A prediction model for type 2 diabetes risk among Chinese people
A prediction model for type 2 diabetes risk among Chinese people
复制标题
DOI：
10.1007/s00125-008-1232-4
复制
发表时间：
2009-03-01
期刊：
影响因子：
8.2
通讯作者：
Hu, F. B.
中科院分区：
文献类型：
--
作者：
Chien, K.;Cai, T.;Hu, F. B.
已经提出了一系列预测2型糖尿病发病的规则。然而，大多数研究都是在白人群体中进行的，目前还不清楚这些模型是否适用于亚洲人群。本研究的目的是构建一个简单的预测中国人糖尿病发病的点数模型。我们在一项队列研究中估计了在基线水平没有糖尿病的中老年参与者患糖尿病的10年风险。使用COX回归系数构建简单的点模型，并利用AUC和净重分类改进以及综合判别改进统计量来确定所得到的预测规则的判别能力。使用五次随机分裂来测试AUC的内部有效性并获得AUC的自举估计。在基线检查时没有糖尿病的2960名参与者中，548人在中位数10年的随访期内患上了2型糖尿病。在多因素模型中，年龄(4分)、空腹血糖升高(11分)、体重指数(8分)、三酰甘油(5分)、白细胞计数(4分)和较高的高密度脂蛋白胆固醇(负4分)对糖尿病发病率有很强的预测作用。该模型的估计AUC值为0.702(95%可信区间为0.676-0.727)。该模型的预测效果优于在其他人群中开发的预测模型，包括预期的心血管疾病蒙斯特、剑桥、圣安东尼亚和弗雷明翰糖尿病风险模型。我们构建了一个预测中国人中糖尿病10年发病率的模型，该模型可用于识别中国人群中糖尿病的高危个体。
A range of prediction rules for the onset of type 2 diabetes have been proposed. However, most studies have been conducted in white groups and it is not clear whether these models apply to Asian populations. The purpose of this study was to construct a simple points model for predicting incident diabetes among Chinese people.We estimated the 10 year risk of diabetes in a cohort study of middle-aged and elderly participants who were free from diabetes at baseline. Cox regression coefficients were used to construct the simple points model and the discriminatory ability of the resulting prediction rule was determined using AUC and net reclassification improvement and integrated discrimination improvement statistics. Fivefold random splitting was used to test the internal validity and obtain bootstrap estimates of the AUC.Of the 2,960 participants without diabetes at the baseline examination, 548 developed type 2 diabetes during a median 10 year follow-up period. Age (four points), elevated fasting glucose (11 points), body mass index (eight points), triacylglycerol (five points), white blood cell count (four points) and a higher HDL-cholesterol (negative four points) were found to strongly predict diabetes incidence in a multivariate model. The estimated AUC for the model was 0.702 (95% CI 0.676-0.727). This model performed better than existing prediction models developed in other populations, including the Prospective Cardiovascular Munster, Cambridge, San Antonia and Framingham models for diabetes risk.We have constructed a model for predicting the 10 year incidence of diabetes in Chinese people that could be useful for identifying individuals at high risk of diabetes in the Chinese population.