Table 1.

Select methodological challenges of research on neighborhood environments and brain health.

Challenge	Example Issues to Consider
Defining neighborhood boundaries: self-reported/perceived, administrative boundaries (e.g., US Census tract), or geographic information system (GIS) buffers around the residence (e.g., ½-mile Euclidian buffer)	Self-reported measures: Difficult to compare self-reported measures across participants. Administrative boundaries: Varying land area, lack of international comparability, may not represent neighborhood area most pertinent to individual. GIS buffers or administrative boundaries: May not represent neighborhood area most pertinent to individual.
Capturing neighborhood exposure: time period, place, degree	What time period of exposure is most important for brain health (childhood, middle age, late life)? Where is exposure most pertinent and does this depend on life stage? How do we quantify degree of exposure? Should we/how to consider accumulated exposure?
Defining the neighborhood construct	Can we develop neighborhood measures that have high validity and reliability?
Self-selection into neighborhoods	Individuals may choose to move to neighborhoods because they offer opportunities for health behaviors (e.g., walking, healthy foods) that affect brain health.
Reverse causation: Association is due to outcome leading to exposure, not vice versa	Alzheimer’s disease and related disorders lead to neighborhood selection (e.g., brain health outcome is related to residential move in late life to accommodate health needs).
Strong correlation of neighborhood characteristics	When highly correlated, how do we know the association found for one variable is not actually demonstrating effect of highly correlated variable?
Neighborhood segregation	Structural racism/exclusionary and discriminatory policies and practices led to residential segregation of racial/ethnic groups and socioeconomic status that is highly correlated and difficult to disentangle from other neighborhood characteristics (e.g., access to parks and healthy foods) [23,24,25].
Spatial considerations	Residential areas that are closer together tend to have similar values (e.g., similar exposures and/or outcomes), which if not accounted for in the analyses can lead to erroneous conclusions. Modifiable area unit problem: the area unit employed to define the neighborhood (e.g., Census tract versus Census block group) can affect the significance of findings.
Studying older adults (e.g., >60 years old)	Older adults are more likely to develop physical and cognitive impairments that can affect study enrollment, attrition, and participation in study procedures (e.g., magnetic resonance imaging). Requiring survival to old age may result in highly select samples who are healthier.
Lag between Alzheimer’s disease and related disorders pathology development and dementia diagnosis	Longitudinal studies of older adults may not sufficiently account for undiagnosed, prodromal disease affecting neighborhood exposure.
Invasive/time consuming procedures to measure brain health (e.g., lumbar puncture, brain imaging) may limit the types of neighborhoods or ranges of neighborhood characteristics captured	Restricting to individuals who consent to/complete invasive brain health procedures are more likely to include individuals of White race and higher socioeconomic status, who typically live in White neighborhoods with higher socioeconomic status. This can limit generalizability and result in selection bias.