Management of iatrogenic bile duct injuries: Multiple logistic analyses of predictive factors effective on morbidity and mortality

Aim: Iatrogenic bile duct injury (IBDI) is a devastating complication of laparoscopic cholecystectomy (LC) and associated with a nearly threefold increase in risk of mortality . The appropriate management procedure can alter the disease course and effect on the high morbidity and mortality rates. Prompt recognition is the emphasized point but the best approach varies with regard to the type of injury and featuring clinical symptoms. We aimed to determine the predictors of prognosis for the success of the treatment in postoperative period. Methods: We retrospectively analyzed the clinical parameters using logistic regression analysis of 105 patients who referred to Ege University School of Medicine Hospital after LC. The data was derived from the computer–based verified system in Ege University and identified through the medical charts. Univariate analysis was performed using Pearson x2 test and Fischer's exact test. A two-tailed p value of < 0.005 was considered significant. The level of injury was classified using Strasberg-Bismuth classification system.The patients were grouped according to the prior attempts and the level of the injury. Clavien-Dindo classification system is used for determining the severity grade of complications. Results: A total of 105 patients was treated for IBDI between 2004-2014. Fortyseven (45%) patients were operated following the first attempt of laparoscopic cholecystectomy, thirty-two (31%) patients were operated after multiple prior attempts following the injury in community hospitals and twenty-six (24%) patients with minor injuries were treated only with interventional techniques etc. ERCP, PTC . The clinical parameters of the patients were analysed using SPSS-PASW statistics 18.0 for Windows. The risk factors (p value < 0.10) in univariate analyse were selected for logistic regression analyse with forward LR method. The predictive risk for mortality was significant with the complicated level of injury (table 1). The predictive effect of level of injury was determined using the 'type A-D'(minor injuries) as reference group. The comparative groups were significantly increased the risk for sepsis (type E1-2 OR: 6.0 CI:1.45-24.9, type E3-4-5 OR: 6.78 CI:1.74- 26.43). The predictive value of level of injury was also significantly showed the risk for mortality when checked against the 'type E3-4-5' group (OR:8.86 CI:1.03-76.06). The predicted probability of sepsis was revealed with the number and type of the x prior attempts (OR:3 CI:1.17-8.05, OR: 10.1 CI: 1.66-62.6 ) before referral to Ege University School of Medicine. Before referral of the patients, ıt was clearly identified that the increasing number and varying procedures for repair are the aggravated factors for sepsis, even for the exclusive procedure of repair; hepaticojejeunostomy (OR:1.69). Therefore, we can strongly recommend to refer the patients with the prompt recognization immediately without intending further surgical procedures. The accompanying vascular injury is one of the strongest predictive factors for mortality and sepsis. As shown on the table 2; increasing level of repair procedure before referral can result with worsen outcomes via vascular injury. After referral of the patients to Ege University as tertiary center; the preferred approach for repair can also vary the risk for mortality and sepsis. As shown on table 3; an accompanying injury (vascular, intestinal,hepatic etc.) was increased the risk for mortality and sepsis (OR:14.5 CI: 1.42-148.5, OR: 12.6 CI:2.40-65.9). The group of nonsurgical procedures (ERCP, PTC etc.) was determined as the reference group. The comparative results showed that performing hepaticojejunostomy had no effect on risk of mortality and sepsis. Therefore, it was obvious to perform hepaticojejeunostomy for repair in tertiary centers. The factors effective on length of hospital stay was analysed using spearman's correlation coefficient test. The median time for referral (r:0.221 p:0.028), number of attempts for percutan drainage (r:0.430 p:0.006) and number of prior attempts (r:0.319 p:0.001) were effective on length of hospital stay. The analyse of the nonparametric correlation factors was inefficient for the length of hospital stay. In multiple logistic regression analyse; Clavien- Dindo classification system was the most significant variable for sepsis (Corrected odds: 4.62 CI: 1.56- 13.8). In the final model; the presence of abscess formation (CO:6.74 CI: 1.21- 37.60) , concomitant vascular injury (CO: 29.41 CI: 4.15 – 208.07) and the level of biluribin in blood (CO: 4.25 CI:1.034- 17.47) were significantly effective for mortality. Conclusion: The sepsis status and Clavien- Dindo grading system correctly relates the prognosis. The presence of abscess formation, vascular injury and biluribin levels in the referral are the potential risk factors. The initial clinical approach has to be determined in the light of predictive factors that are effective on morbidity and mortality.