中文摘要：

目的 比较血管事件总体健康风险（THRIVE）,多中心卒中调查（MSS）,休斯敦动脉内治疗（HIAT）及葡萄糖水平、种族、年龄、性别、收缩压水平、卒中程度（GRASPS）评分4种风险模型预测急性前循环缺血性卒中机械取栓后脑出血与不良结局的效能.方法 前瞻性连续性纳入2013年5月至2016年3月收住南京大学附属金陵医院和厦门大学附属中山医院神经内科发病6 h内的急性前循环大血管闭塞并实施机械取栓的患者153例.采用Logistic回归分析和受试者工作特征（ROC）曲线下面积（AUC）探讨THRIVE、MSS、HIAT、GRASPS评分4种风险模型预测急性前循环缺血性卒中机械取栓后,脑出血与不良结局[包括90 d全因死亡和90 d改良Rankin量表（mRS）评分≥3分]的效能.结果 MSS评分（AUC为0.639,95%CI：0.548~0.730,P=0.004）和GRASPS评分（AUC为0.616,95%CI：0.525~0.706,P=0.017）能够预测脑出血事件,但预测准确性较低;4种模型对机械取栓后90 d内死亡均有预测价值,其中GRASPS评分（AUC为0.783,95%CI：0.706~0.860,P〈0.001）有中等预测准确性,其余3种模型AUC〈0.7,预测准确性低;4种模型均能对90 d不良预后（90 d mRS≥3分）进行预测,其中GRASPS评分及THRIVE评分AUC均〉0.7,GRASPS评分AUC最大（AUC为0.782,95%CI：0.708~0.855,P〈0.01）,两者有中度预测准确性.结论 GRASPS评分对机械取栓后90 d内全因死亡及不良预后均有较好的临床预测价值,THRIVE评分对90 d不良预后有较好的临床预测价值.4种模型对机械取栓后脑出血事件的预测价值仍有待验证.

英文摘要：

Objective To compare the efficacies of 4 risk models （THRIVE[Totaled Health Risks in Vascular Events],MSS[Multicenter Stroke Survey],HIAT[Houston Intra-Arterial Therapy],and GRASPS[Glucose at presentation,Race,Age,Sex,Systolic blood pressure,Severity of stroke at presentation]） in predicting intracranial hemorrhage and poor outcomes in acute anterior circulation ischemic stroke after mechanical thrombectomy.Methods From May 2013 to March 2016,153 consecutive patients with acute anterior circulation vascular occlusion conducted mechanical thrombectomy within 6 hours after onset and admitted to the Departments of Neurology,Jinling Hospital,Nanjing University School of Medicine and Zhongshan Hospital,Xiamen University were enrolled prospectively.Logistic regression analysis and the area under the receiver operating characteristic （ROC） curve （AUC） were used to investigate the efficacies of 4 risk models （the THRIVE,MSS,HIAT,and GRASPS scores） for predicting intracerebral hemorrhage （including any intracranial hemorrhage events and symptomatic intracranial hemorrhage） and poor outcomes （including 90 d all-cause death and modified Rankin Scale[mRS] score≥3） in acute anterior circulation ischemic stroke after mechanical thrombectomy.Results The MSS score （AUC 0.639,95%CI 0.548-0.730,P=0.004） and GRASPS score （AUC 0.616,95%CI 0.525-0.706,P=0.017） could predict any intracranial hemorrhage events,but the predictive accuracy was low.They had the predictive value for death within 90 d after mechanical thrombectomy,and the GRASPS score （AUC 0.783,95%CI 0.706-0.860,P〈0.001） had the moderate predictive accuracy,and the AUC of the other 3 models was all〈0.7,the predictive accuracy was low.The models could predict the poor prognosis at 90 d （90 d mRS≥3）.The AUCs of both the GRASPS score and THRIVE score were 〉0.7.The AUC of GRASPS score was the largest （AUC 0.782,95%CI 0.708-0.885,P〈0.01）.Both had moderate predictive accuracy.Conclusion The GRASPS score had a better clinica