基于糖尿病综合管理队列的卒中风险预测模型的建立与应用

Establishment and application of a stroke risk prediction model based on a comprehensive diabetes management cohort

目的分析糖尿病患者并发卒中的影响因素,构建卒中风险预测模型,以制定个性化干预措施。方法采用全人群队列设计的观察性研究方法,以17个乡镇1016个自然村年龄35周岁及以上的常住居民作为调查对象。构建糖尿病综合管理队列(糖尿病患者共15026例,其中男5512例,女9514例),采用COX回归建立卒中风险预测模型,受试者工作特征曲线下面积衡量模型的辨别能力。结果糖尿病队列中卒中的基线患病率为1.89%(男2.31%,女1.64%);综合干预后,观察期间新发卒中147例,发病密度为7.78‰(男7.71‰,女7.83‰)。最终年龄、体质量指数、吸烟、饮酒、服药依从性、血糖是否控制6个变量进入模型,模型受试者工作特征曲线下面积为0.748(95%CI0.609~0.763)。结论构建的卒中风险预测模型在糖尿病人群中预测能力较好,针对糖尿病患者应加强个体健康干预。

Objective To analyze the factors that influence stroke in diabetic patients,establish a risk prediction model for stroke,and develop personalized intervention measures.Methods An observational study with a whole-population cohort design was conducted in 17 townships and 1016 natural villages with permanent residents≥35 years of age as the subjects.COX regression was used to construct the diabetes management cohort,and the stroke risk prediction model was established.The area under the receiver operating characteristic curve was used to measure the discrimination ability of the model.Results The baseline prevalence of stroke in the diabetes cohort was 1.89%(males,2.31%;females,1.64%).After comprehensive intervention,147 patients had a stroke during the observation period,and the incidence was 7.78‰(males,7.71‰;females,7.83‰).Finally,age,body mass index,cigarette smoking,alcohol consumption,medication use,compliance,and glucose control were entered into the model.The area under the receiver operating characteristic curve was 0.748(95%CI 0.609-0.763).Conclusion The stroke risk prediction model and individual health intervention should be implemented for diabetic patients.