Analysis Of Risk Factors And Development And Validation Of Risk Prediction Models For SLE-associated Pneumonia And Pulmonary Infection

BackgroundPatients with systemic lupus erythematosus(SLE)have a high incidence of pneumonia,which has a great impact on prognosis.Its attributive diagnosis is very important,and it is mainly classified as SLE related and pulmonary infection.Clinically,there is no recognized diagnostic standard for SLE associated pneumonia,and it is often difficult to differentiate and diagnose SLE associated lung diffuse infection.Currently,the early diagnosis and treatment of SLE patients is a major challenge in clinical practice.Therefore,it will be of great significance to further study the clinical characteristics and risk factors of SLE associated lung damage and pulmonary infection,and to establish a predictable model.ObjectiveTo understand the clinical characteristics and risk factors of patients with SLE-associated pneumonia and pulmonary infection,and to establish a prediction model for the risk of SLE-associated pneumonia and pulmonary infection,so as to provide reference for the differentiation of SLE associated pneumonia and pulmonary infection treatment decisions.Methods1.Patients with systemic lupus erythematosus who were admitted into Rheumatology and Immunology Department and Nephrology Department of Guangdong Provincial People’s Hospital from January 2007 to September 2017 were continuously selected as modeling group.Their clinical data were collected retrospectively,and the clinical manifestations,laboratory examination and imaging examination results of SLE patients with pulmonary parenchyma lesion were recorded emphatically,the parameters or nonparametric test analysis comparison of SLE-associated pneumonia and pulmonary infection in patients with clinical features.2.The risk factors of pneumonia and pulmonary infection associated with SLE were screened by univariate correlation analysis.The prediction equations were constructed by multivariate logistic regression,and the receiver-operating characteristic curve(ROC)area under curve(AUC)in the assessment of the discriminant ability of the model.3.Patients from October 2017 to December 2019 were included in the verification group.The established risk prediction model was applied to the validation group,and the predictive power of the model in the validation group was evaluated by ROC-AUC.Results1.Clinical analysis of SLE-associated pneumonia and pulmonary infection.A total of 2305 SLE patients were included in the modeling group.Among them,651 patients had pulmonary parenchyma lesion,including 281 patients with secondary pulmonary parenchyma lesion of SLE and 397 patients with pulmonary infection.The clinical characteristics of the pneumonia group secondary to SLE and the pulmonary infection group were compared.There were statistically significant differences between the two groups in age,SLEDAI score,respiratory symptoms(including cough,shortness of breath,dyspnea,hemoptysis),arthritis,fever,length of stay,HGB,IgA,IgM,CRP and PCT(P<0.05).2.The risk prediction model of SLE-associated pneumonia and pulmonary infection was established.By multiple factors regression analysis,older age,fever,in the hospital for a long time,high SLEDAI score is SLE secondary lung parenchyma lesions are independent risk factors of the disease,in the hospital for a long time,raised the WBC,Increasing CRP,decreasing IgG,fever,no photosensitivity,no arthritis,older age for SLE complicated with lung infection independent risk factors of the disease.Furthermore,the risk prediction model was constructed by regression equation.In the modeling group,the area under the ROC curve of the SLE secondary pneumonia incidence prediction model was 0.725(0.702,0.748),and the corresponding predictive probability cut-off value was 0.167,with a sensitivity of 69.63%and specificity of 66.43%.The area under ROC curve of SLE complicated with pulmonary infection risk prediction model was 0.805(0.785,0.824)in the modeling group,the corresponding predictive probability cut-off value was 0.195,the sensitivity was 78.91%,and the specificity was 69.68%.3.Validation of risk prediction models for SLE-associated pneumonia and pulmonary infection.A total of 356 patients were included in the validation group.Taking the cut-off value of the two groups as the diagnostic cut-off point,the ROC-AUC of the group with pneumonia secondary to SLE was 0.700(0.643,0.753),with a sensitivity of 66.67%and specificity of 73.30%.The ROC of SLE complicated with pulmonary infection group was 0.742(0.688,0.791),the sensitivity was 74.67%,the specificity was 73.76%.ConclusionsThis study suggested that there were significant differences in clinical characteristics and risk factors between patients with SLE-associated pneumonia and lung infection.The two risk prediction models systems have good predictive efficacy for SLE-associated pneumonia and pulmonary infection.They benefit to improve the ability of clinicians to early diagnose and differentiate SLE associated pneumonia and pulmonary infection.