Analysis On Risk Factors And Prediction Model Construction Of Primary Sjogren ’s Syndrome Complicated With Infection

Objecctive:To analyze the clinical characteristics and risk factors of primary Sj(?)gren’s syndrome(pSS)with infection,and to construct a visual prediction model.Methods:The clinical data of pSS patients who met the inclusion and exclusion criteria in the Department of Rheumatology and Immunology of Affiliated Hospital of Yangzhou University from October 2018 to December 2022 were retrospectively collected and divided into infection group(n=64)and non-infection group(n=170).And the distribution of infection sites and pathogens in the infected group were analyzed.Analyze these data between the two groups in the following ways:baseline data,laboratory indicators,disease assessment index,and medication history.The independent t-test was performed for normally distributed continuous variables,the nonparametric test was performed for skewed continuous variables and rank data,the chi-square test was performed for count data.The independent risk factors were screened out by univariate and multivariate logistic regression analysis.Then the corresponding nomogram prediction model was constructived.The predictive performance of the model was evaluated by ROC curve.Result:1.A total of 234 pSS patients were included in the study,of which 64 cases combined with infections,with an incidence of 27.3%.The respiratory system is the most common site,accounting for 53.1%,with pulmonary infections in 29 cases(45.3%)being the most common site;followed by urinary system(31.2%)and skin soft tissue infections(7.8%).2.In 64 pSS patients with infection,56 strains of various types of pathogens were detected,including 43 strains of bacteria(76.8%),5 strains of viral infection(8.9%),4strains of Mycoplasma pneumoniae(7.1%);2 strains of Candida albicans(3.6%);and 2cases of Mycobacterium tuberculosis(3.6%).Gram-negative bacteria were the most common bacteria,with 27 strains(62.8%),of which Escherichia coli and Klebsiella pneumoniae were the most common.Pseudomonas aeruginosa,Acinetobacter baumannii and Proteus were also mainly detected in the respiratory system and urinary system.Gram-positive bacilli were also common,16 strains,accounting for 28.6%.Staphylococcus aureus and Streptococcus pneumoniae are more common in skin and soft tissue infections and respiratory infections.3.Univariate analysis suggested that age,combined chronic disease,white blood cell count,neutrophil count,hemoglobin,NLR(Neutrophil Lymphocyte Ratio),albumin,25(OH)D,RF,D-dimer,CRP,ESR,glycoconjugate,CD4~+T cell and corticosteroid were statistically different between the two groups.After excluding secondary indicators such as white blood cell count,neutrophil count,ESR and CRP,the remaining indicators were subjected to multivariate logistic regression analysis and covariance test.The results suggested that age,lung involvement,history of glucocorticoid use and 25(OH)D were independent risk factors for pSS with infection.4.The prediction model constructed based on multivariate logistic regression analysis,as follows:the probability of pSS with infection was:P=1/1+exp(-Z);Z=3.662+0.026*age+pulmanaryinvolvement*1.45+glucocorticoid*1.38-0.55*25(OH)-D.Then it was found that the area under the curve of the model at the best cut-off value was 0.775(95%CI:0.679-0.831,P<0.001)by plotting the ROC curve,which suggested the model had a better predictive performance.Conslusion:1.Infections were relatively common in patients with pSS.And the respiratory system,urinary system and skin soft tissues were the most common sites of infection;Gram-negative bacteria were the common infecting agents.2.After univariate and multivariate logistic regression analysis,age,lung involvement,history of glucocorticoid use and 25(OH)D were identified as independent risk factors for pSS with infection;the nomogram model based on the above risk factors performed better in predicting the occurrence of infection and could help clinicians identify patients with pSS at high risk of infection.