The influence of health behavior clusters on dietary change.

BACKGROUND: The goal of this study was to identify cancer preventive health behavior clusters and to determine if clusters responded differently to a year-long intervention to increase fruit and vegetable consumption. METHODS: The North Carolina Strategies for Improving Diet, Exercise, and Screening (NC STRIDES) is a health communications intervention (n = 595) among colorectal cancer survivors and a comparison population. Cluster analysis was used to identify nonoverlapping groups based on fruit and vegetable intake (servings/day), physical activity (minutes/day), multivitamin use (yes/no), and body mass index (kg/m(2)). Logistic regression was performed to assess positive change in fruit and vegetable servings, using the healthiest cluster as the reference group. RESULTS: Five clusters were formed; they differed significantly by health behaviors and demographics. Clusters 1 and 2 (those following the « Healthy Choices » and « Eating Well » patterns) were eating more than 5 A Day before the intervention (8.6 and 6.9 servings/day), and did not show any increase. Cluster 3 (« Physically Active ») reported an increase of 1.3 servings/day to reach 5.4 servings/day, and Clusters 4 and 5 (« Average Americans » and « Most Challenged ») improved one serving/day for final intakes of 5.2 and 5.0 servings/day. CONCLUSIONS: These findings illustrate some differences in magnitude of response to a fruit and vegetable intervention based on health behavior profiles. Creating clusters or other categories from baseline health behaviors may help to further improve targeting and/or tailoring in health promotion interventions.