INTRODUCTION

Gastrointestinal (GI) surgery-associated anastomotic leaks (ALs) have been one of the main causes underlying postopera- tive morbidity and mortality irrespective of the continual im- provements in surgical procedures. The frequency and conse- quences of anastomotic failure (partial or complete disruption of anastomosis with leakage of contents) vary widely accord- ing to the sites, definitions, and types of resections within the GI tract (1, 2). Varying rates of ALs are found (1) based on the anastomosis site involved: rectum (8-41%), colon (3-29%), small intestine (1-3%), bile ducts (10-16%), pancreas (9-16%), stomach (1-9%), and esophagus (2-16%).

In systematic reviews of 97 studies, a total of 56 separated def- initions of AL were identified. Combination of clinical features and radiological investigations were used to define and detect anastomotic leak. However, there is no universally accepted definition of anastomotic leak at any site (3).

Studies conducted to identify the incidence and risk factors for anastomotic leak are from the developed world where patient characteristics and availability and utility of diagnostic tools are different from that in the developing world. Several studies have identified risk factors for GI anastomotic leakage with no general consensus on which risk factors consistently feature

(1).

An anastomotic defect in colon causes leakage of colonic con- tent into the abdominal and or pelvic space leading to peritoni- tis, abscess formation, and sepsis that can be fatal. The inci- dence of colorectal anastomotic leak (CAL) varies between 3% and 19%, with associated mortality rates ranging from 10 % to

20 %. Moreover, CAL is a risk factor for local recurrence of colorectal cancer (4).

AL typically becomes clinically apparent between the 5 and the 8 postoperative days, but many exceptions exist (5, 6). The occurrence of gastrointestinal AL is associated with signifi- cantly increased mortality, morbidity, and prolonged hospital stay as well as considerable extra costs (4, 7). Knowledge on the risk factors may influence procedure related decisions and treatment, and possibly reduce the rate of leakage.

The aims of this study were to evaluate the pattern of AL after GI resection and anastomosis, its perioperative risk factors, morbidity and mortality rates related to AL, and to provide surgical professionals, researchers and hospitals with baseline information for further investigation and guideline develop- ment to reduce rate of AL and improve outcome of patients.

MATERIALS AND METHODS

This was a multicenter retrospective study of medical records of patients with GI resection and anastomosis undertaken at Tikur Anbessa, Menelik II, Yekatit 12 and Zewditu Memorial Hospitals in Addis Ababa during the period of January 1, 2017 to December 31, 2018.

During the study period, a total of 556 patients had GI resec- tion and anastomosis. Of these, charts of 414 patients were retrieved and 62 patients were excluded from the study because four were pediatric patients, 12 patients had insufficient data, and 46 patients had bypass procedures. The documents of 352 patients make the basis for the analysis of this study.

Structured questionnaires were prepared for data collection and the study proposal was approved by the Institutional Review Board of Addis Ababa University, College of Health Sciences. At the four hospitals, operating room registers, medical charts, discharge summaries, and death certificates of patients were used for data collection. Demographic fea- tures, comorbid conditions, diagnoses for which GI resection and anastomosis were performed, and presence or absence of perioperative infection, bowel preparation, antibiotic use, blood transfusion, and serum albumin level were determined. Emphasis was also made on the urgency of the operation, operating professional, and duration and type of the proce- dure as well as the duration of hospital stay and postopera- tive complications. The collected data were checked for completeness, and entered into computer and the SPSS ver- sion 23 package was used for statistical analysis.

42

RESULTS

The mean (±SD) age of patients was 48(±17) years; about 75% of patients were ≤ 60 years of age. Majority (72.2%) of patients were male and 257 (73.0%) were urban residents. The most common reasons for gastrointestinal anastomosis were redundant sigmoid colon and sigmoid volvulus (27.8%) fol- lowed by small bowel obstruction (SBO) (14.8%). The rest of the diagnoses in decreasing order of frequency included colo- rectal cancer (12.8%), esophageal cancer (12.2%), and end colostomy for 26 benign and 9 malignant diseases (9.9%), gastric cancer (5.4%), IBD (4.8%), and 10 penetrating and 3 blunt abdominal trauma (3.7%), and others (8.5%) including mesenteric ischemia, ileostomy or abdominal TB (10 each). Cancer accounted for 35.8% of all patients that underwent GI anastomosis (Table 1).

As is shown in Table 2, most patients (79.9%) underwent elec- tive anastomotic procedures, in 224 (63.6%) procedures related to the colon. GI resection and anastomosis were performed in 41.8% of patients without preoperative mechanical bowel prep- aration. Serum albumin level was determined in 70.5% of pa- tients, out of which 15.3% had low serum albumin levels. Evi- dences of infection (pus and GI content in the peritoneum, ab-

scess or fistula) were noted during anastomosis in 6.8% of patients. Blood transfusion was given for 6.8% of patients. The type of operative procedures in decreasing order of frequency were colocolic anastomosis (33.5%), enteroenterostomy (18.2%), ileocolic anastomosis (15.6%), colorectal anastomo- sis (13.9%), esophagectomy (12.2%), gastrectomy (5.4%), and jejunotransverse (and coloanal anastomoses (0. 3% each).

The participating surgical disciplines included general, colo- rectal, and cardiothoracic surgery. Majority of the operations (70.8%) were performed by residents and general surgeons. Colorectal and cardiothoracic surgeons operated on 15.9% and 13.4% of patients respectively.

43

The surgical procedure lasted 90 minutes or more in 94.6%, 2 to 3 hours in 61.4%, more than 3 hours in 9.7%, and more than 5 hours in 0.6% of patients (Table 2).

As depicted in Table 3, anastomotic leak was detected in 35 of 352 patients (9.9%) in this study. AL detection day varied from the 2nd to the 14th postoperative days; and majority of the leaks (21, 60%) were detected on the 5th to 8th postoperative days, and (10, 28.6%) on the 7th postoperative day. Majority of leaks (21, 60%) were detected by GI content or fluid dis- charge via incisional wounds or drainage tubes. Relaparotomy was performed on 8.8% of patients after GI resection and anastomosis, but 71.4% (25/35) of patients that developed anastomotic leak required relaparotomy, and the rest (10/35) were managed conservatively as enterocutaneous fistula whereas only 1.9% (6/311) of patients who did not develop leak required relaparotomy, for wound dehiscence and post- operative collection.

About 60.0% of patients stayed in hospital for one week or more, 54 (15.4%) for more than two weeks, and 34 (9.7%) stayed for more than three weeks. The mean postoperative hospital stay was 12±12.35 days, with a median stay of 8 days and IQR of 7 to 11 days. When conditions of patients on discharge were as- sessed, 25 patients (7.1%) died in hospital after GI anastomosis (Table 3).

On bivariate analysis (Table 4), emergency procedure, presence of infection, absence of mechanical bowel preparation, and low serum albumin level were significantly associated with AL. AL was significantly high with increase in age and prolonged dura- tion of surgery. Sex, duration of procedure, low preoperative HCT level, and pre or intraoperative blood transfusion was not significantly associated with the development of AL.

AL rate following enteroenterostomy (Table 5) was high (17.2%) followed by esophagectomy (16.3%), gastrectomy (10.5%), colorectal anastomosis (10.2%), ileocolic anastomo- sis (9.1%) and colocolic anastomosis (4.2%), but ileorectal anastomosis did not leak (P=0.019). Majority of the proce- dures (240 patients) were performed by residents and general surgeons. There were differences in the leak rates among the different surgical professionals, but the differences were not statistically significant (P=0.434).

Anastomotic leakage rate was found to be a little bit higher in malignant conditions than benign conditions. (13/126, 10.3% vs 22/226, 9.73%), but the difference was not statistically sig- nificant (P=0.357). AL was associated with significantly in- creased mortality, morbidity, and prolonged hospital stay (Table 6). The rate of death in patients who developed AL

was 48.3% compared to the 3.5% death rate in those who did not develop AL. Most patients (71.4%) with AL had relaparotomy compared to the 1.9% relaparotomy rate in those without AL. Majority of patients (62.9%) with AL stayed more than 3 weeks in hospital, whereas only 3.8% of patients without AL stayed that long.

Variables which were statistically significant on bivariate analysis were included in multivariate analysis (Table 6) to see their independent effect on the occurrence of AL. Absence of bowel preparation was strongly associated with AL on bi- variate analysis but became out of the range for significance on multivariate analysis. The variable that had strong inde- pendent association with AL was a low serum albumin level (p<0.0001).

Patients who had low serum albumin were 19 times more likely to develop AL compared to those who had normal serum albumin. The other variable which was independently associated with oc- currence of AL was emergency procedures (p=0.018) where pa- tients were 4.6 times more likely to develop AL than those who underwent elective procedures.

DISCUSSION

Anastomotic leak is perhaps the most dreaded complication following intestinal surgery and is one of the leading causes of postoperative morbidity and mortality despite improvements in surgical care. The rates and complications of AL vary consider- ably depending on the definition, risk factors, site, and type of GI tract resection (1,2).

Majority of studies used a combination of clinical features and radiological investigations to define and detect anastomotic leak. The diagnostic methods commonly used when a leakage is suspected are CT scan, contrast enema, endoscopic examina- tion, and reoperation (8). In our series, except one patient who developed wound dehiscence and the leak was detected on re- operation, all ALs were detected and defined clinically only. There is no universally accepted definition of anastomotic leak at any site (3, 9).

In this study, the rate of AL increased with increase in age of patients and pronged duration of surgery, but the increase was not significant and congruent to the findings by others (4, 7, 10, and 11). In our study, female patients developed AL more fre- quently than males (12.2% vs. 9.1%) even though the variation was not statistically significant (P=0.370). In other studies, AL

occurred more commonly in male patients (12, 13). The varia- tion could be due to small sample size of female patients in our study.

Studies have identified risk factors for GI anastomotic leakage, but there is no universal agreement on which risk factors con- sistently feature (1). Amrika Seshadri (7) reported that serum albumin, need for blood transfusion and others as strongly as- sociated factors for AL. In a retrospective study, male sex, perioperative transfusion, presence of cardiovascular disease and proximal tumor location were predictive factors of anasto- motic leakage after gastrectomy for gastric cancer and the leak- age rate was 1.9% (13). In a prospective study, Nair et al (14) reported a 35.0% rate of anastomotic disruption in patients undergoing emergency small bowel anastomosis, a much high- er rate than our finding. There are clearly many patient and disease factors that contribute to anastomotic leak. AL has been associated with a 6% to 39% mortality rate (15). In this study, emergency procedure, the presence of infection, the ab- sence of bowel preparation, and low serum albumin were sig- nificantly associated with anastomotic leak, but on multivariate analysis only emergency procedures (P=0.018) and low serum albumin(P<0.0001) remained significantly associated with anastomotic leakage. This is because of the fact that patients were operated on without bowel preparation on emergency bases which possibly is a confounding variable. This finding is in line with the findings in several other studies (7, 14, and 16).

DISCUSSION

Anastomotic leak is perhaps the most dreaded complication following intestinal surgery and is one of the leading causes of postoperative morbidity and mortality despite improve- ments in surgical care. The rates and complications of AL vary considerably depending on the definition, risk factors, site, and type of GI tract resection (1,2).

Majority of studies used a combination of clinical features and radiological investigations to define and detect anasto- motic leak. The diagnostic methods commonly used when a leakage is suspected are CT scan, contrast enema, endoscopic examination, and reoperation (8). In our series, except one patient who developed wound dehiscence and the leak was detected on reoperation, all ALs were detected and defined clinically only. There is no universally accepted definition of anastomotic leak at any site (3, 9).

In this study, the rate of AL increased with increase in age of patients and pronged duration of surgery, but the increase was not significant and congruent to the findings by others (4, 7, 10, and 11). In our study, female patients developed AL more frequently than males (12.2% vs. 9.1%) even though the variation was not statistically significant (P=0.370). In other studies, AL occurred more commonly in male patients (12, 13). The variation could be due to small sample size of fe- male patients in our study.

Studies have identified risk factors for GI anastomotic leak- age, but there is no universal agreement on which risk factors consistently feature (1). Amrika Seshadri(7) reported that serum albumin, need for blood transfusion and others as strongly associated factors for AL. In a retrospective study, male sex, perioperative transfusion, presence of cardiovascu- lar disease and proximal tumor location were predictive fac- tors of anastomotic leakage after gastrectomy for gastric can- cer and the leakage rate was 1.9% (13). In a prospective study, Nair et al (14) reported a 35.0% rate of anastomotic disruption in patients undergoing emergency small bowel anastomosis, a much higher rate than our finding. There are clearly many patient and disease factors that contribute to anastomotic leak. AL has been associated with a 6% to 39% mortality rate (15). In this study, emergency procedure, the presence of infection, the absence of bowel preparation, and low serum albumin were significantly associated with anasto- motic leak, but on multivariate analysis only emergency pro- cedures (P=0.018) and low serum albumin (P<0.0001) re- mained significantly associated with anastomotic leak- age.This is because of the fact that patients were operated on without bowel preparation on emergency bases which possi- bly is a confounding variable. This finding is in line with the findings in several other studies (7, 14, and 16).

In general, the rate of AL varies depending on the tissue anas- tomosed, and the rates include, stomach, 1.1-3.3%, small in- testine, 1.0-3.8%, ileocolic, 2.0-6.5%, colocolic, 3.0-5.4%,

48

colorectal, 4.0-26%, and ileorectal, 5.0-19%. AL is known to be correlated with worse prognosis after curative resection for colorectal cancer (15). In the present study, the rates of AL were 16.3% in esophagogastrostomy, 10.5% in gastrectomy and gastrojejunostomy, 17.2% in enteroenterostomy, 9.1% in ileocolic anastomosis, 4.2% in colocolic anastomosis, and 10.2% in colorectal anastomosis. Compared to the preceding report, AL rates in gastric, small bowel and ileocolic anasto- mosis were higher in our study probably due to nutritional deficiencies or emergency procedures. Contrary to our finding where ileorectal anastomosis did not leak, higher incidences of AL, probably related to disease or patient factors, were found in other reports (1,15). The overall AL rate of 9.9% observed in this study is similar to the reported rates that ranged from 1.8% to 15.4% in many studies (2, 10, 16-19). Esophagogas- trostomy, ileocolic, colocolic and colorectal AL rates were comparable with the findings in other series (2, 4, 10, 11, 15, 16, 20, 21). In our series, anastomotic leakage rate was slightly higher in malignant than benign conditions (10.3 % vs 9.73%), but the difference was not statistically significant (P=0.357).

AL presents in a dramatic fashion early or more often in a far subtler fashion, often relatively late in the postoperative period (22). In majority of our patients (60.0%), in agreement with another report (6), AL was detected between the 5th and 8th postoperative days, and on the 14th postoperative day in one patient.

The consequences of AL are peritonitis, fistula or abscess for- mation, postoperative infection, and increased hospital costs and mortality (20, 23, 24). It causes considerable morbidity and mortality to the patient, and it doubles the length of hospi- tal stay (7).

In this study, 40.0% of patients developed peritonitis, and the death rate in patients who developed AL was 48.3% compared to the 3.5% in those who did not develop AL. Previous reports showed a 3% to 39% and 8 to 10-fold mortality rates after AL (2, 4, 7, 20). In our series, the mortality rate was higher than the above reported rates, which may be explained by lack of early suspicion, late clinical detection and or delayed interven- tion since this study could not find usage of diagnostic imag- ing techniques for the detection of leakage.

In our study, the risk of relaparotomy was high in patients with AL; most patients (71.4%) with AL had relaparotomy com- pared to the 1.9% relaparotomy rate in those without AL. Compared to a 3.8% of patients without AL, 62.9% of patients with AL stayed more than 3 weeks in hospital. AL is associat- ed with significantly prolonged hospital stay as well as consid- erable extra costs (4,7).

Conclusion and recommendation:

The incidence of AL in gastrointestinal anastomosis in our series was 9.9%. Low serum albumin and emergency proce- dures are strong risk factors for AL. The occurrence of AL significantly increases the rate of relaparotomy, sepsis, postop- erative mortality and duration of hospital stay.

Awareness of the risk factors, suspicion of AL, and thorough evaluation of patients may impact perioperative decision- making, surgical technique and patient care.

Optimization of nutrition prior to elective GI anastomosis may prevent AL and its far-reaching consequences.

Emergency GI resection and anastomosis should be handled with utmost care. Prospective studies should be conducted to identify determinant factors for AL. We also recommend the four hospitals to have better patient record keeping.

49

ACKNOWLEDGMENT

We would like to thank the administrations of Tikur Anbessa, Menelik II, ZM, and Yekatit 12 Hospitals for cooperation dur- ing data collection. We also thank the Addis Ababa University College of Health Sciences for funding the study.