Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Room 501, No. 17, Xu-Zhou Road, Taipei, 100, Taiwan.
Division of Nephrology, Department of Internal Medicine, National Taiwan University Hospital and College of Medicine, No. 7, Chung-Shan South Road, Taipei, 100, Taiwan.
Sci Rep. 2021 Apr 12;11(1):7894. doi: 10.1038/s41598-021-86955-z.
Higher baseline glomerular filtration rate (GFR) may yield subsequent steeper GFR decline, especially in patients with diabetes mellitus (DM). However, this correlation in patients with chronic kidney disease (CKD) and the presence or absence of DM remains controversial. We conducted a longitudinal cohort study in a single medical center between 2011 and 2018. Participants with CKD stage 1 to 3A were enrolled and divided into DM groups and non-DM groups, and then followed up at least every 6 months. We used a linear mixed regression model with centering time variable to overcome the problem of mathematical coupling in the analysis of the relation between baseline GFR and the changes, and compared the results from correct and incorrect specifications of the mixed models. A total number of 1002 patients with 285 diabetic and 717 non-diabetic persons was identified. The linear mixed regression model revealed a significantly negative correlation between baseline GFR and subsequent GFR change rate in both diabetic group and non-diabetic group (r = - 0.44 [95% confidence interval [CI], - 0.69 to - 0.09]), but no statistical significance in non-diabetic group after within-subject mean centering of time variable (r = - 0.09 [95% CI, - 0.41 to 0.25]). Our study showed that higher baseline GFR was associated with a subsequent steeper GFR decline in the DM group but not in the non-DM group among patients with early-stage CKD. Exact model specifications should be described in detail to prevent from a spurious conclusion.
较高的基线肾小球滤过率（GFR）可能导致随后的 GFR 下降更为陡峭，尤其是在糖尿病患者中。然而，在慢性肾脏病（CKD）患者中，这种相关性以及是否存在糖尿病仍然存在争议。我们在 2011 年至 2018 年期间在一家医疗中心进行了一项纵向队列研究。纳入 CKD 1 至 3A 期的患者，并分为糖尿病组和非糖尿病组，然后至少每 6 个月进行一次随访。我们使用带有时间变量中心化的线性混合回归模型来克服分析基线 GFR 与变化之间关系时的数学耦合问题，并比较了混合模型正确和不正确规格的结果。共确定了 1002 例患者，其中 285 例为糖尿病患者，717 例为非糖尿病患者。线性混合回归模型显示，在糖尿病组和非糖尿病组中，基线 GFR 与随后的 GFR 下降率之间均存在显著负相关（r= -0.44[95%置信区间，-0.69 至-0.09]），但在对时间变量进行个体内均值中心化后，非糖尿病组无统计学意义（r= -0.09[95%置信区间，-0.41 至 0.25]）。我们的研究表明，在早期 CKD 患者中，糖尿病组中较高的基线 GFR 与随后的 GFR 下降更为陡峭相关，但在非糖尿病组中则无此相关性。应详细描述准确的模型规格，以防止得出虚假结论。
