新量表ESSENI在预测脑梗死复发方面的探索性研究

目的用量表预测脑梗死复发是一直以来临床医师关注的课题,ESSEN评分是国际公认的预测量表,但其中未包括影像学等重要因素,预测效度有限。方法本研究选择北京大学人民医院2005至2007年住院的非心源性脑梗死患者,应用电话随访结合电子病例查询调查方法,确定为单发组或复发组,并对其进行各个临床和影像特点的电子表格录入。应用Cox regression筛选脑梗死5年复发的独立危险因素,并根据B值对其进行赋分;结合ESSEN评分制订新的评分量表ESSENI,应用ROC分析比较两种评分的曲线下面积(AUC);应用生存分析计算5年复发率。结果 Cox regression得出,糖尿病、MRI上脑白质病、脑梗死灶多于1个和不同系统梗死灶是脑梗死复发的独立危险因素;根据B值对其分别赋分1、2、6和1.5分,建立ESSENI评分系统,即在ESSEN的9分基础上加上3项影像学因素,共18.5分。并得出≥10分为高危人群,其5年复发率为(68±0.05)%;应用ROC分析,其ESSENI的AUC为0.789;>ESSEN的AUC为0.65。结论新的评分系统ESSENI结合影像学和临床特点,能够更好地提高其长期的脑梗死复发预测度。

Research of predictive power of ESSENI in long term recurrent stroke

Objective Prediction of recurrent stroke is a problem for clinic. ESSEN is a widely used score for prediction of recurrent stroke, but without image factors, it has limited power for prediction. Methods Between 2005 and 2007, we included patients with acute ischemic stroke except cardiogenic cerebral embolism from the Department of Neurology of Peoples＇ Hospital of Peking University, and followed up the patients by telephone and Electronic Medical Record to determine whether they belong to recurrent group or not. We recorded the clinical and image variables. We determined the independent risk factors for recurrent stroke in 5 years by Cox regression, and give each factor score according to B. Then we built up a new ESSENI score, and compare the predictive accuracy by calculating the area under the roc curve. We calculated the 5 year recurrent rate by Kaplan-Meier analysis. Results By Cox regression, DM, leukoencephalopathy in MRI, multiple infarcts, infarcts in different circulations were independent risk factors for recurrent stroke and we give 1, 2, 6, 1.5 respectively according to B score.We built ESSENI score, adding 3 factors to 9 points of ESSEN, with totally 18.5 points. We concluded that patients of more than 10 points are at high risk with the recurrent rate of（68±0.05）% in 5 years. The AUC of ESSENI was 0.789, more than the AUC of ESSEN which was 0.65. Conclusion ESSENI has more predictive power of long term recurrent stroke with not only clinical factors but also image factors.