Operative management of anastomotic leak after sigmoid colectomy for left‐sided diverticular disease: Ileostomy creation may be as safe as colostomy creation

The management of anastomotic leak after sigmoid colectomy for diverticular disease has not been well defined. Specifically, there is a lack of literature on optimal types of reoperations for leaks. The aim of this study was to describe and compare reoperative approaches and their postoperative outcomes.


INTRODUC TI ON
Anastomotic leak after left-sided colon resections remains a frustrating postoperative complication, with rates ranging from 3% to 21%, and association with significant morbidity, mortality, and poor bowel function [1][2][3][4][5][6]. Yet, there are no guidelines on how to manage colorectal anastomotic leaks. Traditionally, operative intervention has been considered standard of care, however, the decision to create a stoma and what type of stoma is most appropriate is unclear [7][8][9]. Of patients managed with end stomas, only 44% undergo subsequent re-establishment of intestinal continuity [7]. On the other hand, selected patients with contained leaks have been managed nonoperatively with percutaneous drainage, transrectal drainage, or advanced endoscopic techniques [8,10]. This balance between the desire to control sepsis and maintain long-term intestinal continuity, as well as the limited existing literature on management of leaks, makes decision-making regarding anastomotic leaks difficult.
This study seeks to address these important knowledge gaps in the management of colorectal anastomotic leaks, specifically focusing on patients who undergo elective sigmoid colectomy for leftsided diverticular disease. Our aim was to determine how patients with anastomotic leaks were managed, specifically focusing on reoperations and creation of ileostomies and colostomies. We also performed a detailed chart review of patients from our institution, enabling us to investigate intraoperative assessment of anastomotic donuts, leak tests, and long-term leak management. We hypothesize that there will be no significant differences in postoperative outcomes after ileostomy or colostomy creation at reoperation for leak.

Overall NSQIP cohort
For our retrospective cohort study, we utilized the American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP), which is a well-regarded multi-institution dataset that collects a sampling of cases from >700 NSQIP-participating sites in the United States. Surgical clinical reviewers at each site record >300 variables per case, including demographics, operative details, and postoperative complications up to 30-days after the index procedure. We also used the NSQIP Colectomy Module, which was developed in 2012 to record additional variables for colorectal procedures.
Using the 2012-2019 NSQIP Participant Use Files (PUF) and Colectomy Module, we identified all patients who underwent elective sigmoid colectomy with primary anastomosis, with or without diverting ileostomy, for diverticular disease. We limited our analysis to patients with diverticular disease because we wanted to minimize potential confounding factors that could have affected leak management, including malignancy that may have required Patients with CPT codes of 44146 and 44208 (open and laparoscopic partial colectomy with anastomosis, with low pelvic anastomosis with colostomy) were excluded because it is not common practice to create a low pelvic anastomosis with diverting colostomy for elective resection of diverticular disease, and the procedure or diagnosis may have been miscoded. Only patients undergoing elective procedures (defined as patient coming from home) and who underwent index operation on the day of admission were included.

Covariates
Demographics collected in the NSQIP dataset include age, sex, race, body mass index (BMI), American Society of Anaesthesiologists (ASA) classification, diabetes requiring oral agents or insulin, current smoker within 1 year of surgery, hypertension requiring medication, steroid or immunosuppressant use for a chronic condition, and preoperative albumin <3.0 g/dl.
Operative characteristics include whether patients received a preoperative mechanical bowel prep and/or oral antibiotic prep, surgical approach, operating time (minutes), and wound class.

Primary and secondary outcomes
The primary outcome of the study was development of anastomotic leak requiring reoperation. Anastomotic leak was defined in the Colectomy Module as "leak of gas, fluid, gastrointestinal contents, or contrast material outside the bowel lumen. The presence of an infection/abscess thought to be related to an anastomosis, even if the leak could not be definitively visualized during an operation or via contrast extravasation, was considered an anastomotic leak if

What does this paper add to the literature?
Literature on colorectal anastomotic leak management is limited, with no existing guidelines. Our study utilized one of the largest cohorts of leaks and found that, in selected patients who undergo reoperations for leaks, ileostomy at the time of reoperation is safe, with comparable results to colostomy.
indicated by the surgeon" [11]. This definition included both clinical and radiological diagnosis of an anastomotic leak, without differentiating between the two.

Single institution cohort
In order to obtain more detailed data, we examined a cohort of

Statistical analysis
Categorical variables were summarized as count (percentage)

RE SULTS
The overall NSQIP cohort included 37,471 patients who underwent sigmoid colectomy with primary anastomosis for diverticular disease. A total of 1003 (2.7%) patients suffered an anastomotic leak, of whom 583 (61.8% of patients with leaks) underwent an operative intervention for the leak.
Characteristics of patients with anastomotic leaks who did or did not undergo reoperation Table 1 illustrates the demographic and operative characteristics of patients who had an anastomotic leak and did not undergo reoperation (n = 361) and those who suffered an anastomotic leak requiring operative intervention (n = 583). Of note, 59 patients were reported as having a leak, but no treatment intervention was documented, therefore they were not included in the analysis. On univariate analysis, patients who underwent reoperation tended to be male (p = 0.03) and current smokers (p = 0.02). Faecal diversion with an ileostomy at the initial operation was associated with decreased likelihood of reoperation for anastomotic leak (1.9% vs. 4.7%, p = 0.01).

Postoperative outcomes after ileostomy versus colostomy creation at reoperation
When comparing patients who underwent ileostomy versus colostomy creation at reoperation, there were no significant differences in demographics or index operative characteristics between the two groups (all p > 0.05) ( Table 2).
Compared to ileostomy creation at reoperation, colostomy creation was associated with fewer days between index procedure and reoperation (6.5 vs. 8, p = 0.01), higher rates of septic shock (22.0% vs. 7.8%, p = 0.002), and higher rates of superficial SSI (13.5% vs. 5.9%, p = 0.045, Table 3). We were unable to determine if septic shock occurred before or after reoperation because most patients were missing data for the "days from operation until septic shock complication" variable. There were no statistically significant differences in ileus, length of stay, discharge to rehab, readmission, or 30day mortality (all p > 0.05).

Single institution experience in management of anastomotic leaks after sigmoid colectomy
Of the 534 patients from our institution who underwent elective sigmoid colectomy with primary anastomosis for diverticular disease and were captured in the local NSQIP files, 16 patients (3.0%) had anastomotic leaks. Table 4 summarizes the demographics, operative characteristics, and how the leak was managed in each patient.  [12] used NSQIP to examine anastomotic leaks after colonic resections, including ileocolic, colocolonic, and colorectal anastomoses from 2012 to 2013. They found that 3.8% of patients leaked, of whom 19.4% were managed medically, 24.5% were managed with nonsurgical intervention, and 56.1% underwent reoperation. They found that creation of a diverting stoma during the initial operation was associated with decreased need for reoperation, which is consistent with our data and other studies [13][14][15][16]. They also found that colocolonic anastomotic leaks had a higher risk of reoperation compared to ileocolic leaks.  In a patient with peritonitis and haemodynamic instability, the decision to reoperate is easy. Yet one of the main questions about leak management is whether to create a diverting ileostomy or end colostomy at reoperation. It is unclear if anastomotic defect size contributes to this decision, as this information is not captured in NSQIP and defect size was poorly documented in our chart review.

DISCUSS ION
Based on qualitative descriptions, it would seem that colostomies were performed in the setting of larger anastomotic disruptions, as operative notes included phrases such as the anastomosis "fell apart just by finger fracture" and "anterior wall dehiscence", compared to phrases for patients who received ileostomies, such as "pinpoint" and "very small hole." In some cases, the anastomosis was not examined in the operating room in order to prevent further disruption. There was no difference in ileus, length of stay, readmission, or mortality between patients who underwent ileostomy or colostomy at reoperation. Therefore, our data suggest it is safe to perform diverting ileostomy, not just a colostomy, in selected patients with left-sided colorectal anastomotic leaks, which is consistent with prior data [17]. Proximal diversion with preservation of the colorectal anastomosis is clearly preferable to end colostomy due to higher rates of ultimate intestinal continuity.
Limitations of this study include that NSQIP does not capture many characteristics probably contributing to how leaks were man-  n/a n/a n/a "Thin anastomotic contrast leak" on CT n/a n/a natural language processing. Surgeons could be incentivized to record these data by monitoring adherence to specific operative note templates as a quality metric. With these data and appropriately powered patient cohorts, we will hopefully learn how best to manage this frustrating complication.

CON CLUS ION
Our study, involving one of the largest cohorts of colorectal anastomotic leaks supplemented by detailed single-institution chart review, seeks to fill an important knowledge gap regarding management of anastomotic leaks. Of the overall NSQIP cohort, 2.7% of patients suffered an anastomotic leak, of whom 61.8% underwent reoperation. About half of patients who underwent reoperation had stomas created, with ileostomies created in one-third and colostomies in two-thirds of those patients. With no difference in mortality, length of stay, or readmission between patients who underwent ileostomy or colostomy at reoperation, we suggest that ileostomy is safe in selected patients with left-sided anastomotic leaks. Our single-institution data suggest that ileostomy creation is associated with a higher likelihood of restoration of intestinal continuity than colostomy creation. Therefore, when faced with a colorectal anastomotic leak, ileostomy creation may be considered.

FU N D I N G I N FO R M ATI O N
No funding was received for this study.

CO N FLI C T O F I NTER E S T S TATEM ENT
SDH has received consulting fees from Shionogi, Takeda, and Guidepoint. The remaining authors have no relevant conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.

E TH I C S S TATEM ENT
The Cleveland Clinic Institutional Review Board approved this study.