Anastomotic leakage in colorectal cancer surgery in Syria: risk factors and improvements following guideline implementation – a retrospective cohort study

Background: Anastomotic leakage (AL) is a serious complication following colorectal cancer surgery, associated with increased morbidity, mortality, and impaired oncologic outcomes. Evidence from resource-limited, conflict-affected settings such as Syria is scarce. This study aimed to identify risk factors for AL and evaluate outcomes after implementation of revised institutional preoperative guidelines. Materials and methods: A retrospective cohort study was conducted at Damascus Hospital, Syria. The first cohort (January 2016–February 2024) included patients undergoing elective sigmoid colon or rectal resection. Based on the findings, preoperative preparation guidelines were updated, and a follow-up cohort (April 2024–October 2025) was evaluated. AL was defined according to the International Study Group of Rectal Cancer criteria. Patient-, tumor-, and procedure-related factors were analyzed, and outcomes before and after guideline implementation were compared. Results: The first cohort ( n = 100) had an AL incidence of 15% and postoperative mortality of 19%. Significant risk factors included hypoalbuminemia, anemia, perioperative blood transfusion, prolonged corticosteroid use, extra-peritoneal anastomosis, and poor tumor differentiation. Following guideline implementation ( n = 19), hypoalbuminemia decreased from 61.3% to 10.5%, perioperative transfusions from 60% to 42.1%, AL incidence from 15% to 10.5%, and mortality from 19% to 15.8%. Comparison with international studies highlighted the high baseline prevalence of hypoalbuminemia and AL, emphasizing the importance of nutritional optimization. Conclusion: AL is strongly associated with modifiable factors, particularly hypoalbuminemia and anemia. Evidence-based guideline implementation targeting these factors significantly improved postoperative outcomes, demonstrating the value of context-specific interventions in resource-limited settings.