QS128 - Development of a Risk Scoring System for Predicting Anastomotic Leak Following Open and Laparoscopic Colorectal Surgery

Purpose/Background: Anastomotic leak (AL) is one of the most feared serious complications in colorectal surgery and correlates with a higher risk of mortality and morbidity. Several scoring systems have been used abroad, but these have applicability limitations. At present, there is no scoring system being used for predicting AL preoperatively for open and laparoscopic colon and rectal surgery. The aim of the study is to develop a risk scoring system that can predict AL occurrence after open and laparoscopic colorectal surgery anastomosis.

Methods/Interventions: The study employed a total enumeration technique and 252 patients with colorectal anastomosis were included in the study from January 1, 2020 to December 31, 2022 in a tertiary government hospital. Twenty-two patients (8.73 per cent) developed AL. Variables included were age, gender, American Society of Anesthesiology (ASA) physical status classification system score, Body Mass Index (BMI) , preoperative serum albumin level, comorbid conditions like hypertension (HTN) and diabetes mellitus (DM), Neutrophil-Lymphocyte Ratio (NLR), neoadjuvant therapy, urgency of surgery, indication for surgery, type of intestinal obstruction, operating surgeon, distance of anastomosis from the anal verge (FAV), additional surgical procedures, blood loss and duration of surgery. Multiple logistic regression analysis was used to determine which of the risk factors of AL were significant to be included in the risk scoring. The scoring system was created based on the log-odds ratios (B coefficient) from the multivariable analysis. A Receiver Operating Characteristic (ROC) curve was generated to determine the Area Under the Curve (AUC) of the score in predicting AL. Diagnostic value parameters were calculated for each cutoff score including, discriminative ability (based on the AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios. An optimal cutoff point was selected based on the Youden’s index. To estimate the internal validity of the risk scoring system the Bootstrap technique in trauma, malignant and non-malignant cases was performed.

Results/Outcomes: Only two variables remained statistically significant in developing the risk scoring system: 1) preoperative albumin level (g/dl): patients with preoperative albumin < 3.5g/dl had about 44 times higher odds of AL than those with albumin ≥3.5 g/dl; and 2) NLR: patients with preoperative NLR >4.65 had 360 times higher odds of AL than those with NLR ≤4.65. Points were created based on the B coefficient: 1) preoperative serum albumin level (g/dL): ≥3.5(0point); < 3.5(4points) and 2) NLR: ≤4.65(0point); >4.65(6points). Using a cutoff of ≥6, the scoring system had good diagnostic value parameters in predicting AL with accuracy=94.84%, sensitivity=95.45%, specificity=94.78%, PPV=63.64%, NPV=99.54%, LR+=188.18 and LR-=0.05. Internal validation was done using bootstrapping technique in trauma, malignant and non-malignant cases which showed excellent discriminative ability (AUC=0.96, 95% CI: 0.94-1.00).

Conclusion/Discussion: A rudimentary scoring system was developed using preoperative serum albumin level and NLR which showed significant accuracy in predicting AL occurrence after open and laparoscopic colorectal surgery. With further studies, this scoring system could perhaps be considered as a useful adjunct in the evaluation and treatment planning of patients for colorectal surgery in the local setting.