卒中危险因素联合脑血流动力学指标预测卒中模型的建立

目的建立卒中危险因素联合脑血流动力学指标综合评估卒中风险的预测模型。方法根据10 565人前瞻性队列人群基线调查资料和7年随访卒中发病资料,分析各危险因素及脑血流动力学积分与卒中发病风险的关系,用Cox比例风险模型,按性别建立传统危险因素联合脑血流动力学积分综合的预测模型。采用另一独立样本队列人群（回代队列）资料,进行回代验证和计算受试者工作曲线（ROC）下面积以检验预测模型的效能。结果①建模队列人群中男女随访分别为28 341.61人年和39 544.12人年,男女卒中累计发病率分别为3.28‰和2.58‰。②卒中预测模型的方程,男性为Y=0.159×心血管病史＋0.375×糖尿病史＋0.634×卒中家族史＋0.609×高血压病史＋0.037×体质量指数＋0.068×年龄-0.022×血流动力学积分值＋0.186×吸烟＋0.408×饮酒。女性为Y=0.096×心血管病史＋0.161×糖尿病史＋0.392×卒中家族史＋0.139×高血压病史＋0.034×体质量指数＋0.057×年龄-0.021×血流动力学积分值。年龄、血流动力学积分是男女卒中的独立预测因素,高血压病史是男性卒中的独立预测因素。③将预测模型方程回代到建模队列,男、女ROC下面积分别为0.839、0.820;回代到回代队列,男、女ROC下面积分别为0.795、0.787。结论传统危险因素联合血流动力学积分综合预测卒中风险具有较好的效能,脑血流动力学积分是卒中较强的预测因子。

Establishment of a stroke predicting model with risk factors for stroke and cerebral hemodynamic indicators

Objective To establish risk factors for stroke in combination with cerebral hemodynamic indicators for comprehensive assessment of a predicting model of stroke risk.Methods The relationship between all the risk factors and cerebral hemodynamic scores and stroke risks was analyzed according to the prospective cohort baseline survey data in 10 565 subjects and the 7-year follow-up data after stroke onset.A comprehensive prediction model of traditional risk factors in combination with the cerebral hemodynamic scores was established using the Cox proportional hazards model.Using another independent sample cohort data,the back substitution test was conducted and the areas under receiver operating characteristic(ROC) curves were used to detect the performance of the prediction model.Results ① 28 341.61 person-year in male and 39 544.12 person-year in female were followed up in the modeling cohort.The cumulative incidences of stroke in males and females were 3.28 ‰ and 2.58 ‰ respectively.②The equation of stroke predictive model for males was Y=0.159 history of CVD+0.375 history of diabetes + 0.634 family history of stroke + 0.609 history of hypertension + 0.037 body mass index(BMI) + 0.068 age-0.022 hemodynamic score + 0.186 cigarette smoking + 0.408 alcohol consumption;the equation of stroke predictive model for females was Y=0.096 history of CVD + 0.161 history of diabetes + 0.392 family history of stroke + 0.139 history of hypertension +0.034 BMI + 0.057 age-0.021 of hemodynamic score were independent stroke risk factors both for male and female and history of hypertension was an independent risk factors for male.③The area under the ROC curve for males and females was 0.839 and 0.820 with the back substitution test of modeling population cohort and it was 0.795 and 0.787 with the back substitution test of the independent population cohort.Conclusion Comprehensive prediction of stroke risk with traditional risk factors in combination with hemodynamic score has a better performance.Cerebral hemodynamic score is a stronger predictor of stroke.