Comparing early and delayed repair of common bile duct injury to identify clinical drivers of outcome and morbidity

Abstract

Background

Outcomes following repair of common bile duct injury (CBDI) are influenced by center and surgeon experience. Determinants of morbidity related to timing of repair are not fully described in this population.

Methods

Patients with CBDI managed surgically at a single center from January 2008 to June 2015 were retrospectively reviewed. Outcomes of patients undergoing early (≤48 h from injury) and delayed (>48 h) repair were compared. Predictive modeling for readmission was performed for patients undergoing delayed repair.

Results

In total, 61 patients underwent surgical biliary reconstruction. Between the early and delayed repair groups, no differences were found in patient demographics, injury classification subtype, vasculobiliary injury (VBI) incidence, hospital length of stay, 30-day readmission rate, or 90-day mortality rate. Patients undergoing delayed repair exhibited increased chance of readmission if VBI was present or if multiple endoscopic procedures were performed prior to repair. A predictive model was constructed with these variables (ROC 0.681).

Conclusion

When managed by a tertiary hepatopancreatobiliary center, equivalent outcomes can be realized for patients undergoing early and delayed repair of CBDI. Establishment of evidence-based consensus guidelines for evaluation and treatment of CBDI may allow identification of factors that drive morbidity and predict clinical outcomes in this population.

Introduction

Injury to the extrahepatic biliary tree is a well-described complication of cholecystectomy.^{1, 2} Though a recent publication describes a decreasing rate of CBDI associated with laparoscopic cholecystectomy,³ CBDI remains a serious concern for patients and surgeons. Despite studies identifying patient^{1, 4, 5} and surgeon-related^{5, 6, 7} factors associated with CBDI including inflammation and conversion to open cholecystectomy, approximately 30% of CBDI are not identified during the index operation⁸ and may not be recognized until several days after the initial injury.

While the Strasberg-Bismuth injury classification system standardizes the anatomical description of biliary injuries^{2, 9} with added descriptions of vasculobiliary injury (VBI)^{5, 10} and extreme VBI,¹¹ no injury description model describes the optimal time for repair of injury based on injury type or severity. In patients with suspected or identified injury, additional surgeries or endoscopic procedures may be required to diagnose the injury, to treat acute intra-abdominal processes such as biliary peritonitis, and to definitively treat biliary injury with reconstitution of bilioenteric flow. Each additional procedure adds to the healthcare cost and presents a procedure-related risk of adverse events, and the deleterious effects on patients' quality of life (QOL) have been shown to extend dramatically beyond the acute injury and treatment phases.^{12, 13} Both early and delayed repair of CBDI are described. Because no guideline exists to guide the timing of repair, the decision for timing of repair in a patient equally eligible for early or delayed repair should be based on the predicted success of the procedure and patient safety. If equivalent technical and morbidity outcomes can be achieved, patient quality of life and efficient healthcare resource utilization should be considered.

Given the frequency with which cholecystectomy is performed worldwide, even a low rate of CBDI presents a tremendous potential healthcare burden. Previous studies of CBDI have identified several factors associated with successful repair. These include treatment by an experienced hepatopancreatobiliary (HPB) surgeon and specialized center as well as multidisciplinary perioperative care.^{9, 14, 15, 16}

Many patients with bile duct injuries are referred for surgical evaluation and intervention to Carolinas Medical Center (CMC), which serves as the central HPB referral center for a 48-hospital healthcare system spanning four states in the southeastern United States. The aim of this paper is to examine the CMC experience with repair of CBDI and to analyze outcomes in patients undergoing acute and delayed repair of CBDI.

Methods

Institutional Review Board (IRB) approval was obtained for retrospective data analysis. Institutional records were queried to identify patients with the diagnosis of CDBI by use of International Classification of Diseases (ICD)-9 codes at a single center from January 1, 2008, to June 30, 2015. Duplicate records were eliminated and individual records were reviewed individually to confirm that all patients included were treated for CBDI. All surgical procedures were performed by one of five attending HPB surgeons.

Patient records were retrospectively entered into a web-based, secure, HIPAA-compliant database. Patient demographics, time elapsed from index procedure, operative findings including injury classification and procedural variables, and postoperative course including 30-day readmission and 90-day mortality were recorded. Frequency of radiographic and endoscopic biliary instrumentation prior to and following definitive repair was recorded. CBDI type was reported according to the Strasberg-Bismuth classification system.^{2, 9} To determine presence of VBI, radiographic images were reviewed and intraoperative ultrasound (IOUS) findings describing hepatic vascular flow were obtained from operative notes. Immediate intraoperative CBDI repairs were assessed with IOUS for VBI, while patients transferred acutely after suspected injury or repaired in delayed fashion were assessed with triphasic computed tomography (CT) imaging of the liver to determine the presence of vascular injury. Any patient with hemodynamic instability on arrival underwent operative exploration prior to imaging.

Patients were divided into two groups based on the timing of definitive biliary repair or reconstruction: acute (≤48 h from index procedure) and delayed (>48 h after index procedure). Continuous and categorical variables were presented as medians and ranges and counts and percentages, respectively. Comparative analysis of continuous variables was performed using Wilcoxon rank-sum test procedure, and categorical variables using Fisher's exact test. All values were considered statistically significant at the P < 0.05 level. Statistical analyses were performed with Stata software version 13 (StataCorp LP, College Station, Texas, USA).

To determine the effect of delayed intervention on comorbidity, injury type, and readmission, patients in the delayed repair cohort were subdivided according to 30-day readmission status, and univariate analyses were used to test all independent variables against the binary 30-day readmission outcome. Variables analyzed for correlation included patient demographics, injury subtype, presence of vascular injury, chronic medical conditions, medications, intraoperative details of the biliary reconstruction surgery, number of therapeutic endoscopic retrograde cholangiopancreatography (ERCP) procedures >1, presence of VBI, requirement for percutaneous transhepatic cholangiography (PTC) or transabdominal drain, and hospital length of stay. Values of P < 0.25 indicated significance. A multiple logistic regression model was then constructed using values identified as significant on univariate analysis. Further stepwise backward elimination with P < 0.10 was performed, and a Homer–Lemeshow test statistic was used to determine the goodness-of-fit of the model.

Using the coefficients generated for the independent variable of 30-day readmission as well as overall model coefficient, a predictive algorithm was generated. To determine the model's predictive accuracy, the individual probability of 30-day readmission for each record was calculated and compared to observed 30-day readmission. Each individual's resultant expected probability of readmission was recorded, and Brier scores and receiver operating characteristic (ROC) curves were generated to determine the ability of the model to discriminate true-positive results and false-positive results as well as the overall predictive capacity of the model. A readmission score was created according to the final logistic regression model. The ERCP variable was stratified according to number of procedures (1 or ≥2). Odds ratio (OR) values of the VBI and ERCP category variables were rounded to the nearest whole integer, and a value of one was subtracted. These scores were assessed using ROC with an area under the curve (AUC) of >0.70 considered to represent a feasible model.^{17, 18, 19}

Results

During the study period, 81 patients were evaluated for CBDI. Of these, 20 patients were excluded from further analysis: 17 were treated definitively with ERCP and required no surgical biliary intervention. One patient was admitted after undergoing Roux-en-Y hepaticojejunostomy to repair iatrogenic CBDI identified during a laparoscopic converted to open cholecystectomy; this patient required no further surgical intervention. Two patients had cystic duct stump leaks (Strasberg-Bismuth type A CBDI) after laparoscopic cholecystectomy with incidental finding of gallbladder adenocarcinoma with positive cystic duct margins; these two patients underwent hepatectomy with portal lymphadenectomy and bile duct resection and were excluded from further analysis. Of the 17 patients treated definitively with ERCP, 5 required laparoscopic abdominal washout and drain placement for biliary peritonitis. During the study period, a total of 199 patients were managed by ERCP and stenting alone and did not require surgical biliary reconstruction. Including patients treated endoscopically and surgically for CBDI, 260 patients in total were treated for CBDI over the study period; endoscopic intervention was sufficient definitive treatment in 76.5% (199/260) of these patients. These were primarily Strasberg-Bismuth Type A and D injuries.

In 5 (8%) of the 61 patients treated surgically, intraoperative consultation with an HPB surgeon resulted in immediate repair of CBDI. A further 22 (36.1%) patients underwent repair ≤48 h of injury after transfer to CMC. In total, 27 (44.3%) patients underwent repair within 48 h of injury. The remaining 34 (55.7%) patients underwent delayed repair. Of all 61 occurrences of CBDI treated surgically, 5 (8.2%) required urgent abdominal exploration for bleeding or septic complications on arrival from outside institutions. Four patients with VBI underwent early repair at a second-look procedure after initial exploration was performed for arrest of hemorrhage (n = 3) or reconstruction of the common hepatic artery and portal vein (n = 1). The remaining patients with VBI (n = 8) underwent delayed repair.

Table 1 describes operative characteristics and outcomes of early and delayed CBDI surgical repair procedures performed within the study period. The original surgical intent precipitating injury was most often cholecystectomy (n = 43). Between patients undergoing early or delayed repair of CBDI, no differences were found in demographics and clinical outcomes. Comparing presurgical values and those obtained 1 month after surgery, BMI decreased in both groups (early, −5.8%, delayed, −4.4%; P = 0.122). Change in serum albumin was analyzed as a surrogate for nutritional status by comparing values prior to surgery and at 1-month follow-up. All other clinical characteristics and operative outcomes were similar.

Table 1.

Operative characteristics and outcomes of early and delayed common bile duct injury surgical repair procedures performed at a single center, 2008–2015 (n = 61)

Characteristic or outcome	Immediate and early repair (≤48 h from injury)	Delayed repair (>48 h from injury)	P Value
Patients, No.	27	34	0.440
Men, No.	12	13	0.624
Age, median, y	58 (18–92)	51 (24–80)	0.316
BMI, kg/m2, median (range)	24.2 (16.9–54.0)	26.6 (15.8–44.9)	0.773
Etiology, No.
Cholecystectomy	16	27	0.446
Non-biliary abdominal procedures	9	7 (20.6)	0.396
Abdominal trauma	2	0	0.107
Hospital length of stay, median, days (range)	7 (2–45)	8 (3–56)	0.444
30-day readmission, No.	3	5	0.680
90-day mortality, No.	3	1	0.092
Intraoperative Repair, No.	5	–	–
ASA classification ≥3, No.	16	20	0.989
EBL, median, mL (range)	100 (15–2000)	250 (25–1000)	0.837
Time to repair, median, days (range)	1 (0–2)	51 (3–1825)b	0.837
Vasculobiliary injury, No.	4	8	0.395
Strasburg-Bismuth classification, No.
A	0	2	0.200
B	0	1	0.369
C	3	4	0.937
D	2	2	0.811
E1	7	3	0.073
E2	7	9	0.962
E3	4	7	0.560
E4	2	4	0.570
E5	0	1	0.369
Xa	2	1	0.107
Preoperative PTC catheter placement, No.	–	17	–
Preoperative percutaneous transabdominal drain placement, No.	–	10	–
Δ albumin, g/dL, median (range)c	−3.6% (−46.4 to 75%)	11.1% (−45 to 64.7%)	0.025

Abbreviations: h, hours; y, years; ASA, American Society of Anesthesiologists; EBL, estimated blood loss; PTC, percutaneous transhepatic cholangiography.

^aInjury caused during endoscopic retrograde cholangiopancreatography.

^bExcluding a single repair performed 1825 days after injury, median time to repair was 69 days.

^c(Preoperative albumin-postoperative albumin)/preoperative albumin. Preoperative albumin measured within 24 h prior to surgery; postoperative albumin collected at 1-month postsurgical follow-up.

Among the 34 patients who underwent delayed repair, VBI (OR 4.2, P = 0.10) and each additional preoperative ERCP beyond that used for diagnosis of CBDI (OR 1.7, P = 0.07) correlated with increased risk of 30-day readmission after the eventual biliary reconstruction. These values were used to construct a predictive model for 30-day readmission after delayed repair of CBDI. For these patients, absence of VBI and only a single diagnostic ERCP prior to repair resulted in a 50/50 chance of readmission. For those without VBI and >1 ERCP prior to repair, the predicted readmission rate was 73.1%. In patients with both VBI and >1 ERCP prior to repair, the predicted rate of readmission was 88.1%. The Brier score for the model, indicating ability of the model to predict average risk of occurrence across a dataset, was 0.497; the ROC for the readmission model, indicating its ability to predict an outcome at the level of an individual patient, was 0.68 (P = 0.03) (Fig. 1).

Figure 1.

Predictive capability of an algorithmic model to identify increased risk for 30-day readmission following delayed biliary reconstruction. From this study population, an algorithm was generated to predict 30-day readmission following delayed surgical repair of common bile duct injury based on multiple endoscopic retrograde cholangiopancreatography (ERCP) procedures and history of vasculobiliary injury. Brier scores and receiver operating characteristic (ROC) curves were generated to determine the ability of the model to discriminate true-positive results and false-positive results as well as the overall predictive capacity of the model. Area under the ROC curve (purple): 0.68 (P = 0.03). Reference line is in green

Discussion

A single-center experience of CBDI surgically managed from 2008 to 2015 is presented. Thorough clinical, radiographic, and endoscopic evaluation is stressed in CMC institutional practices to identify details of the biliary injury and associated vascular injury to select candidates for early repair while addressing acute problems arising in the temporization of CBDI requiring delayed repair. During the study period, 61 patients required operative reconstruction while an additional 199 were treated successfully with endoscopy and stenting alone. Injuries managed endoscopically were primarily Strasberg-Bismuth types A and D. Of 260 total CBDI, 199/260 (76.5%) were treated endoscopically; this compares with other series describing endoscopic success rates of 57.7%²⁰–62.2%.¹³

Delayed repair was necessitated primarily by initial HPB evaluation occurring several days after the index operation as well as the presence of VBI per institutional practice. Though late presentation may be due in part to the geographical expanse of the referral area, delay in injury recognition accounted for approximately 35% of patients who underwent delayed repair. One repair performed 1825 days after injury is not considered to have experienced a delay in diagnosis due to 8 previous ERCPs with stenting prior to surgical referral. The selection of 48 h as the differentiation between early and delayed repair was based on a clear point of differentiation at the 48-h timepoint, beyond which few repairs occurred prior to 4–6 weeks following injury. While early repair is variably described and suggested to be feasible up to 5 days⁸ or 7 days after injury,²¹ 48 h was selected for analysis due to the breakdown of timing in this retrospective analysis.

Large reviews of CBDI suggest that concomitant vascular injury is present in approximately 25% of these patients.^{10, 11, 14} While arterial injury may be considered an indication to delay repair given concern for biliary or bilioenteric anastomotic structure,^{8, 10, 11, 22, 23} liver transplantation may be required for catastrophic vasculobiliary injuries.^{11, 14, 24} The present series includes 4 patients with VBI who underwent early repair: three tangential hepatic vascular injuries requiring intervention to prevent ongoing hemorrhage and a single extreme VBI with segmental portal vein and proper hepatic artery excision. This patient underwent emergent venous allograft reconstruction of the hepatic vasculature within hours of injury as hepatic infarction was considered imminent, with liver transplantation as the only remaining option should revascularization fail. Hepaticojejunostomy was performed at a second-look procedure after confirming arterial and venous patency. Strasberg et al. suggest that extreme VBI may occur in 10% of VBI¹¹; this occurrence constitutes 8.3% of the CMC experience with VBI. Early hepaticojejunostomy after initial damage control surgery is performed only during second-look procedures following resuscitation and when primary repair is not feasible based on tissue quality, the location of injury, or for segmental biliary resections. This is not performed following washout for biliary peritonitis given the concern for two anastomoses in an infected or inflamed field.

Clinically relevant endpoints such as readmission were hypothesized to be influenced by markers of injury severity such as VBI or multiple ERCPs. However, the infrequency of 30-day readmission following reconstruction limits the generation of an accurate predictive model (ROC 0.682). Improving this model with additional patients undergoing delayed repair will allow surgeons to counsel patients and families on the possibility of readmission based on pre-reconstruction or perioperative factors. Identifying patients at risk for specific complications allows modification of perioperative pathways or post-discharge planning and individualizes the discussion of expectations.

Though national perioperative predictive models exist for Roux-en-Y hepaticojejunostomy as provided by the American College of Surgeons National Surgical Quality Improvement Program^® (ACS NSQIP^®) Surgical Risk Calculator,²⁵ these fail to incorporate disease-specific intrinsic determinants of morbidity²⁶ such as VBI, need for pre-intervention PTC or prolonged transabdominal drain, or nutritional deficiencies, which are hypothesized to drive the morbidity of CBDI.^{12, 13} Aggregate-based prediction tools currently fail to incorporate surgeon and center-specific measures such as case volume and focused HPB training, which influence outcomes of CBDI repair and other complex HPB procedures.^{15, 27, 28} For these reasons, novel predictive algorithms for a single procedure performed frequently at CMC were developed.

While consensus guidelines describe perioperative protocols for other HPB procedures,^{29, 30, 31, 32, 33} no evidence-based practice pathway exists to optimize perioperative care of patients undergoing biliary reconstruction. The elective nature of delayed reconstruction may allow the incorporation of evidence-based preoperative patient optimization and perioperative care; this would also facilitate the collection of uniform data to generate disease process-specific predictive platforms. In the absence of a biliary-specific Enhanced Recovery After Surgery (ERAS) pathway, CMC patients are enrolled in the ERAS pathway for hepatectomy.

This study is limited by its retrospective nature and by the small sample size. Therefore, the potential exists for type II error. The size of the catchment area may limit long-term follow-up for assessment of complications such as bilioenteric anastomotic stricture. As this study addresses CBDI treated surgically, this analysis potentially misses Strasberg-Bismuth type A and D injuries treated endoscopically at outlying facilities. This retrospective study also omits QOL measurement. Analysis of changes in serum albumin values prior to surgery and at 1-month follow-up demonstrated a significant difference in median percent change between the groups. This may suggest improvement in nutritional status with internalization of bile flow for nutritional absorption. Prospective analysis of nutritional assessments may provide more conclusive information. Though laparoscopic and robotic CBDI repair was performed, this series is insufficiently powered to determine differences between open and minimally invasive approaches.

Future directions for investigation include QOL assessment for endoscopic management, open repair, and minimally invasive repair. A standardized, evidence-based pathway for perioperative management of patients undergoing bilioenteric reconstruction is suggested with prospective data collection to strengthen existing predictive platforms and potentially identify injury-specific determinants of morbidity.

In conclusion, these data suggest that equivalent perioperative clinical outcomes can be achieved for early and delayed repair of CBDI by experienced surgeons and an experienced center with an algorithm for assessment and management of CBDI; analysis of this population led to the stratification of patients by the 48-h mark for consideration of early repair (Fig. 2). The importance of HPB surgeon assessment and repair of injury has been presented previously.^{14, 15, 16, 28} Though not addressed in this analysis, equivalent perioperative clinical outcomes between early and delayed repair of CBDI led to the hypothesis that QOL and total cost of care for CBDI may be improved if early repair is performed in appropriately selected patients. This would further support expeditious HPB referral for evaluation and treatment. Further analysis is required to investigate the determinants of morbidity and outcomes in patients undergoing CBDI repair.

Figure 2.

Algorithm for evaluation and assessment of bile duct injury. Intraoperative consults for CBDI considered for immediate intraoperative repair are evaluated with intraoperative ultrasonography to assess the patency of hepatic vasculature. Abbreviations: CBDI, common bile duct injury; OR, operating room; CT, computed tomography; VBI, vasculobiliary injury; ERCP, endoscopic retrograde cholangiopancreatography; PTC, percutaneous transhepatic cholangiography

Authors' contributions

All authors had complete access to the study data that support the publication.

Study concept and design: Barnes, Kirks, Iannitti, Vrochides.

Acquisition of data: Barnes, Kirks, Cochran.

Analysis and interpretation of data: All authors.

Drafting of manuscript: Kirks.

Critical revision: All authors.

Approval of final draft to be submitted: All authors.

Funding sources

None.

Conflicts of interest

None declared.