Research Institute, National Health Insurance Service Ilsan Hospital, Goyang, South Korea.
Division of Endocrinology and Metabolism, Department of Internal Medicine, National Health Insurance Service Ilsan Hospital, Goyang, South Korea.
BMJ Open Diabetes Res Care. 2020 Dec;8(2). doi: 10.1136/bmjdrc-2020-001340.
To examine how the risk of cardiovascular disease changes according to degree of change in body mass index (BMI) and low-density lipoprotein (LDL)-cholesterol in patients with diabetes using the health medical examination cohort database of the National Health Insurance Service in Korea. In comparison, the pattern in a non-diabetic control group was also examined.
The study samples were 13 800 patients with type 2 diabetes and 185 898 non-diabetic controls, and their baseline characteristics and repeatedly measured BMI and LDL-cholesterol until occurrence of cardiovascular disease were collected in longitudinal data. We used the variability model that is joint of mixed effects and regression model, then estimated parameters about variability by Bayesian methods.
The risk of cardiovascular disease was increased significantly with high average real variability (ARV) of BMI in the patients with diabetes, but the risk of cardiovascular disease was not increased according to degree of ARV in non-diabetic controls. The Bayesian variability model was used to analyze the effects of BMI and LDL-cholesterol change pattern on development of cardiovascular disease in diabetics, showing that variability did not have a statistically significant effect on cardiovascular disease. This shows the danger of the former simple method when interpreting only the mean of the absolute value of the variation.
The approach of simple SD in previous studies for estimation of individual variability does not consider the order of observation. However, the Bayesian method used in this study allows for flexible modeling by superimposing volatility assessments on multistage models.
本研究旨在利用韩国国家健康保险服务的健康体检队列数据库，探讨糖尿病患者的体重指数（BMI）和低密度脂蛋白胆固醇（LDL-C）变化程度与心血管疾病风险之间的关系。同时，还对非糖尿病对照组的模式进行了研究。
研究样本包括 13800 例 2 型糖尿病患者和 185898 例非糖尿病对照者，收集了他们的基线特征以及在纵向数据中反复测量的 BMI 和 LDL-C，直至发生心血管疾病。我们使用了混合效应和回归模型的联合变异性模型，然后通过贝叶斯方法估计了变异性的参数。
在糖尿病患者中，BMI 的平均真实变异性（ARV）较高时，心血管疾病的风险显著增加，但在非糖尿病对照组中，ARV 程度与心血管疾病风险之间没有显著相关性。贝叶斯变异性模型用于分析 BMI 和 LDL-C 变化模式对糖尿病患者心血管疾病发展的影响，结果显示变异性对心血管疾病没有统计学意义的影响。这表明了在解释变异的绝对值时，仅使用平均值的传统方法存在一定的危险。
之前研究中用于估计个体变异性的简单标准差方法没有考虑观察的顺序。然而，本研究中使用的贝叶斯方法允许通过在多阶段模型上叠加波动率评估来进行灵活建模。
