International Journal of Environmental Research and Public Health Review Methods to Address Self-Selection and Reverse Causation in Studies of Neighborhood Environments and Brain Health Lilah M. Besser 1,*, Willa D. Brenowitz 2, Oanh L. Meyer 3, Serena Hoermann 4 and John Renne 4 1 Department of Urban and Regional Planning, Institute for Human Health and Disease Intervention (I-HEALTH), Florida Atlantic University, Boca Raton, FL 33431, USA 2 Departments of Psychiatry and Behavioral Sciences and Epidemiology & Biostatistics, University of California San Francisco, San Francisco, CA 94121, USA; [email protected] 3 Department of Neurology, University of California Davis, Sacramento, CA 95817, USA; [email protected] 4 Center for Urban and Environmental Solutions (CUES), Department of Urban and Regional Planning, Florida Atlantic University, Boca Raton, FL 33431, USA; [email protected] (S.H.); [email protected] (J.R.) * Correspondence: [email protected] Abstract: Preliminary evidence suggests that neighborhood environments, such as socioeconomic disadvantage, pedestrian and physical activity infrastructure, and availability of neighborhood desti- nations (e.g., parks), may be associated with late-life cognitive functioning and risk of Alzheimer’s disease and related disorders (ADRD). The supposition is that these neighborhood characteristics are associated with factors such as mental health, environmental exposures, health behaviors, and social determinants of health that in turn promote or diminish cognitive reserve and resilience in later life. However, observed associations may be biased by self-selection or reverse causation, such as when individuals with better cognition move to denser neighborhoods because they prefer many Citation: Besser, L.M.; Brenowitz, destinations within walking distance of home, or when individuals with deteriorating health choose W.D.; Meyer, O.L.; Hoermann, S.; residences offering health services in neighborhoods in rural or suburban areas (e.g., assisted living). Renne, J. Methods to Address Research on neighborhood environments and ADRD has typically focused on late-life brain health Self-Selection and Reverse Causation outcomes, which makes it difficult to disentangle true associations from associations that result in Studies of Neighborhood Environments and Brain Health. Int. from reverse causality. In this paper, we review study designs and methods to help reduce bias J. Environ. Res. Public Health 2021, 18, due to reverse causality and self-selection, while drawing attention to the unique aspects of these 6484. https://doi.org/10.3390/ approaches when conducting research on neighborhoods and brain aging. ijerph18126484 Keywords: epidemiological methods; causality; reverse causation; self-selection; bias; neighborhood; Academic Editor: Paul B. Tchounwou built environment; brain health; Alzheimer disease; cognition Received: 8 May 2021 Accepted: 13 June 2021 Published: 16 June 2021 1. Introduction Studies on the influence of neighborhood environments (NE) (i.e., social and built Publisher’s Note: MDPI stays neutral environments (BE)) on brain health are still in their infancy but are growing rapidly with regard to jurisdictional claims in and provide tentative evidence that our community environments may affect the brain published maps and institutional affil- throughout the lifespan [1–3]. Greater neighborhood socioeconomic disadvantage has been iations. associated with worse baseline cognition [4,5], greater decline in cognition over time [6], and total and regional brain volumes from magnetic resonance imaging (MRI) [7] among older adults. Neighborhood racial/ethnic segregation has been linked to poorer cognitive health outcomes in middle and older aged individuals [8–10]. In addition, measures of Copyright: © 2021 by the authors. the BE such as greater land use mix (e.g., mix of retail and residential) [11], access to Licensee MDPI, Basel, Switzerland. retail destinations [12], public transportation availability [13], greater walkability [14] (i.e., This article is an open access article environment conducive to walking by providing multiple destinations and density of distributed under the terms and street connections), and greater greenness/park space access [15–18] have been associated conditions of the Creative Commons with various measures of brain health in older adults including diagnoses of Alzheimer’s Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ disease and related disorders (ADRD). This body of research is typically rooted in the 4.0/). socioecological framework that posits that beyond individual level determinants of brain Int. J. Environ. Res. Public Health 2021, 18, 6484. https://doi.org/10.3390/ijerph18126484 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2021, 18, x FOR PEER REVIEW 2 of 19 sociated with various measures of brain health in older adults including diagnoses of Alz- heimer’s disease and related disorders (ADRD). This body of research is typically rooted Int. J. Environ. Res. Public Health 2021, 18, 6484 2 of 19 in the socioecological framework that posits that beyond individual level determinants of brain health, including age and genetics, there are likely higher-level social determinants of health (SDOH) operating, including neighborhood and community environments. SDOHhealth, affect including environmental age and genetics,exposures there and areaccess likely to support, higher-level resources, social and determinants opportuni- of tieshealth that (SDOH)ultimately operating, affect a population’s including neighborhood morbidity and and mortality. community environments. SDOH affectYet, environmental even the most exposures rigorous studies and access of the to NE support, and brain resources, health andto date, opportunities such as those that thatultimately employ affect population-based a population’s cohorts morbidity with and longitudinal mortality. follow-up or natural experi- ments,Yet, can even still thebe biased most rigorous due to attrition studies of(e.g., the individuals NE and brain with health the tooutcome date, such of interest as those droppingthat employ out at population-based a higher rate), non-representativeness cohorts with longitudinal of the follow-up sample orcompared natural experiments, to the gen- eralcan population, still be biased and due competin to attritiong causes/residual (e.g., individuals confounding. with the Table outcome 1 presents of interest some drop- of theping methodological out at a higher challenges rate), non-representativeness unique to studying NEs of the and sample brain health, compared such to as the defining general thepopulation, neighborhood and competinggeographic causes/residualboundary, capturing confounding. the neighborhood Table1 presents exposure, some defining of the themethodological neighborhood challenges construct,unique typical to reliance studying on NEs studies and of brain older health, adults, such and as the defining lag be- the tweenneighborhood development geographic of pathology boundary, for ADRD capturing and the neighborhooddiagnosis of dementia. exposure, We defining first pre- the sentneighborhood these broader construct, issues to typical set the reliance context on for studies our more of older detailed adults, focus and on the self-selection lag between anddevelopment reverse causation. of pathology Reverse for ADRDcausation and occu the diagnosisrs when the of dementia.outcome precedes We first presentand results these inbroader the exposure issues to(Figure set the 1a). context Cross-sectional for our more studies detailed are focusprone on to self-selection this potential and issue reverse be- causecausation. temporal Reverse ordering causation of exposure occurs and when outcome the outcome cannot precedes be established. and results This in may the expo-also besure an issue (Figure for1a). studies Cross-sectional on outcomes studies with are long prone preclinical to this potentialor subclinical issue periods. because temporalReverse causationordering in of which exposure changes and outcomein brain health cannot predict be established. the neighborhood This may one also resides be an issuein is an for examplestudies of on self-selection. outcomes with However, long preclinical self-selec ortion subclinical can also periods.bias NE-brain Reverse health causation studies in notwhich through changes reverse in brain causation, health but predict through the confounding, neighborhood in one which resides the inlifestyle is an exampleor neigh- of borhoodself-selection. preferences However, influence self-selection residential can ch alsooices bias (Figure NE-brain 1b). healthAny association studies not between through reverse causation, but through confounding, in which the lifestyle or neighborhood prefer- the NE and brain health may instead be attributable to lifestyle behaviors and preferences ences influence residential choices (Figure1b). Any association between the NE and brain that preceded neighborhood choice and that are also associated with the brain health out- health may instead be attributable to lifestyle behaviors and preferences that preceded come. neighborhood choice and that are also associated with the brain health outcome. a Neighborhood Neighborhood/BE Brain health selection characteristics Neighborhood/BE b characteristics BE/lifestyle Health behaviors Brain health Preferences FigureFigure 1. 1.IllustrationIllustration of of relationship relationship between between neighbor neighborhood/builthood/built