Objective To investigate the risk factors for postoperative multi-drug resistant infection in patients with cerebral hemorrhage and establish a Nomogram prediction model.

Methods Clinical materials of 241 patients with surgery for cerebral hemorrhage in the hospital from July 2020 to July 2023 were collected, and they were divided into infection group and non-infection group. Logistic regression models were used to analyze independent influencing factors for the occurrence of postoperative multi-drug resistant infection in patients with cerebral hemorrhage, and a Nomogram prediction model was constructed accordingly. The predictive performance of the Nomogram was evaluated by the consistency index (C-index), the receiver operating characteristic (ROC) curve, and the calibration curve.

Results A total of 241 patients with cerebral hemorrhage were included in this study, among which 56 cases (24.24%) had postoperative multi-drug resistant infection. In the infection group, the preoperative Glasgow Coma Scale (GCS) score, ratio of preoperative vomiting, ratio of preoperative antibiotic treatment, ratio of gastric tube indwelling, ratio of tracheotomy, and ratio of intubation were significantly higher than those in the non-infection group (P < 0.05). Logistic regression analysis revealed that preoperative GCS score ≤8, preoperative vomiting, preoperative antibiotic treatment, gastric tube indwelling, tracheotomy and intubation were the independent risk factors for postoperative multi-drug resistant infection in patients with cerebral hemorrhage (OR > 1, P < 0.05). Values of area under thecurve (AUC) for preoperative GCS score, preoperative vomiting, preoperative antibiotic treatment, tracheotomy and intubation were all above 0.700, indicating these indicators have good predictive value for the occurrence of postoperative multi-drug resistant infection in such patients. Based on these influencing factors, a Nomogram risk model was established. The C-index value of the calibration curve was 0.798, suggesting the Nomogram model has good discriminatory power. The AUC values for the modeling group and validation group in the ROC curve were 0.798 and 0.722 respectively, indicating that the Nomogram model possesses satisfactory predictive efficacy.

Conclusion Nomogram prediction model constructed based on independent risk factors for postoperative multi-drug resistant infection in patients with cerebral hemorrhage can effectively predict the probability of such infections occurring in these patients after surgery.