Affordable Access

deepdyve-link
Publisher Website

Multicollinearity in associations between multiple environmental features and body weight and abdominal fat: using matching techniques to assess whether the associations are separable.

Authors
  • Leal, Cinira
  • Bean, Kathy
  • Thomas, Frédérique
  • Chaix, Basile
Type
Published Article
Journal
American Journal of Epidemiology
Publisher
Oxford University Press
Publication Date
Jun 01, 2012
Volume
175
Issue
11
Pages
1152–1162
Identifiers
DOI: 10.1093/aje/kwr434
PMID: 22534206
Source
Medline
License
Unknown

Abstract

Because of the strong correlations among neighborhoods' characteristics, it is not clear whether the associations of specific environmental exposures (e.g., densities of physical features and services) with obesity can be disentangled. Using data from the RECORD (Residential Environment and Coronary Heart Disease) Cohort Study (Paris, France, 2007-2008), the authors investigated whether neighborhood characteristics related to the sociodemographic, physical, service-related, and social-interactional environments were associated with body mass index and waist circumference. The authors developed an original neighborhood characteristic-matching technique (analyses within pairs of participants similarly exposed to an environmental variable) to assess whether or not these associations could be disentangled. After adjustment for individual/neighborhood socioeconomic variables, body mass index/waist circumference was negatively associated with characteristics of the physical/service environments reflecting higher densities (e.g., proportion of built surface, densities of shops selling fruits/vegetables, and restaurants). Multiple adjustment models and the neighborhood characteristic-matching technique were unable to identify which of these neighborhood variables were driving the associations because of high correlations between the environmental variables. Overall, beyond the socioeconomic environment, the physical and service environments may be associated with weight status, but it is difficult to disentangle the effects of strongly correlated environmental dimensions, even if they imply different causal mechanisms and interventions.

Report this publication

Statistics

Seen <100 times