Association of postoperative infection with early bilirubin clearance and native liver survival: a single - center retrospective cohort study Authors: Xuanyu Meng, Shuheng Liang, Rongjie Li, Li Huang, Qing Tang, Xiang Yun, Xiuqi Chen Published: 2026 ABSTRACT AbstractBackgroundApart from cholangitis, studies focusing on other postoperative infections after the Kasai procedure (KP) in children with biliary atresia (BA) remain limited.This study aimed to investigate the association between post-KP infection and prognosis.MethodsThis retrospective study included children with BA who underwent KP. Postoperative infection was the primary exposure variable of interest in this cohort. The primary outcome measure was post-operative native liver survival, while the secondary outcome measure was mortality attributed to infection. A time-dependent Cox regression model was used to account for the timing of postoperative infection as a time-varying covariate.ResultsOne. A total of 404 cases included in the study. There were no significant differences in gender, age at operation, or gestational age between the two groups (P > 0.05).Two. Early-onset infection is primarily associated with delayed bilirubin clearance (P < 0.01).Time-dependent Cox regression showed that later onset of infection was associated with better native liver survival (HR = 0.777, 95% CI: 0.719–0.839, P < 0.001). After accounting for measured confounders, each one-month later infection onset corresponded to a 22.3% lower hazard of death or liver transplantation. These findings should be interpreted as associations, not causal effects.ConclusionPostoperative infection in biliary atresia is associated with delayed bilirubin clearance and poorer native liver survival. The timing of infection onset is an important factor associated with native liver survival outcomes. Given the observational design, causality cannot be inferred.Keywords: Biliary atresia; Infection; Cholangitis; Liver function, Infants Introduction Biliary atresia (BA) is a disease of unknown etiology characterized by progressive inflammation and fibrotic obstruction of both intrahepatic and extrahepatic bile ducts. The prevalence of BA varies geographically worldwide, with East Asian countries exhibiting the highest incidence rates. Specifically, Japan has an incidence rate of 1.1–1.2/10,000 and Taiwan region of China has a rate of 1.51–1.54/10,000, while the United States has a lower range of 0.44–0.96/10,000 and European countries have similar rates ranging from 0.5 to 0.58/10,000 [1–7]. The primary clinical manifestations of BA include delayed resolution of jaundice after birth (in full-term infants > 2 weeks and preterm infants > 3 weeks), or recurrent episodes of jaundice with progressive worsening [8]. This may be accompanied by lightening of stool color until it becomes clay-colored, as well as liver and spleen enlargement, malnutrition, and delays in growth and development. Infection is a frequent complication following surgery, with cholangitis, bacterial pneumonia, septicemia, fungal infection being the most encountered types. Among these infections, cholangitis has the highest incidence rate, ranging from 40% to 93% [9–11]. Existing studies have shown that cholangitis is associated with shorter native liver survival and higher risk of liver transplantation in affected children [12, 13]. Studies indicate that children who experience early cholangitis have a significantly lower two-year native liver survival rate compared to those without such an occurrence [14, 15]. However, in addition to postoperative cholangitis following BA surgery, there is a risk of other localized and systemic infections such as pneumonia and sepsis [11, 16]. Currently, there is a paucity of relevant research literature examining the association between non-cholangitis postoperative infections and clinical outcomes. This study aims to investigate the risk factors associated with post-KP infection in children with BA and to evaluate the association between infection and native liver survival. Methods Participants Data were accessed for research from January 2012 to August 2022 at the First Affiliated Hospital of Guangxi Medical University. The retrospective single-center cohort study conducted included 404 children with BA who underwent Kasai procedure (KP). The inclusion criteria for BA require confirmation through surgical exploration, which can be classified into three types according to the Kasai classification system. Type I includes dendritic and cloudy common bile duct atresia, while type II involves hepatic duct atresia. Finally, type III refers to hilar atresia or complete extrahepatic bile duct obstruction [17]. Exclusion criteria included patients with other congenital genetic or metabolic disorders, congenital immunodeficiency syndromes, and those who had infection prior to the Kasai procedure. Definition of pertinent clinical indicators and diagnostic criteria Infection: Infection is defined as a local or systemic inflammatory response induced by the invasion of pathogenic microorganisms (e.g., bacteria, viruses, fungi, parasites) [18]. For the present study, the diagnosis of infection required at least one of the following criteria: (1) presence of clinical manifestations including fever, cough, or dyspnea; (2) abnormal levels of inflammatory markers (e.g., white blood cell count, C-reactive protein, procalcitonin); (3) positive results of pathogenic detection assays (e.g., blood culture, sputum culture, galactomannan test). Infection: Infection is defined as a local or systemic inflammatory response induced by the invasion of pathogenic microorganisms (e.g., bacteria, viruses, fungi, parasites) [18]. For the present study, the diagnosis of infection required at least one of the following criteria: (1) presence of clinical manifestations including fever, cough, or dyspnea; (2) abnormal levels of inflammatory markers (e.g., white blood cell count, C-reactive protein, procalcitonin); (3) positive results of pathogenic detection assays (e.g., blood culture, sputum culture, galactomannan test). Cholangitis: Symptoms of a cholangitis attack include unexplained fever, worsening or receding yellowing of the skin and sclera, lighter stool color or reappearance of clay-colored stools, increased levels of serum bilirubin (total and direct), as well as significant increases in leukocytes, neutrophils, C-reactive protein and procalcitonin according to blood tests [19, 20]. By postoperative onset timing, cholangitis is classified into early-onset infection group (within 1 month postoperatively) and late-onset infection group (more than 1 month postoperatively).By occurrence frequency, cholangitis is divided into non-frequent group and frequent group (≥ 3 episodes within 6 months) [21]. Cholangitis: Symptoms of a cholangitis attack include unexplained fever, worsening or receding yellowing of the skin and sclera, lighter stool color or reappearance of clay-colored stools, increased levels of serum bilirubin (total and direct), as well as significant increases in leukocytes, neutrophils, C-reactive protein and procalcitonin according to blood tests [19, 20]. By postoperative onset timing, cholangitis is classified into early-onset infection group (within 1 month postoperatively) and late-onset infection group (more than 1 month postoperatively).By occurrence frequency, cholangitis is divided into non-frequent group and frequent group (≥ 3 episodes within 6 months) [21]. Pneumonia was defined as a lower respiratory tract infection characterized by newly emerged pulmonary infiltrates on chest imaging, accompanied by clinical symptoms such as fever, cough, and tachypnea. This definition encompasses both community-acquired pneumonia (CAP) and hospital-acquired pneumonia (HAP) [22]. Pneumonia was defined as a lower respiratory tract infection characterized by newly emerged pulmonary infiltrates on chest imaging, accompanied by clinical symptoms such as fever, cough, and tachypnea. This definition encompasses both community-acquired pneumonia (CAP) and hospital-acquired pneumonia (HAP) [22]. Sepsis was defined according to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) (Singer et al., 2016), as a life-threatening organ dysfunction caused by a dysregulated host response to infection, identified by an increase in the Sequential Organ Failure Assessment (SOFA) score of ≥ 2 points [23]. Sepsis was defined according to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) (Singer et al., 2016), as a life-threatening organ dysfunction caused by a dysregulated host response to infection, identified by an increase in the Sequential Organ Failure Assessment (SOFA) score of ≥ 2 points [23]. Fungal infection was defined as a pathological process where fungi colonize and propagate in host tissues or body fluids, inducing local or systemic inflammation and tissue damage. Diagnosis required positive fungal culture or molecular detection results from clinical specimens (e.g., bile, blood, sputum), corresponding clinical symptoms, and exclusion of fungal colonization, in accordance with the EORTC/MSG criteria for invasive fungal infections [24]. Fungal infection was defined as a pathological process where fungi colonize and propagate in host tissues or body fluids, inducing local or systemic inflammation and tissue damage. Diagnosis required positive fungal culture or molecular detection results from clinical specimens (e.g., bile, blood, sputum), corresponding clinical symptoms, and exclusion of fungal colonization, in accordance with the EORTC/MSG criteria for invasive fungal infections [24]. Mixed infection: Patients with concurrent cholangitis and other infections (e.g., pneumonia, urinary tract infection, sepsis) confirmed by laboratory or imaging examinations. Mixed infection: Patients with concurrent cholangitis and other infections (e.g., pneumonia, urinary tract infection, sepsis) confirmed by laboratory or imaging examinations. Native liver survival: Native liver survival was defined as the time from Kasai portoenterostomy to liver transplantation or death due to liver failure; patients who remained alive without liver transplantation were censored at the last follow-up [25]. Native liver survival: Native liver survival was defined as the time from Kasai portoenterostomy to liver transplantation or death due to liver failure; patients who remained alive without liver transplantation were censored at the last follow-up [25]. Postoperative medication for Kasai portoenterostomy [19] Corticosteroid therapy: Methylprednisolone was administered intravenously starting postoperatively at a dosage regimen of 10, 8, 6, 5, 4, 3, and 2 mg·kg⁻¹·d⁻¹ for 7 consecutive days. Subsequently, oral prednisolone was given at 2 mg·kg⁻¹·d⁻¹ for 4 weeks, then tapered to 1 mg·kg⁻¹·d⁻¹ for another 4 weeks before discontinuation. Corticosteroid therapy: Methylprednisolone was administered intravenously starting postoperatively at a dosage regimen of 10, 8, 6, 5, 4, 3, and 2 mg·kg⁻¹·d⁻¹ for 7 consecutive days. Subsequently, oral prednisolone was given at 2 mg·kg⁻¹·d⁻¹ for 4 weeks, then tapered to 1 mg·kg⁻¹·d⁻¹ for another 4 weeks before discontinuation. Antibiotic therapy: Intravenous infusion of third-generation cephalosporins was administered for no less than 2 weeks postoperatively, followed by oral administration of third-generation cephalosporins for 3 to 6 months. Antibiotic therapy: Intravenous infusion of third-generation cephalosporins was administered for no less than 2 weeks postoperatively, followed by oral administration of third-generation cephalosporins for 3 to 6 months. Clinical data collection and cohort stratification The clinical data collected in this study included two categories: baseline preoperative data and postoperative follow-up data. Baseline preoperative data: Demographic information (gender, age at operation, height, weight), BA pathological type, preoperative liver function tests including total bilirubin (TBIL), direct bilirubin (DBIL), γ-glutamyl transpeptidase (GGT), alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bile acids (TBA), and albumin (ALB) levels, and preoperative blood routine indicators. All liver function parameters used for risk factor analysis (e.g., multivariate logistic regression) were baseline preoperative values. Baseline preoperative data: Demographic information (gender, age at operation, height, weight), BA pathological type, preoperative liver function tests including total bilirubin (TBIL), direct bilirubin (DBIL), γ-glutamyl transpeptidase (GGT), alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bile acids (TBA), and albumin (ALB) levels, and preoperative blood routine indicators. All liver function parameters used for risk factor analysis (e.g., multivariate logistic regression) were baseline preoperative values. Postoperative follow-up data: Details regarding the occurrence of postoperative infection (onset time, infection type, pathogenic detection results), postoperative glucocorticoid and antibiotic usage duration, postoperative liver function follow-up results, and native liver survival outcomes. Postoperative follow-up data: Details regarding the occurrence of postoperative infection (onset time, infection type, pathogenic detection results), postoperative glucocorticoid and antibiotic usage duration, postoperative liver function follow-up results, and native liver survival outcomes. The subjects were categorized into two groups based on the presence or absence of infection: non-infected group and infected group. Subsequently, subgroup analysis was conducted according to the type of infection, which included fungal infection group, non-fungal infection group, simple cholangitis group, simple pulmonitis group and mixed infection group. The classification was also done according to the timing of infection: early-onset infection group (within 1 month postoperatively), late-onset infection group (more than 1 month postoperatively), non-frequent infection group and frequent infection group. We analyzed the risk factors for postoperative infections in patients with BA, and additionally performed correlation analyses to investigate the associations of age at operation, surgical approach, antibiotic and glucocorticoid duration with jaundice clearance rate and native liver survival. Follow-up and prognostic assessment Biochemical indicators including postoperative liver function were regularly followed up at 4 weeks, 8 weeks, 12 weeks, 6 months, and 12 months after KP. These fixed-time postoperative liver function data were used for between-group comparison (infected vs. non-infected). Notably, the liver function measurements at these fixed time points were not aligned to the exact onset time of postoperative infection in individual patients. The follow-up end points were liver transplantation (due to decompensation of liver failure) or death caused by liver failure. Survival analysis: Short-term survival: the native liver survival time was < 6 months; Intermediate survival: the native liver survival time was between 6 months to two years; Long-term survival: the native liver survival time was > 5 years. Statistical analysis IBM SPSS Statistics 27.0 software was used for all statistical analyses. Measurement data with a normal distribution were expressed as the mean ± standard deviation (x̄±SD), while those with a non-normal distribution were presented as the median [P50 (P25, P75)]. Between-group comparisons of normally distributed data were performed using the independent-samples t-test. Binary categorical variables were expressed as percentages, and inter-group comparisons were conducted using the chi-square test or Fisher’s exact test where appropriate. Binary logistic regression analysis was used to identify the independent risk factors for postoperative infection, with the following model-building strategy: First, univariate logistic regression analysis was performed for all potential clinical variables (P < 0.10 was set as the screening threshold). Variables with statistical significance in univariate analysis and clinically relevant variables (e.g., preoperative liver function, demographic indicators) were then included in the multivariate logistic regression model. No automated stepwise selection was used; a clinical knowledge-guided forward inclusion approach was adopted to construct the final model to ensure the robustness and clinical plausibility of the results. To accurately evaluate the time-dependent association between postoperative infection and native liver survival and avoid time-dependent bias, a time-dependent Cox proportional hazards model was employed. In this model, postoperative infection was defined as a time-dependent covariate. Specifically, the follow-up time of each patient was divided into two intervals by the time of infection onset: an unexposed interval from surgery to the onset of infection, and an exposed interval from the onset of infection to the occurrence of the outcome event or the end of follow-up. The model was adjusted for three predefined clinical confounders, including age at Kasai procedure, preoperative total bilirubin, and surgical weight. No automated stepwise selection was performed, and all fixed covariates were retained in the final model. The proportional hazards assumption was verified via Schoenfeld residual analysis, with no significant violation detected (global P = 0.28, all individual covariates P > 0.05). The patient ID was specified as the subject identifier to generate robust standard errors and correct for the clustering effect of segmented follow-up data. A two-tailed P value < 0.05 was considered statistically significant for all analyses. Native liver survival was analyzed using a time-dependent Cox proportional hazards model to account for the time-varying exposure of postoperative infection. Cox regression was used to examine factors associated with native liver survival. Results Demographic data of the cohort From January 2012 to August 2022, a total of 404 children with BA who underwent KP met the inclusion criteria. In the infection group, there were 240 BA patients, comprising of 118 (49.2%) male and 122 (50.8%) female. The age range for KP was between 39 and 248 days, with a median age of [75.5(63.0, 92.8)] days. The height ranged from 41 cm to 75 cm, with an average of 57.58 ± 4.03 cm, while the weight ranged from 2.2 kg to 9 kg, with an average of 5.09 ± 0.87 kg. BA type I accounted for only12 cases (5%), type II accounted for nine cases (3.8%), and type III accounted for most 219 cases in total (91.2%). In the non-infected group, there were 164 BA patients, comprising of 92 (56.1%) males and 72 (43.9%) females. The age at KP ranged from 40 to 210 days, with a median age of [75.0(62.0, 93.0)] days. The height ranged from 45 cm to 70 cm, with an average of 57.24 ± 3.90 cm, while the weight ranged from 2.1 kg to 8.5 kg, averaging at 4.88 ± 0.94 kg. BA type I was observed in 8 cases (4.9%) while type II and III were seen in 11 (6.7%) and one145 (88.4%) cases respectively. There were no significant differences in gender, age of operation, height and type of BA between the two groups, all P > 0.05. Interestingly, the infected group exhibited a significantly higher weight compared to the non-infected group, with a statistically significant difference between the two groups, P < 0.05 (Table 1). Table 1Comparison of demographic characteristics dataItemNon-infected group(n = 164)Infected group(n = 240)χ²/z/t P Gender [n/(%)] Male92(55.28)118(49.17)4.2450.120 Female72(44.72)122(50.83)Age of operation [n/(%)] ≤42 d1(0.61)7(2.92) 43~60 d30(18.29)44(18.33)-0.1190.905 61~90 d91(55.49)123(51.25) >90 d42(25.61)66((27.50)4.88±0.945.09±0.87-2.2530.025Weight (xˉ±s,kg)57.24±3.9057.58±4.03-0.8180.414Height (xˉ±s,cm)8(4.88)12(5.00)Type [n/(%)]11(6.71)9(3.75)1.1810.404I145(88.41)219((91.25)IIⅢ48(29.27)67(27.92)Corticosteroid Postop.Dur. [n/(%)]41(25.00)75(31.25)≤4 w71(43.29)93((38.75)-0.5080.6118w1(0.61)4(1.67)12w3(1.83)1(0.41)>12wNot administered24(14.63)35(14.58)Antibiotic Postop. Dur. [n (%)]35(21.34)58(24.17)-0.2380.8122 ~ ≤ 4 w62(37.81)83(34.58)4 ~ ≤ 8w43(26.22)64(26.67)8 ~ ≤ 12w12~≤24w Comparison of demographic characteristics data Etiological composition of infection and outcome A total of 240 cases were included in the infection group, of which cholangitis was the most common infection type (90.8%, 218/240), including simple cholangitis (51.6%, 124/240) and cholangitis combined with other infections (39.2%, 94/240). Other infection types (pneumonia, sepsis, fungal infection alone) accounted for only 9.2% (22/240) of all cases, with a small sample size for each single type. Among the 240 infected cases,88 cases were detected positive pathogenic microorganisms, the positive rate of pathogen detection was 36.67%. Among the 88 cases, 58 (65.9%) were positive in blood culture, 32 (36.4%) in sputum culture, 5 in urine culture, 3 in ascites culture and 1 in bile culture. A total of 32 deaths were documented, 27 of which were attributed to infection (corresponding to an infection-related mortality rate of 11.25% [27/240]), while 5 occurred without evidence of infection (4 due to liver failure and 1 to lymphoma), with a mortality rate of 3.05% (5/164) in the non-infection group. Among the infection group 6 cases of cholangitis, and 21 cases of cholangitis with other infections. Positive blood cultures were identified in 10 patients, including Klebsiella pneumoniae + Enterococcus faecalis (1case), Acinetobacter baumannii (2 cases), Enterobacter cloacae (1 case), Escherichia coli (3 cases), and Enterococcus faecalis (3 cases). Additionally, positive sputum cultures were found in 3 patients: Acinetobacter baumannii (1 case), Acinetobacter baumannii + Candida albicans (1 case) and Stenotrophomonas maltophilia (1case). Sepsis was present in 16 fatal cases(Table 2). Table 2Common infectious bacteria and drug susceptibility resultsPathogens of infectionNumber of casesSource of culture specimenDrug-sensitivityKlebsiella pneumoniae17Sputum, blood, urine, ascitesCiprofloxacin, Levofloxacin, Amikacin, gentamicinEscherichia coli16Blood and sputumMeropenem, imipenem, AmikacinEnterococcus faecium12BloodVancomycin, LinezolidBaumanii10Blood and sputumTigecyclineEnterobacter cloacae9BloodMeropenem, imipenem, ciprofloxacinCandida albicans7Sputum, blood, urine, ascitesVoriconazole, DiflucanEnterococcus faecalis5BloodVancomycin, Linezolid, LevofloxacinStenotrophomonas maltophilia3Blood, sputum, bileLevofloxacin, cotrimoxazole, ceftazidimeKlebsiella oxytoca2BloodCefoperazone sulbactam, ceftriaxone, meropenem, imipenemSerratia marcescens2Blood, sputumCefoperazone sulbactam, Ceftriaxone, piperacillin tazobactam, meropenem, imipenemPseudomonas aeruginosa2Blood, sputumMeropenem, imipenem, piperacillin tazobactam, cefoperazone sulbactamMoraxella osloensis1BloodErythromycinCandida parapsilosis1BloodDrug sensitivity test was not performedCandida tropicalis1SputumDrug sensitivity test was not performed Common infectious bacteria and drug susceptibility results Infection risk factors after KP with biliary atresia Binary logistic regression analysis was performed to identify the independent risk factors for postoperative infection, with a clinical knowledge-guided model-building strategy (univariate screening followed by forward inclusion of clinically relevant variables). The results showed that preoperative weight was a significant independent predictor (OR = 1.466, 95% CI: 1.120–1.920, P = 0.005), with higher preoperative weight associated with an increased risk of infection. In contrast, preoperative AST was inversely associated with infection risk (OR = 0.998, 95% CI: 0.996–1.000, P = 0.019), while preoperative total bilirubin showed a positive association (OR = 1.005, 95% CI: 1.000–1.009, P = 0.031) (Table 3) (Table 4). Table 3Multivariate logistic regression analysis resultsVariableβSEWald χ² P OR (95% CI)Constant−1.0420.7252.0670.151—Preoperative total bilirubin0.0050.0024.6560.0311.005 (1.000, 1.009)Preoperative AST−0.0020.0015.5330.0190.998 (0.996, 1.000)Preoperative weight0.3830.1377.7460.0051.466 (1.120, 1.920) Multivariate logistic regression analysis results Table 4Adjusted hazard ratios from the time-dependent Cox model for native liver survivalVariableAdjusted HR95% CIP valuePostoperative infection onset time (per month later)0.7770.719–0.839< 0.001Age at Kasai procedure (per day)1.0010.997–1.0050.591Preoperative total bilirubin (per µmol/L increase)1.0010.999–1.0020.484Preoperative weight (per kg increase)0.9090.805–1.0260.124 Adjusted hazard ratios from the time-dependent Cox model for native liver survival Model diagnostics were performed to ensure the robustness of the regression results. Variance inflation factors (VIF) were calculated for all covariates, and all VIF values were less than 2.0, indicating no significant multicollinearity. The Hosmer–Lemeshow test was non-significant (P = 0.782), confirming good model calibration. The area under the receiver operating characteristic curve (AUC) was 0.78 (95% CI: 0.73–0.83), demonstrating acceptable predictive discrimination. The impact of infection on prognosis Effects of infection on liver function Postoperative liver function parameters were compared between the infected and non-infected groups at fixed follow-up time points (4 weeks, 8 weeks, 12 weeks, 6 months, 12 months post-KP). These measurements were not aligned to the exact onset time of infection in individual patients, and pre-infection and post-infection liver function values were not separately analyzed due to the retrospective study design. The levels of ALB in the infected group were significantly lower post-operation compared to those in the non-infected group at all fixed follow-up time points (4 weeks, 8 weeks, 12 weeks, 6 months and 12 months) (P = 0.026). The levels of TBIL and DBIL in the infected group were significantly higher post-operation compared to those in the non-infected group at all follow-up time points (all P < 0.01). Notably, this observed association between postoperative infection and delayed bilirubin clearance cannot rule out the possibility of reverse causality, that is, impaired bilirubin clearance and poor bile drainage may also increase the risk of postoperative infection. There was no statistically significant disparity in ALT, AST, GGT and TBA between the infected and non-infected group at the one-year post operation (P > 0.05), Fig. 1. The indices of liver function after KP were compared between the infection group and the non-infection group. The levels of albumin (ALB) in the infected group were significantly lower at 4 weeks*, 8 weeks*, 12 weeks*, and 6 months* post-operation compared to those in the non-infected group (P < 0.05). The GGT levels at 12 weeks* and 6 months* post-KP in the infected group were elevated compared to the non-infected group (P < 0.05). The TBA in the infected group was elevated compared to the non-infected group at 6 months* post-surgery (P < 0.05). There was no statistically significant disparity in other liver function indicators between the infected and non-infected group at the one-year post operation (P > 0.05) Time-dependent Cox regression and survival analysis of 5-year native liver survival A total of 404 patients were enrolled. After splitting follow-up for the time-dependent Cox model, 640 records were analyzed, with 366 events (death/transplantation, coded 1) and 274 censored cases (survival, coded 0). No exclusions were due to missing data or invalid survival time. Stratified analysis showed distinct outcomes: stratum 0 (no infection) had 139 events and 261 censored cases (65.3% censoring), while stratum 1 (infection) had 227 events and 13 censored cases (5.4% censoring). The overall censoring rate was 42.8%, ensuring adequate statistical power. The time‑dependent Cox model exhibited excellent fit (Omnibus test χ² = 44.063, P < 0.001).After adjusting for age at Kasai procedure, preoperative total bilirubin, and weight at surgery, later onset of postoperative infection was associated with better native liver survival (HR = 0.777, 95% CI: 0.719–0.839, P < 0.001).Specifically, each one-month later infection onset corresponded to a 22.3% lower hazard of death or liver transplantation in the adjusted model. The hazard ratios for the fixed covariates are provided in Supplementary Table 4. Adjusted survival curves predicted by the time-dependent Cox proportional hazards model showed significantly higher native liver survival rates among patients without postoperative infection, whereas infected patients experienced a rapid decline in survival, particularly within the first 12 months. At 60 months of follow-up, non-infected patients maintained a native liver survival rate exceeding 40%, while the survival rate in infected patients approached 0%. The cumulative hazard function further supported these observations: infected patients demonstrated a rapid and sustained increase in hazard, which exceeded 3.0 by the end of follow-up, compared with a hazard below 1.0 in non-infected patients. These findings indicate a strong association between postoperative infection and poorer native liver survival (Fig. 2). A, Adjusted native liver survival curves predicted by the time-dependent Cox proportional hazards model. B, Adjusted cumulative hazard curves predicted by the time-dependent Cox proportional hazards model Discussion Infectious complications are common following KP for BA. Typically, these include cholangitis, pneumonia, and sepsis. Pathogens include bacteria, cytomegalovirus, Epstein Barr virus, and fungal infections. The highest incidence among them is cholangitis, accounting for 40%-93% [9].In this study, the overall postoperative infection rate was found to be 59.4%(240/404), with cholangitis accounting for a significant proportion at 90.8%(218/240),Other infection types only accounted for 9.2% (22/240), meaning the observed adverse association between postoperative infection and native liver survival was primarily driven by cholangitis.These findings align closely with previous studies conducted in the field [26]. In this study, 88 (36.7%) of infection patients were identified as pathogenic bacterial infections. The most prevalent pathogens identified were Escherichia coli and Enterococcus faecium, followed by Klebsiella pneumoniae and Enterobacter cloacae. These findings are consistent with previous research conducted the previous study [27–29]. In this study, most of the postoperative cases were treated with glucocorticoids, and the incidence of fungal infection was 10.4% (25/240). The incidence of fungal infections was relatively low, which may be attributed to the standardized use of glucocorticoids in children with biliary atresia following KP in this study. The main pathogen of fungal infection is Candida, which is sensitive to drugs such as voriconazole and fluconazole. These findings are consistent with previous reports [30] . The findings of this study indicate that the fatality rate following infection is as high as 11.15%, with sepsis being the primary cause of mortality. The mortality rate following sepsis was found to be 59.26%, indicating a higher fatality rate associated with this condition. These clinical observations indicate that infection is closely correlated with mortality among patients with biliary atresia.However, the positive rate of traditional etiological detection methods, including blood culture, was very low. In this study, the positive rate of etiological detection was only 36.67%. which is consistent with the reported rate (30%-40%) in similar retrospective studies of postoperative infections in BA [28, 29]. With the development of science and technology, metagenomic next generation sequencing (mNGS) can be utilized in post-operative children with BA to enhance the positive detection rate of infectious pathogens [31, 32].Early detection and standardized management are clinically noteworthy for these patients. In this observational cohort, postoperative infection was associated with a reduction in bilirubin clearance. The study design does not allow us to establish a direct causal relationship between infection and impaired bilirubin clearance. Our study found that the jaundice clearance rate was 44.8% among non-infected patients, compared to 31.3% among infected patients. Cholangitis and frequent cholangitis are strongly associated with delayed jaundice clearance, which is in line with the findings of the previous study [33]. Previous studies reported that early bilirubin clearance may be linked to less hepatocellular damage and slower progression of cirrhosis [34]. Notably, this observed association cannot exclude reverse causality: poor bile drainage and delayed bilirubin clearance may create a favorable microenvironment for pathogenic microbial colonization and proliferation, thereby increasing the risk of postoperative infection (especially cholangitis). The present study shows that cholangitis and recurrent cholangitis are predominantly correlated with altered TBA and GGT levels, and albumin levels tend to change in patients with any type of infection. Given that albumin serves as a marker for liver synthesis, it can be inferred that any infection will have an impact on hepatic synthesis. A study showed that achieving a postoperative bile acid level of less than 40 µmol/L could lead to a favorable prognosis [35]. The above associations suggest that monitoring postoperative infection is clinically meaningful for children with biliary atresia. No causal inference can be drawn from this observational study. The results demonstrated that higher preoperative weight was an independent risk factor for postoperative infection following KP in children with biliary atresia (P < 0.05). Specifically, patients with elevated preoperative weight were more prone to early-onset infection after KP, though the frequency of recurrent infections was not increased. Yan’s et al. [36] study demonstrated that children with biliary atresia who developed cholangitis after KP had a higher body weight at the time of surgery. Li’s et al. [37] demonstrated that higher preoperative weight was associated with inferior early biliary drainage outcomes after KP in pediatric patients. The observed association between higher body weight and increased infection risk presents a paradoxical finding. One plausible explanation involves a reverse causality: elevated weight in these infants may not be a protective factor but rather a marker of more advanced disease or a protracted preoperative course. This could imply a baseline state of heightened hepatic inflammation or subclinical infection, which compromises postoperative recovery and predisposes patients to episodes of cholangitis. Furthermore, from a surgical perspective, greater body mass can correlate with increased technical complexity at the hepatic hilum, potentially leading to prolonged operative time and heightened exposure to infection risk, thereby contributing to the incidence of postoperative cholangitis [36, 38–40]. This study found that postoperative infection is strongly associated with 5-year native liver survival.A later onset of postoperative infection was strongly associated with better native liver survival (per-month HR = 0.777, 95% CI: 0.719–0.839, P < 0.001). This observed correlation is of clinical interest, and clinical attention to postoperative infection and its onset time is warranted.The adjusted survival curves and cumulative hazard function predicted by the time-dependent Cox model provided compelling visual evidence of the association between postoperative infection and native liver survival. Patients without infection maintained a significantly higher cumulative survival rate throughout follow-up, whereas those with infection experienced a rapid decline in survival, particularly within the first 12 months. These patterns suggest that the immediate postoperative period may be an important time frame for infection surveillance.The cumulative hazard analysis further reinforced these observations, showing that infection patients had a sustained and substantially higher event risk compared to their non-infected counterparts. From a clinical perspective, these results highlight the need for targeted strategies to reduce the incidence and severity of postoperative infection. This may include optimized antimicrobial prophylaxis, enhanced infection control protocols, and close monitoring of at-risk patients. Additionally, the association between later infection onset and better survival raises the hypothesis that strategies to prolong the infection‑free interval might improve outcomes, but this hypothesis requires confirmation in prospective interventional studies. This study has several limitations. First, as a retrospective single-center cohort study, it is inherently susceptible to selection and information biases. Despite adherence to Chinese [19] and international [9] diagnostic criteria, decadal documentation variations may have compromised data quality, and selection bias may limit the generalizability of the observed infection-native liver survival association. The low pathogen detection rate also weakens conclusions on the prognostic impact of specific pathogens (e.g., fungal vs. bacterial infections). Second, subgroup analyses were limited by small sample sizes introducing potential bias. Although standardized data extraction and sensitivity analyses were performed, residual selection and information biases remained unavoidable. Third, the liver function analysis in this study was based on fixed follow-up time points without alignment to the exact infection onset time, and pre-infection and post-infection liver function values were not separately analyzed, which limits the ability to establish a direct causal relationship between infection and liver function impairment. Furthermore, despite the use of time-dependent Cox regression, residual confounding (e.g., by unmeasured markers of liver function or disease severity at baseline) cannot be excluded. Therefore, all reported associations should be interpreted as statistical associations rather than causal effects. Notably, the large overall sample size and long follow-up duration enhanced internal validity, and consistency with prior studies [13,27] supports the reliability of core conclusions. Future multicenter studies are warranted to validate our findings and enhance generalizability. In conclusion, in this retrospective cohort, postoperative infection (primarily cholangitis) was significantly associated with delayed bilirubin clearance and poorer 5-year native liver survival. The timing of infection onset was an independent factor associated with native liver survival outcomes. These observed associations suggest that infection prevention and early clinical monitoring are clinically relevant, although causality cannot be determined from this observational study. Acknowledgements Not applicable. Abbreviations Biliary atresia Kasai procedure Authors' contributions XY Meng contributed to data collection, statistically analyzed, and drafting of the paper. SH Liang, RJ Li, L Huang and Q Tang contributed the sample collection and revision of the paper. X Yun and XQ Chen contributed the conception and design of study and revised the final paper. All authors reviewed the manuscript. Funding This study was supported by the Guangxi Clinical Medical Research Center for Pediatric Diseases (Guike AD22035121),Excellent Medical Talents Training Program of the First Affiliated Hospital of Guangxi Medical University, Guangxi Natural Science Foundation (no. 2024GXNSFAA010055), Middle/Young aged Teachers’ Research Ability Improvement Project of Guangxi Higher Education (2024KY0093), Clinical Research “Climbing” Program of the First Affiliated Hospital of Guangxi Medical University (YYZS2023011), Difficult and Critical illness Center, Pediatric Clinical Medical Research Center of Guangxi (Gui Ke AD22035219), and the Key Laboratory of Children’s Disease Research in Guangxi’s Colleges and Universities, Education Department of Guangxi Zhuang Autonomous Region. Data availability All data generated or analyzed during this study are included in this article and supporting information. Further inquiries can be directed to the corresponding author. Declarations Ethics approval and consent to participate This retrospective cohort study was approved by the Ethics Committee of the First AffiliatedHospital of Guangxi Medical University (full name: Ethics Committee of the First AffiliatedHospital of Guangxi Medical University) and was conducted in accordance with the ethicalstandards of the Declaration of Helsinki. The Ethics Committee of theFirst Affiliated Hospital of Guangxi Medical University waived the requirement for informedconsent (consent to participate) from parents or legal guardians due to the retrospective natureof the study, which involved no more than minimal risk and used de-identified data. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this article and supporting information. Further inquiries can be directed to the corresponding author.