This retrospective study included 326 patients who underwent PD at the First Affiliated Hospital of Chongqing Medical University from October 2018 to July 2023. Data collection comprised demographic information, laboratory data, surgical records, and postoperative outcomes. The exclusion criteria were as follows: severe cardiovascular and cerebrovascular diseases, hematological diseases, autoimmune diseases, PD combined with other organ resection, and minors. The study conformed to the standards outlined by the Declaration of Helsinki and received approval from the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (Approval no. K2023-322).

PD was performed by experienced surgeons using either open or laparoscopic methods. Digestive tract reconstruction was conducted using the classic Child’s method, with pancreatojejunostomy primarily performed via a duct-to-mucosa anastomosis. All patients received prophylactic antibiotic treatment during the perioperative period. The standard regimen included intravenous administration of cefoperazone-sulbactam (3 g/8 h) initiated 30 min before the skin incision and continued for 24 h postoperatively. Subsequent adjustments were made based on postoperative laboratory markers (e.g., leukocytosis, CRP elevation) and microbiological culture results. For patients with a history of penicillin allergy, levofloxacin (500 mg/24 h) was used as an alternative. Typically, octreotide or somatostatin therapy was administered for 5–7 days following the surgery. After the operation, patients continued to take acid suppressants for several months. Early in the postoperative period, patients received parenteral nutrition before gradually switching to enteral nutrition.

The timing of PBD was determined by the attending surgeon’s clinical judgment, prioritizing factors such as: severity of jaundice (bilirubin levels), nutritional status, comorbidities, and tumor resectability on imaging. Our institutional protocol generally aims for 7–14 days of PBD to allow bilirubin reduction and liver function recovery. The exact duration varied based on rate of bilirubin decline, resolution of cholangitis (if present), and patient’s physiological readiness for major surgery. The primary goal of PBD was not strict normalization but rather reduction to a safer threshold (usually <10 times the normal value in clinical applications). PBD techniques include percutaneous and endoscopic biliary drainage. The percutaneous routes consist of percutaneous transhepatic drainage, while endoscopic routes comprise both endoscopic nasobiliary drainage and endoscopic retrograde cholangiopancreatography with stent placement.

The primary outcomes were defined as the occurrence of any PD complications, including clinical postoperative pancreatic fistula (POPF), intra-abdominal bleeding, intra-gastrointestinal bleeding, Clavien-Dindo grade III (CD3) complications (or higher than grade III), surgical site infection (SSI), postoperative transfusion, admission to the intensive care unit (ICU), reoperation, and readmission. Secondary outcomes included operative time, intraoperative blood loss, total hospital stay, and postoperative hospital stay. Clinical POPF was diagnosed and classified according to the International Study Group on Pancreatic Surgery criteria, which categorizes POPF into biochemical leak, Grade B, and Grade C (16). Clinical POPF incorporates both Grade B and Grade C POPF. Intra-abdominal bleeding was characterized by the presence of hemorrhagic fluid in the abdominal drain or non-cavitary bleeding associated with an unexplained decrease in hemoglobin levels, confirmed by imaging or endoscopy and requiring transfusion, endoscopic or surgical hemostasis. Intra-gastrointestinal bleeding was identified by melena, persistent black stool, or blood in the gastric tube drain, confirmed by imaging or endoscopy, and requiring appropriate hemostatic treatment. SSI was defined according to the Centers for Disease Control and Prevention guidelines (17). Readmission was defined as any hospital admission within 90 days postoperatively. Reoperation was referred to as any unplanned second operation during the hospital stay.

The study cohort was divided into four groups based on the 25th, 50th, and 75th percentiles of preoperative serum bilirubin levels. Baseline characteristics and clinical outcomes of patients were compared in each group. Continuous variables are expressed as medians with interquartile ranges or means with standard deviations, depending on their distribution, and compared using the Kruskal–Wallis rank sum test. Categorical variables are expressed as frequencies and percentages and compared using Pearson’s Chi-squared test or Fisher’s exact test.

Multivariable logistic regression analysis was conducted to analyze the independent association between preoperative serum bilirubin levels and various perioperative complications. Model 0 was unadjusted, while Model 1 was adjusted for age, gender, body mass index (BMI), history of abdominal surgery, and neoadjuvant therapy. Moreover, Model 2 was adjusted for albumin, alanine aminotransferase (ALT), aspartate aminotransferase pressure (AST), main pancreatic duct diameter (MPD), surgical approach, and pathological diagnosis.

A dose–response relationship between a continuous exposure and an outcome can be effectively described using restricted cubic spline functions, which can also be used to statistically and/or visually verify the association’s assumed linearity (18). A RCS model with three knots (at the 10th, 50th, and 90th percentiles) was utilized to explore the dose–response relationship between bilirubin levels and the risk of PD complications.

To evaluate if preoperative serum bilirubin and PBD have a significant effect on PD complications, we examined a 3-way interaction of PD complications × bilirubin (as a continuous variable) × PBD using an interactive restricted spline. First, the 2-way interaction of bilirubin and PBD was tested in the adjusted logistic model. If a significant interaction was observed, further analysis was performed to determine a potential bilirubin threshold. By comparing the outcomes of PBD and non-PBD based on this threshold, patients benefiting from either procedure through bilirubin may be identified. In our study, all statistical analyses were performed using R software (version 4.2.2), along with the use of MSTATA software. A p < 0.05 was considered statistically significant.

A total of 335 patients were initially included in the study. After excluding nine cases, 326 patients remained for analysis (Figure 1). Table 1 displays the patients’ baseline characteristics grouped by preoperative bilirubin quartiles. Age, gender, BMI, hypertension, diabetes, drinking, smoking, neoadjuvant chemotherapy, MPD, and history of abdominal surgery were statistically non-significant between the four groups. PBD, surgical approach, and pathologic diagnosis exhibited statistically significant differences (p < 0.05). The albumin, ALT, and AST also exhibited significant differences (p < 0.05), while the other laboratory makers were statistically non-significant.

The relationship between total serum bilirubin and perioperative PD complications is detailed in Tables 2, 3. Using the lowest quartile as the reference, the highest quartile of total bilirubin (TB) was significantly associated with any complication in all three logistic regression models. The odd ratios (OR) and 95% confidence intervals (CIs) were 2.60 (1.38 and 4.90), 2.67 (1.38 and 5.14), and 4.60 (1.90 and 11.1). Similarly with clinical POPF, OR and 95% CIs were 2.19 (1.09 and 4.38), 2.22 (1.09 and 4.54), and 4.21 (1.40 and 12.6). TB in the highest quartile of Models 0, 1, and 2 were associated with intra-abdominal bleeding compared to reference, with ORs and 95% CIs of 3.59 (1.43 and 9.01), 3.58 (1.41 and 9.14), and 12.8 (3.32 and 49.0). Additionally, the highest quartile of bilirubin was significantly associated with postoperative transfusion in Model 0 (OR: 2.56, 95% CI: 1.26 and 5.18), Model 1 (OR: 2.41, 95% CI: 1.18 and 4.92), and Model 2 (OR: 4.35, 95% CI: 1.59 and 11.9). Although the highest quartile of bilirubin was unrelated to readmission, the second quartile (OR: 4.38, 95% CI: 1.12 and 17.2) and third quartile (OR: 5.43, 95% CI: 1.01 and 29.2) exhibited significant associations in Model 2. Moreover, Table 2 indicates that there is an increased risk of any complication, clinical POPF, intra-abdominal bleeding, SSI, and postoperative transfusion with increasing TB (p-values for trend were 0.002, 0.013, 0.001, 0.044, and < 0.001, respectively). RCS analysis (Figures 2, 3) revealed a significant linear dose–response relationship between bilirubin levels and any complication, clinical POPF, intra-abdominal bleeding, CD3 complication, postoperative transfusion, reoperation, operative time, and total hospital stay (P for overall < 0.05, P for non-linearity > 0.05).

The receipt of PBD was not independently associated with the risk of complications for patients undergoing PD (Supplementary Tables S1, S2). Subgroup analysis revealed that percutaneous biliary drainage was associated with a higher rate of Intra-gastrointestinal bleeding (15.8 vs. 3.6% in endoscopic, p = 0.028, Supplementary Table S3). However, a significant interaction was observed between bilirubin levels and PBD among PD patients (P for bilirubin × PBD interaction = 0.044). This suggests that patients with higher bilirubin exhibited a lower risk of any complication with PBD compared with non-PBD. Spline plots generated from the adjusted logistic model presented that OR for any complication following PD with PBD versus non-PBD varied with bilirubin values. The plots showed a threshold value of around 171 μmol/L, indicating that patients with bilirubin values ≥ 171 μmol/L benefited from PBD compared with non-PBD (OR for bilirubin ≥ 171 μmol/L with PBD versus non-PBD: 3.53; 95% CI: 1.06 and 11.75; p = 0.04). However, patients with bilirubin values < 171 μmol/L exhibited a similar risk of any complication with PBD and non-PBD (OR for bilirubin ≥ 171 μmol/L with PBD versus non-PBD: 0.68; 95% CI: 0.31 and 1.48; p = 0.329) (Figures 4A, 5). Similar results were not observed for other outcomes except operative time (P for bilirubin × PBD interaction = 0.012). Patients with bilirubin values < 171 μmol/L significantly declined in operative time without PBD compared with PBD (Least squares [LS] Mean Difference for bilirubin values < 171 μmol/L with PBD versus non-PBD: –44.17; 95% CI: −81.91 and −6.43; p = 0.023). Patients with bilirubin values ≥ 171 μmol/L exhibited similar operative time with PBD and non-PBD (LS Mean Difference for bilirubin ≥ 171 μmol/L with PBD versus non-PBD: 39.15; 95% CI: −17.19 and 95.50; p = 0.178) (Figures 4B, 6). Moreover, endoscopic biliary drainage is more frequent in patients with TB < 171 μmol/L (Supplementary Table S4).