Short Term Outcomes of Open and Minimally Invasive Approaches to Segmental Colectomy for Benign Colovesical Fistula

Background We speculated that a minimally invasive (MIS) colectomy for colovesical fistula is associated with less morbidity compared to an open colectomy. Methods Multivariate analysis using logistic regression was used to investigate the outcomes of patients who underwent colectomy for benign colovesical fistula during 2012–2017 by surgical approach using the NSQIP database. Results We identified 748 patients underwent partial colectomy for benign colovesical fistula during 2012–2017. Surgeons used the MIS approach in 72.7% of operations, with a conversion rate of 13.1%. The MIS approach was associated with lower morbidity (27.4% vs. 43.1%, AOR: 0.46, P=0.02) compared to the open approach. The mean operation duration was longer in MIS operations compared to open (225 min vs. 201 min, P < 0.01). The robotic approach to colectomy showed no significant difference in morbidity (28.4% vs. 27.2%, P=0.77) but a decrease in conversion rate (8.1% vs. 13.8%, P < 0.01) and an increase in operation length (249 min vs. 222 min, mean difference: 27 min, P < 0.01) compared to a laparoscopic approach. There was no significant difference in the anastomotic leak rate between MIS and open approaches (3.7% vs. 5.4%, P=0.14) and between laparoscopic and robotic approaches (2.8% vs. 3.8%, P=0.99). Conclusions We found a 72.7% utilization rate of MIS approach to colectomy for benign colovesical fistula in the NSQIP hospitals with a 13.6% conversion rate. Patients with MIS approach had significantly lower morbidity compared to open. A robotic approach to partial colectomy has the same morbidity risk with a decreased conversion rate compared to laparoscopic approach.


Introduction
Colovesical fstula is a condition that can be a complication of a variety of diseases and conditions, including diverticulitis, cancer, Crohn's disease, and radiation [1,2]. Complicated diverticulitis with direct extension of a ruptured diverticulum or erosion of a peri-diverticular abscess into the bladder has been reported to be the most common cause, accounting for 70% of cases, followed by Crohn's disease in 10% of cases [1][2][3]. Te fundamental principle of surgical management of colovesical fstula is removal of the fstula and diseased segment of colon [2,4]. Morbidity of open procedures can be as high as 49% with a signifcant reoperation rate (up to 17%) [4][5][6][7]. Surgical treatment for colovesical fstula is evolving to decrease morbidity for the patients through the utilization of minimally invasive (MIS) approaches [4][5][6][7].
Feasibility, safety, and advantages of the MIS approach compared to a traditional open colectomy for diverticulitis have been previously cited [8][9][10]. A majority of patients undergoing an elective colectomy for diverticular disease receive a minimally invasive operation in the US now [8,10]. Recently published data revealed that laparoscopic treatment of complicated diverticulitis with colovesical fstula is feasible and safe, with better outcomes compared to open surgery when performed by skilled laparoscopic surgeons [4,11,12]. However, considering the heterogeneous minimally invasive surgery (MIS) skills of surgeons, such conclusions can only be generalized if similar outcomes can be found in a larger study such as a national database study. Using a nationwide database, this study aims to compare 30 days complications of the MIS approach with the traditional open approach for elective nonmalignant colovesical fstula.

Methods
We performed a retrospective analysis using the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database for the years 2012-2017. We queried adult patients (age eighteen-year-old and more) who underwent partial colectomy for colovesical fstula whose data were submitted to the ACS NSQIP using the Participant Use Data Files (PUF) and the target colectomy fles during 1/1/2012 to 12/31/2017. Te NSQIP database is a nationally validated, prospective, multiinstitutional database extracted from medical records by trained surgical clinical reviewers. ACS-NSQIP details more than 300 data points for deidentifed cases including patient demographics, comorbidities, perioperative characteristics, and 30-day postoperative complications in more than 600 participating institutions of varying sizes and academic afliations [13]. All data points are from deidentifed cases and the database is fully anonymized by the American College of Surgeons, and this study is exempt from IRB approval [13]. Te NSQIP database is available for researchers nationwide at participating hospitals [13].
In this study, we selected adult patients with colovesical fstula who underwent partial colectomy with anastomosis based on the current procedural terminology (CPT) codes of 44140 and 44204 for open and MIS approaches, respectively. Patients who had cancer as the reason of colovesical fstula were excluded from the study. Patients with colovesical fstula were identifed with an International Classifcation of Diseases, Ninth Revision (ICD-9) diagnosis code of 596.1 or International Classifcation of Diseases, Tenth Revision (ICD-10) diagnosis code of N32.1 within the database. Te patients were separated into groups based on whether they underwent open or MIS approaches. Variables compared between groups included demographics (age, race, and gender), comorbidities (such as diabetes, hypertension, and so on.), operative factors (such as operation length and surgical approach), and outcomes (postoperative complications, mortality, hospitalization length, and so on.). Te endpoints were comparing 30 days mortality and morbidity of the patients by surgical approaches. Variables in this study were defned as mentioned by the NSQIP User Guide, which can be referenced for detailed variable defnitions online [13]. Overall morbidity is defned as the presence of at least one postoperative complication of anastomosis leakage, intra-abdominal infection, sepsis, septic shock, ventilator dependency, cardiac arrest, hemorrhagic complication needs transfusion, pulmonary embolism, myocardial infarction, pneumonia, central vascular accident, acute renal failure, progressive renal insufciency, superfcial surgical site infection, deep surgical site infection, unplanned reoperation, deep venous thromboembolism, urinary tract infection, unplanned intubation, prolonged ileus, and wound disruption. Severe morbidity was defned as the presence of at least one of the complications of cardiac arrest, intra-abdominal infection, septic shock, pulmonary embolism, ventilator dependency, acute renal failure, myocardial infarction, anastomosis leakage, and pneumonia. Comparisons of patient characteristics were performed using a chi-square test for categorical variables and an independent t test for continuous variables to determine the diference in proportions for dichotomous and categorical variables between groups in the study. All independent variables that showed a signifcant diference (0.05) in the univariate model were placed in multivariate logistic regression or linear regression models to identify independent risk factors for primary adverse outcomes. Te adjusted odds ratio (AOR) and its confdence interval (CI) and P value were obtained from the fnal model as a measure of the association between the independent predictors and the dependent responses. Te one-way analysis of variance was used to assess the diference in mean for continuous variables. A P value of <0.05 was considered to indicate a statistically signifcant diference for all statistical tests.

Results
A total of 748 patients who underwent open or MIS partial colectomy with anastomosis with a diagnosis of colovesical fstula during 1/1/2012-12/31/2017 were identifed within the NSQIP database. Most patients were Caucasian (78.3%) and male (63.8%). Te most prevalent comorbidities included hypertension (56.4%) and obesity (40.2%). Patients were divided per surgical approach (open vs. MIS) into two groups of patients. Te descriptive statistics and patient demographics by surgical approach have been summarized in Table 1. Open surgery was more commonly performed in patients with COPD, weight loss, hypoalbuminemia, and renal failure on dialysis. Tere was limited information for type of bladder repair and using muscle fap to cover the fstula site. Tree patients required partial cystectomy following resection of the colovesical fstula, and the rest of the patients had repair of bladder following colovesicular fstula without resection. Also, 25 patients had a report for an omental or muscle fap to cover the repaired site of the bladder.
Overall, 544 (72.7%) of operations were conducted with MIS approach. Of these 74 patients (13.6%) had robotic surgery. Overall conversion rate to open for MIS approach was observed in 13.1% of cases (8.1% for robotic and 13.8% for laparoscopic approach). Tere was a steady increase in the utilization of the MIS approach to colovesical fstula from 63.9% for 2012 to 78.8% for 2017. Te median postoperative hospitalization length for MIS and open approaches to colovesical fstula were 5 and 8 days, respectively. Te mean operation duration was longer in MIS approach compared to open operations (225 min vs. 201 min, P < 0.01). Also, the mean operation duration was longer in robotic approach compared to laparoscopic operations (249 min vs. 221 min, P < 0.01).
A risk adjusted analysis of factors associated with 30-day mortality and morbidity in the patients with colovesical fstula who underwent partial colectomy with anastomosis are reported in Tables 2 and 3. Although surgical approach was not associated with mortality of the patients (AOR: 1.71, P � 0.59), MIS approach was signifcantly associated with decreased morbidity (AOR: 0.49, P � 0.01). Also, the American Society of Anesthesiologists (ASA) score more than two was signifcantly associated with an increase in morbidity of the patients (AOR; 2.05, P � 0.01).
30-day mortality, overall morbidity, severe morbidity, and postoperative complications of patients who underwent partial colectomy with anastomosis for colovesical fstula per surgical approach have been reported in Table 4. Overall morbidity, severe morbidity, sepsis, and hemorrhagic complications were signifcantly lower in the MIS approach compared to the open approach (Table 4).
A risk adjusted analysis of postoperative complications of the patients with colovesical fstula who underwent partial colectomy with anastomosis with the laparoscopic and robotic approaches are reported in Table 5. Multivariate analysis revealed there was not any signifcant diference in 30-day postoperative complications between the robotic and laparoscopic approaches. Conversion to open was higher in laparoscopic approach compared to robotic approach (13.8% vs. 8.8%, P < 0.01). However, the mean operation duration was longer in robotic approach than laparoscopic operations (248 min vs. 221 min, P < 0.01). When comparing anastomosis leakage for patients who underwent MIS approach there was no signifcant diference in the anastomotic leak rate of intracorporeal versus extracorporeal anastomosis (4.2% vs. 3.8%, AOR: 1.44, P � 0.70).

Discussion
Our study results show minimally invasive approach to colovesical fstulas is associated with less 30-day morbidity and shorter hospitalization length compared to open approach. We found a decrease in overall and severe morbidity of MIS approach compared to open with no signifcant change in mortality risk. Diverticular fstula is not a contraindication for the MIS approach, and multiple recently published articles revealed the benefts of MIS approach to colovesical fstula [4,12,[14][15][16][17][18]. Along this line we found an increase in utilization of MIS approach to colovesical fstulas during 2012-2017. Although American Society of Colorectal Surgeons (ASCRS) text book for colon and rectum surgery mentioned the benefts of the MIS approach for complicated diverticulitis compared to noncomplicated diverticulitis (page 660 chapter 39), the lack of RCTs does not allow for the drawing of statistically signifcant conclusions on the MIS approach for colovesical fstulas, despite the fact that this approach is considered safe [19].

Surgery Research and Practice
We found a signifcant decrease in the risks of overall morbidity, severe morbidity, sepsis, and hemorrhagic complications using the MIS approach compared to open surgery. In addition, we found the MIS approach is associated with shorter hospitalization compared to open surgery. Also, our study results show that there was a trend toward a decrease in multiple other complications in the MIS group compared to the open group that did not reach the level of statistical signifcance (Table 4). Te benefts of the MIS approach in colorectal surgery have been discussed broadly in the literature [8,10,20,21]. However, comparing the open and MIS approaches may be confounded by selection bias as the baseline characteristics of the two groups of patients with the MIS and open approaches in this study were heterogeneous (Table 1). We found that patients who underwent the MIS approach had less comorbid conditions of COPD, weight loss, and preoperative sepsis. Tis fnding shows there is a trend to operate with an open approach for sicker patients, which might be due to the shorter operation length. Prospective clinical trials need to compare outcomes of the open and MIS approaches to colovesicular fstula in two homogeneous groups of patients.
We found that the robotic approach to colovesical fstula may have advantages over the laparoscopic approach. When comparing the laparoscopic approach to the robotic approach, we could not fnd any signifcant diferences in morbidity and mortality risks. However, there was a trend toward a decrease in severe morbidity of robotic approach compared to the laparoscopic approach that did not reach the level of statistical signifcance in multivariate analysis. Also, the robotic approach had a signifcantly less conversion rate to open compared to laparoscopic approach in this study (8.1% vs. 13.8%). Features of robotic surgery such as three-dimensional vision, restoration of the eye-hand-target axis, better depth perception, and a better defnition of tissue planes that leads to precise dissection can be factors that help overcome some of the challenges of laparoscopic surgery and lead to a decrease in the conversion rate [22,23]. Te benefts of a robotic approach compared to the laparoscopic approach must be weighed against the longer operation length. More research is needed to better understand if the longer operation and probably increased cost in robotic approach is justifed by an improvement in outcomes.
We found longer operation times with shorter hospitalization for the patients underwent the MIS approach to colovesical fstula compared to open surgery. Shorter hospitalization length is one of the general advantage of the MIS approach which can result in signifcant reduce in costs per patient [10,[24][25][26]. However, the benefts of a MIS approach must be weighed against the longer operation time. Advancements in dissection and coagulation devices and increased experience of surgeons in MIS surgery may decrease the length of the procedures [23,27]. However, selected cases who cannot tolerate carbon dioxide insufation for long periods of time may still beneft from an open approach [28].

Study Limitations
Tis study has some limitations. Tis is a retrospective study, and we are unable to draw any causal conclusions and our study results need to be confrmed with a prospective randomized control trial. We could investigate the 30-day postoperative complications of the patients who had operations for colovesical fstula. However, information on the long term outcomes of the patients was not available in the NSQIP database. We compared two groups of patients who had an open and the MIS approach to colovesical fstula. However, the baseline characteristics of these two groups of patients were heterogeneous, and any conclusions may have bios. We attempted to adjust the results for all possible confounders, we could not capture all potentially important explanatory variables such as details of the surgical procedure, reason of conversion to open, and previous abdominal operation. Tere was limited information for type of bladder repair as well as to use of a muscle fap to cover the repaired site due to coding limitation and we could not compare the type of bladder repair (with or without resection) and the beneft of a muscle fap to prevent relapse of the fstula in our study. Tirty despite these limitations, the advantage of using the NSQIP database is the broad national geographic representation across all regions of the country with diferent surgeons MIS skills and this makes it a suitable database to evaluate outcomes in not just tertiary referral centers with specialized surgeons with high MIS skills but a great variety of centers.

Conclusions
Te majority of segmental colectomies for benign colovesical fstula in the NSQIP hospitals are being performed with the MIS approach (72.7%). Our study result shows patients with colovesical fstula who were treated with MIS approaches had signifcantly lower morbidity compared to an open approach. In the majority of the cases, colonic anastomosis in the MIS approach is done with the intracorporeal technique (53.4%) without a signifcant change in the risk of anastomotic leak compared to extracorporeal anastomosis. Te robotic approach to benign colovesical fstula happened in 13.6% of total MIS cases with the same morbidity risk with a modestly decreased conversion rate compared to the laparoscopic approach. Based on these results an MIS approach should be utilized when possible.

Data Availability
Te data used to support the fndings of this study are available at the national database NSQIP.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
Yeganeh Z, Dustin HH, and Kopatsis A conceived and designed the analysis, collected the data, contributed data or analysis tools, wrote the paper, approved the fnal version, and are accountable for all aspects of the work. Kopatsis AP conceived and designed the analysis, performed critical revision, co-wrote and edited the paper, approved the fnal version, and is accountable for all aspects of the work. 8 Surgery Research and Practice