Establishment of risk warning model for surgical site infection

Abstract:

ObjectiveTo establish a risk warning model for surgical site infection（SSI）, provide support for screening high risk population and finding suspected cases. MethodsClinical data of 5 067 patients who underwent abdominal surgery in 6 domestic hospitals from January 2013 to December 2015 were collected retrospectively, all cases were randomly divided into modeling group and validation group according to a 6:4 ratio,  warning model was established by employing logistic regression, the area under the receiver operating characteristic curve (AUC) was used to evaluate discriminant ability of evaluation model, the maximum Youden index was as the optimum cutoff point. ResultsFor the warning model of highrisk patients, AUC was 0.823, sensitivity and specificity were 78.81% and 74.33% respectively, positive predictive value and negative predictive value were 19.67% and 97.78% respectively. For the discriminant model of suspected infection cases, AUC was 0.978, sensitivity and specificity were 93.38% and 95.62% respectively, positive predictive value and negative predictive value were 62.95% and 99.45% respectively.ConclusionThe earlywarning model established in this study has better discrimination ability, which can provide a reference for the development of early warning and discrimination of healthcareassociated infection information system.