Distributed Lag Models: Examining Associations Between the Built Environment and Health

Area of Science:

• Environmental Health
• Spatial Epidemiology
• Biostatistics

Background:

• Built environment influences individual behaviors and health outcomes, necessitating accurate measurement.
• Traditional regression methods often rely on fixed spatial scales, potentially introducing bias.
• The optimal spatial scale for assessing built environment-health associations is often unknown.

Purpose of the Study:

• To propose and validate distributed lag models (DLMs) for analyzing built environment-health associations.
• To circumvent the need for a priori spatial scale selection in built environment research.
• To examine the relationship between convenience store proximity to schools and children's BMI z-scores.

Main Methods:

• Application of distributed lag models (DLMs) to assess health outcomes as a function of distance from built environment features.
• Simulation studies to compare DLM performance against traditional regression methods.
• Analysis of convenience store availability near California public schools and its association with children's body mass index z scores.

Main Results:

• Traditional regression models can yield biased associations (away from the null) when built environment features are spatially correlated.
• Inference based on DLMs demonstrates robustness across various built environment scenarios.
• The study provides an innovative application of DLMs in environmental health research.

Conclusions:

• DLMs offer a flexible and robust approach to modeling built environment-health relationships without prespecified spatial scales.
• Misspecification of spatial scales in traditional methods can lead to biased results.
• This methodology can be applied to investigate environmental factors, such as food access, influencing children's health.