Discordant Perioperative Prophylaxis and Major Morbidity After Pancreatoduodenectomy in Patients Undergoing PTBD: A...

Background: Patients undergoing pancreatoduodenectomy (PD) after preoperative percutaneous transhepatic biliary drainage (PTBD) frequently develop bacterobilia. While bile culture positivity has been variably linked to postoperative infections, the clinical relevance of culture data may be more closely related to perioperative antimicrobial adequacy. We aimed to evaluate whether discordant perioperative antibiotic prophylaxis-defined by mismatch between administered prophylaxis and resistance profiles from preoperative PTBD bile cultures-is independently associated with major postoperative morbidity. Methods: This retrospective cohort study included consecutive patients undergoing PD between January 2020 and October 2025. Major morbidity (primary endpoint) was defined as Clavien-Dindo grade ≥ III. Secondary outcomes included postoperative day 4 inflammatory markers (WBC and CRP), length of stay, and infection-related endpoints. Bile culture findings were categorized by culture status and resistance severity (no growth, low resistance, and high resistance [MDR/XDR/PDR]). Discordant prophylaxis was defined using a predefined coverage-based algorithm incorporating antimicrobial class and susceptibility profiles. Multivariable logistic regression (adjusted for age, dichotomized ASA class, and operative type) and model performance (AUC, DeLong test; Hosmer-Lemeshow calibration) were assessed. Results: A total of 145 patients were analyzed; preoperative bile culture status was no culture (n = 30), culture-negative (n = 59), and culture-positive (n = 56). Bile culture status was not associated with major morbidity (p = 0.406), POD4 inflammatory markers, or length of stay. Resistance severity categories were also not associated with major morbidity (15.3%, 17.4%, and 24.2% across no-growth, low-resistance, and high-resistance groups, respectively; p = 0.77). Discordant prophylaxis occurred in 23 patients (15.9%) and was associated with higher major morbidity compared with concordant coverage (30.4% vs. 18.0%; OR: 1.99, 95% CI: 0.69-5.36; p = 0.25). After adjustment, discordant prophylaxis showed a higher point estimate for major morbidity (adjusted OR: 1.84, 95% CI: 0.63-4.96; p = 0.24), although this did not reach statistical significance. The core clinical model showed poor discrimination (AUC 0.59); adding microbiological variables modestly increased the AUC to 0.63 without significant improvement (DeLong p = 0.46). Model calibration was acceptable (Hosmer-Lemeshow p = 0.88). Conclusions: In this PTBD cohort undergoing PD, bile culture positivity and resistance severity were not independently associated with major postoperative morbidity. Discordant prophylaxis was associated with a numerical increase in major morbidity; however, this finding did not reach statistical significance and should be interpreted cautiously given the limited sample size. These findings support interpreting bile culture data primarily within an antimicrobial stewardship framework to ensure adequate coverage rather than as standalone predictors of severe morbidity and warrant validation in larger prospective cohorts.