Bleeding and vascular complications after percutaneous transhepatic biliary drainage: Incidence and associated factors in a Vietnamese tertiary center

Abstract

BACKGROUND

Percutaneous transhepatic biliary drainage (PTBD) remains an important salvage or urgent drainage option for biliary obstruction when endoscopic retrograde cholangiopancreatography fails or is not feasible. Evidence on bleeding and vascular complications after ultrasound-guided PTBD in low- and middle-income settings remains limited.

AIM

To determine the incidence and identify predictors of bleeding or vascular complications after ultrasound-guided PTBD.

METHODS

We conducted a retrospective single-center cohort study of consecutive adults undergoing index PTBD for benign or malignant biliary obstruction at a tertiary hospital in Southern Vietnam from January 2021 to August 2023. All procedures were performed under real-time ultrasound guidance with color Doppler. Bleeding or vascular complications, defined using standardized criteria, were analyzed as the primary outcome, with clinically significant events reported separately. Associations were evaluated using a reduced Firth penalized logistic regression model including key procedural and disease-related variables, with internal validation by bootstrap resampling.

RESULTS

A total of 234 procedures were analyzed. Overall complications occurred in 33 patients (14.1%) and were predominantly minor. Bleeding or vascular complications occurred in 16 patients (6.8%), including 13 minor bleeding events (5.6%) and 3 clinically significant events (1.3%). No procedure-related mortality was observed within 14 days. Patients with bleeding or vascular complications had more frequent multiple needle passes and longer procedure times. In multivariable analysis, independent predictors included more than one needle pass [adjusted odds ratio (aOR): 7.91, 95% confidence interval (CI): 2.05-30.59], choledocholithiasis (aOR: 18.80, 95%CI: 4.79-73.73), longer procedure time per 10 minutes (aOR: 1.44, 95%CI: 1.05-1.96), and smaller target-duct diameter (aOR: 0.82 per 1 mm increase, 95%CI: 0.69-0.98). The model showed good discrimination (optimism-corrected area under the curve 0.85).

CONCLUSION

Ultrasound-guided percutaneous transhepatic biliary drainage demonstrates a low rate of clinically significant bleeding, with risk driven mainly by procedural difficulty and biliary pathology.

Key Words: Percutaneous transhepatic biliary drainage; Biliary obstruction; Bleeding; Vascular injury; Ultrasound guidance

Core Tip: In this retrospective cohort study, ultrasound-guided percutaneous transhepatic biliary drainage performed without fluoroscopy demonstrated a low rate of clinically significant bleeding (1.3%). Bleeding or vascular complications were primarily associated with procedural difficulty, including multiple needle passes and longer procedure time, as well as anatomical factors such as smaller target-duct diameter. Choledocholithiasis was also strongly associated with increased risk. These findings support the safety and feasibility of ultrasound-guided percutaneous transhepatic biliary drainage in resource-limited settings and highlight key technical factors that may help reduce bleeding complications.

INTRODUCTION

Obstructive jaundice results from interruption of bile flow from the liver to the intestine. The causes include malignant tumors such as cholangiocarcinoma and pancreatic cancer, as well as benign etiologies like choledocholithiasis and postoperative biliary strictures[1,2]. Patients typically present with jaundice, pruritus, or cholangitis, and urgent biliary decompression is required to prevent sepsis and hepatic dysfunction. Endoscopic retrograde cholangiopancreatography remains the first-line drainage method; however, technical failure occurs in 5%-15% of cases, especially in those with altered anatomy or advanced hilar obstruction[3].

In such cases, percutaneous transhepatic biliary drainage (PTBD) serves as an effective alternative, achieving technical success rates exceeding 90% in experienced hands[1,4]. PTBD offers both palliative relief in malignant obstruction and serves as a bridge to surgery in benign disease[2,5]. Despite the advent of endoscopic ultrasound-guided biliary drainage, PTBD remains more widely available and lifesaving in emergency situations, such as acute suppurative cholangitis or septic shock when other options are not feasible[5].

Although PTBD is generally safe, complications are increasingly recognized. These include catheter-related problems (bile leakage, occlusion, dislodgement), infectious events (cholangitis or sepsis), and vascular injury involving hepatic vessels or extrahepatic structures along the puncture tract[1,6,7]. Bleeding occurs in approximately 2%-5% of cases and is often due to vascular trauma along the biliary tract[7]. Venous bleeding typically presents early with dark blood drainage, while arterial bleeding often appears later due to pseudoaneurysm formation along the catheter tract[7]. Although embolization effectively controls most arterial hemorrhages, PTBD remains an invasive procedure with tangible risk. Non-vascular complications such as biliary peritonitis, pleural effusion, and pancreatitis have also been reported[6,7].

Reported overall complication rates vary widely, ranging from 5%-10% in earlier studies to more than 30% in recent multicenter series[1,6]. Higher complication rates are generally observed in patients with complex malignant obstruction, coagulopathy, cirrhosis, renal dysfunction, or when multiple or central punctures are performed. Infection, particularly cholangitis, remains the most frequent and serious event. While the Society of Interventional Radiology recommends maintaining a major complication rate below 10%, this benchmark is frequently exceeded in real-world clinical practice[1,4,8]. It remains particularly challenging to achieve in low- and middle-income countries, where infrastructural constraints inherently limit access to optimal multimodal imaging equipment. In such scenarios, the use of real-time ultrasound guidance with color Doppler becomes an indispensable alternative, as it has been shown to reduce vascular injury and is endorsed by multiple guidelines[5].

In Vietnam, PTBD is widely performed, yet published data on its safety and complication profile remain scarce, especially regarding interventions conducted solely under Doppler ultrasound guidance without fluoroscopic support. Most local studies emphasize technical success without detailed assessment of adverse events. Therefore, the present study aimed to determine the incidence, types, and associated factors of PTBD-related complications in patients with malignant and benign biliary obstruction treated at a major tertiary center in Southern Vietnam. By identifying complication patterns and associated factors, this study seeks to improve procedural safety and inform clinical practice in similar hepatobiliary settings.

MATERIALS AND METHODS

Study design and participants

This was a single-center retrospective cohort study conducted at Nhan dan Gia Dinh Hospital, a tertiary referral center in Southern Vietnam. We reviewed all consecutive adult patients (aged ≥ 18 years) who underwent index PTBD for benign or malignant biliary obstruction from January 2021 to August 2023. Patients were identified from the hospital procedural registry and their medical records were reviewed in detail. We excluded patients who underwent gallbladder drainage rather than transhepatic biliary drainage, those who underwent PTBD together with other percutaneous hepatobiliary interventions during the same session, and bilateral drainage cases in which the punctured side could not be reliably assigned for anatomical analysis. The study protocol was approved by the institutional review board of Nhan dan Gia Dinh Hospital and complied with the Declaration of Helsinki. Because the study used anonymized retrospective routine-care data, the requirement for individual informed consent was waived.

Sample size

A minimum sample size of 130 PTBD cases was estimated using a single-proportion approach with a two-sided 95% confidence interval (CI) level, an expected complication rate derived from Weber et al[9], and the desired precision. The final analytic cohort exceeded this threshold.

PTBD procedure and institutional workflow

During the study period, routine fluoroscopic guidance was temporarily unavailable at our institution; therefore, PTBD was performed under real-time ultrasound guidance with color Doppler. Patients were positioned according to the planned access route, and a peripheral dilated intrahepatic duct was punctured under ultrasound guidance using a 16-gauge needle while actively avoiding visible vascular structures. After successful duct puncture, a guidewire was advanced, the tract was serially dilated, and an 8-Fr or 10-Fr pigtail catheter was inserted using the Seldinger technique. Procedural variables recorded for analysis included puncture side, procedure duration, number of needle passes, catheter size, bile appearance, and technical success. PTBD was typically performed by a team involving both a senior and a junior operator from the hepatopancreatobiliary service; however, the retrospective record did not permit reliable attribution of the key technical steps to one individual operator, so operator-specific experience was not analyzed.

Data collection and variables

Demographic and clinical variables included age, sex, body mass index, acute cholangitis, and major comorbidities. Pre-procedural laboratory variables included white blood cell count, bilirubin, aminotransferases, platelet count, international normalized ratio (INR), and prothrombin time (PT) when available. Documented history of antiplatelet or anticoagulant use was recorded when available. In routine practice, urgent drainage for acute cholangitis was generally not deferred because of antithrombotic therapy, whereas patients with biliary obstruction without cholangitis usually had antithrombotic agents withheld according to contemporaneous recommendations when clinically feasible. However, the exact withholding interval, reversal strategy, and adherence to this practice could not be reconstructed reliably for each patient from retrospective charts; therefore, antithrombotic exposure was described in baseline analyses but was not entered into the primary multivariable model. Imaging variables included right and left intrahepatic duct diameters. A target-duct diameter on the punctured side was defined by matching the documented puncture side with the corresponding right- or left-sided duct diameter. Etiology was reviewed from the clinical record and categorized for analysis, with choledocholithiasis retained as a key binary exposure.

Outcomes and definitions

The primary outcome was a composite of bleeding or vascular complications after PTBD. This composite included minor bleeding, major bleeding requiring intervention, pseudoaneurysm, and major vascular puncture. Bleeding was defined as clinical evidence of hemorrhage or a hemoglobin decrease attributed to PTBD in the absence of another clear source. Minor bleeding referred to self-limited events not requiring endovascular or surgical hemostasis. Major bleeding referred to life-threatening hemorrhage or bleeding that required an invasive hemostatic intervention. Pseudoaneurysm was diagnosed on post-procedural imaging, and major vascular puncture referred to a recognized puncture of a major vessel during or after the procedure. For descriptive purposes, clinically significant bleeding or vascular events were defined as major bleeding requiring intervention, pseudoaneurysm, or major vascular puncture. Other procedure-related complications were recorded separately according to predefined criteria. Technical success was defined as successful catheter placement within the intended biliary system with immediate bile return or drainage at the end of the index procedure; a subsequent reduction in bilirubin was not required for technical success. All-cause mortality within 14 days after PTBD was recorded as a secondary short-term outcome.

Statistical analysis

Continuous variables are presented as median and interquartile range, and categorical variables as n (%). Group comparisons used the Wilcoxon rank-sum test for continuous variables and Fisher’s exact test (or Pearson’s χ² test when appropriate) for categorical variables. Missing data were managed via an available-case analysis without the use of statistical imputation. In instances of incomplete records, patients were excluded pairwise for that specific variable rather than from the entire cohort. Because only 16 bleeding or vascular events occurred, the primary multivariable analysis was limited to four clinically selected variables in order to improve parsimony and the events-per-variable (EPV) ratio. The main Firth penalized logistic regression model included number of needle passes (> 1 vs 1), choledocholithiasis, procedure time scaled per 10 minutes, and target-duct diameter on the punctured side scaled per 1 mm. Two sensitivity analyses were also performed: An exploratory model that additionally included sex, and a pre-procedural hemostasis model that examined choledocholithiasis, sex, target-duct diameter, platelet count, and INR among patients with available laboratory data. Because only three clinically significant bleeding or vascular events were observed, no separate adjusted multivariable model was fit for that severe-event subset. Model performance was assessed by the apparent area under the receiver-operating characteristic curve (AUC), bootstrap optimism-corrected AUC, and bootstrap-corrected calibration slope using 200 resamples. Statistical significance was defined as a two-sided P < 0.05. Analyses were performed using R software (version 4.3).

RESULTS

A total of 246 adults underwent PTBD from January 2021 to August 2023. Twelve cases were excluded (8 concurrent hepatobiliary procedures and 4 bilateral drainage procedures with unclear laterality), leaving 234 index PTBD procedures for analysis. The study flow diagram summarizing patient inclusion and exclusion is presented in Figure 1.

Figure 1 Study flow diagram of patient selection and inclusion for analysis. PTBD: Percutaneous transhepatic biliary drainage.

Procedure-related complications occurred in 33 of 234 procedures (14.1%). The most frequent events were minor self-limited bleeding (13/234, 5.6%) and catheter dislodgement (13/234, 5.6%), followed by biliary perforation (4/234, 1.7%). Major bleeding requiring surgery, pseudoaneurysm, and major vascular puncture each occurred in 1 of 234 procedures (0.4%). Overall, bleeding or vascular complications occurred in 16 of 234 patients (6.8%); however, only 3 of 234 events (1.3%) represented clinically significant bleeding or vascular injury (major bleeding, pseudoaneurysm, or major vascular puncture). No patient died within 14 days of the index procedure. The detailed distribution of complications by type is shown in Figure 2.

Figure 2 Procedure-related complications. A and B: Catheter dislodgement on cholangiogram in a case of biliary obstruction due to a Klatskin tumor with bilateral biliary drainage. Image A is a cholangiogram before contrast injection, with orange arrows indicating the positions of the left and right biliary drainage catheters. Image B is a cholangiogram after contrast injection: The left catheter remains in the correct position, and contrast fills the left intrahepatic ducts up to the hepatic hilum. However, the right catheter has become dislodged from the biliary tract, resulting in contrast extravasation into the right subphrenic space (orange arrowhead), with no visualization of the right intrahepatic bile ducts; C-F: Pseudoaneurysm images on axial, arterial phase abdominal multi-slice computed tomography scan in a case of biliary obstruction due to a pancreatic head tumor with right hepatic duct drainage. Image C (non-contrast phase) shows the distal tip of the catheter located within the common bile duct (orange arrow). Images D (non-contrast), E (arterial phase), and F (venous phase) at the level of the pancreatic head reveal a large retroperitoneal hematoma containing a pseudoaneurysm, without evidence of contrast extravasation. The lesion appears as a low-attenuation area on non-contrast phase, becomes hyperattenuating in the arterial phase (orange asterisk), and shows no change in attenuation or shape across phases. It is surrounded by and communicates with a hypoattenuating (non-contrast), smooth-walled sac; G: Catheter dislodgement on axial, arterial phase abdominal multi-slice computed tomography scan in a case of biliary obstruction due to a pancreatic head tumor with left hepatic duct drainage. The orange arrow indicates the catheter located within the liver parenchyma, adjacent to the surface of the left hepatic lobe, rather than within the bile ducts, which are dilated on the left side.

Patients with bleeding or vascular complications differed from those without events across several clinical and procedural characteristics, as shown in Table 1. They were more often male (75.0% vs 40.4%, P = 0.009), had higher pre-procedural white blood cell counts (median 15.0 K/μL vs 9.7 K/μL, P = 0.010), underwent longer procedures (30 minutes vs 25 minutes, P = 0.013), and more frequently required more than one needle pass (50.0% vs 9.6%, P < 0.001). Choledocholithiasis was also more common in the bleeding group (43.8% vs 9.2%, P < 0.001). In contrast, baseline hemostatic parameters were similar between groups, including platelet count (312.5 × 10⁹/L vs 305.0 × 10⁹/L, P = 0.576), INR (1.19 vs 1.19, P = 0.974), and PT (14.25 seconds vs 14.00 seconds, P = 0.601). The diameter of the target duct on the punctured side was smaller in the bleeding group (10.0 mm vs 11.0 mm), although this difference did not reach statistical significance (P = 0.131). Documented pre-procedural antithrombotic use was uncommon (7/234, 3.0%) and was not associated with bleeding in univariable analysis (12.5% vs 2.3%, P = 0.075). An illustrative case of severe post-PTBD bleeding due to extrahepatic intercostal vessel injury is shown in Figure 3, highlighting that major hemorrhage can originate along the chest-wall access tract rather than exclusively from intrahepatic vessels and emphasizing the need for prompt recognition and hemostatic intervention.

Figure 3 Extrahepatic intercostal hemorrhage after right intercostal percutaneous transhepatic biliary drainage. A: Contrast-enhanced computed tomography demonstrates active extravasation along the access tract (blue arrow) adjacent to the percutaneous transhepatic biliary drainage catheter (orange arrow); B: Intraoperative image confirms chest-wall or intercostal bleeding.

Table 1 Baseline and procedural characteristics by bleeding or vascular complication status, n (%)/median (interquartile range).

Characteristic	No bleeding (n = 218)	Bleeding/vascular complication (n = 16)	Overall (n = 234)	P value
Demographics and clinical variables
Age (years)	62 (52, 69)	67 (52, 83)	63 (52, 70)	0.221
Male sex	88 (40.4)	12 (75.0)	100 (42.7)	0.009
Body mass index (kg/m2)	21.0 (18.7, 23.0)	20.0 (18.3, 22.0)	21.0 (18.7, 23.0)	0.319
Acute cholangitis (TG18)	130 (59.6)	13 (81.2)	143 (61.1)	0.113
Hypertension	49 (22.5)	5 (31.2)	54 (23.1)	0.537
Diabetes mellitus	26 (11.9)	3 (18.8)	29 (12.4)	0.428
Coronary artery disease	16 (7.3)	3 (18.8)	19 (8.1)	0.129
Chronic kidney disease	3 (1.4)	0 (0)	3 (1.3)	1.000
Cirrhosis	2 (0.9)	0 (0)	2 (0.9)	1.000
Documented antithrombotic use	5 (2.3)	2 (12.5)	7 (3.0)	0.075
Hemostasis and laboratory variables
Platelet count (× 109/L)1	305.0 (230.0, 401.5)	312.5 (180.8, 411.0)	305.0 (224.0, 403.0)	0.576
INR2	1.19 (1.06, 1.36)	1.19 (1.05, 1.31)	1.19 (1.06, 1.36)	0.974
PT (seconds)2	14.00 (12.80, 16.52)	14.25 (12.78, 17.77)	14.00 (12.80, 16.62)	0.601
Platelet < 150 × 109/L1	18 (8.5)	3 (18.8)	21 (9.3)	0.174
INR >1.52	32 (15.4)	3 (18.8)	35 (15.6)	0.721
WBC before PTBD (K/μL)	9.7 (7.5, 14.5)	15.0 (10.4, 19.3)	10.0 (7.7, 15.0)	0.010
Total bilirubin before (μmol/L)	246 (134, 350)	252 (78, 320)	248 (125, 350)	0.436
Direct bilirubin before (μmol/L)	141 (88, 200)	140 (44, 185)	141 (87, 199)	0.388
AST before (U/L)	138 (79, 201)	110 (80, 195)	136 (80, 201)	0.851
ALT before (U/L)	117 (54, 191)	103 (46, 221)	114 (54, 192)	0.704
Imaging and procedural variables
Right intrahepatic duct diameter (mm)	11.0 (8.0, 14.0)	10.0 (7.0, 11.0)	10.5 (8.0, 13.0)	0.200
Left intrahepatic duct diameter (mm)	11.0 (9.0, 14.0)	10.5 (7.5, 11.0)	11.0 (9.0, 13.0)	0.095
Target-duct diameter on punctured side (mm)	11.0 (9.0, 14.0)	10.0 (7.0, 11.0)	11.0 (8.0, 13.8)	0.131
Choledocholithiasis	20 (9.2)	7 (43.8)	27 (11.5)	< 0.001
Procedure duration (minutes)	25 (20, 35)	30 (25, 45)	25 (20, 35)	0.013
More than one needle pass	21 (9.6)	8 (50.0)	29 (12.4)	< 0.001
Right-sided drainage	150 (68.8)	12 (75.0)	162 (69.2)	0.781
Technical success	211 (96.8)	15 (93.8)	226 (96.6)	0.438
Catheter size > 8 Fr	4 (1.8)	0 (0)	4 (1.7)	1.000

¹Platelet data were available for 227 patients (211 without bleeding or vascular complications and 16 with bleeding or vascular complications).

²International normalized ratio and prothrombin time data were available for 224 patients (208 without bleeding or vascular complications and 16 with bleeding or vascular complications).

P values were derived using the Wilcoxon rank-sum test for continuous variables and Fisher’s exact test (or Pearson’s χ² test when appropriate) for categorical variables. TG18: Tokyo Guidelines 2018; PT: Prothrombin time; INR: International normalized ratio; WBC: White blood cell count; PTBD: Percutaneous transhepatic biliary drainage; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase.

Given the limited number of events, the primary multivariable analysis was simplified to four clinically selected variables to improve the EPV ratio. In this reduced Firth penalized logistic regression model (16 events, 4 predictors; EPV = 4.0), more than one needle pass [adjusted odds ratio (aOR): 7.91, 95%CI: 2.05-30.59; P = 0.003], choledocholithiasis (aOR: 18.80, 95%CI: 4.79-73.73; P < 0.001), and longer procedure time per 10-minute increase (aOR: 1.44, 95%CI: 1.05-1.96; P = 0.023) were independently associated with bleeding or vascular complications, while larger target-duct diameter was protective (aOR: 0.82 per 1-mm increase, 95%CI: 0.69-0.98; P = 0.026). These findings are summarized in Table 2, and model performance remained acceptable despite sparse events, with an apparent AUC of 0.873, an optimism-corrected AUC of 0.848, and an optimism-corrected calibration slope of 0.886.

Table 2 Factors independently associated with bleeding or vascular complications after percutaneous transhepatic biliary drainage (reduced Firth penalized logistic regression model).

Characteristic	Adjusted OR	95%CI	P value
More than one needle pass	7.91	2.05-30.59	0.003
Choledocholithiasis	18.80	4.79-73.73	< 0.001
Procedure time (per 10 minutes)	1.44	1.05-1.96	0.023
Target-duct diameter on punctured side (per 1 mm)	0.82	0.69-0.98	0.026

Model characteristics: n = 234, bleeding or vascular events = 16, number of predictors = 4, events-per-variable = 4.0. Internal validation: Apparent area under the receiver-operating characteristic curve = 0.873; optimism-corrected area under the receiver-operating characteristic curve = 0.848; optimism-corrected calibration slope = 0.886. Sensitivity analyses are described in the results text. OR: Odds ratio; CI: Confidence interval.

In an exploratory sensitivity model that additionally included sex (5 predictors; EPV = 3.2), male sex remained associated with bleeding or vascular complications (aOR: 6.92, 95%CI: 1.61-29.81; P = 0.009). However, this finding is interpreted cautiously as exploratory rather than causal. In a separate pre-procedural hemostasis sensitivity model incorporating platelet count and INR, neither platelet count nor INR was independently associated with bleeding or vascular complications.

DISCUSSION

In this retrospective single-center cohort of 234 consecutive index PTBD procedures performed during a period when routine fluoroscopy was unavailable, overall procedure-related complications occurred in 14.1%, while bleeding or vascular complications occurred in 6.8%. Most bleeding or vascular events were minor and self-limited; only three patients (1.3%) experienced clinically significant events and no patient died within 14 days after PTBD. These overall rates are within the range reported in contemporary PTBD series and remain below the major-complication benchmark recommended by the Society of Interventional Radiology[4,10]. Published reviews and technique primers describe major bleeding in approximately 2% to 5% of PTBD cases, although definitions and case mix vary across studies[6,11].

The composite bleeding or vascular endpoint should be interpreted alongside its component events. Although the composite was retained for regression because only 16 total events occurred, the burden of severe vascular injury was much lower than the composite rate alone might suggest. Because only three clinically significant events were observed, any multivariable model restricted to severe events alone would have been statistically unstable and potentially misleading. Accordingly, these severe events are better described descriptively rather than modeled separately.

The ultrasound-guided workflow used in this cohort should also be interpreted in the context of institutional resource constraints. Routine fluoroscopy was temporarily unavailable, so this approach reflected local practice during the study period rather than a comparative protocol. Real-time ultrasound with color Doppler is nevertheless recommended for access planning and has been shown to reduce radiation exposure while remaining feasible in urgent or bedside settings[5,12,13]. Our data therefore support the feasibility of this workflow in one hepatobiliary service, but they should not be interpreted as evidence of equivalence or superiority vs fluoroscopy-assisted PTBD, endoscopic retrograde cholangiopancreatography, or endoscopic ultrasound-guided biliary drainage.

One illustrative case also reinforces that post-PTBD hemorrhage does not always arise from intrahepatic vessels. Extrahepatic chest-wall or intercostal vessel injury along the access tract is a recognized mechanism of severe bleeding and warrants prompt escalation when shock, bright-red drainage, or hemothorax is suspected[6,14].

In the primary Firth model, bleeding or vascular complications were associated with more than one needle pass, longer procedure time, choledocholithiasis, and smaller target-duct diameter on the punctured side. This pattern suggests that baseline anatomy and intraprocedural difficulty are tightly linked. The inverse association with target-duct diameter provides an anatomical correlate for the practical observation that smaller or less favorable ducts are harder to access and may increase the likelihood of vascular traversal or repeated manipulation. For the same reason, number of needle passes and procedure time should be interpreted primarily as markers of technical difficulty rather than purely pre-procedural predictors. This interpretation is consistent with prospective multicenter data identifying repeated puncture as an important determinant of hemorrhagic risk and with studies showing greater technical difficulty and bleeding risk in less dilated ducts[15,16].

Choledocholithiasis was also strongly associated with bleeding or vascular complications. A plausible explanation is that stone-related obstruction often coexists with cholangitis, inflammatory edema, friable ductal mucosa, and urgent drainage scenarios that collectively increase technical difficulty and susceptibility to bleeding. At the same time, the CI remained wide, so this finding should not be overinterpreted as a definitive causal effect and requires confirmation in larger cohorts.

Platelet count, INR, and PT showed no clear differences between groups and were not significantly associated with the composite bleeding or vascular outcome in descriptive analyses; platelet count and INR were likewise not significant in the hemostasis sensitivity model. However, these negative findings do not eliminate confounding by coagulation management. Only seven patients had documented antithrombotic use, and exact case-level data on withholding interval, reversal, or peri-procedural optimization were not consistently reconstructable from retrospective charts. In routine practice, urgent PTBD for cholangitis was generally not delayed, whereas patients with biliary obstruction without cholangitis usually had antithrombotic therapy withheld according to contemporaneous recommendations when feasible. Because these management details could not be verified reliably for each procedure, residual confounding by hemostatic management remains an important limitation.

The association between male sex and bleeding should also be interpreted cautiously. When sex was included in an exploratory sensitivity model, male sex remained associated with bleeding or vascular events, but it was not included in the reduced primary model because the biological basis for the association is unclear and prioritizing parsimony improved the EPV ratio. We therefore regard sex as an exploratory signal rather than a causal determinant.

We did not observe a clear laterality signal in this cohort. The literature on access laterality remains heterogeneous; a systematic review and meta-analysis suggested higher risk of arterial injury in some left-sided subgroup analyses, whereas randomized and contemporary comparative studies have not shown consistent overall differences in adverse events[17-19].

This study has several limitations. It was retrospective, single-center, and based on a modest number of bleeding or vascular events. The primary model was intentionally simplified, yet it remains susceptible to imprecision and residual confounding. Operator-specific experience, trainee involvement, and exact periprocedural antithrombotic management could not be reconstructed reliably. Follow-up was limited to 14 days and may therefore have underestimated delayed hemobilia or pseudoaneurysm formation. Strengths of the study include a consecutive cohort, explicit separation of minor and clinically significant bleeding events, inclusion of platelet count, INR/PT, and documented antithrombotic exposure when available, incorporation of a target-duct anatomical variable, and use of penalized regression with internal validation. Overall, the data suggest that bleeding after PTBD in this cohort was driven less by a single laboratory abnormality than by the interaction between unfavorable anatomy, inflammatory etiology, and procedural difficulty.

CONCLUSION

In this retrospective single-center cohort of ultrasound-guided PTBD performed during a period when routine fluoroscopy was unavailable, overall complication rates were acceptable and clinically significant bleeding or vascular events were uncommon. Bleeding or vascular complications were associated mainly with more than one needle pass, longer procedure time, choledocholithiasis, and smaller target-duct diameter on the punctured side. These findings describe the safety profile of one institutional workflow and should not be interpreted as comparative evidence.

ACKNOWLEDGEMENTS

The authors would like to express their sincere gratitude to the Department of Hepato-Pancreato-Biliary Surgery at Nhan dan Gia Dinh Hospital, Ho Chi Minh City, Viet Nam for providing invaluable support throughout the study. We are deeply grateful to all patients and their families whose medical data contributed to this research. Without their cooperation, this study could not have been conducted.