脑出血后早期血肿扩大预测评分的建立及其效果的初步评价  

Establishment of prediction scale for early hematoma expansion after intracerebral hemorrhage and its preliminary effectiveness evaluation

作　　者：孔祥宇 董军[1] 钱志远[1] Kong Xiangyu;Dong Jun;Qian Zhiyuan(Department of Neurosurgery, Second Affiliated Hospital of Soochow University, Suzhou 215004, China)

机构地区：[1]苏州大学附属第二医院神经外科,苏州215004

出　　处：《中华神经医学杂志》2019年第5期487-493,共7页Chinese Journal of Neuromedicine

基　　金：江苏省研究生实践创新计划(SJCX18_0855);苏州市科技发展计划(SYSD2018102).

摘　　要：目的建立脑出血后早期血肿扩大预测评分并初步评价其效果。方法回顾性收集苏州大学附属第二医院神经外科自2016年1月至2018年5月收治的317例脑出血患者的临床资料,采用多元Logistic回归分析筛选脑出血后早期血肿扩大的危险因素,取其中P<0.05的独立危险因素建立预测评分,以偏回归系数(β值)为权重进行赋分,并简化取整。应用受试者工作特征(ROC)曲线评价预测评分的鉴别能力,应用Hosmer-Lemeshow拟合优度检验及校准曲线评价其准确性,应用决策曲线分析法评价其临床实用性,最后将建立的预测评分应用于本组患者进行验证。结果多元Logistic回归分析显示抗凝药物史、超早期血肿扩大率≥2.7 mL/h、格拉斯哥昏迷评分(GCS)≤8分、非增强CT征象(岛征、黑洞征、混杂征、液平)出现一种或多种为脑出血后早期血肿扩大的独立危险因素(P<0.05),其中GCS评分≤8分赋值分值为2.0分,超早期血肿扩大率≥2.7 mL/h赋值分值为2.5分,非增强CT征象(岛征、黑洞征、混杂征、液平)出现一种或多种赋值分值为2.0分,抗凝药物史赋值分值为4.5分,预测评分总分为0~11分。预测评分的ROC曲线下面积为0.854(95%CI:0.803~0.904,P=0.000),校准度高(χ^2=3.323,P=0.344),决策曲线分析的净获益率高,可选阈概率范围大。预测评分应用于本组患者显示高危组(预测评分≥4.5分)预测早期血肿扩大的敏感度为0.77,特异度为0.85,准确度为0.83。结论本研究建立的脑出血后早期血肿扩大预测评分可以为临床上精准识别脑出血后早期血肿扩大高危个体、指导临床诊疗及临床试验等提供一些参考和帮助。Objective To establish a prediction scale of early hematoma expansion (HE) after intracerebral hemorrhage (ICH) and evaluate its prediction effectiveness. Methods A retrospective analysis of clinical features of 317 ICH patients, admitted to our hospital from January 2016 to May 2018, was performed. Risk factors for early HE after ICH were obtained by multivariate Logistic regression analysis, and independent risk factors with P<0.05 were used to establish the prediction scale;the assigned scores for each item were derived by parameter estimates (β coefficients) and increased proportionately to the nearest integer. Receiver operating characteristic (ROC) curve was used to evaluate the discriminating ability of the prediction scale, Hosmer-Lemeshow goodness of fit test and calibration curve were used to evaluate its accuracy, and the decision curve analysis was used to evaluate its clinical practicability. Finally, the established prediction scale was applied to this group of patients for verification. Results History of anticoagulants, ultra-early hematoma growth (uHG)≥2.7 mL/h, Glasgow coma scale (GCS) scores≤8, and non-enhanced CT (NCCT) signs (island sign, black hole sign, blend sign, niveau formation) existing one or several kinds were independent risk factors for early HE after ICH (P<0.05);the assigned values of uHG≥2.7 mL/h, GCS scores≤8, NCCT signs (island sign, black hole sign, blend sign, niveau formation) existing one or several kinds, and history of anticoagulants were 2.5, 2.0, 2.0, and 4.5, respectively;the total prediction scale scores were 0-11. The area under ROC curve was 0.854 (95%CI: 0.803-0.904, P=0.000), and the calibration was high (χ^2=3.323, P=0.344);decision curve analysis showed high net benefit and wide range of optional threshold probability of the scale. After validating in the development cohort, the sensitivity, specificity and accuracy of the high-risk group (prediction scale≥4.5) were 0.77, 0.85 and 0.83, respectively. Conclusion The prediction scale for early HE afte

关 键 词：脑出血 早期血肿扩大 预测评分 非增强CT征象 

分 类 号：R743.34[医药卫生—神经病学与精神病学]