Development and Validation of a Novel Prognostic Scoring Model for Ischemic Colitis

PURPOSE: This study was conducted to identify prognostic factors affecting the course of ischemic colitis and to develop a prognostic scoring model. METHODS: We analyzed medical records of consecutive patients with ischemic colitis treated between October 2002 and September 2008 at Severance Hospital, Seoul, Republic of Korea. Patients were excluded if results of endoscopy were unavailable. Patients were classified as having severe ischemic colitis on the basis of outcome (improvement delayed for more than 2 weeks, complications requiring surgery, or death). Univariate analyses and multivariate logistic regression analyses with backward stepwise selection were used to identify clinical, endoscopic, and laboratory variables associated with severe ischemic colitis. A novel prognostic scoring model was derived from the data, with probability of severe ischemic colitis and risk index determined for 8 risk groups based on independent risk factors identified by multivariate analyses. Predictive power was tested by means of 10-fold cross-validation, with area under the receiver operating characteristic curve representing discrimination accuracy. RESULTS: Analyzable data were available for 153 of 173 consecutive patients. Ischemic colitis was classified as severe in 20 patients. Multivariate analyses showed the following significant independent predictors of severe ischemic colitis: tachycardia (adjusted odds ratio = 4.6; 95% CI, 1.4–14.7), shock within 24 hours after admission (adjusted odds ratio = 6.5; 95% CI, 2.0–21.2), and endoscopic evidence of ulceration (adjusted odds ratio = 9.9; 95% CI, 2.0–48.8). Probability of severe ischemic colitis and risk index were 74 times higher for patients with all 3 risk factors (group 8) than for patients with none (group 1). Internal validation showed the area under the receiver operating characteristic curve to be 0.91 (95% confidence interval, 0.86–0.96). CONCLUSIONS: Endoscopic findings and instability of vital signs were associated with the disease course of ischemic colitis. A novel scoring model based on presence of tachycardia, shock within 24 hours after admission, and endoscopic evidence of ulceration provides a method of assessing patient prognosis and should be further validated.