Objective To explore the influencing factors of respiratory bacterial infection in patients with primary biliary cholangitis(PBC),and to establish and evaluate a nomogram prediction model.Methods Ac-cording to the presence or absence of respiratory bacterial infection,a total of 269 PBC patients hospitalized in the Department of Hepatology,the Third People's Hospital of Kunming from September 2012 to September 2022 were divided into bacterial infection group and non-bacterial infection group,and the general data,co-morbidities and laboratory tests of the patients were collected for retrospective analysis.The variables were screened by univariate analysis,and the influencing factors were analyzed by multivariate LR partial likelihood estimation method Logistic regression,based on which a nomogram prediction model was constructed.The re-ceiver operator characteristic curve(ROC)and calibration curve were used for model evaluation.Results The prevalence of respiratory bacterial infection was 14.5％.Univariate and multivariate LR partial likelihood esti-mation Logistic regression analysis showed that anti-gp210 antibody positive(OR=2.598,95％CI:1.193～5.657,P=0.016),anti-Ro-52 antibody positive(OR=2.860,95％CI:1.321～6.193,P=0.016),and anti-Ro-52 antibody positive(OR=2.860,95％CI:1.321～6.193,P=0.008)and high neutrophil count(OR=1.494,95％CI:1.224～1.824,P＜0.001)were independent risk factors for respiratory bacterial infec-tion in PBC patients.High hemoglobin level(OR=0.982,95％CI:0.969～0.996,P=0.009)was an indepen-dent protective factor.Hosmer-lemeshow test results(x2=3.718,P=0.882)showed that the nomogram pre-diction model constructed by the above factors had a good degree of fit.the receiver operator characteristic curve(ROC)showed that the area under the curve(AUC)0.769(95％CI:0.682～0.857,P＜0.001).The sensitiv-ity was 64.1％and the specificity was 80.9％.Bootstrap method was used for internal validation with 1000 re-peated samples.The results showed that the mean absolute error was 0.015,indicating that the accuracy of the fitting model was acceptable,and there was a good agreement between the actual value and the predicted value of respiratory bacterial infection.Conclusion The nomogram prediction model constructed in this study has good discrimination ability and accuracy,which can provide a reference for clinicians to preliminarily judge the risk of respiratory bacterial infection in patients with PBC,which is helpful to achieve early intervention and treatment and improve the prognosis of patients.
维普
万方数据
