review and editing. Mrinal Deb Barma: investigation, writing – review and editing. Mandira Sarma: investigation, writing – review and editing. Ameet Mandot: investigation, writing – review and editing. Wasim Khot: investigation, writing – review and editing. investigation, writing – review and editing. Shalimar: investigation, writing – review and editing. Akash Shukla: investigation, writing – review and editing. Arun Vaidya: investigation, writing – review and editing. Mithra Prasad: investigation, writing – review and editing. Dibyalochan Praharaj: investigation, writing – review and editing. Anil C. Anand: investigation, writing – review and editing. Radha Krishan Dhiman: investigation, writing – review and editing. Surender Singh: investigation, writing – review and editing. Narendra Singh Choudhary: investigation, writing – review and editing. Anand Kulkarni: investigation, writing – review and editing. Sandeep Satsangi: investigation, writing – review and editing. Sahaj Rathi: investigation, writing – review and editing. Arka De: investigation, writing – review and editing. Madhumita Premkumar: investigation, writing – review and editing. Sunil Taneja: investigation, writing – review and editing. Virendra Singh: investigation, writing – review and editing. Archana Angrup: investigation, writing – review and editing. Neelam Taneja: investigation, writing – review and editing. Nusrat Shafiq: investigation, writing – review and editing. Ajay Duseja: investigation, writing – review and editing, supervision.

Funding

The authors have nothing to report.

Supporting information

Figure S1: Flow chart depicting classification and prognostication of ascitic fluid infections stratified by mode of acquisition of infection and presence/absence of shock.

Figure S2: Proportion of multidrug‐resistant organism (MDRO) among different bacterial species.

Figure S3: Love plot showing covariate balance before and after propensity score matching.

Table S1: Centres involved in the study.

Table S2: Multivariable logistic regression analysis for predicting multidrug‐resistant infection.

Table S3: Univariable logistic regression analysis for predictors of carbapenem‐resistant organism (CRO).

Table S4: Multivariable logistic regression analysis for predictors of carbapenem‐resistant organism (CRO).

Table S5: Details of organism involved stratified by MDR status.

Table S6: Baseline characteristics of patients with and without community‐acquired SBP.

Table S7: Microbiological profile of ascitic fluid infections stratified by mode of infection acquisition.

Table S8: Details of organism involved stratified by mode of acquiring SBP.

Table S9: Univariable cox regression analysis for predictors of 30‐day mortality.

Table S10: Multivariable cox regression analysis for predictors of 30‐day mortality.

Table S11: Propensity scores matching analysis for comparison of culture‐negative neutrocytic ascites with spontaneous bacterial peritonitis (culture positive with ascitic fluid neutrophil count ≥ 250 cells/mm³).

Table S12: Covariate balance before and after propensity score matching.

Verma N., Mehtani R., Valsan A., et al., “Epidemiology, Predictors of Antimicrobial Resistance and Empiric Treatment Strategies for Spontaneous Bacterial Peritonitis in Cirrhosis: A Multicentre SBP‐INDIA Study,” Alimentary Pharmacology & Therapeutics 64, no. 4 (2026): 479–498, 10.1111/apt.70716.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.