Background: Diabetes mellitus (DM) is one of the most serious global health problems. In Greece, DM constitutes a public health problem and is highly associated with decreasing levels of physical activity, increasing obesity rates, population ageing, and unhealthy lifestyle and dietary behaviors.
Materials: In this study we evaluated the sera from 800 type 2 diabetic patients recruited during a three year period of time and 200 age matched controls without any clinical history of diabetes. For each subject we measured levels of fasting glucose (GLU), total cholesterol (TCHOL), triglycerides (TRG), high density lipoproteins (HDL-C), and glycosylated hemoglobin (GHbA1c) and calculated levels of low density lipoproteins (LDL-C). The aims of our study were to find characteristics of lipid parameters in the population under study, to find gender differences in the parameters, to evaluate correlations between pairs of lipid parameters, and to compare the lipid parameters between patients and healthy controls focusing on patient gender. For this purpose we analyzed the data using descriptive statistics, x-square test, logistic regression and ROC curve analysis.
Results: According to our results, 70.0% of diabetic patients presented at least one lipid abnormality. Elevated LDL-C, elevated TCHOL, elevated TRG, and reduced HDL-C levels were noted in 28.37%, 36.37%, 39.01%, and 30.12% of the patients, respectively. The combination of elevated TRG and reduced HDL-C was the most preva- lent of the combined lipid abnormalities. Moreover, there are statistically significant differences in the levels of HDL-C, TCHOL, TRG, and GLU between men and women. In contrast, no differences were observed in levels of GHbA1c.
Conclusions: We identified an important linear relationship between LDL-C and TCHOL (LDL-C = -28.69 + TCHOL * 0.75, adjusted R2 = 76.96%. Finally, we calculated optimal thresholds for GLU and GHb1Ac levels using two methodologies: overall accuracy maximization or sensitivity-specificity minimization for the identification of patient from healthy controls. Differences in the optimal thresholds between men and women were not observed.