Lipid ratios and appropriate cut off values for prediction of diabetes: a cohort of Iranian men and women.

BACKGROUND: Dyslipidemia is a risk factor for incident type 2 diabetes; however, no study has specifically assessed the lipid ratios (i.e. total cholesterol (TC)/high density lipoprotein cholesterol (HDL-C) and triglyceride (TG)/HDL-C) as predictors of diabetes. We aimed to compare the independent association between the different lipid measures with incident diabetes over a median follow up of 6.4 years in Iranian men and women.

METHOD: The study population consisted of 5201 non diabetic (men = 2173, women = 3028) subjects, aged > or =20 years. The risk factor adjusted odds ratios (ORs) for diabetes were calculated for every 1 standard deviation (SD) change in TC, log-transformed TG, HDL-C, non-HDL-C, TC/HDL-C and log-transformed TG/HDL-C using multivariate logistic regression analysis. Receiver operator characteristic (ROC) curve analysis was used to define the points of the maximum sum of sensitivity and specificity (MAXss) of each lipid measure as a predictor of diabetes.

RESULT: We found 366 (146 men and 220 women) new diabetes cases during follow-up. The risk-factor-adjusted ORs for a 1 SD increase in TG, TC/HDL-C and TG/HDL-C were 1.23, 1.27 and 1.25 in men; the corresponding risks in females were 1.36, 1.14, 1.39 respectively (all p < 0.05, except TC/HDL-C in females which was marginally significant, p = 0.07). A 1 SD increase of HDL-C only in women decreased the risk of diabetes by 25% [0.75(0.64-0.89)]. In both genders, there was no difference in the discriminatory power of different lipid measures to predict incident diabetes in the risk factor adjusted models (ROC approximately 82%). TG cutoff values of 1.98 and 1.66 mmol/l; TG/HDL-C cutoff values of 4.7 and 3.7, in men and women, respectively, TC/HDL-C cutoff value of 5.3 in both genders and HDL-C cutoff value of 1.18 mmol/l in women yielded the MAXss for defining the incidence of diabetes.

CONCLUSION: TC/HDL-C and TG/HDL-C showed similar performance for diabetes prediction in men population however; among women TG/HDL-C highlighted higher risk than did TC/HDL-C, although there was no difference in discriminatory power. Importantly, HDL-C had a protective effect for incident diabetes only among women.