列线图在人感染 H7 N9禽流感死亡风险预测中的应用

摘要

目的：应用列线图模型构建人感染H7 N9禽流感死亡风险预测的评估模型。方法采用分层随机抽样的方法选取浙江省人感染H7 N9禽流感定点治疗医院2013年3月至2015年3月收治的102例人感染H7N9禽流感确诊病例为研究对象，使用调查表对患者的流行病学资料、临床资料进行回顾性分析。采用单因素和多因素Logistic逐步回归分析方法寻找与H7N9禽流感死亡风险相关的独立危险因素，并使用列线图建立人感染H7 N9禽流感死亡风险的预测评估模型，采用一致系数（C-index）和受试者工作特征曲线（ROC）评价模型的预测精度。结果多因素Logistic逐步回归分析发现，年龄≥60岁（χ2＝3．98，OR＝2．99，95％CI：1．05～9．21）、中性粒细胞百分比高于正常（χ2＝6．66，OR＝5．06，95％ CI：1．56～18．83）、C-反应蛋白≥120 mg／L（χ2＝8．63，OR＝5．15，95％ CI：1．79～16．31）、手卫生差（χ2＝6．83，OR＝10．29，95％CI：2．18～81．49）和潜伏期≤5 d（χ2＝7．23， OR＝4．75，95％CI：1．59～15．80）为死亡的独立危险因素（P值均＜0．05）。在Logistic回归模型基础上绘制的列线图模型包含的这5个独立危险因子的影响等级和评分依次为，手卫生差：1级，100．0分；CRP≥120 mg／L：2级，76．5分；中性粒细胞百分比高于正常：3级，70．5分；潜伏期≤5 d：4级，62．0分；年龄≥60岁：5级，51．0分。列线图模型的C-index、ROC曲线下面积分别为0．833和0.817，模型拟合的效果较好。结论列线图模型能有效地进行人感染H7N9禽流感死亡风险的个体化预测和评估。%Objective To develop and validate a mortality risk prediction model for patients infected with avian influenza A H 7N9 virus.Methods A stratified and random sampling method was adopted for selection of subjects .A total of 102 patients infected with avian influenza A H7N9 virus, who were admitted to the designated hospitals in Zhejiang Province during March 2013 and March 2015, were enrolled.Standard questionnaires were used to collect data about demographic , epidemiologic and clinical characteristics , and the data were retrospectively reviewed . Univariate analysis and stepwise logistic regression analysis were used to identify the mortality risk factors of patients infected with avian influenza A H7N9 virus, and nomogram was applied to develop the risk prediction model .The accuracy of the prediction model was assessed using Concordance index (C-index) and receiver operating characteristic (ROC) curve. Results Stepwise multiple logistic regression analysis showed that age ≥60 years (χ2 =3.98, OR=2.99, 95%CI:1.05-9.21, P<0.05), increased initial neutrophil count (χ2 =6.66,OR=5.06, 95%CI:1.56-18.83, P<0.05), C-reactive protein≥120mg/L (χ2 =8.63, OR=5.15, 95%CI:1.79-16.31, P<0. 01), poor hand hygiene (χ2 =6.83, OR =10.29, 95%CI:2.18-81.49, P <0.01) and 5 days of incubation period or shorter (χ2 =7.23, OR=4.75, 95%CI:1.59-15.80, P<0.01) were independent risk factors for mortality of patients .Based on the above study , a risk prediction model of nomogram was developed.Poor hand hygiene (grade A, 100.0 points) ranked on the top of all risk factors, followed by C-reactive protein≥120 mg/L (grade B, 76.5 points), increased initial neutrophil count (grade C, 70.5 points), 5 days of incubation period or shorter (grade D, 62.0 points) and age ≥60 years (grade E, 51.0 points).The C-index and the area under the curve were 0.833 and 0.817 for the nomogram model , respectively;and the nomogram model fitted well .Conclusion Nomogram model can effectively predict and estimate the risk of death for patients infected with avian influenza A H 7N9 virus.