Association between blood lipids and diabetes mellitus in older Chinese adults aged 65 years or older: a cross-sectional analysis of residents' electronic health records.

ABSTRACT

AIM: This study aimed to investigate how blood lipids are associated with diabetes among older Chinese adults. METHODS: 3,268,928 older Chinese adults without known diabetes were included. Logistic regression and restricted cubic spline (RCS) models were conducted to study associations between blood lipids (total cholesterol [TC], triglycerides [TG], low-density lipoprotein cholesterol [LDL-C], and high-density lipoprotein cholesterol [HDL-C]) and diabetes. RESULTS: 202,832 diabetes cases were included. Compared with the lowest quintiles, TC, TG, and LDL-C in the highest quintiles showed a higher diabetes prevalence risk and HDL-C presented a lower risk in multivariate-adjusted logistic regression models. Odds ratios (ORs) and 95% confidence intervals (95% CIs) for the highest quintiles of TC, TG, and HDL-C were 1.39 (1.37-1.41), 2.56 (2.52-2.60), and 0.73 (0.72-0.74), respectively. For LDL-C, 3-5% lower risk was found in the second and third quintiles, and 4-23% higher risk was found in the fourth and fifth quintiles. RCS curves showed a non-linear relationship between each blood lipid parameters and diabetes (P-non-linear < 0.001). TG and HDL-C curves presented monotonically increasing and L-shaped patterns, respectively, whereas TC and LDL-C curves exhibited a J-shaped pattern. When TC < 4.04 mmol/L or LDL-C < 2.33 mmol/L, ORs of diabetes increased with the decrease of corresponding indexes. However, after excluding participants with lower LDL-C, the J-shaped association with TC disappeared. CONCLUSIONS: This study demonstrates non-linear associations between lipids and diabetes. Low cholesterol levels are associated with a high risk of diabetes. The cholesterol paradox should be considered during lipid-lowering treatments.