We completed a systematic search of the epidemiologic literature on built environment and obesity and identified 63 relevant papers, which were then evaluated for the quality of between-study evidence. We were able to classify studies into one of two primary approaches for defining place and corresponding geographic areas of influence: those based on contextual effects derived from shared pre-determined administrative units and those based on individually unique geographic buffers. The 22 contextual papers evaluated 80 relations, 38 of which did not achieve statistical significance. The 15 buffer papers evaluated 40 relations, 24 of which did not achieve statistical significance. There was very little between-study similarity in methods in both types of approaches, which prevented estimation of pooled effects. The great heterogeneity across studies limits what can be learned from this body of evidence.
Copyright 2009 Elsevier Ltd. All rights reserved.