OBJECTIVE To investigate the clinical characteristics and risk factors for the postoperative abdominal infection in the patients with colorectal cancer and establish the risk prediction model.

METHODS Totally 179 colorectal cancer patients who received surgical procedures in Dongying District People′s Hospital from Jan 2020 to Dec 2022 were enrolled in the study and were divided into the infection group with 43 cases and the no infection group with 136 cases according to the status of postoperative abdominal infection. The clinical data were retrospectively collected from the patients, multivariate logistic regression analysis was performed for the risk factors for the postoperative abdominal infection, the risk prediction model was established, and the predictive value of the model was evaluated.

RESULTS The result of logistic regression analysis showed that the postoperative abdominal infection was associated with the complication with diabetes mellitus, hypoproteinemia, postoperative stoma and drainage tube indwelling time no less than 10 days(P < 0.05). The established prediction model was as follows: Logit(P)=-2.410+complication with diabetes mellitus×1.031+hypoproteinemia×0.919+postoperative stoma×0.987+drainage tube indwelling time×1.000; Hosmer-Lemeshow test of goodness of fit showed that the model had favorable fitting effect(Chi-Square=3.412, P=0.578), the result of calibration curve showed that the predicted probability approximated to the actual probability, indicating that the model had favorable capability of differentiation, calibration and prediction. Receiver operating characteristic (ROC) curve analysis showed that the area under curve (AUC) of the model was 0.828 in prediction of postoperative abdominal infection in the colorectal cancer patients.

CONCLUSION The postoperative abdominal infection in the colorectal cancer patients is associated with the complication with diabetes mellitus, hypoproteinemia, postoperative stoma and drainage tube indwelling time. The risk prediction model has high predictive value.