Social & Scientific Systems, Inc., Silver Spring, MD.
Westat, Rockville, MD.
Ann Epidemiol. 2014 Feb;24(2):83-9. doi: 10.1016/j.annepidem.2013.10.008. Epub 2013 Oct 18.
Although mean concentrations of hemoglobin A1c (A1C), fasting plasma glucose, and 2-hour plasma glucose differ by demographics, it is unclear what other characteristics of the distributions may differ, such as the amount of asymmetry of the distribution (skewness) and shift left or right compared with another distribution (shift).
Using kernel density estimation, we created smoothed plots of the distributions of fasting plasma glucose (N = 7250), 2-hour plasma glucose (N = 5851), and A1C (N = 16,209) by age, race-ethnicity, and sex in the 2005-2010 National Health and Nutrition Examination Survey, a nationally representative sample of U.S. adults including people with and without diabetes. We tested differences in distributions using cumulative logistic regression.
The distributions were generally unimodal and right-skewed. All distributions were shifted higher and more right-skewed for older age groups (P < .001 for each marker). Compared with non-Hispanic whites, the distribution of fasting plasma glucose was shifted higher for Mexican-Americans (P = .01), whereas the distribution of A1C was shifted higher for non-Hispanic blacks (P < .001). The distribution of fasting plasma glucose was shifted higher for men (P < .001) and the distribution of 2-hour plasma glucose was shifted higher for women (P = .01).
We provide a graphic reference for comparing these distributions and diabetes cut-points by demographic factors.
尽管血红蛋白 A1c(A1C)、空腹血浆葡萄糖和 2 小时血浆葡萄糖的平均值因人口统计学特征而异,但分布的其他特征(如分布的不对称程度[偏度]和与另一个分布相比的左移或右移[移位])是否不同尚不清楚。
使用核密度估计,我们根据年龄、种族和性别,对 2005-2010 年全国健康和营养检查调查(National Health and Nutrition Examination Survey,NHANES)中 7250 名空腹血浆葡萄糖(N=7250)、5851 名 2 小时血浆葡萄糖(N=5851)和 16209 名 A1C 的分布进行了平滑绘图,该调查是美国成年人的全国代表性样本,包括有和没有糖尿病的人。我们使用累积逻辑回归检验了分布的差异。
分布通常为单峰且右偏。对于年龄较大的年龄组,所有分布均升高且右偏(每个标志物的 P<0.001)。与非西班牙裔白人相比,墨西哥裔美国人的空腹血浆葡萄糖分布升高(P=0.01),而非西班牙裔黑人的 A1C 分布升高(P<0.001)。空腹血浆葡萄糖的分布向更高处移动,男性的 2 小时血浆葡萄糖分布向更高处移动(P<0.001)。
我们提供了一个图形参考,用于根据人口统计学因素比较这些分布和糖尿病切点。