Cluster Analysis Defines Distinct Dietary Patterns for African-American Men and Women
Accepted 3 July 2008.
Abstract
Background
African Americans have a higher prevalence of diet-related chronic diseases than the general US population. Recent evidence suggests that examining dietary patterns rather than specific nutrients may better allow public health professionals to translate national dietary goals into practical dietary recommendations that are culturally relevant and sex-specific.
Objective
To examine and characterize the dietary patterns of African-American men and women using cluster analysis.
Design
Participants older than the age of 18 years were recruited from a variety of locations and completed a self-administered questionnaire.
Subjects/setting
Seven-hundred sixty-three African-American men and women in North Central Florida.
Statistical analyses performed
Cluster analysis was done with the VARCLUS procedure and Ward's clustering approach in the Statistical Analysis System program. The pseudo-t2 statistic was then used to identify the optimal numbers of clusters of men and women with distinctive food-consumption patterns.
Results
Cluster analyses identified six clusters of men with 13 distinctive dietary patterns and six clusters of women with 15 distinctive dietary patterns. Pairwise comparisons of group means indicated notable differences in intakes between clusters of women in all but two food clusters. Pairwise comparisons of group means indicated notable differences in intakes between clusters of men in all but two food clusters.
Conclusions
The study illustrated that cluster analysis can be used to identify groups of African Americans with unique patterns of food consumption. They also suggest that a single “typical” eating pattern does not exist for African Americans and that dietary pattern vary by sex.
Address correspondence to: Delores C. S. James, PhD, RD, Department of Health Science Education, PO Box 118210, Room 5 Florida Gym, University of Florida, Gainesville, FL 32611-8210
ABSTRACT