A16200 HyperTriglyceridemia is Identified As a Risk Factor of HyperGlycemia Via Bayesian Network Inference

Objectives: Whereas both hypertriglyceridemia and hyperglycemia are known as risk factors for cardiovascular diseases, little is known regarding the causal association between them. In this study, we made an attempt to explore whether elevated triglyceride can increase the risk for hyperglycemia or vice versa. Methods: Datasets used for the analysis were obtained from a large-scale retrospective 4-year cohort study including 19886 subjects (age:68.95 ± 7.33; 9097 men) who took health examination in 2013 and 2016 at zhuji, Zhejiang province, China, respectively. We first performed Bayesian network inference to capture the intrinsically causal relationships among body mass index (BMI), smoking and drinking habits, total cholesterol (TC), triglycerides, fasting blood glucose (FBG), blood pressure (BP) and estimated glomerular filtration rate (eGFR) on health examination dataset in 2013 and tested them on dataset in 2016. Multivariate regression analysis was conducted to compare the results from Bayesian network inference with the adjustments of age, BMI, baseline TC, baseline triglycerides, baseline FBG, baseline BP and baseline eGFR. Results: Bayesian network inference demonstrated that elevated triglycerides level would cause the increase of FBG. Consistent with results from network inference, hypertriglycerides is an independent predictor for hyperglycemia (OR: 1.156 per mmol/L increase; 95% CI: [1.113,1.200]) identified by multivariate regression analysis. Conclusion: Elevated triglycerides increases the risk of high glucose, suggesting the potential significance for the prevention of diabetes mellitus through efficient triglycides control. Additionally, Bayesian network inference is an efficient way to identify risk factors for clinic related events, which is free of time-consuming and costly cohort studies.