The Association of Neighborhood Socioeconomic Status with Executive Function and Processing Speed in Cognitively Normal Mexican American Elders from the Health and Aging Brains Study: Health Disparities Cohort Free

Dementia is increasing in frequency worldwide. According to the Centers for Disease Control and Prevention (CDC), in the USA, the prevalence in adults over the age of 65 will rise from 5.6 million in 2014 to 14 million in 2060, with the largest increases in Latino/Hispanic (hereafter Hispanic) and African Americans [1]. Dementia is a group of disorders that influence the daily lives of individuals through changes in memory, behavior, language, and other cognitive abilities [2], with a significant economic burden for individuals, families, and society. Research suggests that many individual factors (e.g. age, education, diabetes, hypertension, obesity, among others) are associated with cognitive decline in elders [3, 4]. Recently, investigators have focused on social determinants of health that have a broader impact on individual’s health and specifically those that are a risk factor for cognitive decline [5, 6]. A growing body of research has shown that living in a deprived neighborhood affects physical health, increases cardiovascular morbidity and mortality rates, and has links with lower cognitive function regardless of individual level characteristics [5‒8]. Neighborhood deprivation is a key contributor of health inequalities seen in Hispanic and other minorities [9]. Multiple mechanisms have been proposed to explain the neighborhood effects on health. Neighborhood socioeconomic status (NSES) has been linked to psychosocial stress [10], poor health behaviors, low access to health care, dangerous environments [11], green space availability, and air pollution [12]. The mechanisms responsible for the link between NSES and cognition are not well understood. Living in a disadvantaged neighborhood is associated with a higher prevalence of cardiovascular risk factors [13]; these factors have a well-established association with brain health and cognition [14].

The area deprivation index (ADI), a validated measure of state and national neighborhood disadvantage in the USA, has been associated with cognitive impairment [15] and brain structural changes in Alzheimer’s disease (AD) signature regions [16]. Furthermore, in 2020, Powell et al. analyzed 447 autopsy samples and found that living in disadvantaged neighborhoods at the time of death was associated with increased neurofibrillary tangles and amyloid plaques, key pathological markers of AD [17].

The ADI has been associated with cognitive outcomes, particularly with executive functioning measured by the trail making test, part B [18, 19]. Socioeconomic status has been associated with lower processing speed [20]. Beck et al. [21] in 2018 used multiple mediation analyses to examine the direct and indirect effects of childhood socioeconomic status in multiple cognitive domains in midlife. They showed that men from lower child socioeconomic status performed worse during midlife in many cognitive measures including processing speed. Despite the rapid growth of the US Mexican American (MA) population, studies examining neighborhood disadvantaged and cognitive performance have primarily focused on non-Hispanic whites (NHWs), with few studies focusing on MAs [22, 23]. Research has shown that Hispanics living in the most disadvantaged areas have worse cognitive performance than other Hispanics living in the least deprived neighborhoods, and NHWs living in similar disadvantaged conditions [24]. This gap must be addressed to develop prevention, care, and treatment interventions that focus on this underrepresented population.

The current study examines whether NSES is cross-sectionally associated with performance on measures of executive functioning and processing speed among NHWs, and MAs from the Health and Aging Brain: Health Disparities Study (HABS-HD). We hypothesized that living in the most disadvantaged neighborhoods would be associated with poorer cognitive performance. Based on prior research [25], we also hypothesized that associations would be stronger in MAs than in NHWs.

Data from the HABS-HD, a community-based longitudinal study of cognitive aging, conducted by the Institute for Translational Research at the University of North Texas Health Science Center was analyzed. Participants are recruited from the Dallas/Fort Worth metropolitan area in Texas. The HABS-HD study inclusion criteria include age 50 and older, self-identified as MAs or NHWs, willingness and capable of undergoing blood draws and neuroimaging procedures, and fluency in Spanish or English. Exclusion criteria include Type 1 diabetes, current inflammatory conditions like lupus, current diagnoses of cancer, mental illness (except depression), a traumatic brain injury with loss of consciousness in the last year, and other severe medical conditions like end stage renal failure, which preclude participation. HABS-HD participants undergo a clinical interview, neuropsychological testing, functional examination, MRI of the head, amyloid PET scan, and blood draw for clinical and biomarkers analysis. All study components are conducted in English or Spanish based on participant preference. Validated Spanish versions of cognitive tests were used, and native Spanish speakers administered the test. A complete description of the HABS-HD study has been published elsewhere [26].

The HABS-HD study was approved by the North Texas Regional Institutional Review Board. Each participant, or legal representative for those with cognitive impairment, signs a written informed consent. All HABS-HD data is available to the scientific community through the UNTHSC Institute for Translational Research (ITR) website [27]. From March 2017 to June 2022, 2,380 subjects were enrolled in the study baseline visit. To analyze the relationship of ADI and cognition in a cognitively unimpaired cohort, 380 participants with a diagnosis of mild cognitive impairment (MCI), and 155 participants with an AD diagnosis were excluded. Of the remaining 1,845 participants, 316 were excluded due to insufficient demographic information (zip code) to obtain the ADI. Additionally, 217 participants who did not have complete neuropsychological tests data were excluded. The final sample of 1,312 met the following criteria: (1) provided zip code to obtain ADI, (2) underwent all tests of interest, (3) did not meet criteria for MCI or AD, and (4) had complete data on covariates including age, sex, education, ethnicity, and APOE4 status. Of the final sample, 652 were NHWs, 660 were MAs, 63.7% were female, and the mean age was 65.7 years old. Participants excluded from the analysis did not significantly differ from the final sample participants in age nor ADI scores. Participants excluded from the study had less education, performed worse in all cognitive measures, were less likely to be female, and had a higher percentage of APOE4 positivity (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000539035)

Cognitive diagnoses are assigned using an algorithm (decision tree) that is verified at consensus review. Unimpaired Cognition (UC) = no cognitive complaints, clinical dementia rating (CDR) sum of boxes score of 0 and cognitive test scores broadly within normal limits (i.e. performance no more than 1.5 standard deviations below the mean of the normative range on any test). For the current study, only those meeting the criteria for unimpaired cognition were included.

The Area Deprivation Index is based on 17 census variables including education, employment, income, occupation, and housing, to create a composite measure of neighborhood socioeconomic disadvantage [28]. The ADI allows for comparisons at the census-block group level, and is publicly available at https://www.neigborhoodatlas.medicine.wisc.edu/. For analysis, we used the national ADI percentile ranking as a continuous variable, with scores ranging from 1 to 100. A score/ranking of 1 indicates the lowest level of neighborhood disadvantage (least deprived), and a ranking/score of 100 indicates the highest level of neighborhood disadvantage (most deprived). ADI quartiles were used to assess the differences in baseline characteristics and cognitive scores, comparing the least deprived neighborhoods (quartile 1) to the most deprived neighborhood (quartile 4) within groups, and the differences in quartile 4 between groups.

The trail making test (TMT) is widely used to assess executive function and reflects a variety of mental abilities such as attention, mental flexibility, visual scanning, psychomotor speed, and abstraction [29]. The TMT consists of parts A and B. In part A, the participants connect a series of 25 numbers in circles in order, while in part B, circles contain numbers and letters that participants need to connect in numerical and alphabetical order, alternating between numbers and letters. The total time for completion is the variable of interest. Trail A primarily tests motor speed, and trail B is a test of higher-level cognitive skills [30]. The digit-symbol substitution test (DSST) assesses processing speed, attention, and visuoperceptual functions [31]. It has been shown that executive functioning also contributes to the performance in this test [32].

Age, sex, total years of education, and APOE4 status were entered as covariates. Participants self-reported age, sex, ethnicity, and total years of education during the interview.

To facilitate effect size comparisons between variables, we used neuropsychological tests z-scores obtained from normative data for Texas MA Adults [33]. The signs of z-scores for trails A and B, where a higher score indicates a worse outcome, were inverted before statistical comparisons with ADI were performed. After this transformation, a higher z-score indicated a better performance for all tests. Following the transformation of the scores, the cross-sectional association between ADI and the cognitive measures was examined.

Baseline differences between MAs and NHWs were analyzed using t tests for continuous variables, and χ² tests for categorical variables. Previous research showed that individuals in the top 20% most deprived neighborhoods had the most impact on health outcomes [28]. Therefore, the differences of baseline characteristics between participants living in the least deprived neighborhoods (ADI quartile 1) and those living in the most deprived neighborhoods (ADI quartile 4) within ethnic groups were compared. The differences between ethnic groups among participants in ADI quartile 4, were also analyzed using t-tests and chi-squared tests.

Moderation analysis was conducted using PROCESS macro (Hayes 2013) to assess the effect of the interaction of ADI and ethnicity on cognitive scores. Then, after checking for linearity (Fig. 1-3), a hierarchical linear regression, stratified by ethnicity, was performed to assess the association between ADI as a continuous variable, and each cognitive test score as a separate outcome. Model 1 was adjusted for age and APOE4, Model 2 was adjusted for age, APOE4, and sex, and Model 3 was adjusted for age, APOE4, sex, and education.

The total analyzed sample of 1,312 participants (n = 652 – NHW and n = 660 – MA) had a mean age of 65.7 ± 8.4 years, and 64% were female. Baseline characteristics are presented in Table 1. MAs were significantly younger, less educated, had higher ADI scores, and had a higher proportion of female participants when compared to NHWs. MA participants performed worse in trail B. A higher percentage of NHWs were APOE4 carriers.

Table 2 shows the differences between ADI quartiles 1 and 4 within groups. NHWs living in the most disadvantaged neighborhoods (ADI quartile 4) had fewer years of education, and scored worse in trails A and trails B, when compared with those living in the least deprived neighborhoods (ADI quartile 1). For MAs, those living in ADI quartile 4 were older, less educated, and had lower scores in trails A, trails B, and DSS, than their counterparts in ADI quartile 1.

The ethnic group differences between NHWs and MAs living in the most deprived neighborhoods (Table 3) showed that a higher percentage of MAs live in the most disadvantaged neighborhoods (χ² = 312.93, 95% CI = 42.12–51.14, p = <0.01). MAs in ADI quartile 4 were younger, less educated, and had lower scores in DSS, when compared with NHWs in ADI quartile 4.

Moderation analysis (online suppl. Table 2) showed that the interaction between ADI and ethnicity was not significant for trail A and DSS (p > 0.05). However, the interaction between ADI and ethnicity was significant for trails B (b = −0.01, SE = 0.002, t = −3.86, p < 0.01). The simple slope of ADI on trail B was significant for Hispanic race (b = −0.013, SE = −0.001, t = −7.37, p < 0.01) but not for NHW race (b = 0.003, SE = 0.001, t = −1.92, p = 0.054).

Table 4 presents results from three linear regression models for the cross-sectional association between ADI and transformed cognitive scores. After adjusting for age, APOE4 status, sex, and education, ADI did not predict cognitive scores in NHWs. In MAs, the relationship between ADI and trails A (B = −0.004, p < 0.015), trails B (B = −0.005, p = 0.017), and DSS (B = −0.003, p = 0.02) remained significant after adjusting for covariates.

This cross-sectional data analysis demonstrated an association between neighborhood disadvantage and poorer cognition focused on aspects of executive function and processing speed in this bi-ethnic cohort. Both NHWs and MAs living in the most disadvantaged neighborhoods performed worse on trails A, trails B, and the digit symbol substitution. These findings are consistent with previous work examining the association of neighborhood disadvantage and cognition [19, 34]. However, when NHWs were compared to MAs living within the same level of neighborhood deprivation (ADI quartile 4), performance on DSS was the only measure that showed a significant difference between the ethnic groups.

For the two groups, the relationship between cognition and ADI was differentially impacted by demographic factors and the presence of APOE4. After adjusting for covariates (demographic variables and APOE4 status), ADI was associated with lower scores in executive function and processing speed only in MAs. These findings did not corroborate previous research in NHW populations. In a cross-sectional analysis of the English Longitudinal Study of Ageing, Lang et al. [8] concluded that higher neighborhood socioeconomic deprivation was associated with lower cognitive functioning. Similar results were reported by Zuelsdorff et al. [28] in a sample composed of 88.6% NHWs. Researchers found that subjects living in the 20% most disadvantaged neighborhoods performed lower in many cognitive domains, including executive function. The small proportion of NHW participants within the most disadvantaged neighborhood (ADI quartile 4), may have prevented finding a potential relationship between ADI and cognition in this ethnic group. Another consideration is the low power due to small sample size, which may explain the difference in findings between MAs and NHWs. A new study reported that low-income white people tend to live in better neighborhoods than middle-class Hispanic [35], which may explain the lower percentage of NHWs living in ADI quartile 4. Even at the same ADI level (quartile 4), MAs had significantly fewer years of education, and higher prevalence of CVRF like diabetes, both of which are related to cognition. These differences in individual factors may explain some of the differences found between ethnic groups. Phelan and Link suggested that it is not only socioeconomic status that is a cause of health disparities, but that ethnicity also plays a role in health and mortality [36]. Racial discrimination produces stress and could contribute to deficits in cognition. Researchers have found evidence showing minorities perceived stress as a mediator between SES and lower cognitive scores in multiple cognitive domains [37]. Negative effects in executive function have been reported in individuals who experienced or observed subtle discrimination which contributes to academic, employment, and health inequalities [38].

Many potential pathways for neighborhood deprivation to impact cognition have been postulated. The lack of recreation centers, parks, and green spaces could promote lower physical activity [39]. Lower social activity and support, and higher levels of anxiety and depression as a result of the absence of cultural centers, community centers, and the presence of social stressors (e.g. violence, substance abuse) may increase the risk of cognitive impairment [40]. Cardiovascular risk is influenced by the socioeconomic environment and is associated with an increased risk of dementia [41]. In an elderly cohort, socioeconomic factors such as income, education, and a measure of neighborhood disadvantage, were associated with asymptomatic cardiovascular disease, independent of age, gender, and sex [42]. A novel mechanistic pathway, epigenetic age, has been studied. Smith et al. [43] in 2017 concluded that neighborhood status may influence methylation and gene expression, even after controlling for individual socioeconomic factors. The weathering hypothesis as described by Simmons and colleagues, whereby chronic exposure to social and economic disadvantage accelerates adverse health outcomes through DNA methylation changes, may help in the understanding of the results from this study [44]. Currently, the HABS-HD study is capturing genomic data that will allow us to assess the role of epigenetic factors in the association of ADI and cognition. More research is needed to elucidate the pathophysiological pathways through which neighborhood disadvantage affects cognition across the individual life span.

Limitations of this study include the cross-sectional design; thus, causal relationships must be interpreted with caution. Residential history and changes in the living environment through life were not considered, so the association of early-life socioeconomic disadvantage with poor cognitive health was not accounted for. As the HABS-HD study is longitudinal, and is adding a younger cohort, future research will investigate the impact of these variables over time. Other possible confounding factors like income, smoking status, physical activity, diet, kidney disease, among others, need to be addressed in future research. As explained above, the percentage of NHW participants living in the most deprived neighborhoods is small when compared to MAs, which may have affected comparisons. We excluded participants with MCI and AD diagnoses which may have contributed to bias in the associations. Finally, we did not do a sensitivity analysis imputing covariate missing data, and this may affect the results.

This study in a bi-ethnic sample provides evidence of the association of neighborhood disadvantage and cognitive performance. Residing in a neighborhood with higher deprivation was associated with lower cognitive scores in MAs but not in NHWs. This is important as processes that affect neighborhood status and cognition may differ for MAs when compared to other ethnic groups. Cognitive risk is not only a result of behavior or personal traits, but also where individuals live. Findings highlighted the need for preventive strategies and interventions targeting neighborhood conditions and modifiable risk factors of at-risk populations like MAs. Future research that includes other cognitive domains and individual socio-economic factors is warranted to study the longitudinal relationship of NSES and the increased risk for cognitive decline.

The research team thanks the local Fort Worth community and participants of the HABS-HD study.

This study protocol was reviewed and approved by the North Texas Regional Institution Review Board approval number 2016-128. Each participant signed a written informed consent.

L.A.J. has a financial interest in CX Precision Medicine, Inc. S.E.O. has multiple patents on precision medicine for neurodegenerative diseases and is the scientific founder of CX Precision Medicine, Inc. RV, AB, RM, and JH declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Research reported in this publication was supported by the National Institute on Aging of the National Institutes of Health under Award Numbers R01AG054073, R01AG058533, P41EB015922, U19AG078109, and R35AG071916. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

R.V. and J.H. conceived the idea. R.V., A.B., and R.M. equally contributed to drafting the manuscript. J.R.H., L.A.J., and S.E.O. provided critical review, editing, and input. All authors contributed to the article and approved the submitted version.

All HABS-HD data are available to the scientific community through the UNTHSC Institute for Translational Research (ITR) website: https://apps.unthsc.edu/itr/. Further inquiries can be directed to the corresponding author.