Methodological advances in designing a longitudinal urban exposome study: the UrbanX exposome study for the city of Paris within the CONSTANCES cohort

Abstract Background The complexity of urban ecosystems and populations presents major challenges in studying how urban settings shape populations’ wellbeing and health over space and time. The urban exposome methodological framework has been earlier defined as the continuous spatiotemporal monitoring of population indicators associated with urban external and internal exposome domains, using small area-level analysis. We present the design of a novel urban exposome methodological framework and outline the application of its exposomics tools as nested within a specific longitudinal setting, that is that of the CONSTANCES cohort in France. Methods CONSTANCES is a population-based cohort including >200 000 volunteers recruited between 2012 and 2020 and followed-up to date annually with a medical examination every 4 years. The urban Paris population and its setting is considered here, using IRIS small area level (IRIS division of the territory typically involves a population of 1800 to 5000 inhabitants). Using the human exposome domains’ classification, we describe a novel methodological approach for tackling the spatiotemporal profile dynamics of the urban Paris exposome and investigate its prospective association with a health outcome, that is body mass index (BMI). Participants from the inclusion dataset are clustered into exposome-based spatial networks using hierarchical clustering on principal components, applying spatial constraints. Mixture effects of co-occurring and space- and time-varying exposome variables in their prospective association with BMI may be estimated using Bayesian kernel machine regression—lagged distribution models, random forest and/or penalized regression models. Discussion The application of an urban exposome methodological framework nested within a longitudinal cohort population presents with opportunities to advance our understanding of chronic disease process in complex urban settings. Advanced biostatistical algorithms that account for repeated measures of exposures and outcomes are warranted. This approach would be also applicable in multiple urban population studies nested within different cohorts.