Background: Metabolic syndrome (MetS) has been traditionally defined as abnormal values of at least three of the following components [systolic blood pressure (SBP), diastolic blood pressure (DBP), waist circumference (WC), HDL cholesterol, triglycerides (TG), and fasting glucose (FG)]. However, there is concern that cut-points used to define abnormality are inappropriate for certain populations and which components are present matters. Methods: Using structural equation modeling (SEM), we estimated a continuous MetS scores as a latent variable based on the actual values of SBP, DBP, WC, HDL, TG, and FG by gender and race subgroups for 7791 participants from NHANES III, 1988-94. Data were linked with National Death Index to determine mortality status and cause of death up to December 31, 2006. Cox proportional hazard ratios were estimated to test the association between each MetS continuous component, MetS scores estimated from SEM, and traditionally defined MetS (yes/no) with cardiovascular disease (CVD) mortality adjusted for age, race, gender, smoking status, education, physical activity, alcohol consumption, and medication use for diabetes, hypertension, or high cholesterol. Results: For men, continuous MetS components individually associated with CVD mortality were SBP [Hazard ratio (HR) = 1.24, 95% confidence interval (CI) = 1.11 [[Unable to Display Character: –]] 1.38], DBP (HR = 1.36, 95% CI = 1.15 [[Unable to Display Character: –]] 1.61), and TG (HR = 1.01, 95% CI = 1.01 [[Unable to Display Character: –]] 1.02). In women, SBP (HR = 1.21, 95% CI = 1.11 [[Unable to Display Character: –]] 1.32) and DBP (HR = 1.28, 95% CI = 1.06 [[Unable to Display Character: –]] 1.54) were associated with CVD mortality. Traditionally defined MetS or number of MetS components were not significantly associated with CVD mortality [men (MetS HR = 1.10, 95% CI = 0.73 [[Unable to Display Character: –]] 1.66; # of MetS components HR = 1.05, 95% CI = 0.89 [[Unable to Display Character: –]] 1.23) and women (MetS HR = 0.97, 95% CI = 0.68 [[Unable to Display Character: –]] 1.38; # of MetS components HR = 1.10, 95% CI = 0.97 [[Unable to Display Character: –]] 1.25)]. However, significant associations between MetS score and CVD mortality were found in men (HR = 1.43, 95% CI = 1.07 [[Unable to Display Character: –]] 1.90) and women (HR = 1.34, 95% CI = 1.07 [[Unable to Display Character: –]] 1.69). In subgroup analyses, the association between the MetS score and CVD mortality was greater among Mexican-American women (HR = 2.84, 95% CI = 1.46 [[Unable to Display Character: –]] 5.53) and non-Hispanic black women (HR = 2.14, 95% CI = 1.48 [[Unable to Display Character: –]] 3.08) compared to non-Hispanic white women (HR = 1.26, 95% CI = 1.001 [[Unable to Display Character: –]] 1.59). When comparing models using the MetS score with those using traditionally defined MetS or number of abnormal MetS components, goodness-of-fit and concordance were better for all models with the MetS score based on Harrell’s C-statistics, Gönen & Heller’s K-statistics, Akaike information criterion, and Bayesian information criterion. Conclusions: When investigating CVD risk and associations with MetS components, MetS score was a better predictor and risk assessment tool for CVD mortality than traditional defined MetS.
Read full abstract