Risk factor analysis and construction of prediction models for short-term postoperative complications in patients undergoing gastrointestinal tract surgery

PURPOSE: To identify risk factors associated with short-term postoperative complications in patients with gastrointestinal cancer and develop and validate prediction models to predict the probability of complications. METHODS: A total of 335 patients enrolled in the primary cohort of this study were divided into training and validation sets in a chronological order. Using univariate and multivariate logistic regression analyses, the risk factors for postoperative complications were determined, and nomogram prediction models were constructed. The performance of the nomogram was assessed with respect to the receiver operator characteristic and calibration curves. RESULTS: Patients with complications had a stronger postoperative stress response and a longer duration of daily fluid intake/output ratio >1 after surgery. Logistic analysis revealed that body mass index (BMI), body temperature on POD4 (T.POD4), neutrophil percentage on POD4 (N.POD4), fasting blood glucose on POD4 (FBG.POD4), and the presence of fluid intake/output ratio <1 within POD4 were risk factors for POD7 complications, and that BMI, T.POD7, N.POD7, FBG.POD4, FBG.POD7, and the duration of daily fluid intake/output ratio >1 were risk factors for POD30 complications. The areas under the curve of Nomogram-A for POD7 complications were 0.867 and 0.833 and those of Nomogram-B for POD30 complications were 0.920 and 0.918 in the primary and validation cohorts, respectively. The calibration curves showed good consistency in both cohorts. CONCLUSION: This study presented two nomogram models to predict short-term postoperative complications in patients with gastrointestinal cancer. The results could help clinicians identify patients at high risk of complications within POD7 or POD30.