| Introduction. Anastomotic leak (AL) is one of the most devastating complications of colon and rectal surgery with an incidence of up to 21% in rectal anastomoses. Despite many studies on risk factors, there is no agreement on a validated predictive model for AL. The purpose of this study was to identify patient, clinical and surgical factors that may predispose to AL and to develop a predictive model.;Methods. This was a prospective observational study in a cohort of 616 patients undergoing colorectal resection in a single institution. The main outcome was AL within 30 days postoperatively. Variables collected were: age, gender, body mass index (BNI), life style (smoking history and alcohol consumption), Charlson Comorbidity Index (CCI), diabetes, anemia, radiation and chemotherapy, immunomodulator medications, hemoglobin, albumin, preoperative diagnoses, surgical procedure(s), American Society of Anesthesiologists (ASA) class, bowel preparation, surgical technique (laparoscopic vs. open), anastomotic technique (staple vs. hand sewn), anastomotic fashion, type of anastomosis (colon or rectal), number of major arteries ligated at surgery, surgeon's experience, presence of infectious complication(s) at surgery, and intraoperative adverse events. Outcome variables were: postsurgical complications, gastrointestinal function restoration, pain management, and readmission and reoperation rates. Data were prospectively collected prior to, during, and after surgery up to 30 days. Functional status was measured by self report with the SF-36 General Health Survey. Path analysis and binary logistic regression were done to assess the association between the variables and AL. Bootstrapping was performed to test the predictive model's stability.;Results. Of the 616 patients, the median age was 63 years, 53.4% were female, and the mean BMI was 25.9 kg/m2. Indications for surgery were neoplasm (55%), inflammatory bowel disease (16%), diverticulitis (14%), and other (14%), and 80.3% had laparoscopic and 19.5% open surgery. AL occurred in 6% (37) patients, major complications occurred in 21.9% (135) and the mortality rate was 0.9% (6). Comparison between patients with Al vs. no-AL was done. In multivariate logistic regression with the dependent variable being the occurrence of AL, significant independent predictors were: anastomoses <10 cm from the anal verge; CCI ≥ 3; number of major arteries ligated; intraoperative complications; presence of an infectious condition at surgery; surgeon's experience and male gender. Three predictive models for AL (preoperative, intraoperative, and scoring) demonstrated stability within this sample with bootstrap estimates for the area under the curve (AUC) and 95% Confidence Intervals [CI] of 0.84 [.841, .845]; 0.82 [.753, .885]; and 0.78 [.776, .781], respectively. The AL indices for each model (preoperative, intraoperative, and scoring) showed good bootstrap estimates as well with AUC and 95% CI of 0.81 [.813, .817]; 0.81 [.809, .814]; and 0.78 [.776, .781] respectively. Predictors of creating a diverting stoma in multivariate analysis were: neoadjuvant therapy; type of anastomosis; and anastomoses <10cm from the anal verge. Case match analysis for AL confirmed the overall findings of the study. None of the SF-36 General Health Survey domain scores were associated with AL.;Conclusion. In addition to the well known risk factors of anastomoses <10 cm from anal verge and male gender, in this study we found that CCI ≥ 3, number of major arteries ligated, intraoperative complications, and presence of an infectious condition at surgery also were risk factors for AL and they should be considered prior to and during the surgical care of colorectal patients. This study cannot provide information regarding the protective role of creating a stoma; therefore surgeons should continue to use their experienced discretion as to whether or not to create a diverting stoma. Finally, predictive models such as those demonstrated in the study should be considered for use in practice in an attempt to decrease AL rate and improve outcomes after surgery. |
