Abstract
To discuss the mechanisms of infection complications in different degrees after percutaneous nephrolithotomy (PCNL) through predicting and comparing post-PCNL infections based on nomograms, a retrospective cohort study was conducted among 969 cases who underwent PCNL from Dec 5, 2016 to Dec 25, 2017 in Kunming, Yunnan Province. We examined clinical features, urine routine, blood routine, blood biochemistry, imaging studies and operative information and recorded the examination results before surgery for univariate and multivariate logistic regression. We applied receiver operating characteristic curves, calibration curves, accuracy, specificity, sensitivity, positive predictive value and negative predictive value to evaluate and compare the models. Nomograms were used to visualize the different degrees of postoperative infection complications. The risk scores of the three groups were compared by diabetes mellitus distribution. Our results suggest that the more severe the infection is, the more accurate the model predicts and that the occurrence of severe infection mostly is related to the patients' homeostasis. Hence, we developed an online post-PCNL sepsis dynamic nomogram which can achieve visualization and dynamically predict the incidence of sepsis in postoperative patients.