Early warning model to detect anastomotic leakage following colon surgery: a clinical observational study

PURPOSE: We aimed to develop a predictive tool for anastomotic leakage (AL) following colon cancer surgery by combining a clinical early warning score (EWS) with the C-reactive protein (CRP) level. METHODS: The records of 1,855 patients who underwent colon cancer surgery at the Oxford University Hospitals NHS Foundation Trust between January 2013 and December 2018, with or without AL, were retrospectively reviewed. EWS and CRP levels were assessed daily from the first postoperative day until discharge. AL was defined as an anastomotic defect observed at reoperation, the presence of feculent fluid in a pelvic drain, or evidence of AL on computed tomography. The tool incorporated postoperative EWS and CRP levels for the accurate early detection of AL. RESULTS: From postoperative days 3 to 7, the mean CRP level exceeded 200 mg/L in patients with AL and was under 200 mg/L in those without AL (P<0.05). From postoperative days 1 to 5, the mean EWS among patients with leakage exceeded 2, while scores were below 2 among those without leakage (P<0.05). Receiver operating characteristic curve analysis identified postoperative day 3 as the most predictive of early leakage, with cutoff values of 2.4 for EWS and 180 mg/L for CRP; this yielded an area under the curve of 0.87 (sensitivity, 90%; specificity, 70%). CONCLUSION: We propose using an EWS of 2.4 and a CRP level of 180 mg/L on postoperative day 3 following colon surgery with anastomosis as threshold values to prompt investigation and treatment of AL.