ARS | Publication request: Metabolic risk profiles created using cluster analysis are differentially associated with physical activity: The ARIC study

Submitted to: Obesity
Publication Type: Abstract Only
Publication Acceptance Date: August 1, 2007
Publication Date: September 1, 2007
Citation: Monda, K., Bray, M., Adair, L., Asao, K., Kao, W.H., Boerwinkle, E., Pankow, J., North, K. 2007. Metabolic risk profiles created using cluster analysis are differentially associated with physical activity: The ARIC study [abstract]. Obesity. 15:126-OR(Suppl.)

Technical Abstract: Conditions such as hypertension, dyslipidemia, glucose intolerance, and obesity tend to cluster together and predict cardiovascular disease, type 2 diabetes, and premature mortality. This clustering has led to multiple definitions of the Metabolic Syndrome (MetS). While the definitions agree on the essential components, they differ in detail and criteria. Physical activity (PA) reduces the severity of metabolic risk factors and is a primary therapy for those with MetS. Our purpose was to use cluster analysis to create groups of individuals with similar metabolic risk profiles and to investigate whether PA is differentially associated with the clusters. For comparison, we also investigated the association of PA to MetS using a standard definition. Cluster analysis was used to separately group African American (AA) (n = 2870) and white (n = 9738) men and women aged 45-64 based on the following risk factors: waist circumference, total cholesterol, HDL, LDL, triglycerides, fasting glucose and insulin, and systolic and diastolic BP from the Atherosclerosis Risk in Communities Study baseline visit. Cross-sectional associations with PA from sports-related activities (as quartiles) were assessed using multinomial logistic regression models, controlling for center, sex, age, smoking, and body mass index. The resulting 3 clusters in both whites and AA were characterized by low risk (whites 38.8%; AA 39.9%), dyslipidemia (whites 32.8%; AA 39.0), or hypertension (whites 28.4%; AA 21.2%). We found for whites that those in the hypertensive (OR = 1.39 (1.17-1.66)) and dyslipidemic (OR = 1.42 (1.21-1.67)) clusters were more likely to be in the lowest quartile of PA than those in the low risk cluster. For AA we found similar associations for those in the hypertensive (OR = 1.52 (1.05-2.21)) but not in the dyslipidemic cluster. Lower PA was associated with having the MetS in whites (OR = 1.22 (1.04-1.42)), but not in AA. Individuals with higher risk profiles as determined by cluster analysis had lower levels of PA. Using a standard definition of MetS, the association with PA was only significant in whites. Aggregating risk factors in new ways to emphasize specific characteristics may be beneficial in investigating lifestyle associations, especially in African Americans.