Effective index of mass screening for diabetes mellitus.

虽然空腹血糖（FBG）被广泛用于筛查，但它对发现餐后高血糖患者并不敏感。本研究以糖尿病的另一项筛查试验糖蛋白A_<1c> (HbA_<1c>)为研究对象，采用75g口服糖耐量试验（OGTT）比较两种筛查方法对餐后高血糖的准确性，并探讨HbA_<1c>的理想临界值。已有研究报道，HbA_<1c>的6.5%临界值与FBS=7.0 mmol/L的临界值一样准确。但是，截断点太高，因此，非常具体但不敏感。本研究通过降低HbA_<1c>的临界值，观察发现新发糖尿病患者的诊断准确性和归因的变化。受试者包括231名男性和323名女性，他们接受了所有三项测试，即FBG、HbA_<1c>和75g OGTT。从ROC曲线分析，与FBG不同，HbA_<1c>的准确度不会随着截断点的降低而降低。当截断点为6.5%时，受试者的真阳性和真阴性比例分别为2.4%和1.1%。在研究人群中，19%的空腹血糖检测不到的餐后高血糖可通过HbA_<1c>检测到。当临界值降至5.6%时，受试者的真阳性和真阴性比例分别为2.4%和1.1%，80%的餐后高血糖可通过HbA_<1c>发现。敏感性为0.91，特异性为0.82，阳性预测值为0.50。如果截断点降低到5.6%以上，则正预测值迅速降低，意味着成本效益降低。从本研究来看，HbA1c对截断点的变化具有鲁棒性，因此降低截断点并不会导致准确性降低。5.6%的临界值似乎是发现更多餐后高血糖患者的合理选择。

Although fasting blood glucose (FBG) is well-used in mass screening, it is not sensitive to find postprandial hyperglycemia patients. Focusing on another screening test for diabetes mellitus, Glycohemoglobin A_<1c> (HbA_<1c>), this study compared the accuracy of both screening tests for postprandial hyperglycemia using 75g oral glucose tolerance test (OGTT), and discussed the ideal cut off point of HbA_<1c>. Previous study reported that the cut off point of 6.5% for HbA_<1c> is as accurate as that of FBS=7.0 mmol/L. However, the cut off point is too high and therefore, very specific but not sensitive. In this study the authors lowered the cut off point of HbA_<1c> and observed the change of accuracy of diagnosis and attribution to finding new patients of diabetes mellitus. The subjects were 231 males and 323 females who underwent all three tests, i.e., FBG, HbA_<1c>, and 75g OGTT.From the analysis of the ROC curve, the accuracy of HbA_<1c> is not lowered as lowering the cut off point, unlike FBG. When the cut off point was 6.5%, the true positive and negative proportions of the subjects were 2.4% and 1.1%, respectively. In the study population, 19% of postprandial hyperglycemia who could not be found by FBG was detectable by HbA_<1c>. When the cut off point was lowered to 5.6%, the true positive and negative proportions of the subjects were 2.4% and 1.1%, respectively, and 80% of postprandial hyperglycemia could be found by HbA_<1c>. At the point, the sensitivity, the specificity, and positive predictive value were 0.91, 0.82 and 0.50, respectively. If the cut off point was lowered beyond 5.6%, positive predictive value was rapidly lowered, meaning lowering cost performance.From the study, HbA1c was robust against the change of cut off point, and therefore, lowering the cut off point did not cause of lower accuracy. The cut off point of 5.6% seemed to be reasonable to find more postprandial hyperglycemia patients.