糖皮质激素联合免疫抑制剂致系统性红斑狼疮患者感染预测模型  

Predictive model of infection in patients with systemic lupus erythematosus induced by glucocorticoids combined with immunosuppression

作　　者：单慧亭 孟岩[3] 陈迹[2] 潘慧敏[2] 杨建华 SHAN Huiting;MENG Yan;CHEN Ji;PAN Huimin;YANG Jianhua(College of Pharmacy,Xinjiang Medical University,Xinjiang Urumqi 830011,China;Pharmaceutical Department,The First Affiliated Hospital of Xinjiang Medical University,Xinjiang Urumqi 830011,China;Department of Rheumatology and Immunology,The First Affiliated Hospital of Xinjiang Medical University,Xinjiang Urumqi 830011,China)

机构地区：[1]新疆医科大学药学院,新疆乌鲁木齐830011 [2]新疆医科大学第一附属医院,药学部,新疆乌鲁木齐830011 [3]新疆医科大学第一附属医院,风湿免疫科,新疆乌鲁木齐830011

出　　处：《中国医院药学杂志》2023年第14期1588-1592,共5页Chinese Journal of Hospital Pharmacy

基　　金：新疆维吾尔自治区自然科学基金青年科学基金项目(编号:2019D01C327)。

摘　　要：目的:分析系统性红斑狼疮(systemic lupus erythematosus,SLE)患者联合使用糖皮质激素与免疫抑制剂发生感染的危险因素并建立风险预测模型。方法:回顾性分析联合使用糖皮质激素与免疫抑制剂的348例SLE住院患者临床资料,其中感染组124例、非感染组224例,并对2组患者进行倾向性评分匹配,使组间混杂因素均衡。匹配后运用单因素分析和多因素logistic回归模型分析SLE并发感染的危险因素,并建立发生感染的风险预测模型,采用受试者工作特征(receiver operating characteristic,ROC)曲线评价预测模型,使用70例SLE患者数据进行模型验证。结果:倾向性评分匹配后,感染组与非感染组分别为123例,糖皮质激素剂量(>7.5 mg)、发热、淋巴细胞计数减少(<1×10^(9)L^(-1))和血小板减少(<100×10^(9)L^(-1))、低蛋白血症(<30 g·L^(-1))是SLE患者感染的独立危险因素。建立logistic回归模型logit(P)=0.8X_(1)+2.384X2+1.131X_(3)+2.059X_(4)+0.848X_(5)-1.127,构建各危险因素的ROC曲线,模型预测概率的曲线下面积约0.872,其在ROC曲线对应的最佳截断值为0.5064。使用符合纳入标准的70例患者资料进行模型外部验证,结果显示预测模型的准确率为77.1%。结论:该研究建立的SLE患者感染风险预测模型具有较高的准确率,有助于早期识别SLE患者的感染风险,对临床制订综合治疗策略有重要意义。OBJECTIVE To develop a predictive infection risk model for patients with systemic lupus erythematosus(SLE)treated with combined glucocorticoids and immunosuppressive drugs.METHODS A retrospective analysis was conducted on the clinical data of 348 SLE inpatients who were treated with a combination of glucocorticoids and immunosuppressants,including 124 cases in the infected group and 224 cases in the non infected group.The propensity score matching was performed on the two groups of patients to balance the confounding factors between the groups..Univariate analysis and multifactor logistic regression models were used to explore risk factors for co-infection in SLE and to develop a predictive model for the risk of infection.The receiver operating characteristic(ROC)curves were used to evaluate the model’s accuracy.Finally,the model was validated using independent data from 70 SLE patients.RESULTS After using propensity score matching,123 cases were assigned to each infected and non-infected group.Glucocorticoid dose(>7.5 mg),fever,reduced lymphocyte count(<1×10^(9)L^(-1)),thrombocytopenia(<100×10^(9)L^(-1)),and hypoproteinemia(<30 g·L^(-1))were identified as independent infection risk factors in patients with SLE.The logistic regression model was fit as logit(P)=0.8X_(1)+2.384X_(2)+1.131X_(3)+2.059X_(4)+0.848X5-1.127.External validation of the independent dataset showed that the model’s prediction accuracy was 77.1%.CONCLUSION The infection risk prediction model established in this study has a high accuracy rate,which helps to identify the infection risk of SLE patients at an early stage and is essential for the clinical formulation of comprehensive treatment strategies.

关 键 词：糖皮质激素 免疫抑制剂 系统性红斑狼疮 感染 预测模型 

分 类 号：R969[医药卫生—药理学]