Sage Journals: Discover world-class research

Abstract

Neighborhood socioeconomic status is linked to risk of Alzheimer’s disease and related dementias, but studies utilizing latent variable methods to clarify how built and social environment resources may relate to cognitive outcomes are limited. We applied exploratory and confirmatory factor analysis to data from the 2010 wave of the Health and Retirement Study (N = 17,642) to derive built and social environment resource factors and then examined associations with memory using structural equation modeling. Results revealed greater built and social environment resources were associated with better memory performance. Effects were modified by race/ethnicity such that environmental resource factors were more robustly associated with memory among non-Latino White compared to non-Latino Black and Latino participants. Results highlight that the presence of built and social environmental resources may support memory functioning, but disparities in the distribution of these resources must be addressed to ensure benefits are conferred equally across racial/ethnic groups.

Keywords

built environment social environment cognitive function racial/ethnic health disparities

Introduction

The prevalence of Alzheimer’s disease and related dementias (ADRD) is rapidly rising as the population ages, with more than 13 million older adults expected to be living with ADRD by the year 2060 (Rajan et al., 2021). The growing prevalence of ADRD represents a significant public health concern, given that disease-modifying treatments are currently limited, and there is an urgent need to shift towards identifying modifiable factors that may prevent or slow cognitive decline (Livingston et al., 2020). Environmental factors have been increasingly emphasized in recent research initiatives, as we have come to understand that the context in which people live, work, and play may be responsible for shaping risk for ADRD across the life course (Adkins-Jackson et al., 2023). Limited exposure to health-promoting resources (e.g., healthcare access, green space) and overexposure to health-depleting risk factors (e.g., toxins, pollution) may increase risk for the development of ADRD in late-life, and it has been well established that these factors are not equally distributed across neighborhood environments occupied by different racial/ethnic groups within the United States (Anderson, 2017).

Neighborhood socioeconomic status (NSES) represents one way of characterizing exposure to environmental resources and is one of the most widely utilized measures of the environmental context in the larger aging literature (Besser et al., 2017). NSES is usually calculated as an aggregate measure of the economic (e.g., % unemployed, % below the poverty line, family income) and social (e.g., % with a college degree) characteristics of a place where a person lives. Previous research has demonstrated that older adults who reside in higher NSES environments generally perform better on cognitive tests and exhibit slower rates of cognitive decline when compared to those that reside in low NSES environments (Sheffield & Peek, 2009; Zuelsdorff et al., 2020). These studies have illustrated that older adults residing in neighborhoods composed of educated and financially stable individuals tend to exhibit better cognition, but ultimately, they have provided limited insight into which specific physical and social environmental features are most strongly linked to late-life cognitive health. Moreover, there is some evidence to suggest that reliance on compositional NSES markers to serve as a proxy for the availability of environmental resources (e.g., grocery stores, pharmacies, etc.) may be somewhat problematic. Indeed, research has shown that higher poverty levels are paradoxically associated with a greater concentration of certain physical and social resources in some neighborhood environments (Small & McDermott, 2006). These limitations underscore the need to more directly characterize distinct types of environmental resources, especially given that these may represent critical points of intervention for promoting healthy cognitive aging.

Environmental resources are often conceptualized across two domains: the built environment and the social environment. The built environment (BE) is generally thought to consist of the physical features of the environment (e.g., parks, recreation centers, grocery stores), and the presence of these BE resources has been directly linked to health behaviors including physical activity and healthy eating that play a role in preventing cognitive decline (Frank et al., 2022; Vásquez et al., 2016). On the other hand, the social environment (SE) can be conceptualized as the relational and cultural dimensions of the environment, including social cohesion, sociocultural norms, and social disorder (Besser et al., 2017; Finlay et al., 2022). However, it is worthwhile to note that some researchers have also considered aspects of social infrastructure (e.g., social organizations, social destinations) as part of the social environment, given their potential to foster social interaction and exchange of social support (Diez Roux & Mair, 2010). Indeed, access to social infrastructure (e.g., restaurants, libraries, and museums) has been associated with higher levels of social engagement, which in turn is positively associated with cognitive functioning in late life (Finlay et al., 2022).

Although several existing studies have linked individual BE and SE resources to cognitive outcomes, the patterns of association have varied considerably (Besser et al., 2017; Peterson et al., 2021). Some research has shown that cognitive performance is linked to distinct BE and SE features including greater mixed land use (Chan et al., 2023), retail access (Besser et al., 2018), better housing conditions (James III & Sweaney, 2010), and the density of social organizations and institutions (e.g., arts and cultural sites, libraries; Clarke et al., 2012; Finlay et al., 2022). However, other studies have found null or negative associations between similar BE/SE constructs and cognition (Besser et al., 2018; Chan et al., 2023; Clarke et al., 2012). These inconsistencies may reflect heterogeneity in study designs, population characteristics, and geographic contexts. However, a recent systematic review suggests that these mixed findings may also stem from the wide variation in indicators used to operationalize the broad and multi-faceted domains of the BE and SE (Besser et al., 2017). Focusing solely on isolated elements of the BE or SE may also fail to capture the cumulative effects of these domains on cognitive health.

Data-driven latent variable techniques–such as exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and structural equation modeling (SEM)–offer a more integrated approach to examining the shared and distinct contributions of BE and SE features on cognitive functioning. These modeling approaches also account for measurement error, providing less biased estimates of environmental constructs compared to using individual indicators. However, prior studies using latent variable modeling techniques have often combined variables measured at both the individual- (e.g., occupation, frequency of social participation) and community-level (e.g., health care services), and imposed orthogonality constraints (Clay et al., 2022; Smith et al., 2013). Including individual-level variables that do not objectively reflect environmental resources may dilute the validity of these constructs, while forcing factors to be uncorrelated overlooks the interdependence between features of the BE and SE. Finally, existing studies have also largely treated these domains as distinct and examined their effects independently. This is problematic because features of BE and SE are often interrelated and may be mutually reinforcing (Finlay et al., 2022). Research that comprehensively measures BE and SE features at the same geographic level and utilizes an SEM framework which allows interdependence between BE and SE factors is therefore critical to understanding their shared effects on cognitive health outcomes.

It is also important to recognize that the availability of BE and SE resources significantly varies across racial/ethnic groups. In the United States, residential segregation and redlining have deprived neighborhoods predominantly occupied by racially/ethnically minoritized individuals of critical BE and SE resources (Adkins-Jackson et al., 2023). There is some evidence to suggest that greater exposure to environmental hazards and reduced access to supportive resources as a result of institutionalized racism have contributed to poorer cognitive outcomes among Black and Latino older adults (Besser et al., 2018; Clarke et al., 2012). However, prior research studies have largely relied upon regionally constrained samples (e.g., Chicago Community Adult Health Study, Multi-Ethnic Study of Atherosclerosis) and did not directly examine racial/ethnic differences in levels of exposure to BE and SE features. Additional research in a geographically diverse, national sample is therefore needed to clarify how BE and SE resources are distributed across racial/ethnic groups, and to determine how the relationships between these environmental resources and cognition differ in ways that may be contributing to racial/ethnic health disparities in ADRD.

The present study sought to improve characterization of environmental resources and clarify whether BE and SE are associated with cognition in a large, national sample of racially/ethnically diverse adults enrolled in the Health and Retirement Study (HRS). We aimed to develop a more comprehensive and theoretically-sound measure of environmental resources by applying EFA to first identify latent BE and SE constructs, followed by CFA to validate their structure in an independent sample. We hypothesized that two distinct latent constructs would emerge that account for unique variance in the distribution of BE and SE resources. The second aim was to utilize structural equation modeling (SEM) to examine associations between BE, SE, and cognition. Memory was chosen as the primary cognitive outcome of interest, given that it is a strong pre-clinical indicator of future ADRD risk and functional decline in later life (Boraxbekk et al., 2015). Examining domain-specific effects on memory also fills a gap in the current literature given that most research has been limited to examining environmental effects on general cognitive composites (Clarke et al., 2012). We hypothesized that environmental factor scores would be associated with memory function, but that the relative strength of the association would differ across BE and SE resource factors. We also explored whether patterns of association differed by race/ethnicity and hypothesized that racially/ethnically minoritized participants would be exposed to lower levels of BE and SE resources compared to non-Latino White participants, and the beneficial effects of environmental factors on memory function would not be equally conferred across racial/ethnic groups.

Methods

Participants

Data for this cross-sectional study were drawn from the 2010 wave of the publicly available Health and Retirement Study (HRS), a longitudinal, nationally representative cohort study of Americans above the age of 50. Detailed documentation of the HRS can be found online (https://hrs.isr.umich.edu/). The study was approved by the University of Michigan Institutional Review Board, and all participants provided informed consent. The HRS oversamples for Black and Latino adults and, in 2017, began offering harmonization with restricted contextual data resources (e.g., census-based and geographically linked data), allowing the effects of environmental factors on memory to be examined and compared across racial/ethnic groups.

The current sample utilized HRS participants who were interviewed in 2010 (n = 22,807) to maintain measurement consistency with the National Neighborhood Data Archive (NaNDA) datasets, which provide geographically linked data for 2010 census tracts. Our exploratory factor analysis sample consisted of 19,553 participants who were not currently living in a nursing home and had geographically linked data available. The final analytic sample for our confirmatory factor analysis and structural equation modeling consisted of 17,642 participants who also had memory assessments and race/ethnicity data available (See Supplemental Figure 1 for a study flow diagram).

Measures

Built Environment and Social Environment

Data representing built and social environment domains were drawn from HRS Contextual Data Resources (Health and Retirement Study, 2024), the National Neighborhood Data Archive (NaNDA; Finlay et al., 2020), and the US Decennial Census and American Community Survey (Ailshire et al., 2020). NaNDA variables were measured in 2010 and ACS estimates from 2006-2010 were leveraged to ensure geographic data was temporally matched as close as possible to the time of survey and cognitive data collection. To account for disparate environmental resources as a function of population size, extracted amenities of interest were restricted to population density estimates (i.e., counts per 1000 population in the census tract). Environmental indicator variables were not normally distributed and were Blom transformed to permit maximum likelihood estimation (Blom, 1958).

BE and SE features were selected based on the Diez Roux and Mair framework (2010), which details the physical and social attributes of the environment that are theoretically linked to health outcomes, as well as based on upon evidence from prior studies demonstrating associations between these features and cognitive outcomes (Besser et al., 2017; Finlay et al., 2022). BE features utilized in the analyses included: commercial amenities (i.e., clothing store density, department store density), recreational amenities (i.e., the density of fitness centers, golf courses, ski resorts, bowling alleys, etc.), food access (i.e., eating place density, supermarket density), health care access (i.e., outpatient health care density, pharmacy density), and housing (i.e., median home value, median gross rent). SE features reflected aspects of social infrastructure that could be objectively measured using administrative data and included: the density of religious organizations (i.e., churches, synagogues, etc.), social organizations (i.e., social clubs, veterans’ organizations, etc.), civic/social services (i.e., senior centers, job training, etc.), and art or cultural institutions (i.e., performing arts organizations, museums/libraries, etc.). See Supplemental Table 1 for a full list of datasets and variable descriptions.

Memory Outcome

Episodic memory performance was measured using a version of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) Immediate and Delayed Word Recall task. We utilized imputed raw scores from the RAND longitudinal file to minimize bias due to study attrition (RAND HRS Longitudinal File 2020, 2024). An episodic memory composite score was used in analyses and was generated by averaging z-scored immediate and delayed recall totals.

Covariates

We controlled for baseline age, gender (binarily-coded with male as the reference group), years of education (self-reported, top-coded at 17), depressive symptoms (total score from the Center for Epidemiological Studies – Depression [CESD] scale; [Radloff, 1977]), chronic health condition burden (sum of seven binarily-coded, self-reported health conditions including diabetes, cancer, lung disease, heart problems, hypertension, arthritis, and stroke), income, and net wealth (assets minus debts). Race and ethnicity were coded into three binary variables (non-Latino Black, non-Latino other, and Latino) with non-Latino White as the reference group.

Statistical Analysis

Exploratory and Confirmatory Factor Analysis

Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were performed to derive latent environmental resource constructs from the selected BE/SE variables. Two equivalent subsamples were generated using the SOLOMON method (Lorenzo-Seva, 2022) to permit cross-validation of factor structures. There were 5,858 unique census tracts in the whole sample, thus each subsample included 2,929 tracts. The EFA subsample was evaluated to determine suitability for factor analysis using the Kaiser–Meyer–Olkin test and Bartlett’s test of sphericity. Scree plots and eigenvalues were examined to gauge the maximum number of suitable factors. We applied oblique (oblimin) rotation using the fa function from the psych package in R since BE and SE variables were expected to be correlated with one another. Variables with lower loadings were sequentially excluded until all variables had loadings of at least 0.4 to ensure each variable meaningfully contributed to the environmental factor. Alternative models (e.g., 2-, 3-, and 4-factor) were compared based on their theoretical interpretability and goodness-of-fit statistics including lower χ², lower Bayesian Information Criterion (BIC), Tucker-Lewis Index [TLI] > 0.95, comparative fit indices (CFI) > 0.96, Root Mean Square Error of Approximation (RMSEA) < 0.05, and standardized root mean square residual (SRMR) < 0.08. Once the best-fitting model was selected from the EFA, we tested the replicability of the model through CFA using the lavaan package in R.

Structural Equation Models

SEM was employed to estimate the association between our latent environmental factors and memory function while accounting for measurement error and the clustering of observations within census tracts and households. Income and wealth covariates were Blom transformed to meet assumptions of multivariate normality. All other continuous predictors were grand-mean centered. We modeled environmental resource factors alone and simultaneously with the other factors. This approach allowed us to first assess whether each environmental factor had a significant association with memory, then determine their independence. Covariates were added sequentially (Model 1: adjusted for sociodemographic and health variables, Model 2: adjusted for Model 1 covariates plus income and wealth) to examine how individual socioeconomic characteristics attenuated environmental effects. Clustering by census tract and household was addressed through the use of robust standard errors applied via the survey package in R. Multigroup SEM with configural invariance (i.e., identical factor structures) was employed to compare differences in patterns of association across racial/ethnic groups. Group differences in sample characteristics were tested via analysis of variance for continuous variables and chi-square tests for categorical variables (Table 1). Tukey’s post hoc test was conducted to determine which groups significantly differed. Findings were considered significant at the p < 0.05 level and all analyses were conducted in R version 4.2.1.

Table 1.

Sample Characteristics of the Analytic Sample of HRS Respondents

Variables	Overall	Non-Latino White	Non-Latino Black	Non-Latino Other	Latino	Group Comparison
Variables	N = 17642	n = 11677	n = 3275	n = 507	n = 2183	Group Comparison
Age, mean (SD)	65.96 (10.85)	67.85 (10.92)	62.56 (9.90)	62.14 (9.86)	61.89 (9.41)	NLW > NLB = NLO = L
Gender, % female	57.90%	56.90%	61.90%	55.60%	57.70%	NLB > L > NLW > NLO
Years of education, mean (SD)	12.70 (3.17)	13.33 (2.51)	12.43 (2.83)	13.51 (2.97)	9.54 (4.58)	NLO = NLW > NLB > L
Chronic health conditions, mean (SD)	1.84 (1.36)	1.89 (1.36)	1.93 (1.38)	1.60 (1.41)	1.49 (1.28)	NLB = NLW > NLO = L
Depressive symptoms, mean (SD)	1.50 (2.02)	1.29 (1.89)	1.84 (2.09)	1.62 (1.99)	2.10 (2.38)	L > NLB = NLO > NLW
Income, mean (SD)	$62,760.00 ($86,504.01)	$71,874.38 ($95,538.09)	$44,988.34 ($61,241.75)	$68,496.85 ($73,982.87)	$39,335.86 ($57,306.89)	NLW = NLO > NLB = L
Wealth, mean (SD)	$259,622.23 ($81,7456.41)	$353,289.70 ($95,2185.15)	$56,199.92 ($35,1935.53)	$214,156.70 ($48,6465.73)	$74,328.59 ($40,7865.20)	NLW > NLO > L = NLB
Commercial Infrastructure factor, mean (SD)	0.00 (0.91)	0.06 (0.90)	−0.13 (0.86)	0.03 (1.01)	−0.11 (0.93)	NLW = NLO > L = NLB
Social Infrastructure factor, mean (SD)	0.00 (0.90)	0.08 (0.89)	−0.16 (0.84)	0.06 (0.96)	−0.22 (0.93)	NLW = NLO > NLB = L
Housing Costs factor, mean (SD)	0.00 (0.91)	0.05 (0.87)	−0.27 (0.87)	0.31 (1.11)	0.04 (1.02)	NLO > NLW = L > NLB
Memory Immediate Recall, mean (SD)	5.36 (1.65)	5.51 (1.65)	5.10 (1.60)	5.30 (1.66)	4.94 (1.59)	NLW > NLO > NLB > L
Memory Delayed Recall, mean (SD)	4.29 (1.97)	4.51 (1.96)	3.75 (1.95)	4.24 (1.98)	3.95 (1.86)	NLW > NLO > L > NLB

Note. NLW = Non-Latino White, NLB = Non-Latino Black, NLO = Non-Latino Other, L = Latino, SD = Standard Deviation; All group comparisons were significant at the p < 0.001 level.

Results

Environmental Resource Constructs

Exploratory and Confirmatory Factor Analysis

The EFA results showed that data were suitable for factor analysis (Bartlett’s test: p < 0.05; KMO = 0.84) and the final and best-fitting model (TLI = 0.96, RMSEA = 0.05, BIC = 17.46) included three factors representing the constructs of commercial infrastructure, social infrastructure, and housing costs (Supplemental Table 2). The commercial infrastructure and housing costs factors best represented the BE. The social infrastructure factor captured aspects of the SE. The commercial and social infrastructure factors were moderately correlated (0.53), while social and housing and housing and commercial factors had weaker correlations (0.12, 0.11, respectively). Five variables that were theorized to represent these constructs were dropped from the final factor analysis solution due to insufficient loadings (i.e., pharmacy density, grocery store density, performing arts organization density, recreational amenity density, and religious organizations density). We validated the model fit through CFA in the second half of the sample (See Supplemental Table 3). Except for library density (0.30), all loadings remained above the 0.40 threshold. The covariance between the commercial and social infrastructure factors increased to 0.84, while the covariance between the social and housing costs factors (0.27) and commercial and housing costs factors (0.17) remained smaller. The model fit was satisfactory (CFI = 0.89, TLI = 0.85, RMSEA = 0.10, SRMR = 0.07) but decreased compared to the EFA, likely due to the inferior loading of library density. CFA across the whole sample produced similar factor loadings and covariances (Figure 1).

Figure 1.

Comprehensive Structural Model of Associations Between Memory, Environmental Resource Factors, and Individual Covariates. Note. Commercial = Commercial Infrastructure Factor, Housing = Housing Costs Factor, Social = Social Infrastructure Factor, df = Degrees of Freedom, BIC = Bayesian Information Criterion, RMSEA = Root Mean Square Error of Approximation, SRMR = Standardized Root Mean Square Residual, CFI = Comparative Fit Index, TLI = Tucker-Lewis Index, Med. = Median, CESD = Center for Epidemiological Studies Depression Scale, Hispanic = Hispanic/Latino; Solid Lines Indicate a Statistically Significant Path (p < 0.05) and Dotted Lines Indicate Non-significant Paths (p ≥ 0.05). Standard Errors are Reported in Parentheses after the Main Path Coefficients. All Coefficients are Standardized. *p < 0.05 **p < 0.01 ***p < 0.001

Sample Characteristics

Descriptive characteristics are presented in Table 1. The sample had an average age of 66 (SD = 10.85), 13 (SD = 3.17) years of education, and was majority female (57.90%). The racial/ethnic composition of the sample was 66% non-Latino White (n = 11,677), 19% non-Latino Black (n = 3,275), 3% non-Latino Other (n = 507), and 12% Latino (n = 2,183). There were significant differences across racial/ethnic groups for all variables. Non-Latino White and non-Latino other participants had more years of education, higher income, greater wealth, and better memory performance on average compared to non-Latino Black and Latino participants. Non-Latino White and non-Latino other participants also had higher commercial and social infrastructure factor scores compared to non-Latino Black and Latino participants. For the housing factor, non-Latino other participants had the highest factor scores, followed by non-Latino White and Latino groups, who were equivalent to each other, and then non-Latino Black participants.

Relationship Between Memory and Environmental Resource Constructs

Whole sample models revealed that housing costs were positively associated with memory performance (B = .06, p < .001) independent of sociodemographic variables and health (Table 2). The association between housing costs and memory was slightly attenuated by adjustment for income/wealth but remained significant (B = .03, p = .001). Social infrastructure was also positively associated with memory performance (B = .03, p = .004) but this association was fully attenuated by adjustment for individual income and wealth (B = .01, p = .09). There were no significant associations between commercial infrastructure and memory performance (Bs ≤ .02, ps > .05). Figure 1 depicts our comprehensive model which includes the relative associations of all three environmental constructs with memory performance. Only the housing costs factor was positively associated with memory independent of the other two factors (B = .03, p = .002).

Table 2.

Path Coefficients for Environmental Resource Factors on Memory

Model	Regression Paths	B	B (SE)	Model Fit
Model	Regression Paths	B	B (SE)	ChiSq	CFI	RMSEA
Model 1: Adjustment for sociodemographics and health^a
1.1	C → M	0.02	0.02 (0.01)	1627.50***	0.86	0.07
1.2	S → M	0.03**	0.02 (0.01)	1625.94***	0.86	0.07
1.3	H → M	0.06***	0.06 (0.01)	1617.35***	0.86	0.07
Model 2: Adjustment for model 1 covariates plus income and wealth
2.1	C → M	0.01	0.01 (0.01)	1999.92***	0.85	0.07
2.2	S → M	0.01	0.01 (0.01)	1999.54***	0.85	0.07
2.3	H → M	0.03**	0.03 (0.01)	1997.98***	0.85	0.07

Note. B = standardized coefficient, B = unstandardized coefficient, SE = standard error, C = Commercial Infrastructure Factor, S = Social Infrastructure Factor, H = Housing Costs Factor, M = Memory, CFI = Comparative Fit Index, RMSEA = Root Mean Square Error Of Approximation. Scaled ChiSq is reported. Robust and scaled CFI and RMSEA are reported.

^aCovariates included baseline age, gender, education, health conditions, and depressive symptoms. Significant effects are indicated in bold. *p < 0.05 **p < 0.01 ***p < 0.001.

Effect Modification by Race/Ethnicity

Multigroup SEM revealed patterns of association between environmental resource constructs and memory differed across racial/ethnic groups. Housing costs (Table 3) had a positive main effect on memory performance after full covariate adjustment within non-Latino White (B = .03, p = .003) and non-Latino Black participants (B = .04, p = .046), but not among Latino (B = .02, p = .37) or non-Latino other participants (B = .02, p = .62). Before adjustment for income/wealth, the housing costs factor was also positively associated with memory independent of the other two environmental resource constructs for both non-Latino White (B = .05, p < .001) and non-Latino Black participants (B = .05, p = .005). However, after adjustment for all covariates, only the association within non-Latino White participants was significant independent of commercial infrastructure and social infrastructure factors (B = .02, p = .009).

Table 3.

Path Coefficients for Environmental Resource Factors on Memory Across Racial/Ethnic Groups

Model	Regression paths	Non-Latino White (n = 11677)		Non-Latino Black (n = 3275)		Non-Latino Other (n = 507)		Latino (n = 2183)
Model	Regression paths	B	B (SE)	B	B (SE)	B	B (SE)	B	B (SE)
Model 1: Adjustment for sociodemographics and health^a
1.1	C → M	0.01	0.01 (0.01)	−0.01	−0.01 (0.02)	0.02	0.02 (0.04)	0.02	0.02 (0.02)
1.2	S → M	0.02*	0.02 (0.01)	0.00	0.00 (0.02)	−0.02	−0.02 (0.04)	0.02	0.02 (0.02)
1.3	H → M	0.05***	0.05 (0.01)	0.05**	0.05 (0.02)	0.04	0.04 (0.04)	0.04	0.03 (0.02)
1.4	C → M	−0.02	−0.02 (0.02)	−0.03	−0.03 (0.04)	0.22*	0.20 (0.09)	0.05	0.04 (0.06)
	S → M	0.03	0.03 (0.02)	0.02	0.02 (0.04)	−0.24*	−0.22(0.09)	−0.03	−0.03 (0.06)
	H → M	0.05***	0.04 (0.01)	0.05**	0.05 (0.02)	0.03	0.03 (0.04)	0.04	0.03 (0.02)
Model 2: Adjustment for model 1 covariates plus income and wealth
2.1	C → M	0.01	0.01 (0.01)	−0.01	−0.01 (0.02)	0.02	0.02 (0.04)	0.02	0.02 (0.02)
2.2	S → M	0.01	0.01 (0.01)	0.00	0.00 (0.02)	−0.03	−0.02 (0.04)	0.01	0.01 (0.02)
2.3	H → M	0.03**	0.02 (0.01)	0.04*	0.03 (0.02)	0.02	0.02 (0.04)	0.02	0.02 (0.02)
2.4	C → M	−0.01	−0.01 (0.02)	−0.03	−0.03 (0.04)	0.21*	0.20 (0.09)	0.06	0.05 (0.06)
	S → M	0.01	0.01 (0.02)	0.02	0.02 (0.04)	−0.23*	−0.21 (0.09)	−0.04	−0.04 (0.06)
	H → M	0.02**	0.02 (0.01)	0.03	0.03 (0.02)	0.01	0.01 (0.04)	0.02	0.02 (0.02)

Note. C = Commercial infrastructure, S = Social infrastructure, H = Housing costs, M = Memory, CFI = Comparative Fit Index, RMSEA = Root Mean Square Error Of Approximation. Models .1-.3 estimate independent effects of environmental factors while model .4 includes all three factors simultaneously. Models are constrained to have the same factor structure (configural invariance) across groups.

^aCovariates included baseline age, gender, education, health conditions, and depressive symptoms. Significant effects are indicated in bold. *p < 0.05 **p < 0.01 ***p < 0.001.

Social infrastructure had a positive main effect on memory performance only within non-Latino White participants (B = .02, p = .03). However, this association was no longer significant after adjusting for individual income and wealth (B = .01, p = .22). Non-Latino other participants exhibited a negative association between social infrastructure and memory when all three factors were included in the same model (B = −.23, p = .02). Commercial infrastructure had no main effects on memory performance within any racial/ethnic group (Bs ≤ .02, ps > .05). However, when all three environmental constructs were modeled simultaneously, commercial infrastructure was positively associated with memory among non-Latino other participants (B = .21, p = .04). There were no significant associations between environmental factors and memory performance among Latino participants (Bs ≤ .06, ps > .05).

Discussion

This study aimed to improve our understanding of how environmental resources measured by BE and SE factors may be related to memory functioning and whether associations differed across race/ethnicity. EFA identified three total environmental resource factors that were validated by CFA in a separate subsample. The commercial infrastructure and housing costs factors were reflective of the BE and a social infrastructure factor was reflective of the SE. Higher housing costs and social infrastructure, but not commercial infrastructure, were associated with better memory performance. Multi-group SEM revealed that the effects of housing costs and social infrastructure on memory were less robust among Black and Latino compared to non-Latino White participants, reflecting how exposure to fewer environmental resources may constrain cognitive benefits and reinforce racial/ethnic disparities in ADRD risk. Taken together, our findings highlight housing and social infrastructure as BE and SE resources that may support cognitive health. Results also emphasize the need for more community-based interventions that address disparities in the distribution and access to important environmental resources among racially/ethnically minoritized community members.

Measurement of BE and SE Resource Factors

We believe the present study builds upon the NSES literature by improving the operationalization of BE and SE environmental resources that may be supportive of preventing ADRD in later life. While other studies have examined the effects of contextual features (Clarke et al., 2012; Finlay et al., 2022), these studies have typically treated BE and SE resources as individual, observable variables rather than latent constructs. Our EFA approach allowed each variable to be classified into the most appropriate contextual factor to capture shared variance while accounting for measurement error to improve the estimation of the effect of BE or SE resource exposure. We additionally completed a CFA on a separate subsample to rigorously test the model fit and validated that these constructs could be reliably captured across different contexts. Our study is distinct given that previous research using data-driven approaches has modeled these environmental resources at differing geographic scales (e.g., individual, zip code, census tract), which may ultimately make it more challenging to infer at which spatial scale these effects are most salient (Clay et al., 2022; Smith et al., 2013). The present results suggest that census tract level BE and SE resources are linked to current levels of cognitive functioning and help clarify which aspects of the environment may be important to focus on from a dementia intervention perspective.

Our results revealed a three-factor solution consisting of an SE factor, referred to as social infrastructure, and two distinct BE factors, housing costs and commercial infrastructure, which is consistent with ecological theories suggesting that a multidimensional approach best captures simultaneous exposure to both BE and SE resources in the environment (Diez Roux & Mair, 2010; Finlay et al., 2022). Although previous studies have utilized a latent variable approach to modeling environmental resources, their statistical methods required environmental factors to be orthogonal, which may have artificially forced independence between BE and SE constructs when these should be considered as interrelated and mutually reinforcing dimensions (Wahl & Lang, 2003). Allowing factors to correlate in our model likely improved the ecological validity of our latent constructs and suggests that environmental interventions for each domain may have shared mechanisms by which they operate to impact health. For example, building more retail storefronts may also increase the density of social infrastructure through providing commercial spaces to house third places (e.g., coffee shops), community organizations, or non-profit social services that ultimately support health behaviors.

It is notable that the housing costs factor was weakly correlated with the other BE and SE factors. While we initially conceptualized median rent and median home value as proxies for the quality of physical housing stock and expected them to be highly related to BE resources (e.g., restaurants, shopping centers), these variables instead loaded strongly onto a separate factor. This likely reflects their strong association with neighborhood socioeconomic conditions (i.e., income/wealth of residents, affordability of housing), and it is important to acknowledge that these variables have been included as indicators of NSES in other work (Clay et al., 2022; Zuelsdorff et al., 2020). Housing costs are context-dependent and may not directly measure housing quality, as the same rent could reflect very different housing conditions across geographic areas. At the same time, higher housing costs may capture aspects of neighborhood desirability or housing demand that are influenced by unmeasured BE characteristics, such as lower pollution levels and better public transit access (Chen et al., 2023; Wang et al., 2022). Ultimately, by including housing costs we sought to capture some of the more complementary contextual aspects of BE features that are important for cognitive health. However, the weak correlation between housing costs and the other factors emphasizes that it is unclear whether these variables primarily reflect NSES, BE features, or some combination of both, highlighting the need for additional research efforts.

BE, SE, and Memory Associations

Higher scores on the housing costs factor were associated with better average memory performance independent from individual-level covariates. Residential environments with higher home values generally have higher rates of homeownership and residential stability which can facilitate larger social networks and more opportunities for the exchange of social support that may be beneficial for cognitive health (Peterson et al., 2021; Rohe & Stewart, 1996). Conversely, lower housing costs may reflect greater exposure to hazardous housing conditions (e.g., lead, mold, pests, physical deterioration), industrial pollution, economic disinvestment, and environmental stressors (e.g., crime, physical disorder), which can affect brain health directly through toxin exposure or indirectly through increasing psychological distress, which has been associated with poor memory outcomes (Sharifian et al., 2020; Shih et al., 2006; Zaheed et al., 2019). Given the suspected link between housing costs and NSES, lower home values may also indicate constrained educational and employment opportunities, potentially reducing cognitive reserve (Hamlin et al., 2024). Future studies should incorporate subjective assessments of neighborhood characteristics (e.g., sidewalk quality, physical disorder) and housing decay to more directly measure housing quality.

Similarly, we observed that higher levels of social infrastructure were associated with better memory performance. This adds to accumulating evidence of a positive effect of social infrastructure on cognition, likely by providing cognitive stimulation and neural resilience to stress (Clarke et al., 2015; Finlay et al., 2022). Importantly, our study extends prior studies that have focused on global cognition by demonstrating a domain-specific effect on memory, which may be particularly sensitive to stress-related effects on the brain (Lupien & Lepage, 2001). Our results also add to the existing literature by specifically demonstrating that effects were attenuated after accounting for individual income and wealth, suggesting that the positive effects of social institutions and clubs on cognitive health may be limited to those who have the economic resources to pay to utilize them considering many often have fees or dues (Finlay, Yu, et al., 2021). Finally, our results demonstrated that the effect of social infrastructure on cognition was also mitigated by the inclusion of housing costs in our comprehensive model which suggests neighborhood-level economic factors may influence access to and utilization of social resources. From a public policy perspective, increasing affordable social programming at local libraries, museums, and senior centers could serve as a potential intervention to increase social engagement and lower ADRD risk at the community level.

In contrast, commercial infrastructure did not exhibit beneficial effects on memory. This is consistent with prior work showing negative or null effects of urban density on cognition (Besser et al., 2018; Clarke et al., 2012). One possible explanation may be that a greater density of establishments is not always desirable or directly supportive of health. Greater density can hurt housing prices in some cases and residents may oppose greater economic development in favor of more residential zoning. This could be a partial explanation for why the commercial infrastructure and housing costs factors were only weakly correlated in our study. Additionally, our BE characterization was not exhaustive and other types of commercial infrastructure (e.g., grocery stores, recreational facilities, parks, public transit) have been linked to protective lifestyle behaviors, such as eating healthy foods and exercising, that could support cognitive health (Finlay, Esposito, et al., 2021). Furthermore, a greater presence of commercial establishments does not indicate whether residents are utilizing these facilities and engaging in more positive health behaviors as a result. Additional investigation is needed to clarify what role commercial establishments may play in cognitive aging and how individual factors (e.g., financial resources, subjective perceptions) shape utilization of BE resources.

We found evidence that associations between aspects of the environment and memory differed across racially/ethnically marginalized adults. We observed significant positive associations between housing costs and memory within Black participants, although this association was not robust to full covariate adjustment. Higher housing costs may thus be reflective of greater access to environmental resources that are supportive of memory functioning in Black older adults. However, beneficial effects were contingent upon individual (i.e., income/wealth) and environmental characteristics (i.e., commercial/social infrastructure). For example, living in a residential environment with higher home values may only enhance cognitive health if the physical and social amenities are financially accessible and culturally appropriate in order to facilitate use by Black older adult residents (Finlay, Yu, et al., 2021). No other BE and SE factors were significantly associated with memory among Black participants and contextual variables had no effect on cognition for Latino participants. This pattern likely reflects fundamental differences in the level and quality of environmental resources, given that we observed that non-Latino Black and Latino participants generally had lower factor scores for social infrastructure, commercial infrastructure, and housing costs compared to non-Latino White participants. It is possible that Black and Latino older adults were not exposed to high enough levels of housing costs and social infrastructure to experience substantial benefits to memory performance, which is consistent with null or negative associations observed in the present study and previous work (Clarke et al., 2012; Finlay, Yu, et al., 2021). To reduce health inequities in dementia, community-based prevention efforts must address the drastically different levels of exposure to environmental resources as a function of residential segregation and the ongoing racial wealth gap (Sharkey & Faber, 2014). These interventions could include investing in public infrastructure (e.g., public transit, parks, community centers) in historically redlined communities to facilitate greater access to health-promoting resources. However, simultaneous efforts must be made to support homeownership (e.g., community land trusts, subsidized mortgage lending) among long-time residents to prevent displacement and ensure they can reap the benefits of environmental improvements.

Interestingly, we observed positive effects on cognition for commercial infrastructure, and negative effects for social infrastructure in the comprehensive models for non-Latino other participants. We are cautious to interpret these findings as (1) no main effects were observed when factors were examined independently, (2) the sample size was much smaller compared to the other subgroups, and (3) this group is heterogeneous in terms of racial identity, and as such, individual experiences within environmental contexts may vastly differ. Follow-up studies within larger samples of Asian, Indigenous/Native American, and Alaskan/Pacific Islander participants are needed to characterize how unique exposures to environmental resources are associated with cognitive aging outcomes within these groups.

Strengths and Limitations

The strengths of this study included using a large, geographically diverse cohort recruited through population-based methods with over-sampling for Black and Latino adults. Our utilization of several secondary data sources allowed for a more comprehensive and objective measurement of census tract level BE and SE features representing environmental resources. We created theoretically supported and data-driven measures of BE and SE using EFA and CFA in two random subsamples, which contributed greater evidence for the reliability and ecological validity of our environmental constructs. Additional strengths included examining the independent effects of environmental factors within the same model, accounting for individual economic resources, exploring racial/ethnic differences in patterns of association using multigroup SEM, and utilizing a domain-specific measure of cognition, memory, which is particularly sensitive to early changes indicative of dementia risk.

However, there are several important limitations to note. Our analysis was cross-sectional, limiting inference about the causal effects of environmental factors on memory. Individuals may move residences and features of the environment can change over time (Sol et al., 2024), so there is also a need to better understand how the length of exposure to environmental factors may influence effects. Participants who were living in a nursing home or missing geographic and memory outcome data were not included in the analysis, which may have biased our sample by excluding individuals with physical or cognitive problems who could not complete these measures. Additionally, our results may be impacted by selective attrition, in which participants who were less healthy were more likely to drop out and not participate in the 2010 wave of the HRS, which may contribute to an underestimation of some of the effects. Nevertheless, it is notable that we still observed associations with memory in a sample that may have lower levels of cognitive impairment. While our findings showed that housing costs were associated with memory outcomes beyond the effects of commercial and social infrastructure, other unmeasured contextual features (e.g., transit accessibility, green space, social cohesion) could simultaneously be contributing to this association. Additionally, we employed a data-driven latent variable approach in the interest of measuring environmental features more reliably, but there are cases in which measuring the relationships between individual indicators and cognitive outcomes may be more appropriate (e.g., individual features are weakly correlated, leading to less reliable estimates of the BE or SE construct). Future studies should continue to more comprehensively examine BE and SE features using both latent and observed variable approaches and investigate synergistic effects on cognitive outcomes (Diez Roux & Mair, 2010). Because of the geographic heterogeneity of our sample, we are also limited in our ability to infer context-specific patterns, as environments have unique geographic, historical, and social factors that may affect cognition differently. The application of methods such as latent class analysis could help to identify subgroups of residential environments with similar patterns of resources.

Conclusions

This research characterized latent environmental resource factors and their complex associations with memory outcomes. Associations between environmental factors and memory varied by race/ethnicity, illustrating how systemically reduced access to environmental resources among non-Latino Black and Latino older adults may contribute to cognitive health disparities. Our findings have implications for urban planning policy and community-level interventions by demonstrating that housing and social infrastructure may be key factors in designing environments supportive of healthy cognitive aging. Adopting policies that ensure equitable access to environmental resources (e.g., reduced transportation fees, investments in public infrastructure) and stable homeownership (e.g., community land trusts, property tax exemptions) may help to address racial disparities in ADRD risk.

Supplemental Material

Supplemental Material - Built and Social Environment Resources are Associated with Memory Outcomes of Adults Enrolled in the Health and Retirement Study

Supplemental Material for Built and Social Environment Resources are Associated with Memory Outcomes of Adults Enrolled in the Health and Retirement Study by Abbey M. Hamlin, Lourdes S. Romañach Álvarez, Elizabeth Muñoz, Ashley Chikkala and Alexandra L. Clark in Research on Aging.

Footnotes

Acknowledgements

The proposal and analyses for the present project were preregistered; .

ORCID iDs

Abbey M. Hamlin

Lourdes S. Romañach Álvarez

Alexandra L. Clark

Ethical Approval

The study was approved by the University of Michigan Institutional Review Board (HUM00061128) and all participants provided written informed consent.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Health and Retirement Study is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. The National Neighborhood Data Archive is sponsored the National Institutes of Health/National Institute of Nursing Research and the National Institute on Minority Health and Health Disparities (NINR/NIMHD 1U01NR020556) as well as the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR, 90RTHF0001). This work was further supported by National Institute on Aging awards to A.L.C. (R03 AG085241; U19AG078109); an Alzheimer’s Association Award to A.L.C. (AARG-22-723000); and a National Science Foundation Graduate Research Fellowship Program (DGE-2137420) award to A.M.H.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Survey data from the Health and Retirement Study (HRS) is publicly available for download on the HRS website (https://hrs.isr.umich.edu/about). Access to geographically-linked data is restricted and can be applied for via a restricted data access agreement ().

Supplemental Material

Supplemental material for this article is available online.

Author Biographies

Abbey M. Hamlin, MA, is a doctoral student in the Department of Psychology at the University of Texas at Austin.

Lourdes S. Romañach Álvarez is a doctoral student in the Department of Human Development and Family Sciences at the University of Texas at Austin.

Elizabeth Muñoz, PhD, is an associate professor in the Department of Human Development and Family Sciences at the University of Texas at Austin.

Ashley Chikkala is an undergraduate student in Psychology at the University of Texas at Austin.

Alexandra L. Clark, PhD, is an assistant professor in the Department of Psychology at the University of Texas at Austin.

References

Adkins-Jackson

P. B.

George

K. M.

Besser

L. M.

Hyun

Lamar

Hill-Jarrett

T. G.

Bubu

O. M.

Flatt

J. D.

Heyn

P. C.

Cicero

E. C.

Zarina Kraal

Pushpalata Zanwar

Peterson

Kim

Turner II

R. W.

Viswanathan

Kulick

E. R.

Zuelsdorff

Stites

S. D.

Babulal

(2023). The structural and social determinants of Alzheimer’s disease related dementias. Alzheimer's and Dementia: The Journal of the Alzheimer's Association, 19(7), 3171–3185. https://doi.org/10.1002/alz.13027

Ailshire

J. A.

Mawhorter

S. L.

Choi

E. Y.

(2020). Contextual data resource (CDR): US decennial census and American community survey data, 1990-2018, Version 2.0. [Dataset]. USC/UCLA Center on Biodemography and Population Health.

Anderson

(2017). Racial residential segregation and the distribution of health-related organizations in urban neighborhoods. Social Problems, 64(2), 256–276. https://doi.org/10.1093/socpro/spw058

Besser

L. M.

McDonald

N. C.

Song

Kukull

W. A.

Rodriguez

D. A.

(2017). Neighborhood environment and cognition in older adults: A systematic review. American Journal of Preventive Medicine, 53(2), 241–251. https://doi.org/10.1016/j.amepre.2017.02.013

Besser

L. M.

Rodriguez

D. A.

McDonald

Kukull

W. A.

Fitzpatrick

A. L.

Rapp

S. R.

Seeman

(2018). Neighborhood built environment and cognition in non-demented older adults: The Multi-Ethnic Study of Atherosclerosis. Social Science & Medicine, 200(1), 27–35. https://doi.org/10.1016/j.socscimed.2018.01.007

Blom

(1958). Statistical estimates and transformed beta-variables [Stockholm College]. https://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-75457

Boraxbekk

C.-J.

Lundquist

Nordin

Nyberg

Nilsson

L.-G.

Adolfsson

(2015). Free recall episodic memory performance predicts dementia ten years prior to clinical diagnosis: Findings from the betula longitudinal study. Dementia and Geriatric Cognitive Disorders Extra, 5(2), 191–202. https://doi.org/10.1159/000381535

Chan

O. F.

Liu

Guo

Chui

C. H. K.

H. C.

Song

Cheng

Chiu

R. L. H.

Webster

Lum

T. Y. S.

(2023). Neighborhood built environments and cognition in later life. Aging & Mental Health, 27(3), 466–474. https://doi.org/10.1080/13607863.2022.2046697

Chen

Jiao

Farahi

(2023). Disparities in affecting factors of housing price: A machine learning approach to the effects of housing status, public transit, and density factors on single-family housing price. Cities, 140(1), Article 104432. https://doi.org/10.1016/j.cities.2023.104432

10.

Clarke

P. J.

Ailshire

J. A.

House

J. S.

Morenoff

J. D.

King

Melendez

Langa

K. M.

(2012). Cognitive function in the community setting: The neighbourhood as a source of ‘cognitive reserve. Journal of Epidemiology & Community Health, 66(8), 730–736. https://doi.org/10.1136/jech.2010.128116

11.

Clarke

P. J.

Weuve

Barnes

Evans

D. A.

Mendes de Leon

C. F.

(2015). Cognitive decline and the neighborhood environment. Annals of Epidemiology, 25(11), 849–854. https://doi.org/10.1016/j.annepidem.2015.07.001

12.

Clay

O. J.

Ball

K. K.

Wheeler

K. M.

Crowe

Marsiske

Dean

L. T.

Thorpe

R. J.

Jones

Owens

J. H.

Rebok

G. W.

Willis

S. L.

(2022). Evaluating social determinants of health domains and their predictive validity within black/African American and white older adults from the ACTIVE trial. Journal of Aging and Health, 11S-18S(9_suppl), 11S–18S. https://doi.org/10.1177/08982643221111205

13.

Diez Roux

A. V.

Mair

(2010). Neighborhoods and health. Annals of the New York Academy of Sciences, 1186(1), 125–145. https://doi.org/10.1111/j.1749-6632.2009.05333.x

14.

Finlay

Esposito

Langa

K. M.

Judd

Clarke

P. J.

(2022). Cognability: An Ecological Theory of neighborhoods and cognitive aging. Social Science & Medicine, 309(1), Article 115220. https://doi.org/10.1016/j.socscimed.2022.115220

15.

Finlay

Esposito

Colabianchi

Zhou

Judd

Clarke

P. J.

(2021). Neighborhood active aging infrastructure and cognitive function: A mixed-methods study of older Americans. Preventive Medicine, 150(1), Article 106669. https://doi.org/10.1016/j.ypmed.2021.106669

16.

Finlay

Esposito

Gomez-Lopez

Khan

Clarke

P. J.

Chenoweth

(2020). National neighborhood data archive (NaNDA): Arts, entertainment, and recreation organizations by census tract, United States, 2003-2017 [Dataset]. Inter-university Consortium for Political and Social Research (ICPSR). https://doi.org/10.3886/E115543V2

17.

Finlay

Clarke

P. J.

Judd

Esposito

(2021). Neighborhood cognitive amenities? A mixed-methods study of intellectually-stimulating places and cognitive function among older Americans. Wellbeing, Space and Society, 2(1), Article 100040. https://doi.org/10.1016/j.wss.2021.100040

18.

Frank

L. D.

Bigazzi

Hong

Minaker

Fisher

Raine

K. D.

(2022). Built environment influences on healthy eating and active living: The NEWPATH study. Obesity, 30(2), 424–434. https://doi.org/10.1002/oby.23352

19.

Hamlin

A. M.

Weigand

A. J.

Clay

O. J.

Marsiske

Wallace

Dadson

Thomas

K. R.

Clark

A. L.

(2024). The independent and interactive effects of economic stability and health care access on 10-Year cognitive trajectories of Black/African American and white older adults from the ACTIVE study. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 80(2), Article gbae196. https://doi.org/10.1093/geronb/gbae196

20.

Health and Retirement Study . (2024). Cross-Wave Geographic Information (Detail) restricted data set [Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740)]. https://hrs.isr.umich.edu/data-products/restricted-data/available-products/9706

21.

James III

R. N.

Sweaney

A. L.

(2010). Housing dissatisfaction and cognitive decline in older adults. Journal of Housing for the Elderly, 24(1), 93–106. https://doi.org/10.1080/02763890903547104

22.

Livingston

Huntley

Sommerlad

Ames

Ballard

Banerjee

Brayne

Burns

Cohen-Mansfield

Cooper

Costafreda

S. G.

Dias

Fox

Gitlin

L. N.

Howard

Kales

H. C.

Kivimäki

Larson

E. B.

Ogunniyi

Mukadam

(2020). Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet (London, England), 396(10248), 413–446. https://doi.org/10.1016/S0140-6736(20)30367-6

23.

Lorenzo-Seva

(2022). SOLOMON: A method for splitting a sample into equivalent subsamples in factor analysis. Behavior Research Methods, 54(6), 2665–2677. https://doi.org/10.3758/s13428-021-01750-y

24.

Lupien

S. J.

Lepage

(2001). Stress, memory, and the hippocampus: Can’t live with it, can’t live without it. Behavioural Brain Research, 127(1), 137–158. https://doi.org/10.1016/S0166-4328(01)00361-8

25.

Peterson

R. L.

George

K. M.

Tran

Malladi

Gilsanz

Kind

A. J. H.

Whitmer

R. A.

Besser

L. M.

Meyer

O. L.

(2021). Operationalizing social environments in cognitive aging and dementia research: A scoping review. International Journal of Environmental Research and Public Health, 18(13), 13. https://doi.org/10.3390/ijerph18137166

26.

Radloff

L. S.

(1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. https://doi.org/10.1177/014662167700100306

27.

Rajan

K. B.

Weuve

Barnes

L. L.

McAninch

E. A.

Wilson

R. S.

Evans

D. A.

(2021). Population estimate of people with clinical Alzheimer’s disease and mild cognitive impairment in the United States (2020–2060). Alzheimer’s and Dementia: The Journal of the Alzheimer's Association, 17(12), 1966–1975. https://doi.org/10.1002/alz.12362

28.

RAND HRS Longitudinal File 2020 (2024). RAND HRS Longitudinal File 2020 (Version 2). [Dataset]. RAND Center for the Study of Aging.

29.

Rohe

W. M.

Stewart

L. S.

(1996). Homeownership and neighborhood stability. Housing Policy Debate, 7(1), 37–81. https://doi.org/10.1080/10511482.1996.9521213

30.

Sharifian

Spivey

B. N.

Zaheed

A. B.

Zahodne

L. B.

(2020). Psychological distress links perceived neighborhood characteristics to longitudinal trajectories of cognitive health in older adulthood. Social Science & Medicine, 258(1), Article 113125. https://doi.org/10.1016/j.socscimed.2020.113125

31.

Sharkey

Faber

J. W.

(2014). Where, when, why, and for whom do residential contexts matter? Moving away from the dichotomous understanding of neighborhood effects. Annual Review of Sociology, 40(40), 559–579. https://doi.org/10.1146/annurev-soc-071913-043350

32.

Sheffield

K. M.

Peek

M. K.

(2009). Neighborhood context and cognitive decline in older Mexican Americans: Results from the Hispanic established populations for epidemiologic studies of the elderly. American Journal of Epidemiology, 169(9), 1092–1101. https://doi.org/10.1093/aje/kwp005

33.

Shih

R. A.

Glass

T. A.

Bandeen-Roche

Carlson

M. C.

Bolla

K. I.

Todd

A. C.

Schwartz

B. S.

(2006). Environmental lead exposure and cognitive function in community-dwelling older adults. Neurology, 67(9), 1556–1562. https://doi.org/10.1212/01.wnl.0000239836.26142.c5

34.

Small

M. L.

McDermott

(2006). The presence of organizational resources in poor urban neighborhoods: An analysis of average and contextual effects. Social Forces, 84(3), 1697–1724. https://doi.org/10.1353/sof.2006.0067

35.

Smith

R. J.

Lehning

A. J.

Dunkle

R. E.

(2013). Conceptualizing age-friendly community characteristics in a sample of urban elders: An exploratory factor analysis. Journal of Gerontological Social Work, 56(2), 90–111. https://doi.org/10.1080/01634372.2012.739267

36.

Sol

Morris

E. P.

Lee

J. H.

Zaheed

A. B.

Palms

J. D.

Scambray

Clarke

P. J.

Zahodne

L. B.

(2024). Histories of neighborhood socioeconomic status contribute to race differences in later-life cognition. Alzheimer’s and Dementia: The Journal of the Alzheimer's Association, 20(5), 3342–3351. https://doi.org/10.1002/alz.13786

37.

Vásquez

Botoseneanu

Bennett

J. M.

Shaw

B. A.

(2016). Racial/Ethnic differences in trajectories of cognitive function in older adults: Role of education, smoking, and physical activity. Journal of Aging and Health, 28(8), 1382–1402. https://doi.org/10.1177/0898264315620589

38.

Wahl

H.-W.

Lang

F. R.

(2003). Aging in context across the adult life course: Integrating physical and social environmental research perspectives. Annual Review of Gerontology and Geriatrics, 23(1), 1–33. https://doi.org/10.1891/0198-8794.23.1.1

39.

Wang

Zhao

Wang

Zhang

(2022). Examining the effects of the built environment on housing rents in the Pearl River delta of China. Applied Spatial Analysis and Policy, 15(1), 289–313. https://doi.org/10.1007/s12061-021-09412-4

40.

Zaheed

A. B.

Sharifian

Kraal

A. Z.

Sol

Hence

Zahodne

L. B.

(2019). Unique effects of perceived neighborhood physical disorder and social cohesion on episodic memory and semantic fluency. Archives of Clinical Neuropsychology, 34(8), 1346–1355. https://doi.org/10.1093/arclin/acy098

41.

Zuelsdorff

Larson

J. L.

Hunt

J. F. V.

Kim

A. J.

Koscik

R. L.

Buckingham

W. R.

Gleason

C. E.

Johnson

S. C.

Asthana

Rissman

R. A.

Bendlin

B. B.

Kind

A. J. H.

(2020). The Area Deprivation Index: A novel tool for harmonizable risk assessment in Alzheimer’s disease research. Alzheimer’s & Dementia, 6(1), Article e12039. https://doi.org/10.1002/trc2.12039

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB