Evaluating causal relationships between urban built environment characteristics and obesity: a methodological review of observational studies

Background: Existing reviews identify numerous studies of the relationship between urban built environment characteristics and obesity. These reviews do not generally distinguish between cross-sectional observational studies using single equation analytical techniques and other studies that may support more robust causal inferences. More advanced analytical techniques, including the use of instrumental variables and regression discontinuity designs, can help mitigate biases that arise from differences in observable and unobservable characteristics between intervention and control groups, and may represent a realistic alternative to scarcely-used randomised experiments. This review sought first to identify, and second to compare the results of analyses from, studies using more advanced analytical techniques or study designs. Methods: In March 2013, studies of the relationship between urban built environment characteristics and obesity were identified that incorporated (i) more advanced analytical techniques specified in recent UK Medical Research Council guidance on evaluating natural experiments, or (ii) other relevant methodological approaches including randomised experiments, structural equation modelling or fixed effects panel data analysis. Results: Two randomised experimental studies and twelve observational studies were identified. Within-study comparisons of results, where authors had undertaken at least two analyses using different techniques, indicated that effect sizes were often critically affected by the method employed, and did not support the commonly held view that cross-sectional, single equation analyses systematically overestimate the strength of association. Conclusions: Overall, the use of more advanced methods of analysis does not appear necessarily to undermine the observed strength of association between urban built environment characteristics and obesity when compared to more commonly-used cross-sectional, single equation analyses. Given observed differences in the results of studies using different techniques, further consideration should be given to how evidence gathered from studies using different analytical approaches is appraised, compared and aggregated in evidence synthesis.