Anastomotic insufficiency leads to clinical strains for patients, and significantly increases morbidity and mortality. On average, hospital stay is extended by 12 days while healthcare-related expenses are increased by 30,000 USD when patients suffer from an anastomotic leak. In experienced centers, the approximated incidence of anastomotic insufficiency is 3,3% for colon and 8.6% for colorectal procedures. Multiple subgroups of patients with increased risk for anastomotic leaks have been described in previous publications. Meticulous preoperative recognition of patients with increased risk for anastomotic insufficiency is clinically beneficial, as it would permit improved ressource preparation, enhanced patient education and superior surgical decision-making. However, it is often difficult for clinicians to balance the plethora of crucial risk factors for anastomotic leaks for a single patient. Machine learning methods have been exceptionally effective at incorporating various clinical variables into one unified risk prediction model. To the authors' best knowledge, there does not yet exist a credible prediction model or a conclusive prediction score for anastomotic insufficiency after colon and colorectal anastomosis. The aim of the Prediction of Anastomotic Insufficiency risk after Colorectal surgery (PANIC) study is to establish and externally validate an efficient machine-learning-based prediction tool based on multicenter data from a range of international centers.
