Risk Factors And Prediction Model For Inpatient Surgical Site Infection After Abdominal Surgery

Objective:To identify risk factors and develop surgical site infections prediction model on inpatients undergoing abdominal surgery.New measures are proposed for targeted interventions in clinical work and to reduce the use of healthcare resources and to improve the access of patients as well.Methods:A retrospective analysis of the inpatient records undergoing abdominal surgery in Shaanxi Provincial People’s Hospital from January 1st,2018 to January 1st,2021.The cases were collected in strict inclusion and exclusion criteria,the abdominal SSI prediction model was established accordingly.According to clinical diagnosis for surgical site infection,patients undergoing abdominal surgery were divided into infected and uninfected groups.Inpatient SSI risk factors were evaluated using univariate analysis and multivariate logistic regression,with P≤0.05 indicating a statistically significant difference to determine independent risk factors for abdominal SSI in patients.These factors were incorporated into the predictive model for abdominal SSI using R 4.2.1,the predictive efficacy was assessed by plotting model operating characteristic curves(ROC)and deriving AUC values.The Bootstrap self-sampling method was used to internally validate the prediction model with a frequency setting of 1000 to produce a C-index value to assess the model’s discrimination.The degree of calibration of the model was assessed by Hosmer-Lemeshow goodness-of-fit test and a graphical calibration method to visualize it.Finally,a web-based system of abdominal SSI prediction models was developed based on the patient’s abdominal SSI prediction model derived in this study.Results:A total of 2975 patients met the inclusion criteria.SSI occurred in 140 patients(4.71%)included in this study.Independent risk factors for the development of abdominal SSI in patients included:A history of diabetes(OR=6.308,95%CI:3.614～11.009,P<0.001),days of preoperative antibiotic use antibiotic use≥6(OR=4.229,95%CI:1.725～10.37,P=0.002),NRS 2002 score≥ 3(OR=2.083,95%CI:1.031～4.208,P=0.041),PCT≥ 0.05 μg/L(OR=1.693,95%CI:1.041～2.752,P=0.034),LDL≥3.37 mmol/L(OR=1.785,95%CI:1.073～2.969,P=0.026),intraoperative blood loss ≥ 200 mL(OR=32.655,95%CI:15.363～69.416,P<0.001),general anaesthesia(OR=3.578,95%CI:1.255～10.204,P=0.017),NNIS score of≥2(OR=2.226,95%CI:1.029～4.812,P=0.042),incision grade(OR=2.774,95%CI:1.372～5.608,P=0.005),surgical season(P≤0.05)and surgical site(P<0.05).The overall area under the ROC of the predictive model was 0.924,which is significantly higher than the NNIS score(AUC=0.661).The C-index value of the Bootstrap self-sampling method was 0.850(95%CI 0.845～0.853).which indicated a high degree of model discrimination.Both the model performance curve and the deviation correction curve in the graphical calibration method are close to the ideal model curve with an absolute error of 0.018 by using R 4.2.1.The Hosmer-Lemeshow goodness of fit test wasχ2=6.664,P=0.574 which indicated a high degree of model calibration.Conclusion:The analysis of this study showed the following independent risk factors revealed the abdominal SSI:a history of diabetes,days of preoperative antibiotic use antibiotic use≥ 6,NRS 2002 score of≥ 3,PCT≥ 0.05 μg/L,LDL≥3.37 mmol/L,intraoperative blood loss>200 mL,general anaesthesia,NNIS score of≥ 2,incision grade Ⅲ～Ⅳ,surgical season and surgical site.The prediction model is shown to be a highly accurate predictive model for preventing postoperative SSI in patients undergoing abdominal surgery.The model web system is also widely applicable and convenient and has high clinical application value.