Depressive Symptoms but not Chronic Stress Mediate the Link Between Income and Cognition in Latino and Black Older Adults: Findings from the HABS-HD Study

Abstract

Objectives:

The mechanisms linking economic resources to cognition in older ethnic/racial minorities remains unclear. This study examined whether depressive symptoms and chronic stress mediate the link between income and both episodic memory and executive functioning in Black and Latino older adults.

Methods:

We analyzed data from 579 Black and 1,056 Latino older participants in the Health and Aging Brain Study-Health Disparities. Mediation models estimated the total, direct, and indirect effects of income on memory/executive functioning through depressive symptoms and chronic stress.

Results:

Among Black adults, depressive symptoms significantly mediated the association between income and both memory (B = 0.025, CI [0.007, 0.052]) and executive functioning (B = 0.036, CI [0.016, 0.069]), while chronic stress did not significantly mediate the link between income and memory (B = 0.003, CI [−0.019, 0.025]) or executive functioning (B = 0.009, CI [−0.014, 0.036]). For Latino adults, depressive symptoms also significantly mediated the link between income and both memory (B = 0.025, CI [0.010, 0.044]) and executive functioning (B = 0.021, CI [0.008, 0.040]), whereas chronic stress was not a significant mediator for the income on memory (B = −0.002, CI [−0.011, 0.004]) or executive functioning (B = 0.003, CI [−0.005, 0.012]) associations.

Discussion:

These initial results suggest that the development of interventions targeted towards depressive symptoms in Black and Latino older adults may help promote well-being and alleviate the detrimental consequences of income inequalities on cognition in older adulthood.

Introduction

In the United States (U.S.), African American/Black and Hispanic/Latino/Latine (hereafter Black and Latino) older adults are the two largest ethnic/racial minoritized populations who also experience elevated rates of Alzheimer’s disease and related dementias (ADRD) compared to their non-Latino white (NLW) counterparts (Matthews et al., 2019). The ethnic/racial disparities in ADRD rates experienced by Black and Latino older adults may be directly influenced by economic inequalities (Li et al., 2023) that decrease their opportunity to access ADRD prevention and intervention treatments (Petersen et al., 2021). As such, it has become critical to identify modifiable mechanisms that may help alleviate the impact of enduring economic inequalities among these two ethnic/racial minoritized populations. The National Institute on Aging (NIA) introduced a Health Disparities Research Framework that includes levels of influence (i.e., environmental, sociocultural, behavioral, biological) that may contribute to health disparities among older minoritized populations, and emphasized that the discovery of underlying mechanisms linked to the risk of these disparities may be more precisely captured if multiple levels of influence are examined concurrently (Hill et al., 2015). Although there may be multiple contributing factors to ADRD risk among Black and Latino older adults (Livingston et al., 2024), depressive symptomatology and chronic stress are consistently higher among Latino and Black adults compared to NLW adults (Brown et al., 2020; Hooker et al., 2019), are risks of poor cognition in late-life (Jimenez et al., 2024; Zahodne et al., 2014a), and are consequences of economic inequalities (Lantz et al., 2005). This study examined the role of depressive symptoms and chronic stress as potential mechanisms through which economic inequalities may operate to impact cognitive functioning in Black and Latino older adults.

Socioeconomic status (SES) is a measure of individual-, household-, and/or community-level social and economic position that reflects inequalities in access to resources and opportunities across the life course (Braveman et al., 2005). Both individual SES indicators (e.g., occupation, education, income, wealth) and the aggregate of these indicators in adulthood have been linked with cognitive function in adulthood (Filigrana et al., 2023; González et al., 2013, 2015). There is also evidence to suggest that adult SES may be a pertinent contributor to cognition above and beyond the effects of childhood SES (Filigrana et al., 2023; González et al., 2013). Despite this, SES indicators are often included as covariates to account for their potential confounding effects (Morris et al., 2022) or not directly measured with race/ethnicity instead used as a proxy (Zahodne et al., 2019; Zaheed et al., 2021).

The measurement of SES is complex and there are few widely used and validated SES composite measures (Braveman et al., 2005), thereby contributing to the ongoing debate regarding the best approach to characterize relevant socioeconomic resources that impact health outcomes in older adulthood (Braveman et al., 2005). Research indicates that, although SES indicators are often used interchangeably, they confer differential economic resources across racial/ethnic populations (Braveman et al., 2005). For example, Braveman and colleagues (2005) found that among Black, NLW, and Latino adults of Mexican origin/descent who had comparable educational levels, Black and Latino adults had lower average incomes when compared to their NLW counterparts across all educational levels. The contribution of SES indicators to health outcomes also varies depending on the health outcome measured, resulting in inconsistent findings about their impact on health outcomes (Braveman et al., 2005). Research suggests that these inconsistencies may be partially due to the distinct pathways through which specific SES indicators influence different health outcomes (Braveman et al., 2005). In the context of ADRD research, higher educational and occupational attainment are thought to be linked with lower risk of ADRD through enhanced cognitive maintenance (i.e., cognitive reserve; Stern et al., 1999). Among minoritized populations, years of education or attainment may not fully capture the inequitable distribution of economic resources that contribute to differences in educational quality and ultimately to cognitive health (e.g., Manly et al., 2002). Income and wealth are economic resources that may be critical to consider in the context of cognitive health among Black and Latino older adults, given their direct link to risk of ADRD (Walsh et al., 2025) and psychological and physiological health outcomes that contribute to risk of ADRD (Lantz et al., 2005; Lorant, 2003). More importantly, economic resources have direct effects on opportunity to access ADRD prevention and intervention resources (Petersen et al., 2021), and Black and Latino adults also face significantly higher economic burden related to ADRD when compared to their NLW counterparts (Mudrazija et al., 2025). Black and Latino adults also report lower economic resources (i.e., income and wealth) compared to NLW older adults, which are likely attributable to systemic inequalities that impede their opportunity towards upward economic mobility. Nonetheless, economic inequalities among Black and Latino populations in the U.S. are long-standing and therefore may be more challenging to directly intervene on. Thus, it is crucial that we identify population-specific modifiable mechanisms that may be more immediate targets of intervention to help mitigate the consequences of economic inequalities on cognitive function among these underserved groups.

Chronic stress may be one potential modifiable mechanism linking economic inequalities to cognition in adulthood (Muñoz et al., 2021; Marquine et al., 2022; Lantz et al., 2005). Chronic stress, or the prolonged exposure to stress, results in adverse physiological changes (e.g., hypercortisolism, chronic elevation of inflammatory cytokines; McEwen, 1998; Liu et al., 2017) that consequently increase susceptibility to neurodegeneration and other conditions (Liu et al., 2017) associated with heightened risk of ADRD. Chronic stress burden is a commonly used measure of chronic stress that captures the accumulation of multiple ongoing life stressors (Bromberger & Matthews, 1996; Muñoz et al., 2021). Some research indicates that elevated chronic stress burden is linked to worse verbal learning but not psychomotor speed or verbal fluency (Marquine et al., 2022; Muñoz et al., 2021). In relation to economic inequality, research indicates that individuals who are economically disadvantaged may be more susceptible to chronic stress burden (Lantz et al., 2005). Importantly, economically disadvantaged individuals—particularly those from ethnically/racially minoritized populations—may experience “weathering”, an accelerated decline in health resulting from the cumulative effects of chronic stress (Geronimous et al., 1992), which may increase their risk of developing chronic illnesses linked to poor cognitive health (Troxel et al., 2003).

Depressive symptomatology may be another potential pathway linking low economic resources to poor late-life cognition, likely through feelings of hopelessness, apathy, and anhedonia related to limited access to material resources (Lorant, 2003). For example, in a predominately Black (77%) and white (16%) older adult sample, depressive symptoms significantly mediated the link between different forms of economic disadvantage (e.g., debt, poverty) and cognitive performance based on a brief cognitive screener collected over the telephone (Zhou et al., 2021). In another older adult sample based in China, depressive symptoms significantly mediated the association between self-reported income-expenditure status and a brief cognitive screener (Yuan et al., 2020). Additionally, elevated depressive symptoms are associated with lower baseline learning and memory, attention, executive functioning, processing speed, and memory decline (Shimada et al., 2014; Zahodne, et al., 2014a). The mechanisms linking elevated depressive symptoms to increased risk of ADRD are likely multifactorial, potentially involving HPA axis dysfunction secondary to chronic stress, or elevated depressive symptoms as direct triggers to HPA axis dysfunction (Dafsari & Jessen, 2020). Alternatively, elevated depressive symptoms may also increase the risk of ADRD via prolonged neurotransmitter imbalances (Dafsari & Jessen, 2020), or negative behavioral changes such as appetite and weight loss (Saha et al., 2016).

Although most of the research linking elevated depressive symptoms and stress to poor cognition has involved predominately NLW samples, there is research indicating that Black and Latino older adults may be more vulnerable to negative cognitive outcomes due to these risk factors compared to their NLW counterparts. For example, Black older adults report higher depressive symptoms compared to their NLW counterparts (Hooker et al., 2019). Research also demonstrates that elevated depressive symptoms contribute to lower baseline cognition and decline among Black older adults. For example, Zahodne and colleagues (2014a), found that elevated depressive symptoms were linked with worse episodic memory, task-switching, and inhibition among Black older adults but not among their NLW counterparts (Zahodne et al., 2014a). Another study found that the link between higher depressive symptoms and lower executive functioning was exacerbated by increasing age for Black and Latino older adults but not for their NLW counterparts (Niu et al., 2024). The existing evidence suggests that Black older adults experience a higher number of ongoing life stressors than their NLW counterparts (Morris et al., 2022). Regarding cognitive aging, one longitudinal study of Black older adults demonstrated that higher levels of stress were linked with more rapid declines in memory, learning, visuospatial functioning, and global cognitive functioning (Turner et al., 2017).

Existing evidence also demonstrates that Latino older adults report higher levels of depressive symptoms than both Black and NLW older adults (Hooker et al., 2019). Elevated depressive symptoms have been linked with poor baseline levels and greater decline in memory among Latino older adults (Jimenez et al., 2024). In relation to stress, there is evidence that higher levels of chronic stress and related stress measures are linked with lower baseline levels of cognitive functioning among middle-aged and older Latino adults (Marquine et al., 2022; Muñoz et al., 2021). In a large sample of Latino adults (n = 3,265), higher chronic stress burden was linked with worse verbal learning and other measures of stress were negatively associated with verbal memory, word fluency, and psychomotor speed measures (Muñoz et al., 2021). In another study leveraging the same sample, Marquine and colleagues (2022), found that higher stress was linked with lower performance on a measure of global cognition that included aspects of episodic learning and memory, language, and processing speed (Marquine et al., 2022).

Research suggests that depressive symptoms and stress may partially explain differences in cognitive functioning across racial/ethnic groups (Letang et al., 2021; Zaheed et al., 2021; Zahodne et al., 2019). For example, in a mixed-race older adult sample (age range = 65–104; n = 1,313 Black, n = 793 Latino), depressive symptoms significantly mediated the link between Latino ethnicity and episodic memory performance at baseline but did not significantly mediate the association between Black race and memory function at baseline (Zahodne et al., 2019). In a sample of Black and NLW middle to older adults (age range = 51–101; 21.14% Black), chronic stress explained 1.14% of the differences in baseline memory but not rate of change in memory between Black and NLW older adults (Morris et al., 2022). In a cross-sectional study of NLW, Black, and Latino young adults (age range = 22–36; 16.19% Black, 7.11% Latino), independent stress and SES paths significantly mediated the link between Black race and episodic memory, working memory, and cognitive flexibility at baseline (Letang et al., 2021). Interestingly, for Latino young adults, the stress-only pathway, but not the independent SES pathway, was a significant mediator between Latino ethnicity and episodic memory, working memory, and cognitive flexibility (Letang et al., 2021). These studies consistently utilize race/ethnicity as proxies for SES and, in some cases, demonstrate that SES serves as an independent mechanism that explains cognitive health disparities among ethnically/racially minoritized younger adults. Direct measurements of SES such as income are linked to an individual’s opportunity to access an array of health-promoting resources and thus have been theoretically conceptualized as “fundamental causes” of disease that triggers the initial cascade of negative health outcomes (Link & Phelan, 1995). It is possible that relying solely on race/ethnicity as a predictor and proxy of SES may obscure unique population differences and important nuances captured by direct measurements of SES. Directly examining SES indicators as primary drivers of cognitive functioning in older adults, particularly within ethnic/racial minoritized groups, may provide valuable insights into the distinct mechanisms relevant to each population. This approach may contribute to the development of culturally tailored interventions that address the specific needs of individual ethnically/racially diverse older populations.

Examining the potential mediating role of stress and depressive symptoms on episodic memory and executive functioning—two cognitive domains often affected in the early stages of AD—may be particularly important among Black and Latino older adults. Indeed, in our ongoing efforts to identify relevant modifiable mechanisms underlying health disparities among ethnic/racial minoritized populations, it is crucial to consider race/ethnicity in the context of SES, given that these two social constructs may intersect and differently contribute to health outcomes throughout the life course (Zahodne et al., 2016). Thus, the current study seeks to investigate whether chronic stress and depressive symptoms mediate the link between economic resources and cognition for both Latino and Black older adults. Existing literature demonstrates that Black older adults report higher levels of stress (Zaheed et al., 2021; Zahodne et al., 2019) compared to Latino older adults, and Latino older adults report higher depressive symptoms than Black older adults (Zaheed et al., 2021; Zahodne et al., 2019). As a result, we hypothesized that chronic stress would mediate the link between income and cognition for Black older adults, and depressive symptoms would mediate the income-cognition link for Latino older adults. Findings from this study may help identify mechanisms for developing culturally appropriate interventions to improve well-being among ethnic/racial minority populations and, in turn, help minimize the impact of low economic resources on cognitive functioning in older adulthood.

Methods

Participants and Procedures

In this secondary analysis, we leveraged cross-sectional data from the Health and Aging Brain Study—Health Disparities (HABS-HD) at the University of North Texas Health Science Center (UNTHSC) which uses a non-probability sampling approach and community-based participatory research method to collect participants. The HABS-HD study was approved by the UNTHSC Institutional Review Board.

Eligibility for the larger HABS-HD study were as follows: 50 years or older, no type 1 diabetes, cancer within the past 12 months, or alcohol/substance abuse disorder diagnoses; no severe medical conditions, mental illness, or traumatic brain injury that resulted in a loss of consciousness within the past 12 months. A more comprehensive description of the HABS-HD methodology has been previously published (O’Bryant et al., 2021).

Analytic Sample

Baseline data for 2,001 HABS-HD participants were accessed on 09/10/2024. The final analytic sample included 1,635 older adults. Inclusion criteria for the current study included the following: participants were without dementia as defined by a Clinical Dementia Rating Scale Global Score (CDRS-GS) ≤ .50 (Morris, 1997) and had to have at least one cognitive factor score (See Supplementary Figure 2). Self-identified race and ethnicity was established using the NIH guidelines consistent with the U.S. Census Bureau. Participants who self-reported Black race could be of mixed-racial background or of any ethnicity other than Latino, and participants who self-reported Latino ethnicity could be from any racial background.

Measures

Cognitive Outcomes

Individual memory and executive functioning test scores were converted to z-scores and used as indicators of separate latent memory and executive functioning variables. The memory latent variable consisted of the total learning and delayed recall trials from the Spanish-English Verbal Learning Test (González et al., 2002). The executive functioning latent variable consisted of the total scores from Digit Span Backwards (Wechsler et al., 1997), Digit Symbol Substitution (Wechsler, 1981), Trail Making Test Parts A & B (Reitan et al., 1956) and the animals category fluency test (Spreen & Straus, 1998). Trails A and B were log transformed to address skewness and reversed coded so higher numbers indicated better performance.

Income

Self-reported annual household income in U.S. dollars was utilized as an economic resource indicator and was treated as a continuous variable. Raw income data is presented in Table 1 for interpretability, but the log-transformed income variable is used for all other analyses given the high right skew.

Table 1.

Demographic Characteristics by Latino and Black Race/Ethnicity for the Final Analytic Sample

Variable	Black (n = 579)	Latino (n = 1,056)	W/t/χ2	p	Cohen’s d
Demographic
Age, M (SD)	62.84 (7.91)	62.86 (7.82)	−0.05	0.96	--
Years of education, M (SD)	15.00 (2.63)	10.23 (4.63)	26.58	<.001	1.27
Female, n (%)	379 (65%)	705 (67%)	0.28	0.59	--
Annual income, $, Median (IQR)	60,000 (27,000–100,000)	28,248 (14,400–50,000)	421,586	<.001	0.47
Modifiable Risk Factors
Geriatric depression scale, M (SD)	5.32 (5.13)	6.40 (6.03)	−3.81	<.001	−0.19
Chronic stress total, M (SD)	8.33 (7.28)	7.13 (6.52)	3.29	.001	0.17
Cognition
Memory	0.07 (0.96)	−0.04 (0.93)	2.31	0.02	0.12
Executive functioning	0.24 (0.84)	−0.14 (0.96)	7.87	<.001	0.42

Note. SD = standard deviation; M = mean; IQR = interquartile range; post-hoc tests were used to explore significant omnibus tests. Based on results from two sample t-test, Wilcoxon rank sum, Welch’s two sample t-test, and Pearson’s Chi squared. Missing data for the overall sample: 0.18% GDS, 3.91% memory and 3.91% executive functioning. Memory and executive functioning based on our two cognitive outcomes.

Depressive Symptoms

Depressive symptoms were measured continuously via the Geriatric Depression Scale (GDS-30). The GDS-30 is a continuous self-report depression screening tool developed for older adults, and therefore primarily focuses on affective and cognitive symptoms rather than somatic symptoms (Yesavage et al., 1982). It can range from 0–30 and higher GDS-30 scores reflect greater depressive symptoms. There is some evidence that it has comprable sensitivity and specificity to the shorter GDS-15, and similar validity indices as the Center for Epidemiological Studies Depression scale, another commonly used measure among community-based samples (Wancata et al., 2006).

Chronic Stress

Similar to previous studies among middle-aged and older Latino and Black adults, we used the Chronic Stress Burden Scale to measure chronic stress (Bromberger & Matthews, 1996; Muñoz et al., 2021). The Chronic Stress Burden Scale is an eight-item questionnaire (range 0–40) that captures the cumulative impact of chronic stress. Participants indicate current stressors in various life domains (e.g., “Have you experienced ongoing financial strain?”) that have persisted for a minimum of 6 months and rate each stressor as not very stressful, moderately stressful, and very stressful. Each stressor and appraisal rating were then summed into a total score to assess overall burden of stress (0–8). All analyses included chronic stress as a continuous variable.

Covariates

Age, self-reported gender (0 = male, 1 = female), and years of education were included as covariates in each path to remove their potential confounding effects on chronic stress, depressive symptoms, and the cognitive outcomes given the substantial theoretical evidence linking these factors to our indicators and outcomes.

Analytic Strategy

We analyzed data using RStudio (version 4.2.2). All continuous variables were z-score standardized for inclusion in Structural Equation Modeling (SEM). In structural equation models, missing data were handled using full information maximum likelihood with all available data. SEM was used to explore Confirmatory Factor Analysis (CFA) and the mediation effect linking income (log transformed) to the latent cognitive variables through depressive symptoms and chronic stress, respectively (see supplementary materials for additional details on analyses).

Sociodemographic, depressive symptoms, chronic stress, and cognitive variables for the entire sample and for each ethnic/racial group were computed. Two sample t-test, Wilcoxon rank sum, Welch’s two sample t-tests, and Pearson’s chi-squared tests were used to compare group differences, and effect sizes were calculated to identify the magnitude of all reported significant results.

Structural Equation Modeling

The initial two-dimensional CFA model had adequate model fit (CFI = 0.976, RMSEA = 0.080, SRMR = 0.036). After establishing a measurement model, regression paths were specified to the cognitive latent variables. Four separate mediation models that included covariates of interest tested the hypothesized associations (Figure 1), between the predictor (income), mediators (chronic stress, depressive symptoms), and latent variables (memory and executive functioning). More specifically, we used two separate models for Black older adults and Latino older adults to investigate potential differences in the associations between income, chronic stress, depressive symptoms, and the latent cognitive variables.

Figure 1. Theoretical Diagram of All Paths Estimated by the Structural Equation Model.

Note. Models A and B were fit separately for Latino and Black older adults, for a total of four models. EF = Executive Function (a) Model examining the direct effect from income to memory (Path c1)/EF (Path c2) and indirect effect through depressive symptoms (Path a*Path b1 for memory; Path a*Path b2 for EF). (b) Model examining the direct effect from income to memory (Path c1)/EF (Path c2) and indirect effect through chronic stress (Path a*Path b1 for memory; Path a*Path b2 for EF). The c’ paths represent the strength of the association between income and memory/executive functioning after controlling for indirect effects through depressive symptoms/chronic stress. Age, gender, and years of education covariates are not displayed for simplicity.

Model fit was determined via the Root Mean Square Error of Approximation (RMSEA), Comparative Fit Index (CFI) and Standardized Root Mean Square Residual (SRMR). Adequate model fit was defined as RMSEA < 0.08, CFI ≥ 0.95 and SRMR < 0.08. To address potential nonnormality in the data we used a Maximum Likelihood Estimation with Robust Huber-White Standard Errors (MLR). Bootstrapping was conducted until 5,000 or more samples were obtained to establish 95% confidence intervals. A significant indirect effect was established if the confidence intervals did not include zero.

Results

Descriptive Statistics

Race/ethnicity stratified demographic and health characteristics are reported in Table 1. Compared to Black older adults, Latino older adults had significantly lower years of education and annual median income, reported significantly higher depressive symptoms and lower total chronic stress. Latino older adults also had lower memory and executive functioning scores compared to Black older adults (ps < .05; Cohen’s d values ranged from −0.19–1.27).

The majority of older Latino adults reported being of Mexican origin/descent (n = 1,013). Other backgrounds reported were Puerto Rican (n = 9), Cuban (n = 2), and other, not specified (n = 32).

Mediation Analyses

Table 2 represents the total, direct, and indirect effects of income on memory and executive functioning via depressive symptoms and chronic stress. Among Black older adults, both the models for depressive symptoms (Robust CFI = 0.959, Robust RMSEA = 0.067, SRMR = 0.033) and chronic stress (Robust CFI = 0.957, Robust RMSEA = 0.069, SRMR = 0.034) as mediators had adequate model fit. Regarding the depressive symptoms mediation model, there was a significant total effect (c) of income on episodic memory (B = 0.113, SE = 0.056, p = 0.043) and executive functioning (B = 0.265, SE = 0.089, p = 0.003). For path a, higher income was significantly linked with lower depressive symptoms (B = −0.157, SE = 0.041, p < .001). For paths b, lower depressive symptoms were linked with higher memory (B = −0.157, SE = 0.053, p = 0.003) and executive functioning (B = −0.234, SE = 0.057, p < .001). Bootstrapped analysis indicated that depressive symptoms was a significant mediator for the link between income and memory (B = 0.025, [0.007, 0.052]) and executive functioning (B = 0.0367 [0.016, 0.069]).

Table 2.

Total, Direct and Indirect Effects of Income on Cognitive Outcomes for Black and Latino Older Adults

	Black
	Indirect (a × b)	Direct (c')	Total (c' + a × b)
Income à depressive symptoms à memory	0.025, [0.007, 0.052]	0.088 (0.055)	0.113 (0.056) *
Income à depressive symptoms à executive functioning	0.036 [0.016, 0.069]	0.229 (0.086) **	0.265 (0.089) **
Income à chronic stress à memory	0.003 [−0.019, 0.025]	0.109 (0.055) *	0.112 (0.055) *
Income à chronic stress à executive functioning	0.009 [−0.014, 0.036]	0.251 (0.088) **	0.260 (0.088) **
	Latino
	Indirect (a × b)	Direct (c')	Total (c = c' + a × b)
Income à depressive symptoms à memory	0.025 [0.010, 0.044]	0.093 (0.044) *	0.117 (0.043) *
Income à depressive symptoms à executive functioning	0.021 [0.008, 0.040]	0.183 (0.051) ***	0.204 (0.051) ***
Income à chronic stress à memory	−0.002 [−0.011,0.004]	0.119 (0.042) **	0.116 (0.043) **
Income à chronic stress à executive functioning	0.003 [−0.005, 0.012]	0.200 (0.051) ***	0.202 (0.050) ***

Note. Mediation models controlling for covariates: age, years of education, and gender are presented. Path a (not shown): income → depressive symptoms/chronic stress; Path b (not shown): depressive symptoms/chronic stress → memory/executive functioning; Path c’: income + depressive symptoms/chronic stress → memory/executive functioning; Path c: income → memory/executive functioning; Path a × b: income → depressive symptoms/chronic stress → memory/executive functioning.

^*p < .05.

^**p < .01.

^***p < .001.

For the chronic stress mediation model, there was a significant total effect (c) of income on memory (B = 0.112, SE = 0.055, p = 0.043) and executive functioning (B = 0.260, SE = 0.088, p = 0.003). For path a, higher income was significantly associated with lower chronic stress (B = −0.256, SE = 0.053, p <.001). For paths b, lower chronic stress was linked with higher memory (B = −0.011, SE = 0.041, p = 0.787) and executive functioning (B = −0.036, SE = 0.046, p = 0.430), although these associations were not statistically significant. Bootstrapped analysis indicated that chronic stress was not a significant mediator in the association between income and memory (B = 0.003 [−0.019, 0.025]) and executive functioning (B = 0.009 [−0.014, 0.036]).

For Latino older adults, both the model including depressive symptoms as the mediator (Robust CFI = 0.964, Robust RMSEA = 0.067, SRMR = 0.027) and chronic stress as the mediator (Robust CFI = 0.962, Robust RMSEA = 0.069, SRMR = 0.027) indicated good model fit. For the depressive symptoms model, there was significant total effect of income on memory (B = 0.117, SE = 0.043, p = 0.006) and executive functioning (B = 0.204, SE = 0.051, p <.001). For path a, higher income was significantly negatively associated with lower depressive symptoms (B = −0.154, SE = 0.043, p <.001). For paths b, lower depressive symptoms were significantly linked with higher episodic memory (B = −0.161, SE = 0.032, p <.001) and executive functioning (B = −0.138, SE = 0.036, p <.001). Bootstrapped analysis for depressive symptoms as the mediator indicated significant indirect effects for both memory (B = 0.025 [0.010, 0.044]) and executive functioning (B = 0.021 [0.008, 0.040]).

For the chronic stress mediation model, total effects of income on memory (B = 0.116, SE = 0.043, p = 0.006) and executive functioning were significant (B = 0.202, SE = 0.050, p <.001). For path a, higher income was significantly associated with lower chronic stress (B = −0.090, SE = 0.036, p = 0.013). For paths b, lower chronic stress was linked with lower memory (B = 0.025, SE = 0.036, p = 0.474) and higher executive functioning (B = −0.030, SE = 0.040, p = 0.456), although these associations were not statistically significant. Bootstrapped analysis indicated that chronic stress did not significantly mediate the association between income and either memory (B = −0.002 [−0.011,0.004]) or executive functioning (B = 0.003 [−0.005, 0.012]).

Discussion

The present study aimed to extend the current literature focused on the impact of income on cognitive functioning in late life by examining potential mechanisms underlying this association. For Black and Latino older adults, the association between income and memory, as well as income and executive functioning, was mediated by depressive symptoms. Chronic stress did not significantly mediate the link between income and episodic memory or executive functioning for either ethnic/racial group. While more research is needed to support the temporal process underlying these associations, this initial pattern of findings suggests that depressive symptoms represent a unique mechanism underlying economic inequalities in cognition among Black and Latino older adults.

Among Black older adults, depressive symptoms mediated the relationship between income and memory. This finding extends previous research that links low income to low memory by demonstrating that this relationship may be functioning via somatic (e.g., fatigue) and affective (e.g., guilt) depressive symptoms that lead to neurodegeneration in memory related brain regions among Black older adults (O’Shea et al., 2018). One study in older adults found a significant association between elevated somatic symptoms and decreased total, right, and left hippocampal volumes, whereas elevated affective symptoms were linked with decreased total entorhinal cortex volumes (O’Shea et al., 2018). There is also conflicting evidence on which type of depressive symptoms are most often reported by Black older adults (Sauceda et al., 2021). Understanding these distinctions in depressive symptoms among Black older adults may shed light on the exact mechanisms involved in the income and episodic memory link.

Depressive symptoms also mediated the relationship between income and executive functioning among Black older adults. Executive functioning is a multi-faceted cognitive domain that relies on multiple brain regions and is directly impacted by economic resources and depressive symptoms (Wang et al., 2017; Zhang et al., 2015). This study extends prior research by demonstrating that both depressive symptoms and income in late life may share a common brain mechanism (i.e., pre-frontal cortex) through which they may impact executive functioning among Black older adults (Ding et al., 2021; Pizzagalli & Roberts, 2022). Future research investigating this brain mechanism among Black adults will provide concrete insights into these processes.

Chronic stress was not a significant mediator in the link between income and memory or executive functioning for Black older adults. Our nonsignificant findings could be related to our measurement of stress, which focused on chronic stress burden or the cumulative effects of stress exposure (i.e., objective stressful experience) and appraisal (i.e., perceived stress) from multiple general life domains (Lazarus & Folkman, 1987). A previous study of young adults demonstrated that the relationship between Black race and episodic memory, working memory, and cognitive flexibility was significantly mediated by perceived stress (Letang et al., 2021). Another study of Black and NLW older adults found that higher stress exposure levels among Black older adults partially explained racial differences in memory performance, and lower stress appraisal levels among Black older adults buffered against the negative effects of stress exposure on memory (Morris et al., 2022). Taken together, these past studies highlight that stress exposure and appraisal may differentially contribute to cognitive function in Black older adults (Letang et al., 2021; Morris et al., 2022). The chronic stress measure we used also did not incorporate life stressors such as perceived discrimination, which has demonstrated to be impacted by SES factors (Stepanikova & Oates, 2017) and contribute to poorer cognition among Black older adults (Barnes et al., 2012). Perceived discrimination may be important for future studies to consider as a relevant social stressor for Black older adults living in southern states like Texas, which have deep-rooted histories of racism.

Among Latino older adults, income was associated with episodic memory and executive functioning through depressive symptomatology. Findings are consistent with prior research indicating that depressive symptoms significantly mediate the association between proxies for socioeconomic inequalities (i.e., Latino ethnicity) and episodic memory (Zaheed et al., 2021; Zahodne et al., 2019). This study highlights the importance of directly examining socioeconomic factors in addition to race/ethnicity. In a diverse mixed-race sample, the perceived stress and episodic memory link was initially mediated by depressive symptoms, but upon controlling for monthly household income, depressive symptoms no longer significantly mediated this association (Zaheed et al., 2021). The present study extends prior research by including executive functioning, another important cognitive predictor of ADRD risk, and utilizing a direct economic resource measure within a large sample of Latino older adults. The negative impact of income on cognitive functioning through depression may be related to brain atrophy due to overproduction of cortisol that leads to cell death in pre-frontal and hippocampal regions or increased formation of amyloid-beta plaque in the brain (Dafsari & Jessen, 2020). Future studies should examine the mechanistic role of the pre-frontal cortex and hippocampus in the associations examined in the present study.

The current finding that chronic stress was not a significant mediator in the relationship between income and either of the cognitive outcomes among the Latino subsample is somewhat inconsistent with the existing literature (Letang et al., 2021). Letang and colleagues (2021) found that perceived stress significantly explained the racial/ethnic differences in episodic memory, working memory, and cognitive flexibility between Latino and NLW young adults (Letang et al., 2021). This inconsistency may partly be a result of better management of emotions and coping with stressful experiences, both of which may develop with age (Carstensen et al., 2003). Alternatively, our findings may be linked to the measurement of the chronic stress variable, which combined both stress exposure and stress appraisal. There is evidence that Latino older adults report higher levels of chronic stress exposure, but lower levels of stress appraisal compared to NLW older adults (Brown, 2020). It is possible that these unique variations in stress experience among Latino older adults could have contributed to the lack of findings with chronic stress. Disentangling the complexities of the stress process on cognition in late life among Latino older adults was beyond the scope of the current study but represents a critical direction for future research.

The current study includes several strengths such as its large sample of ethnically/racially and linguistically diverse older adults, the direct measurement of income (one type of economic resource and socioeconomic factor), the use of cognitive latent variables within the mediation models, which allow for the reduction of measurement error, the examination of two potential mechanisms to explain economic inequalities on cognitive outcomes, and the use of structural equation modeling which allows for the examination of complex associations. Although our study has several strengths, several limitations are worth considering. First, given the cross-sectional nature of the present study, we were unable to ascertain the bi-directional relationship between stress and depressive symptoms. Although there is research to support distinct mechanisms through which depressive symptoms and chronic stress affect cognitive function, there is also evidence indicating that they may share common pathways given their bidirectional relationship (Dafsari & Jessen, 2020; Liu et al., 2017). Alternatively, some longitudinal research suggests that depressive symptoms are a pathway through which stress impacts memory (Zaheed et al., 2021). As such, it is possible that the stress to depression link may also explain the association between economic inequalities and cognitive functioning. Future longitudinal studies should consider exploring these complex associations (e.g., utilizing serial mediation approaches), as establishing temporal precedence may confirm whether stress and depressive symptoms work through distinct or sequential pathways in the income and cognition link. Future longitudinal studies are also needed to establish the causal ordering of depressive symptoms and cognitive function. Although some longitudinal evidence has demonstrated an association between higher baseline depressive symptoms and higher memory decline (Zahodne et al., 2014b), it is crucial to consider the literature that supports depressive symptoms as a psychological consequence of memory decline (Yin et al., 2024). The present study also did not capture information regarding the duration of depressive symptoms experienced by participants and focused on depressive symptoms captured at one time point rather than a current or lifetime clinical diagnosis of depression. Also, the observed mediation effects of depressive symptoms for both ethnic/racial background groups were relatively small (i.e., 0.021–0.036), which make replication of these findings critical to establish their clinical significance. Nonetheless, it is important to note that older adults tend to report lower depressive symptoms, and these findings highlight that even at small levels, they may have critical implications for late-life cognitive functioning. We fit a multi-group mediation model in a supplementary analysis, (see Supplementary Table 1) to determine if the mediating effect of depressive symptoms is equivalent or different for Black compared to Latino adults. Results from the likelihood ratio test were statistically significant indicating that the indirect effects differed across both groups and that the effect was larger for Black compared to Latino adults. Thus, although depressive symptoms mediated the income to cognition link among Black and Latino adults, the mechanism may be stronger for Black adults and suggests that other mechanisms should be examined for Latino adults. We also only measured one type of economic resource, but other individual-level (e.g., wealth) and area-level (e.g., neighborhood income) may be important to examine in future studies. Lastly, we did not consider other mechanisms, such as behavioral or biological levels of influence that the National Institute on Aging Health Disparities Research Framework identifies as relevant for minoritized populations. Future studies should build on the present study and explore additional mechanisms linking economic inequalities and poor late-life cognition among Black and Latino older adults.

Conclusion

There is an increasing demand to better characterize modifiable mechanisms that can help explain long standing economic inequalities that play a role in cognitive health disparities among Latino and Black older adults. We found that individual annual household income was linked with memory and executive functioning through depressive symptoms, but not chronic stress. These findings elucidate the unique contribution of depressive symptomatology in the complex relationship between economic inequalities and late-life cognition. Although longitudinal evidence is needed to establish the temporal process of these links, our findings indicate that depressive symptomatology among Black and Latino older adults may be a key priority for interventions focused on alleviating the adverse effects of economic resources on cognitive aging in these populations.

Data Availability

Data used in preparation of this article were obtained from the HABS-HD database (https://apps.unthsc.edu/itr/researchers). The present study was not preregistered, but a data use agreement was approved by the University of North Texas Health Science Center’s Institute for Translational Research prior to all data analyses.