Evolving bile pathogen profiles and antimicrobial resistance in patients undergoing ERCP, with a focus on acute biliary pancreatitis: a 12-year retrospective study

Abstract

Background

Long-term surveillance data on biliary pathogen profiles and antimicrobial resistance (AMR) are limited. Moreover, risk factors for multidrug resistance (MDR) in acute biliary pancreatitis (ABP) remain unclear.

Methods

We retrospectively analyzed 1,377 bile culture results from patients who underwent ERCP at a tertiary care center between January 2012 and December 2023. Pathogen distribution and antimicrobial resistance trends were evaluated in the full cohort. In a predefined ABP subgroup (n = 180), risk factors for multidrug resistance (MDR) were identified using univariate and multivariable logistic regression.

Results

A total of 1,377 patients undergoing ERCP with intra-procedural bile culture were included, among whom 859 (62.4%) had positive cultures, yielding 956 non-duplicate isolates. Gram-negative bacteria predominated overall, followed by Gram-positive bacteria and fungi. Over the 12-year study period, significant temporal shifts in biliary pathogen distribution were observed, characterized by a decreasing proportion of Escherichia coli and a gradual increase in Enterococcus species. Antimicrobial resistance profiles demonstrated dynamic changes over time, including an increasing prevalence of extended-spectrum β-lactamase (ESBL)–producing Enterobacterales, whereas resistance to carbapenems remained relatively low. In the acute biliary pancreatitis (ABP) subgroup, multidrug-resistant (MDR) organisms were identified in a substantial proportion of bile culture–positive cases.

Conclusions

This study provides long-term epidemiological insights into changes in biliary pathogen profiles and antimicrobial resistance patterns among patients undergoing ERCP. Within the ABP subgroup, several factors were found to be statistically associated with the presence of multidrug-resistant organisms. These findings highlight the potential value of integrating microbiological surveillance with individualized risk assessment to inform antimicrobial decision-making in ABP, while emphasizing that further prospective validation is required.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-026-04760-y.

Background

Biliary tract infections have become an increasing concern in the context of antimicrobial resistance (AMR) over the past decade [1]. Patients undergoing endoscopic retrograde cholangiopancreatography (ERCP) are particularly vulnerable, as biliary obstruction, bile stasis, and endoscopic manipulation facilitate bacterial colonization of bile and subsequent infection [2].

Microbiological surveillance of bile cultures in ERCP patients provides valuable data on pathogen distribution and resistance patterns in real-world clinical settings. However, the composition of biliary pathogens and their resistance profiles are strongly influenced by local epidemiology, antibiotic prescribing practices, and underlying patient conditions, resulting in marked regional heterogeneity and temporal variation [3]. For example, a center in northern China reported that Gram-negative bacilli accounted for 76.8% of bile isolates between 2011 and 2018, with Gram-positive cocci comprising 22.5% [4]. In contrast, a more recent study observed a year-on-year increase in Gram-positive isolates and a decline in Gram-negative organisms in bile samples over the past five years [5]. Most published studies on biliary pathogen distribution and resistance patterns are cross-sectional or involve relatively small cohorts, and long-term continuous surveillance data remain limited.

Among patients undergoing ERCP, acute biliary pancreatitis (ABP) represents a distinct and clinically significant subgroup. ABP is one of the most common forms of acute pancreatitis worldwide and often involves biliary obstruction and infection. Although empirical antibiotics are frequently administered in ABP cases, there is limited evidence regarding pathogen profiles and MDR risk factors in this specific population [6]. Identifying predictors of MDR infections in ABP is critical for optimizing early therapy and preventing adverse outcomes.

Therefore, the present study had two main objectives: (1) to characterize the overall pathogen spectrum and antimicrobial resistance trends in bile cultures from patients undergoing ERCP over a 12-year period (2012–2023), and (2) to evaluate clinical predictors of MDR infection specifically in the ABP subgroup. This dual-level design enables both broad epidemiological insight and focused clinical insight for a high-risk patient population.

Materials and methods

Study design and patient population

This retrospective study included all inpatients who underwent ERCP and bile culture at our center from January 2012 to December 2023. We predefined a subgroup of patients with ABP, diagnosed based on the revised Atlanta criteria and imaging-confirmed biliary etiology. The full cohort was used to analyze pathogen distribution and antimicrobial resistance trends. The ABP subgroup was used for risk factor analysis of MDR infection.

Inclusion and exclusion criteria

Patients were eligible for inclusion if they met both of the following criteria:

1. At least one valid bile culture obtained during ERCP with a definitive microbiological result (positive or negative).

2. Availability of key clinical information required for analysis.

Exclusion criteria were:

1. Pre-existing chronic infection prior to admission.

2. Chronic organ failure at baseline (Chronic organ failure was defined as established end-stage or decompensated organ dysfunction present before the index admission, including but not limited to end-stage renal disease requiring dialysis, decompensated cirrhosis/chronic liver failure, or advanced heart/respiratory failure requiring long-term oxygen therapy or recurrent hospitalization).

3. Age < 18 years.

4. Bile samples considered to be contaminated based on microbiological assessment and clinical judgement.

Data collection and variables

For the ABP subgroup, clinical variables potentially associated with MDR infection were collected, including age, comorbidities, duration of antibiotic therapy, ICU admission, and use of third-generation cephalosporins.

Variables were selected based on clinical relevance and literature review and tested in univariate analysis before entering multivariable models. Variables that were strongly collinear or considered to be outcome-dependent (e.g., length of stay) were excluded.

The variable 'duration of antibiotic therapy prior to bile culture' was defined as the cumulative number of days the patient received systemic antibiotics for the current episode of ABP prior to the index ERCP. This calculation included documented antibiotic use at referring hospitals for transferred patients and use at our center prior to the procedure.

Bile sampling and microbiological processing

Bile samples were collected intra-procedurally during ERCP under sterile conditions after successful cannulation of the bile duct, either through direct aspiration via the catheter or via nasobiliary drainage when placed as part of routine clinical management. Whenever feasible, bile sampling was performed prior to modification to antimicrobial therapy. Samples were immediately transported to the microbiology laboratory for culture and antimicrobial susceptibility testing according to standard protocols.

Bile samples were routinely inoculated into aerobic culture media. However, anaerobic cultures were not performed routinely during the study period, and thus the pathogen profiles reported herein primarily represent aerobic and facultative anaerobic bacteria.

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing (AST) was performed using an automated microdilution system (VITEK 2 Compact, bioMérieux, Marcy-l’Étoile, France) according to the manufacturer’s instructions. Minimum inhibitory concentrations were interpreted in accordance with the Clinical and Laboratory Standards Institute (CLSI) breakpoints valid for the corresponding year of testing.

A panel of commonly used agents covering the predominant Enterobacterales and Enterococcus species was analyzed, including third-generation cephalosporins (e.g., cefotaxime, ceftriaxone, and ceftazidime), β-lactam/β-lactamase inhibitor combinations (e.g., piperacillin–tazobactam, and ampicillin–sulbactam), fluoroquinolones (e.g., levofloxacin), carbapenems (e.g., imipenem), and aminoglycosides, as well as vancomycin, linezolid, and tigecycline for Gram-positive cocci. These representative antibiotics were selected to describe temporal trends in resistance among the most frequently isolated Enterobacterales and enterococci.

Isolates suspected of producing extended-spectrum β-lactamases (ESBLs) based on the screening profile were confirmed using a phenotypic enzyme inhibition test with cefotaxime/clavulanate or ceftazidime/clavulanate.

Definition of non-duplicate isolates

To avoid overrepresentation of repeated cultures, non-duplicate isolates were defined as the first isolate of a given bacterial species recovered from a single ERCP procedure in each patient. When multiple bacterial species were isolated from the same bile sample, each species was included separately in the analysis. Repeated isolation of the same species from the same patient during the same procedure was excluded.

Definition of multidrug resistance

The definition of MDR followed the international expert proposal by Magiorakos et al. MDR was defined as acquired non-susceptibility to at least one agent in three or more antimicrobial categories. This definition was applied uniformly to Enterobacterales, Enterococcus spp., and non-Enterobacterales isolates in the present study.

Definition and presentation of ESBL and VRE phenotypes

Extended-spectrum β-lactamase (ESBL) production among Enterobacterales was determined according to CLSI-recommended phenotypic criteria valid for the corresponding year of testing. Vancomycin-resistant Enterococcus (VRE) was defined based on antimicrobial susceptibility testing results interpreted using CLSI breakpoints. ESBL- and VRE-positive isolates were summarized descriptively and evaluated for temporal trends as part of the resistance phenotype analysis.

Statistical considerations and EPV clarification

For multivariable logistic regression analyses, the number of covariates included in the final models was limited to minimize the risk of overfitting, given the number of outcome events. Candidate variables were selected based on clinical relevance and data availability rather than automated stepwise procedures. Outcome-dependent variables, such as length of hospital stay, were excluded from multivariable models and were considered only in descriptive or univariable analyses. The results of regression analyses should therefore be interpreted as exploratory and hypothesis-generating.

Furthermore, while our multivariable model identified key risk factors, the number of MDR events (n = 44) relative to the number of variables analyzed (n = 5) results in an EPV of approximately 8.8. This is slightly below the ideal threshold of 10, which may increase the risk of model overfitting. Consequently, the associations observed—particularly the high OR for third-generation cephalosporins—should be interpreted with caution and warrant further validation in larger prospective cohorts.

Statistical analysis

All statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA) and R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). Continuous variables are presented as mean ± standard deviation (SD) or median with interquartile range (IQR), as appropriate, and were compared between groups using the Student’s t-test or the Mann–Whitney U-test. Categorical variables are expressed as counts and percentages and were compared using the χ2 test or Fisher’s exact test. To assess time-dependent changes, the 12-year study period (2012–2023) was divided into three consecutive 4-year intervals: 2012–2015, 2016–2019, and 2020–2023. This stratification ensured comparable segment lengths and facilitated trend analysis across early, middle, and recent phases of the study period. Temporal trends in detection rates and resistance proportions over the three predefined time periods were assessed using the Cochran–Armitage trend test. Univariable and multivariable logistic regression analyses were used to identify independent risk factors for MDR and for bile culture positivity in ABP patients; odds ratios (ORs) and 95% confidence intervals (CI) were reported. Forest plots and trend line graphs were generated in R using the ggplot2 package. A two-sided P value < 0.05 was considered statistically significant.

Results

Overall distribution and temporal trends of bile pathogens

A total of 1,377 patients who underwent bile culture, including cases of cholangitis and biliary pancreatitis, were analyzed. Among them, 859 cultures were positive, yielding a positivity rate of 62.38%, with 956 non-duplicate isolates identified. Gram-negative bacteria were the most commonly isolated pathogens (74.06%), followed by Gram-positive bacteria (23.95%) and fungi (1.99%). The most frequently detected species were Escherichia coli (44.87%) and Enterococcus faecium (19.14%). (Table 1).

Table 1.

Distribution of major bile pathogens and resistance phenotypes across three time intervals (2012–2023)

Pathogen category/species	2012–2015	2016–2019	2020–2023
	Isolates (n) and Proportion (%)	Isolates (n) and Proportion (%)	Isolates (n) and Proportion (%)
Gram-negative bacteria
Escherichia coli	174(50.73)	127(55.22)	132(34.46)
Klebsiella pneumoniae	28(8.16)	44(19.13)	60(15.67)
Enterobacter cloacae	12(3.50)	4(1.74)	72(18.8)
Pseudomonas aeruginosa	26(7.58)	0(0)	0(0)
Acinetobacter baumannii	13(3.79)	4(1.74)	12(3.13)
Gram-positive bacteria
Enterococcus faecium	68(19.83)	35(15.22)	80(20.89)
Enterococcus faecalis	18(5.25)	8(3.48)	20(5.22)
Fungi
Candida albicans	4(1.17)	8(3.48)	7(1.83)
Total	343(100)	230(100)	383(100)

Over the 12-year period, the proportion of E. coli showed a significant decreasing trend (Cochran–Armitage trend test, χ2 = 20.09, P < 0.001), whereas the detection rate of Enterococcus faecium demonstrated a significant upward trend (Cochran–Armitage trend test, χ2 = 3.96, P = 0.046) (Fig. 1).

Fig. 1.

Temporal shifts in the composition of biliary pathogens across three periods (2012–2023)

Antimicrobial resistance patterns of the major pathogens

A total of 433 Escherichia coli and 183 Enterococcus faecium isolates were included in the antimicrobial resistance analysis across the three predefined study periods (2012–2015, 2016–2019, and 2020–2023).

Among E. coli isolates, resistance profiles exhibited pronounced time-dependent changes. Resistance to several third- and fourth-generation cephalosporins, including cefotetan, ceftriaxone, cefuroxime, ceftazidime, and cefepime, increased or fluctuated significantly across the three study periods, remaining at relatively high levels during the middle and recent periods (all P < 0.05). In contrast, resistance to imipenem, amikacin, piperacillin, ampicillin–sulbactam, and trimethoprim–sulfamethoxazole remained largely stable throughout the study period, without clear increasing or decreasing trends (P > 0.05). Notably, resistance to levofloxacin showed a continuous decline over time and was significantly lower in the most recent period (P = 0.001). In addition, resistance to piperacillin–tazobactam and tobramycin decreased markedly from the early to the middle period and remained low thereafter (both P < 0.05), whereas the downward trend in gentamicin resistance did not reach statistical significance (P = 0.057) (SupplementaryTable 1; Fig. 2).

Fig. 2.

Heatmap showing 12-year trends in antimicrobial resistance of Escherichia coli to key antibiotics

For Enterococcus faecium, resistance to glycopeptides and oxazolidinones remained consistently low throughout the study period. No significant temporal changes were observed in resistance rates to tigecycline, vancomycin, or linezolid (P > 0.05). In contrast, resistance to several commonly used agents demonstrated clear time-dependent variation. Resistance to gentamicin, penicillin, and ampicillin gradually decreased over time, whereas resistance to tetracycline, nitrofurantoin, and levofloxacin increased significantly in the most recent period (all P < 0.05). Resistance to erythromycin and ciprofloxacin remained persistently high across all three periods, without statistically significant upward or downward trends (P > 0.05) (SupplementaryTable 2; Fig. 3).

Fig. 3.

Heatmap showing 12-year trends in antimicrobial resistance of Enterococcus faecium to key antibiotics

Exact values are provided in Supplementary Table 1.

Exact values provided in Supplementary Table 2.

Pathogen distribution in patients with ABP

A total of 180 patients met the inclusion criteria for ABP, with a mean age of 55.02 ± 16.95 years; 50.0% were male. The proportion of severe ABP was 17.8% based on the revised Atlanta classification. Among these patients, 64 had positive bile cultures during ERCP, corresponding to a positivity rate of 35.6%.

The leading pathogens in ABP patients were E. coli (29.7% of isolates) and Enterococcus faecium (24.3%), followed by Klebsiella pneumoniae, Enterobacter cloacae, and Candida species. (Fig. 4).

Fig. 4.

Distribution of biliary pathogens in patients with ABP

Antimicrobial resistance profiles in ABP patients

Among Gram-negative isolates from ABP patients, E. coli and K. pneumoniae showed resistance rates exceeding 75% to third-generation cephalosporins, including ceftriaxone, cefuroxime and ceftazidime, as well as to piperacillin. Resistance to levofloxacin was also high, whereas most isolates remained susceptible to imipenem and amikacin. Enterobacter cloacae exhibited a resistance rate of 60.0% to ceftazidime but retained good susceptibility to carbapenems and aminoglycosides (Supplementary Table 3; Fig. 5).

Fig. 5.

Heatmap showing resistance profiles of Gram-negative isolates in ABP

Among the Gram-positive isolates, Enterococcus faecium was predominant. It exhibited high resistance rates to penicillin and ampicillin (77.8%), as well as to levofloxacin and erythromycin (both exceeding 80%). However, all isolates remained fully susceptible to tigecycline, vancomycin, and linezolid (100% susceptibility) (Supplementary Table 4; Fig. 6).

Fig. 6.

Heatmap showing resistance profiles of Gram-positive isolates in ABP

Exact values are provided in Supplementary Table 3.

Exact values provided in Supplementary Table 4.

Exploratory analysis of factors associated with multidrug resistance

Given the sample size constraints, the following multivariable analysis should be interpreted as exploratory. Of the 64 ABP patients with positive bile cultures, 44 (68.8%) were infected with MDR organisms according to the predefined criteria. In the present cohort, univariable analysis identified several potential risk factors for MDR, including duration of antibiotic therapy, the use of third-generation cephalosporins, and the presence of severe pancreatitis (Table 2). Variables with P < 0.10 in univariable analysis were entered into a multivariable logistic regression model. Multivariable analysis identified duration of antibiotic therapy (OR 1.15, P = 0.007) and prior use of third-generation cephalosporins (OR 3.02, P < 0.001) as independent risk factors for MDR. Other variables in the model, including severe pancreatitis and pancreatic necrosis, did not reach statistical significance (P > 0.05).(Fig. 7).

Table 2.

Univariate analysis of risk factors associated with MDR in ABP patients

Variable	MDR group (n = 44)	Non-MDR group (n = 20)	Test Statistic	P-value
Age (years)	59.87 ± 17.71	55.60 ± 15.09	0.939	0.351
Number of antibiotics (n)	1(1–2)	1(1–1)	−2.014	0.044
Duration of antibiotic therapy (days)	9(5–16)	4(3–4)	−4.246	0.001
Male (%)	26(59.1)	10(50.0)	0.239	0.625
Immunodeficiency (%)	8(18.2)	3(15.0)	0.058	0.809
Use of third-generation cephalosporins (%)	28(63.6)	5(25.0)	6.973	0.008
Hypoxemia (%)	24(54.5)	10(50.0)	0.026	0.871
History of ERCP (%)	18(40.9)	3(15.0)	3.741	0.053
Prior cholecystectomy (%)	12(27.3)	6(30.0)	0.108	0.743
Diabetes mellitus (%)	12(27.3)	3(15.0)	0.446	0.504
Severe pancreatitis (%)	26(59.1)	6(30.0)	3.926	0.048
Pancreatic necrosis (%)	30(68.2)	4(20.0)	11.411	0.001

Fig. 7.

Forest plot of logistic regression analysis of risk factors for MDR

Predictors of bile culture positivity in ABP

In this study, we examined 11 commonly cited clinical variables as potential predictors of positive bile culture in ABP, including age, sex, body mass index, history of ERCP, hypoxemia, diabetes mellitus, prior cholecystectomy, SIRS, and pancreatic necrosis (Table 3).

Table 3.

Univariate analysis of risk factors associated with positive bile culture

Variable	Positive group (n = 64)	Negative group (n = 116)	Test Statistic	P-value
Age (years)	58.47 ± 17.22	53.12 ± 16.58	2.044	0.042
Body mass index (kg/m2)	23.53 ± 3.03	24.41 ± 3.83	−1.569	0.119
Male (%)	36(56.25)	54(46.6)	1.552	0.213
Hypoxemia (%)	34(53.13)	54(46.6)	0.713	0.398
History of ERCP	21(32.81)	8(6.9)	20.496	0.001
Prior cholecystectomy (%)	16(25.0)	15(12.9)	4.214	0.040
Diabetes mellitus (%)	15(23.4)	12(10.3)	5.545	0.019
SIRS (%)	14(21.9)	12(10.3)	4.437	0.035
Severe pancreatitis (%)	32(50.0)	36(31.0)	6.311	0.012
Pancreatic necrosis (%)	34(53.1)	18(15.5)	28.396	0.001
Immunodeficiency (%)	9(14.1)	7(6.0)	3.282	0.070

Univariable analysis identified several variables associated with bile culture positivity. These candidates were then included in a multivariable logistic regression model. Pancreatic necrosis, history of ERCP, and prior cholecystectomy emerged as independent risk factors for a positive bile culture in ABP patients (P < 0.05 for all; Fig. 8).

Fig. 8.

Forest plot of multivariable logistic regression analysis identifying risk factors for positive bile culture

Discussion

This study systematically analyzed 1,377 intraoperative bile culture results collected over a 12-year period, comprehensively characterizing the spectrum of biliary pathogens and the long-term trends in their antimicrobial resistance. On this basis, we further evaluated the factors associated with the development of MDR in patients with ABP. It should be noted that the overall ERCP bile culture cohort was used to assess temporal trends in pathogen spectrum and resistance, whereas the analysis of MDR risk factors was confined to the ABP subgroup. This design not only reflects the general epidemiological characteristics but also addresses clinical relevance in a specific high-risk population.

In the overall ERCP bile culture cohort, we observed significant changes in pathogen composition and resistance profiles over time: the proportion of Gram-positive bacteria (exemplified by Enterococcus faecium) steadily increased each year, whereas Gram-negative bacteria (predominantly Escherichia coli) showed a downward trend. This shift is consistent with previous domestic studies. For example, a surveillance study in Sichuan Province covering 2017–2023 reported a similar temporal change in biliary pathogens, accompanied by a gradual increase in resistance [7]. Furthermore, another study from a large tertiary hospital found that the isolation rate of Gram-positive bacteria in bile increased annually and that higher resistance rates were significantly associated with a history of prior biliary intervention and antibiotic exposure [8]. In recent years, multiple studies have shown that Enterococcus has gradually transitioned from an opportunistic pathogen to a dominant genus in the biliary microbiota, typically exhibiting high levels of drug resistance, suggesting that empirical antimicrobial therapy for biliary infections may need to be adjusted accordingly [9–11].

Notably, our study found a significantly higher detection rate of Enterobacter cloacae compared to reports from other regions, clearly exceeding the national average and indicating marked geographic variation in the biliary pathogen spectrum. Although E. cloacae has historically been considered a low-virulence environmental bacterium, its incidence in hospital infections has been steadily rising in recent years and is often accompanied by high levels of antimicrobial resistance [12]. Molecular epidemiological studies in China further show that this species exhibits high genetic diversity, with the proportion of MDR strains reaching 54.3% and carbapenem resistance up to 8.7% [13]. This finding suggests that in certain regions, some traditionally “non-dominant” bacteria may have gradually evolved into resistant pathogens of significant clinical importance.

Compared to reports in Western literature, the overall proportion of Gram-positive bacteria in our study was relatively high [1]. This difference is likely related to characteristics of the Chinese patient population, including a higher degree of population aging, rising prevalence of underlying diseases such as diabetes, and more widespread prior antibiotic exposure [14, 15]. These factors suggest that international empirical treatment recommendations based on Western patient data may need to be adjusted for Chinese patients to account for local epidemiological characteristics.

In our study, only a small number of bile cultures grew fungi, mainly Candida species. Considering the clinical presentations and laboratory data, we believe that in most cases these positive cultures reflected biliary colonization rather than true invasive fungal infection, since the patients lacked evidence of fungal cholangitis or disseminated candidiasis [16–18]. A positive bile culture for Candida alone is usually not sufficient to justify initiating antifungal therapy unless accompanied by persistent unexplained fever, immunosuppressive status, or other signs of invasive fungal infection. These findings suggest that, in the absence of clear clinical indications, routine antifungal coverage for Candida isolated from bile may not be necessary [19, 20].

Our 12-year analysis revealed a continuous evolution in the composition of biliary pathogens and their resistance patterns over time, along with pronounced regional differences. These results indicate that the microbiological profile of biliary infections is not static, and empirical treatment regimens based on historical or short-term data may fail to reflect the current actual risks. Therefore, in patients undergoing ERCP, ongoing and systematic biliary pathogen surveillance is not only diagnostically valuable but also crucial for dynamically updating local empirical antimicrobial strategies. Such surveillance provides an important basis for more precise antibiotic selection at the regional level.

Beyond the overall trends, further analysis of the ABP subgroup showed that a positive bile culture was significantly associated with a history of prior biliary surgery, especially ERCP procedures. Previous studies have shown that ERCP, particularly after endoscopic sphincterotomy, can cause duodenal fluid reflux and alter the biliary microenvironment, thereby affecting the composition of the biliary microbiota [11, 21]. Moreover, Enterococcus exhibits traits such as biofilm formation, carriage of multiple resistance determinants, and the ability to survive long-term on the surface of biliary stents [22]. Consistently, Zhang et al. found that patients with biliary metal stents had a significantly higher isolation rate of Enterococcus compared to those without stents [5]. These findings suggest that biliary interventions and the stent materials themselves may create a niche for long-term colonization by such organisms and that with the widespread use of ERCP and stent placement, the biliary microenvironment could gradually shift toward a more drug-resistant and difficult-to-eradicate microbial community. Therefore, the increase in the detection rate of Enterococcus faecium observed in our study should be interpreted with caution. Although Enterococcus species are common biliary tract colonizers, particularly in patients with indwelling stents or those who have undergone biliary interventions, whether treatment is required for these isolated Enterococcus strains should be based on the patient's clinical presentation and the severity of the infection. However, in patients presenting with clinical symptoms, the role of Enterococcus as a pathogen cannot be overlooked. In our study cohort, the increase in Enterococcus faecium was often associated with the failure of empirical cephalosporin therapy against Enterococcus, suggesting treatment failure. This “selective pressure” is likely to promote the transformation of Enterococcus faecium from a benign colonizer to a dominant opportunistic pathogen. Pancreatic necrosis was also identified as an independent risk factor for positive bile cultures in ABP patients, which is in line with previous reports that necrotizing pancreatitis carries a higher risk of infection [23]. This suggests that some ABP patients constitute a high-risk group for infection and may require broader initial antimicrobial coverage in empirical therapy, whereas for lower-risk patients a relatively narrow-spectrum regimen may suffice [24, 25].

In our exploratory analysis of factors associated with multidrug resistance, we found that prolonged antibiotic use was an independent risk factor for the development of MDR in ABP patients. The median duration in the MDR group was 9 days (IQR 5–16), which is longer than the typical 24–72 h window for urgent ERCP in early-phase ABP. This discrepancy reflects the clinical profile of our tertiary referral center, which often manages patients transferred after initial treatment failure or those with delayed complications like infected necrosis. This underscores that for patients requiring late or complex interventions, the cumulative antimicrobial pressure is a primary driver of resistance, necessitating more vigilant stewardship. This finding is consistent with earlier reports that early, prolonged, or broad-spectrum antibiotic exposure can promote the emergence of resistant organisms [26, 27]. For example, Sawyer et al. demonstrated that shortening the duration of antimicrobial therapy in intra-abdominal infections significantly reduced the occurrence of resistant strains without adversely affecting clinical outcomes [28]. Similarly, other studies have recommended a risk-stratified empirical treatment strategy for biliary infection patients, with prompt de-escalation once pathogen identification and susceptibility results are available [29, 30], thereby better aligning with antimicrobial stewardship (AMS) principles.

Our study also identified prior exposure to third-generation cephalosporins as another independent risk factor for MDR in ABP patients. Notably, over the past 12 years, the resistance rate of Gram-negative bacteria to third-generation cephalosporins in our cohort has risen markedly. According to recent global antimicrobial resistance surveillance data and priority pathogen assessments, Enterobacteriaceae producing extended-spectrum β-lactamases (ESBLs) or resistant to third-generation cephalosporins (including E. coli) have been listed by the World Health Organization as priority resistant pathogens. In some low- and middle-income countries, resistance to third-generation cephalosporins is particularly problematic [31].

Given the limited number of MDR events, the estimates from our multivariable model should be interpreted with caution, as they may be unstable. Nonetheless, the observed associations are clinically plausible and align with known mechanisms of resistance selection. They provide a focused signal that warrants proactive investigation and stewardship in clinical settings, pending validation by larger prospective studies.

In the context of regional epidemiology and medical practice in China, the long-term, widespread use of third-generation cephalosporins as empirical therapy—combined with frequent referrals between different levels of healthcare facilities and increased biliary interventions—may have collectively amplified antibiotic selective pressure, thus promoting the emergence and spread of resistant strains. Based on our local data, routine empirical use of third-generation cephalosporins may warrant reconsideration in this setting, and they should be used for a limited duration only when clearly indicated.

From a clinical perspective, our findings underscore that antibiotic therapy for ABP should be guided by the evolving local pathogen spectrum while emphasizing individualized risk stratification. For high-risk patients, an initial broad-spectrum coverage aligned with local epidemiological patterns is advisable, with rapid de-escalation once pathogen and susceptibility results are obtained. In contrast, for low-risk patients, a narrow-spectrum regimen of limited duration is likely sufficient [25, 32, 33].

Limitations

This study has several limitations. First, its retrospective observational design precludes causal inference regarding antibiotic exposure and MDR, and residual confounding (e.g., by disease severity) cannot be excluded. Second, as a single-center study including only ERCP patients, the findings may not generalize to milder cases managed conservatively. The prolonged antibiotic exposure in our MDR group (median 9 days) reflects our tertiary referral population but may not represent early-phase ABP. Furthermore, excluding patients with chronic organ failure to ensure severity assessment validity may have underestimated the overall MDR prevalence. A key methodological limitation concerns the multivariable model identifying MDR risk factors. With only 44 MDR events for 5 predictor variables, the events-per-variable ratio was below the recommended threshold, increasing the risk of model overfitting and unstable estimates. Therefore, the reported associations—especially the strong odds ratio for third-generation cephalosporin use—must be interpreted with caution and are primarily hypothesis-generating. They highlight candidate risk factors requiring validation in larger prospective studies before clinical implications can be drawn. Third, while resistance trends were described, we did not perform molecular analyses (e.g., resistance gene detection, sequencing), so the genetic mechanisms and transmission dynamics of resistance remain unexplored. Finally, the clinical and economic impacts of different antimicrobial strategies were not assessed, which is important for informing stewardship programs and should be examined in future research [34, 35].

Conclusion

This study characterizes the long-term evolution of biliary pathogen distribution and antimicrobial resistance patterns in patients undergoing ERCP, providing updated epidemiological evidence from real-world clinical practice. In patients with ABP, several clinical and treatment-related factors were statistically associated with multidrug-resistant organisms, underscoring the heterogeneity of infection risk within this population.

Rather than supporting standardized treatment recommendations, these findings suggest that antimicrobial management in ABP may benefit from an individualized, risk-informed approach that incorporates local microbiological data. Further multicenter prospective studies are warranted to validate these associations and to determine their applicability in broader clinical settings.

Abbreviations

ABP

Acute biliary pancreatitis

AP

Acute pancreatitis

SIRS

Systemic inflammatory response syndrome

ICU

Intensive care unit

ERCP

Endoscopic retrograde cholangiopancreatography

AST

Antimicrobial susceptibility testing

AMR

Antimicrobial resistance

MDR

Multidrug resistance

AMS

Antimicrobial stewardship

ESBL

Extended-spectrum β-lactamase

VRE

Vancomycin-resistant Enterococcus

E. coli

Escherichia coli

E. faecium

Enterococcus faecium

E. faecalis

Enterococcus faecalis

GNB

Gram-negative bacteria

GPB

Gram-positive bacteria

TZP

Piperacillin–tazobactam

FEP

Cefepime

CAZ

Ceftazidime

CTX

Cefotaxime

CRO

Ceftriaxone

IPM

Imipenem

LVX

Levofloxacin

VAN

Vancomycin

LZD

Linezolid

TGC

Tigecycline

OR

Odds ratio

CI

Confidence interval

SD

Standard deviation

IQR

Interquartile range

Authors’ contributions

Zhao Chengsi: Conceptualization, methodology, data analysis, writing-original draft preparation, visualization. Wei Jie: Investigation, data collection, validation, writing, review and editing. Yao Weijie: Resources, formal analysis, investigation. Wang Zuozheng: Supervision, project administration, funding acquisition. All authors read and approved the final manuscript.

Funding

This research was supported by the Yinchuan Science and Technology Support Program (2025SF02); the High-level Scientific and Technological Innovation Leading Talent Project of the Ningxia Hui Autonomous Region (2021GKLRLX04); the Yinchuan Science and Technology Support Program (2024SF043); and the Natural Science Foundation of Ningxia (2024AAC03703).

Data availability

The datasets analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Ningxia Medical University General Hospital (Approval No: 2019–466). Because this was a retrospective study based on anonymized clinical data and bile samples collected during routine ERCP procedures for standard clinical care, the requirement for informed consent was waived by the Ethics Committee. All patient identifiers were removed prior to analysis to ensure data confidentiality and privacy protection.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.