The Best Obesity Indices to Use in a Single Factor Model Indicating Metabolic Syndrome: a Population Based Study.

Abstract

Although metabolic syndrome (MetS) is a major health problem worldwide, there is no universal agreement on its definition. One of the major disagreements is dealing with the issue of obesity in this definition. This study was conducted to determine a preferably better index of obesity which can be interrelated with other components of MetS in a single factor model of MetS. Out of 6140 participants of a cohort study of subjects aged 10-90 years in northern Iran, the baseline data of 5616 participants aged 18-75 was considered. Confirmatory factor analysis was conducted using AMOS software to evaluate a single factor model of MetS in which blood pressure, triglyceride (TG), high density lipoprotein (HDL), fasting blood sugar (FBS) and obesity measures including waist circumference (WC), body mass index (BMI), waist to hip ratio (WHR) and waist to height ratio (WHtR) were used as indicators of metabolic syndrome. Four single factor models differing from each other by obesity indices were evaluated. The models were evaluated in all 5616 subjects and 4931 subjects without diabetes mellitus according to sex separately. All single factor models had appropriate fit indices with CFI > 0.95, GFI > 0.95 and RMSEA < 0.08 in non-diabetic population, wherein all models obtained the best values of fit indices in men and good fit indices in women. In the general population of men, the single factor models built based on WHR (Chi-square=6.9, df=2, P-value=0.031, RMSEA = 0.028, CI = 0.007-0.052, CFI = 0.994, GFI = 0.999 and AIC = 22.9) and WHtR (Chi-square = 9.97, df = 2, P-value = 0.007, RMSEA = 0.036, CI = 0.016-0.059, CFI = 0.992, GFI = 0.998 and AIC = 25.97) were fitted properly with data while in th general population of women, the model based on WHR obtained better fit indices (Chi-square = 7.5, df = 2, P-value = 0.023, RMSEA = 0.033, CI = 0.011-0.060, CFI = 0.994, GFI = 0.998 and AIC = 23.5). Models based on WHtR obtained better regression weights than WHR. While single factor validity of MetS was confirmed in almost all models, the best models were different according to sex and population of study.