ExCoGIS (Extant Cohorts and Geographic Information Systems) initiative
Ecological model of the effects of urban environments on cognitive health
Several cohort that have been established to study individual-level determinants of cognitive health or collected cognitive function data have joined the ExCoGIS initiative from its inception (January 2018). These include: the Australian Imaging, Biomarkers and Lifestyle (AIBL) on ageing study; the Personality And Total Health (PATH) through life project; the Sydney Memory and Ageing Study (Sydney MAS); the Australian Diabetes, Obesity & Lifestyle Study (AusDiab); and the ALzheimer and FAmilies (ALFA) project. Each of these cohorts has specific methodological features and geographical settings that uniquely contribute to achieving the aims of this initiative. Collectively, analyses across cohorts will enable assessment of the generalizability of findings.
The proposed work under the ExCoGIS initiative consists of three sequential stages
Creation of spatial environmental variables to be added to extant cohort datasets.
Objective GIS-based data on urban design characteristics of residential neighbourhoods will be linked to each cohort included in the ExCoGIS initiative. These will include median household income, dwelling density, street intersection density, land use mix, and number and distance to public transport stops, parks, public open spaces, health-related destinations, community services (e.g., libraries), general services (e.g., post offices), food outlets and retail. We will also quantify the amount of live green vegetation (Normalized Difference Vegetation Index), the concentration of air pollutants linked to ill health (PM2.5, PM10 and NO2) and traffic-related noise. All the above variables will be created for participants’ residential addresses and, if applicable, workplace addresses. Neighbourhood areas will be operationalized as 0.5km and 1km buffers surrounding residential and work addresses. Data on pedestrian and cycling infrastructure will be collected using the online version of the Microscale Audit of Pedestrian Streetscapes (MAPS) Global. Using images from Google Earth Street-View, trained observes will use MAPS Global to collect information on traffic safety, quality of footpaths and amenities for pedestrians and cyclists along the route that connects a residential/workplace address to the nearest commercial area.
Cohort-specific data analyses addressing research questions focusing on specific aspects of the proposed ecological model of the effects of urban environments on cognitive health.
The aims of the ExCoGIS initiative revolve around the estimation of associations between levels of, and changes in, urban environmental factors, lifestyle behaviours, blood and brain imaging biomarkers and cognitive health. Generalized Additive Models (GAMs) and 2-level Generalized Additive Mixed Models (GAMMs) with random intercepts (accounting for spatial clustering arising from the adopted sampling strategy) will be employed to estimate the shape and strength of the cross-sectional associations between attributes of the urban environment, cognitive function and related lifestyle behaviours and biological mechanisms using cohorts with cross-sectional data. For cohorts with multiple assessments of cognitive function across time (but no spatial sampling), several analytical approaches will be adopted depending on the data available. For example, we will use 2-level GAMMs with random intercepts and slopes accounting, respectively, for the dependence of repeated measures on the same individuals and individual-differences in exposure effects. Mediation analyses will be conducted to quantify the environmental effects on cognitive function mediated by lifestyle behaviours and biological mechanisms. Moderating effects of personality, genotype and other factors will be estimated by adding appropriate interaction terms to the main effect models described above. Missing data (for time points prior to a participants’ withdrawal from a study) will be imputed using Multiple Imputations by Chained Equations and analyses will be conducted on 100 imputed datasets. Directed Acyclic Graphs will be employed to select confounders to be included in the regression models.
Synthesis of findings providing an overview of the validity of the proposed ecological model of the effects of urban environments on cognitive health
We will provide an overview of the validity of the proposed ecological model of the effects of urban environments on cognitive health based on our findings and an updated systematic review of the literature.
Status / Timeline
The creation and development of spatial environmental variables (Stage 1) to be added to the cohorts that joined the ExCoGIS initiative from its inception commenced in May 2018 and is expected to be completed by June 2019.
Key Investigators (as per grant application)
• Prof Mark Nieuwenhuijsen, ISGlobal (Barcelona, Spain) and ACU (Melbourne, Australia)
• Prof Kaarin Anstey, UNSW and NeuRA (Sydney, Australia)
• A/Prof Anthony Barnett, ACU (Melbourne, Australia)
• Prof Takemi Sugiyama, ACU (Melbourne, Australia)
• Prof Nicola Lautenschlager, University of Melbourne (Melbourne, Australia)
• Prof Bin Jalaludin, UNSW (Sydney, Australia)
• Prof David Dunstan, Baker Heart and Diabetes Institute and ACU (Melbourne, Australia)
• Prof Perminder Sachdev, UNSW, Centre for Healthy Brain Ageing (Sydney, Australia)
• Prof Jonathan Shaw, Baker Heart and Diabetes Institute (Melbourne, Australia)
• Prof Olivier Salvado, CSRIO, Biomedical Informatics Group (Brisbane, Australia)
• Dr Marta Crous-Bou, Barcelonaβeta Brain Research Center, Fundació Pasqual Maragall (Barcelona, Spain)