65岁及以上重症下呼吸道感染患者28 d和90 d预后预测模型的构建  

Establishing prediction models of 28-day and 90-day mortality in elderly patients with severe lower respiratory tract infections

作　　者：齐天琪 李曼玉 杨靖娴[1] 吴晓瑞[2] Qi Tianqi;Li Manyu;Yang Jingxian;Wu Xiaorui(Department of Clinical Laboratory,Aerospace Central Hospital,Beijing 100049,China;Department of Respiratory and Critical Care Medicine,Aerospace Central Hospital,Beijing 100049,China)

机构地区：[1]航天中心医院检验科,北京100049 [2]航天中心医院呼吸与危重症医学科,北京100049

出　　处：《国际呼吸杂志》2023年第12期1450-1457,共8页International Journal of Respiration

基　　金：北京市卫生健康委员会重点专科医学检验科培育类项目(京卫医[2021]135号)。

摘　　要：目的基于重症下呼吸道感染(LRTIs)患者的临床资料、早期免疫功能参数和动态衍生炎症指标构建入院28 d和90 d预后模型、绘制列线图并评价模型。方法本研究为队列研究。采用非随机抽样法,纳入2020年9月至2023年3月航天中心医院收治的年龄≥65岁的101例重症LRTIs患者,随访患者入院28 d和90 d生存情况,分为28 d死亡组23例和存活组78例,90 d死亡组40例和存活组61例。收集患者的临床资料,检测入院24 h、第3天和第7天的血常规、血生化,入院24 h的淋巴细胞亚群和免疫球蛋白水平。利用单因素分析和多因素logistic回归分析筛选预后危险因素构建模型,Bootstrap法进行内部验证并绘制列线图,以ROC曲线下面积(AUC)、校准曲线和决策曲线分析(DCA)评价模型的区分度、一致性和临床有效性。结果使用血管活性药物、C反应蛋白/白蛋白(入院第7天)升高、IgG和IgM升高是28 d预后的危险因素(均P<0.05)。28 d模型AUC值0.88,内部验证AUC值0.85;当预测概率为0.1~0.5时,预测值略高于实测值;DCA显示临床有效性高。使用血管活性药物、并发多器官功能障碍综合征、CD8^(+)细胞和CD16^(+)CD56^(+)细胞计数降低是90 d预后的危险因素(均P<0.01)。90 d模型AUC值0.96,内部验证AUC值0.94;预测值与实测值一致性好;DCA显示临床有效性高。结论构建的65岁及以上重症LRTIs患者入院28 d和90 d预后模型区分度、一致性好,临床有效性高。Objective To establish and evaluate prediction models of 28-day and 90-day mortality in elderly patients with severe lower respiratory tract infections(LRTIs)by integrating clinical data,immune parameters,and derived inflammatory markers.Methods This was a cohort study.A total of 101 elderly hospitalized patients diagnosed with severe LRTIs in Aerospace Central Hospital from September 2020 to March 2023 were collected as subjects by nonrandom sampling and divided into survival group and death group according to their 28-day and 90-day clinical outcomes.The factors affecting the prognosis of severe LRTIs patients were analyzed by multivariate logistic regression analysis,and nomogram prediction was constructed.The area under receiver operating characteristic curve(AUC),calibration plot,and decision curve analysis(DCA)were used to evaluate the two models.Results Logistic regression analysis showed that using of vasoactive agents,increased level of CRP/ALB,IgG,and IgM were independent prognostic factors of severe LRTIs patients in 28-day prediction model.The AUC of train group and validation group were 0.88 and 0.85 respectively.The predicted value was slightly higher than the true value when the predicted probability was within the range of 0.1 to 0.5.The DCA showed good clinical net benefit.Using of vasoactive agents,complication of multiple organ dysfunction syndrome,and decreased counts of CD8^(+)lymphocytes and CD16^(+)CD56^(+)cells were independent prognostic factors in 90-day model.The AUC of train group and validation group were 0.98 and 0.94.The predicted value was approximately equal to true value.The DCA also showed good clinical net benefit.Conclusions The 28-day and 90-day prediction models showed good discrimination,consistency,and clinical validity to predict the death of elderly patients with severe LRTIs.

关 键 词：重症下呼吸道感染 列线图 免疫功能参数 衍生炎症指标 预测模型 

分 类 号：R56[医药卫生—呼吸系统]