SIRI与基底节区脑出血早期神经功能恶化的关系及Nomogram预测模型建立  

Relationship between SIRI and early neurological deterioration in basal ganglia cerebral hemorrhage and construction of Nomogram predictive model

作　　者：费小斌[1] 周新民[1] 薛鑫诚[1] 龚凯[1] 高恒[1] Fei Xiaobin;Zhou Xinmin;Xue Xincheng;Gong Kai;Gao Heng(Department of Neurosurgery,the Affiliated Jiangyin Hospital of Southeast University Medical College,Jiangyin 214400,China)

机构地区：[1]东南大学医学院附属江阴医院神经外科,214400

出　　处：《国际外科学杂志》2020年第10期679-683,F0003,共6页International Journal of Surgery

摘　　要：目的探讨全身炎症反应指数(SIRI)对基底节区脑出血(BGH)早期神经功能恶化(END)的预测价值,并建立END发生的Nomogram预测模型。方法采用回顾性队列研究方法。收集2016年1月—2018年12月东南大学医学院附属江阴医院收治的146例BGH患者资料,根据是否发生END将病例分为END组(n=34)和非END组(n=112)。符合正态分布的计量资料以均数±标准差(Mean±SD)表示,两组比较采用t检验;不符合正态分布的计量资料以M(P25,P75)表示,组间比较采用Mann-Whitney U非参数检验。计数资料以例数和百分数(%)表示,组间比较采用χ2检验或Fisher确切概率法。通过单因素、多因素Logistic分析SIRI与END发生的关系,并筛选出其他影响END的独立危险因素。同时应用Nomogram法对各个因素进行评分,构建预测模型。最后绘制受试者工作特征曲线(ROC),评价SIRI及模型对END发生的预测价值。结果单因素分析发现,END发生与血肿体积、不规则血肿、血肿破入脑室、血糖、淋巴细胞及SIRI水平密切相关(P<0.05);多因素Logistic回归分析显示,血肿体积(P<0.001)、血肿破入脑室(P=0.012)、SIRI水平(P=0.023)是END发生的独立危险因素;ROC确定SIRI=5.40×109/L为诊断END发生的最佳截点。基于上述分析结果建立Nomogram预测模型,结果显示结合SIRI和其他标准变量的模型(AUC=0.869,95%CI:0.804~0.935,P<0.001)比未结合SIRI的模型(AUC=0.811,95%CI:0.734~0.889,P<0.001)和仅基于SIRI的模型(AUC=0.716,95%CI:0.622~0.810,P<0.001)对END发生具有更佳的预测价值。结论SIRI为END发生的独立危险因素,临床上可结合其共建Nomogram预测模型,以提高对END发生的早期识别和筛选能力。Objective To investigate the association between systemic inflammation response index(SIRI)and early neurological deterioration(END)in patients with basal ganglia hemorrhage(BGH),and then set up a prediction Nomogram model for END.Methods The retrospective cohort study was conducted.A total of 146 patients with BGH from January 2016 to December 2018 were chosen in the Affiliated Jiangyin Hospital of Southeast University Medical College.The patients were divided into the END group(n=34)and non-END group(n=112),according to whether END occurred or not.The normally distributed data were presented as the mean±standard deviation(Mean±SD),and the groups were compared using the t test.The non-normally distributed data were expressed as M(P25,P75),and this data was analysed via the Kruskal-Wallis test.Categorical variables were described as numbers of patients(%)and compared using chi-square analysis or Fisher exact test,as appropriate.Univariate analysis and multivariate logistic regression analysis were used to identify the risk factors of END occurrence,and the relationship with SIRI.Then,each factor was scored by Nomogram method to construct the prediction model.Receiver operating characteristic curve(ROC)was drawn to assess the predictive value of SIRI and Nomogram model in the occurrence of END.Results Univariate analysis showed that the occurrence of END was associated with hematoma volume,presence of intraventricular hemorrhage,blood glucose,lymphocyte count and SIRI(P<0.05).Multivariate logistic regression analysis showed that hematoma volume(P<0.001),presence of intraventricular hemorrhage(P=0.012)and SIRI(P=0.023)are independent risk factors for END occurrence.ROC curve analysis showed that SIRI has certain predictive value for END occurrence,and the optimal cut-off value was SIRI=5.40×109/L.Then these risk factors were incorporated into the Nomogram.Statistically analysis showed the model had a good predictive value,and the model combining the SIRI and other prognostic factors(AUC=0.869,95%CI:0.804-0.935,P

关 键 词：脑出血 危险因素 全身炎症反应指数 基底节脑出血 早期神经功能恶化 Nomogram模型 

分 类 号：R743.34[医药卫生—神经病学与精神病学]