Lifecourse socioeconomic status and subjective cognitive decline among diverse older adults

Abstract

Objectives

Lifecourse socioeconomic conditions are important determinants of age-related cognitive decline and dementia. We extend this research by examining whether these associations hold for subjective cognitive decline (SCD), an early marker that may reflect dementia risk across socioeconomic contexts.

Methods

In 2 multiethnic older-adult cohorts of long-term Kaiser Permanente Northern California members without prior dementia, we constructed dichotomous measures of financial, social, and cultural capital in childhood and late life. SCD was measured using the psychometrically validated Everyday Cognition 12-item scale (ECog-12). Generalized linear models estimated associations of each lifecourse socioeconomic status (SES) measure with SCD in the entire sample and stratified by sex/gender and race/ethnicity. Causal mediation analyses explored whether depressive symptoms (PROMIS Depression) mediated findings. All models adjusted for age, and moderation models adjusted for sex/gender or race/ethnicity.

Results

Low childhood (b = 0.098, 95% CI: 0.056, 0.140) and late-life (b = 0.080, 95% CI: 0.051, 0.110) financial capital, low childhood (b = 0.054, 95% CI: 0.023, 0.085) and late-life social capital (b = 0.062, 95% CI: 0.018, 0.107), and low late-life (b = 0.068, 95% CI: 0.030, 0.106) cultural capital were each independently associated with greater SCD. Low late-life cultural capital had the strongest negative impact among Black participants (b = 0.143, 95% CI: 0.085, 0.202). Depressive symptoms partially mediated associations of childhood and late-life financial and social capital with SCD, accounting for approximately 34%–64% of the associations.

Discussion

Multiple aspects of lifecourse SES contribute to SCD, with variation by sex/gender and race/ethnicity in lifecourse cultural capital. Depression may be a key pathway linking lifecourse SES disadvantage to SCD.

Evidence increasingly suggests that among older adults, the perception that one’s cognition has declined, often referred to as subjective cognitive decline (SCD), is a risk for the development of neurodegenerative diseases such as Alzheimer’s disease (Jessen et al., 2014, 2020). A 2014 review found that SCD doubles the risk of developing cognitive impairment or dementia, with 27% of those with SCD progressing to mild cognitive impairment and 14% to dementia within 4 years (Mitchell et al., 2014). SCD has also been linked to Alzheimer’s and related disorders including brain atrophy (Perrotin et al., 2015) and elevated amyloid levels (Amariglio et al., 2021). As such, SCD can be part of the early prodromal stage of Alzheimer’s disease (AD) and associated disorders. However, beyond biological factors, socio-contextual and environmental factors may also influence perceptions of cognition (Brigola et al., 2015; Hill et al., 2016).

Social determinants of health are non-medical factors such as the conditions in which people are born, grow, work, live, and age (World Health Organization, 2025), and they are increasingly recognized as important drivers of cognitive health. These conditions are often unequally distributed across historically marginalized communities and contribute to health inequities (Adkins-Jackson et al., 2023). While these factors operate throughout the lifespan, certain life stages may represent sensitive periods during which socioeconomic exposures exert stronger effects on later cognitive outcomes.

To date, relatively little research has examined how social determinants of health, especially indicators of SES, are associated with SCD. One previous study found that poverty (as defined by the (US Census Bureau, 2023) assessed at a single timepoint in mid- to late-life (aged 45+) was associated with greater risk of SCD complaints (de Havenon et al., 2024). Another found lower household income was associated with higher odds of SCD among adults aged 45+ (Peterson et al., 2019). A third study found that low socioeconomic status (a composite of educational attainment and household income) among African American women in midlife predicted SCD in late life (Simons et al., 2023). Research on associations between SES and SCD using a lifecourse approach is limited. In the Japan Gerontological Evaluation Study, low childhood SES (self-perceived social status at age 15) was associated with higher SCD prevalence compared to those with high childhood SES (Nishizawa et al., 2019). Childhood SES may influence late-life SCD by shaping cognitive reserve, chronic stress exposure, and later-life socioeconomic positioning, influencing both subjective and objective cognition (Faul et al., 2021; Greenfield & Moorman, 2019; Kaplan et al., 2001).

Our group previously constructed a comprehensive set of socioeconomic and contextual variables representing both early- and late-life, guided by Bourdieu’s (1986) framework of socioeconomic capital, including financial, cultural, and social domains (Bourdieu, 1986; Peterson et al., 2021b). Financial capital reflects access to economic resources. Cultural capital reflects both material and symbolic resources transmitted through generations and can be captured through educational attainment. Finally, social capital represents access to resources through social connections, operationalized as having trusting and supportive relationships. Utilizing this approach, we previously showed that higher late-life financial and cultural capital were associated with better objective cognition (Peterson et al., 2021b). In the present study we will extend this work to examine these same SES indicators across the lifespan on subjective cognitive decline.

There are many potential pathways by which aspects of SES across the lifespan may impact SCD in late life. If SCD is an early sign of the cognitive decline spectrum (i.e., prodromal disease) then the same mechanisms that impact objective cognition should impact SCD. For example, SES factors across the life span could impact access to healthcare and access to activities that provide cognitive stimulation, which may have negative effects on late-life cognition. However, lower SES also increases exposure to stress and adversity. Previous research has identified depressive symptoms as a potential pathway by which socioeconomic factors may impact SCD (Brown et al., 2022; Simons et al., 2023). Lower financial, cultural, or social capital may increase exposure to chronic stress and reduced access to resources that buffer against mental health challenges, thereby increasing the risk of depressive symptoms. In turn, depression can negatively shape self-appraisals of cognition, potentially amplifying perceptions of decline even in the absence of measurable impairment. Given prior evidence, testing for mediation is essential to determining how financial, cultural, and social capital impact SCD directly and indirectly through depressive symptoms. Finally, we also examined whether associations differed by sex/gender and race/ethnicity, as prior research suggests that the health effects of socioeconomic conditions vary across groups (American Psychological Association, 2010, 2017). Racially minoritized individuals and women may experience greater cumulative structural disadvantage and diminished returns of socioeconomic resources, potentially amplifying the impact of SES on SCD.

The objective of this study was to evaluate the associations of lifecourse socioeconomic factors with SCD among an ethnoracially diverse and representative group of older adults. Unlike previous studies that primarily relied on single indicators or composite socioeconomic indices, we assess financial, social, and cultural capital (each reflecting distinct and meaningful forms of socioeconomic resources) at two critical life stages: childhood and late life. We also evaluate potential moderation by sex/gender (women, men) and race/ethnicity (Asian, Black/African American, Hispanic/Latino, White), and potential mediation by depressive symptoms. We hypothesized that lower early-life and late-life capital across all domains will be associated with greater SCD, and that the strengths of these associations may differ by life stage. We further hypothesized partial mediation by depressive symptoms, reflecting the emotional toll of socioeconomic disadvantage on SCD. We also explored whether associations differed by sex/gender and race/ethnicity.

Method

Participants

We harmonized two large and ethnoracially diverse cohorts: the Kaiser Healthy Aging and Life Experiences (KHANDLE) study and the Study of Healthy Aging in African Americans (STAR). Both studies aim to evaluate how lifecourse sociocultural and health factors influence cognitive outcomes in late-life among diverse older adults. At the time of the current analyses, KHANDLE includes 2,285 community-dwelling older adults who reside in the San Francisco Bay Area and Sacramento Valley. Enrollment began in 2017, and eligible individuals were long-term members of Kaiser Permanente Northern California (KPNC), were age ≥ 65 on January 1, 2017, had participated in voluntary additional health checkups (the KPNC Multiphasic Health Checkups [MHC]) between 1964 and 1985, and did not have a diagnosis of dementia in their electronic medical record at the time of enrollment. Additional details, including exclusion criteria and sampling approaches of the KHANDLE cohort, can be found elsewhere (Hayes-Larson et al., 2022). Enrollment for the STAR cohort began in 2018 and is ongoing. This work includes the first 764 participants. Eligible individuals were long-term members of KPNC, had the same eligibility criteria as KHANDLE, and self-identified as Black/African American. Additional details of the STAR cohort can be found elsewhere (Peterson et al., 2021a).

Measures

Subjective cognitive decline

The Everyday Cognition (ECog) short form (ECog-12) is a 12-item informant-rated questionnaire designed to assess perceived changes in everyday cognitive functioning (Tomaszewski Farias et al., 2011). The original 39-item ECog scale has been validated as a multidimensional and psychometrically sound measure of everyday function in older adults, with minimal evidence of measurement bias across sociodemographic groups (Farias et al., 2008, 2020, 2021). The ECog-12 offers a more efficient assessment while retaining the strong psychometric properties of the original 39-item measure (Tomaszewski Farias et al., 2011). The ECog-12 evaluates everyday cognition over six domain-specific factors: divided attention, language, memory, organization, planning, and visuospatial abilities. Each question is rated on a 4-point scale: 1 = “Better or no change,” 2 = “Questionable/occasionally worse,” 3 = “Consistently a little worse,” and 4 = “Consistently much worse.” We used a composite score by averaging the 12-item responses (range = 1–4) at baseline. Higher ECog-12 scores indicate greater SCD.

Lifecourse socioeconomic status

We assessed childhood and late-life SES using self-reported measures from KHANDLE Wave 1 and STAR Wave 1 (study baselines), following the approach of Peterson et al. (2021a), which draws on Bourdieu’s (1986) theory of capital. This framework (Table 1) maps three forms of capital—financial, cultural, and social—across childhood and late-life. All SES indicators were coded as binary variables, with 1 indicating low capital and 0 indicating high capital. Childhood financial capital was considered low if the participant endorsed either of the following: (1) ever skipped a meal or went hungry because there was not enough money to buy food, or (2) reported their family was poor relative to others from birth to age 16, and their parents did not own their childhood home. Late-life financial capital was considered low if (1) the participant receives supplemental security income, cash assistance from a state/county welfare program, help from family/friends, and has household income less than $55,000 annually, or (2) is often worried about not having enough money to pay for living expenses and has a household income less than $75,000 annually. Childhood cultural capital was considered low if the participant reported one or both parents with no more than an 8th grade education. Late-life cultural capital was considered low if the participant had at most a high school diploma or passed the general educational development (GED) test. Childhood social capital was considered low if the participant rarely or never had someone to whom they could talk, trust, and confide. Late-life social capital was considered low if the participant did not have a special person that they could share confidences and feelings with or depend on.

Table 1.

Theoretical framework for measuring lifecourse socioeconomic status (SES) in childhood and late life.

Form of capital	Childhood	Late adulthood
Low financial capital	Q: “When you were growing up, how often did you skip a meal or go hungry because there was not enough money to buy food?” A: Ever had this experience vs never Q: “Think about your family when you were growing up, from first to age 16. Would you say your family during that time was pretty well off financially, about average, or poor?” A: Poor vs well off/about average Q: “During your childhood, did your family mainly: rent, pay a mortgage/own a home, have other living arrangements?” A: Parents rented vs owned home/other living arrangements	Q: “What was the total income (before taxes) for you and your spouse in the past year?” Q: “Which sources of income do you or your spouse receive?” A: Receives supplemental security income or cash assistance from state/county welfare program or help from family/friends or has household income <US$55,000 Q: “How often do you worry about not meeting your expenses with your current income?” A: Always/often vs Never Sometimes AND has household income <US$75,000
Low cultural capital	Q: “What was the highest educational level of your (mother/father) or the (woman/man) who raised you?” A: One or both parents had ≤8th grade education.	Q: “What is the last grade or highest level in school you completed and got credit for?” A: ≤High School diploma/GED
Low social capital	Q: “During your childhood, how often was there someone in whom you could talk to, trust and confide?” A: None/a little of the time vs some/most/all of the time.	Q: “Is there any one special person you know that you feel very close and intimate with—someone you share confidences and feelings with, someone you feel you can depend on?” A: No vs Yes

Note. Based on the theory of capital developed by Bourdieu (1986) and adapted by Peterson et al. (2021a).

Depressive symptoms

The KHANDLE cohort assessed depressive symptoms using the 4-item Patient-Reported Outcomes Measurement Information System (PROMIS) computer adaptive testing (CAT) v1.0 administered via the NIH Toolbox, while the STAR cohort used an 8-item PROMIS-57 that was not computer adaptive (Pilkonis et al., 2011, 2014). The specific items included in these measures were not identical, as the CAT dynamically selects items from the PROMIS depression item bank based on participant responses, whereas the PROMIS-57 short form uses a fixed set of eight items (HealthMeasures, 2023; PROMIS Depression Scoring Manual, 2019). However, both approaches are derived from the same PROMIS depression item bank and are calibrated using Item Response Theory (IRT), which links individual items to a common underlying latent construct of depressive symptoms (Cella et al., 2007; HealthMeasures, 2023; PROMIS Depression Scoring Manual, 2019; Reeve et al., 2007). This allows scores from different item sets and administration formats (e.g., CAT and short forms) to be placed on a common metric and compared across cohorts. We therefore harmonized the two depression measures at Wave 1 using the PROMIS depressive symptoms scoring manual. We converted these scores into theta scores (HealthMeasures, 2023), which were standardized to the U.S. adult population (Reference Populations, n.d.). Theta scores were centered at zero for the average U.S. adult; values below zero indicate fewer depressive symptoms than the average U.S. adult. Consistent with prior work done in these cohorts (Jimenez et al., 2023), depressive symptom theta scores were analyzed as a continuous variable.

Covariates/potential moderators

Covariates were obtained from KHANDLE Wave 1 and STAR Wave 1. The set of covariates used in the analyses included age at interview in years, self-reported sex/gender (women, men), and self-reported race/ethnicity (Asian, Black/African American, Hispanic/Latino, White). Self-reported sex/gender data were supplemented with physician-recorded information from electronic medical records when missing. We use the term sex/gender to acknowledge that this variable reflects both biological and sociocultural factors.

Statistical analysis

We conducted descriptive analyses to summarize means and prevalences of participant characteristics and lifecourse SES measures in the overall harmonized sample and stratified by sex/gender and race/ethnicity. Because the score was right-skewed (i.e., most participants reported little to no SCD), we rescaled it by dividing each participant’s ECog-12 score by the maximum possible value (=4), resulting in a transformed ECog-12 score ranging from 0 to 1. We then fit multivariable generalized linear models with a Beta family distribution and logistic link to estimate the total effects of each lifecourse SES measure (childhood/late-life financial, cultural, and social capital) on the transformed ECog-12 score.

We tested for moderation by sex/gender and race/ethnicity to further examine how lifecourse SES contributes to sex/gender and racial/ethnic disparities in SCD. We included an interaction term between each lifecourse SES measure and a potential modifier (sex/gender or race/ethnicity), then evaluated its statistical significance using a Type 3 Sum of Squares test at p < .10. For interactions that met this threshold, we conducted stratified analyses to further explore group-specific associations.

We also implemented two multivariable generalized linear models with a Beta family distribution and logistic link to examine the independent effects of childhood and late-life SES on SCD. In the first model, we included only the childhood SES measures (childhood financial, social, and cultural capital) to estimate their independent effects on SCD. In the second model, we included both childhood and late-life SES measures simultaneously to estimate the independent effects of late-life SES, accounting for earlier life SES. This two-model approach was used because including all SES measures in a single model would likely underestimate the influence of childhood SES, as late-life SES may act as a mediator in the pathway between early life conditions and later cognitive perceptions.

To explore mediation by depressive symptoms, we used the SAS procedure PROC CAUSALMED (Valente et al., 2020) to estimate (1) the effect of lifecourse SES on SCD, controlling for the potential mediator depressive symptoms (i.e., the natural direct effect); (2) the effect of depressive symptoms on SCD if lifecourse SES was held at a specific level (i.e., the natural indirect effect (of the lifecourse SES measure on SCD through depressive symptoms, independent of the direct effect); and (3) the percentage of the relationship between lifecourse SES and SCD mediated by depressive symptoms. Although PROC CAUSALMED does not accommodate Beta-distributed outcomes, we conducted sensitivity analyses using Beta regression models to assess the robustness of the mediation findings. Results were highly consistent with those obtained from PROC CAUSALMED using the linear outcome model. Demographic variables (age at interview, sex/gender, and race/ethnicity) were included in the exposure-outcome, mediator-outcome, and exposure-mediator models to control for confounding.

All models adjusted for baseline age, sex/gender, and race/ethnicity. Moderation models adjusted for baseline age and the non-moderating sociodemographic variable. All statistical analyses were performed using SAS 9.4.

Results

Descriptive statistics

In the harmonized sample of 3,049 participants, we excluded two missing sex/gender and four missing race/ethnicity resulting in an analytic sample of 3,043 participants. As shown in Table 2, 62% were women, and 44.6% identified as Black, 20.3% as White, 17.7% as Asian, and 17.4% as Hispanic/Latino. The average age was 74 years (SD = 7.9). The average depressive symptoms theta score was negative, indicating fewer depressive symptoms compared to the average U.S. adult. Men had a lower average depressive symptoms theta score than women, and Black participants had the lowest average theta score, followed by Asian, White, and Hispanic/Latino participants.

Table 2.

Baseline demographic characteristics and distribution of measures of lifecourse socioeconomic status among the harmonized sample of KHANDLE + STAR and stratified by sex/gender and race/ethnicity.

Characteristic	Overall, N = 3,043	By sex/gender		By race/ethnicity
		Women, n = 1,887	Men, n = 1,156	Asian, n = 539	Black, n = 1,357	Hispanic/Latino, n = 528	White, n = 619
Age, mean (SD)	74.0 (7.9)	73.7 (8.1)	74.5 (7.5)	75.3 (6.4)	71.6 (8.5)	75.5 (6.5)	76.9 (7.1)
Sex/gender, N (%)
Women	1,887 (62.0)	1,887 (100)	–	293 (54.4)	927 (68.3)	306 (57.9)	361 (58.3)
Men	1,156 (38.0)	–	1,156 (100)	246 (45.6)	430 (31.7)	222 (42.1)	258 (41.7)
Race/ethnicity, N (%)
Asian	539 (17.7)	293 (15.5)	246 (21.3)	539 (100)	–	–	–
Black	1,357 (44.6)	927 (49.1)	430 (37.2)	–	1,357 (100)	–	–
Hispanic/Latino	528 (17.4)	306 (16.2)	222 (19.2)	–	–	528 (100)	–
White	619 (20.3)	361 (19.1)	258 (22.3)	–	–	–	619 (100)
Depressive symptoms theta score, mean (SD)	−0.16 (0.75)	−0.12 (0.76)	−0.23 (0.74)	−0.17 (0.78)	−0.27 (0.77)	0.003 (0.74)	−0.05 (0.65)
ECog-12 score, mean (SD)	1.36 (0.39)	1.37 (0.38)	1.35 (0.38)	1.42 (0.42)	1.35 (0.38)	1.35 (0.36)	1.36 (0.37)
Lifecourse socioeconomic status, N (%)
Childhood financial capital
Low	360 (12.0)	188 (10.1)	172 (15.1)	58 (11.0)	153 (11.4)	96 (18.4)	53 (8.7)
High	2,640 (88.0)	1,675 (89.9)	965 (84.9)	468 (89.0)	1,185 (88.6)	427 (81.6)	560 (91.3)
Late-life financial capital
Low	1,056 (36.2)	743 (41.0)	313 (28.3)	111 (21.5)	570 (43.9)	182 (36.5)	193 (31.9)
High	1,864 (63.8)	1,069 (59.0)	795 (71.7)	406 (78.5)	729 (56.1)	317 (63.5)	412 (68.1)
Childhood social capital
Low	751 (27.6)	439 (26.1)	312 (29.8)	169 (35.6)	267 (21.5)	149 (33.2)	166 (29.6)
High	1,975 (72.4)	1,241 (73.9)	734 (70.2)	306 (64.4)	975 (78.5)	300 (66.8)	394 (70.4)
Late-life social capital
Low	312 (10.5)	189 (10.2)	123 (10.9)	59 (11.2)	133 (10.1)	64 (12.3)	56 (9.2)
High	2,670 (89.5)	1,667 (89.8)	1,003 (89.1)	469 (88.8)	1,190 (89.9)	455 (87.7)	556 (90.8)
Childhood cultural capital
Low	1,378 (50.5)	854 (50.4)	524 (50.7)	244 (50.4)	638 (53.4)	318 (67.7)	178 (30.7)
High	1,350 (49.5)	841 (49.6)	509 (49.3)	240 (49.6)	556 (46.6)	152 (32.3)	402 (69.3)
Late-life cultural capital
Low	491 (16.2)	323 (17.2)	168 (14.6)	44 (8.2)	229 (16.9)	129 (24.5)	89 (14.4)
High	2,547 (83.8)	1,560 (82.8)	987 (85.4)	494 (91.8)	1,125 (83.1)	398 (75.5)	530 (85.6)

Regarding lifecourse SES, 12% of participants had low childhood financial capital, 36% had low late-life financial capital, 51% had low childhood cultural capital, 16% had low late-life cultural capital, 28% had low childhood social capital, and 10% had low late-life social capital. Women were more likely than men to report low late-life financial capital (41% vs 28%) and low late-life cultural capital (17% vs 15%). Across racial/ethnic groups, lifecourse socioeconomic disadvantage was most common among minoritized groups. For example, low late-life financial capital was reported by 44% of Black and 37% of Hispanic/Latino participants compared to 32% of White and 22% of Asian participants. Similarly, low childhood cultural capital was most prevalent among Hispanic/Latino (68%), Black (53%), and Asian participants (50%) and least prevalent among White participants (31%).

The overall average ECog-12 score was 1.36 (SD = 0.39), indicating low levels of SCD. Women reported slightly higher ECog-12 scores, and Asian participants had the highest mean ECog-12 score, while Black, Hispanic/Latino, and White participants had lower and similar mean ECog-12 scores.

Financial capital and subjective cognitive decline

Compared with high childhood financial capital, low childhood financial capital was associated with greater reported SCD (worse ECog-12 scores) (b = 0.103, 95% CI: 0.058, 0.147; Table 3, Model 1), with no evidence of moderation by sex/gender (p = .47) or by race/ethnicity (p = .17). Low late-life financial capital was also associated with worse ECog-12 (b = 0.090, 95% CI: 0.059, 0.121; Table 3, Model 2), with no evidence of moderation by sex/gender (p = .13) or race/ethnicity (p = .10).

Table 3.

Beta regression models separately estimate the association between measures of socioeconomic status (SES) in childhood and late life with subjective cognitive decline (ECog-12).

Model	Lifecourse SES (ref = High capital)	β	(95% CI)
Model 1	Low childhood financial capital	0.103***	(0.058, 0.147)
Model 2	Low late-life financial capital	0.090***	(0.059, 0.121)
Model 3	Low childhood social capital	0.061***	(0.028, 0.095)
Model 4	Low late-life social capital	0.068**	(0.021, 0.115)
Model 5	Low childhood cultural capital	0.013	(−0.018, 0.045)
Model 6	Low late-life cultural capital	0.071***	(0.031, 0.112)
Model 7	Combined model for childhood SES
	Low childhood financial capital	0.074**	(0.023, 0.125)
	Low childhood social capital	0.060***	(0.024, 0.095)
	Low childhood cultural capital	0.000	(−0.032, 0.033)
Model 8	Combined model for late-life SES
	Low late-life financial capital	0.074***	(0.040, 0107)
	Low late-life social capital	0.068*	(0.014, 0.122)
	Low late-life cultural capital	0.049*	(0.001, 0.097)

Note. All models adjusted for baseline age, sex/gender, and race/ethnicity.

^* p < .05. **p < .01. ***p < .001.

Social capital and subjective cognitive decline

Low childhood social capital was associated with greater SCD compared with high childhood social capital (b = 0.061, 95% CI: 0.028, 0.095; Table 3, Model 3), with no moderation by sex/gender (p = .17) or by race/ethnicity (p = .13). Low late-life social capital was also associated with worse ECog-12 (b = 0.068, 95% CI: 0.021, 0.115; Table 3, Model 4), with no evidence of moderation by sex/gender (p = .91) or by race/ethnicity (p = .61).

Cultural capital and subjective cognitive decline

Childhood cultural capital was not associated with SCD ratings (b = 0.013, 95% CI: −0.018, 0.045; Table 3, Model 5), nor was there evidence of moderation by race/ethnicity (p = .36). However, we found evidence that suggested that the association between childhood cultural capital and SCD was modified by sex/gender (p = .09). In sex/gender-stratified models, women with low childhood cultural capital tended to endorse greater SCD (b = 0.03, 95% CI: −0.010, 0.071), whereas men with low childhood cultural capital tended to endorse lower SCD (b = −0.015, 95% CI: −0.066, 0.036); however, neither association was statistically significant (Figure 1, Models 5a and 5b; Supplementary Tables S1, Models 5a and 5b). These sex/gender differences indicate a potentially stronger adverse effect of low childhood cultural capital on SCD for women compared to men.

Figure 1.

Stratified beta regression associations of childhood and late-life cultural capital with subjective cognitive decline by sex/gender and race/ethnicity.

Low late-life cultural capital was associated with greater SCD ratings (b = 0.071, 95% CI: 0.031, 0.112; Table 3, Model 6). This association was not moderated by sex/gender (p = .83) but was moderated by race/ethnicity (p = .01). Among Black participants, low late-life cultural capital was significantly associated with worse ECog-12 (b = 0.143, 95% CI: 0.085, 0.202; Figure 1, Model 6b; Supplementary Tables S1, Model 6b). In contrast, the association was null among White participants (b =−0.003, 95% CI: −0.095, 0.088; Figure 1, Model 6d; Supplementary Tables S1, Model 6d) and Hispanic/Latino participants (b = 0.006, 95% CI: −0.073, 0.084; Figure 1, Model 6c; Supplementary Tables S1, Model 6c). Among Asian participants, the association trended in the same direction as that observed in Black participants (b = 0.092, 95% CI: −0.045, 0.229; Figure 1, Model 6a; Supplementary Tables S1, Model 6a) but did not reach statistical significance, although it may suggest a similar adverse pattern. Overall, these findings indicate a stronger negative impact of low late-life cultural capital on SCD among Black participants compared to other racial and ethnic groups.

Combined lifecourse SES

In the combined childhood SES model (Table 3, Model 7), lower childhood financial capital (b = 0.074, 95% CI: 0.023, 0.125) and lower childhood social capital (b = 0.060, 95% CI: 0.024, 0.095) were independently associated with worse ECog-12, whereas childhood cultural capital was not associated with ECog-12 (b = 0.000, 95% CI: −0.032, 0.033). These results are consistent with the single-factor childhood models (Table 3, Models 1, 3, and 5), which also showed significant associations for childhood financial and social capital but not for childhood cultural capital.

In the combined model including all childhood and late-life SES measures (Table 3, Model 8), lower late-life financial capital (b = 0.074, 95% CI: 0.040, 0.107), lower late-life social capital (b = 0.068, 95% CI: 0.014, 0.122), and lower late-life cultural capital (b = 0.049, 95% CI: 0.001, 0.097) remained significantly associated with greater SCD ratings. These findings corroborate the results from the individual models (Table 3, Models 2, 4, and 6), which showed that lower financial, social, and cultural capital in late-life were generally linked to greater SCD.

Potential mediation by depressive symptoms

Mediation analyses are presented in Figure 2 and Supplementary Tables S2. The natural direct effect of childhood financial capital on SCD was statistically significant, and the natural indirect effect was attenuated but remained significant, suggesting partial mediation by depressive symptoms (Figure 2; Supplementary Tables S2). An estimated 34% (95% CI: 13.4, 53.8) of the association between childhood financial capital and SCD was mediated by late-life depressive symptoms. A similar pattern was observed for late-life financial capital, with depressive symptoms estimated to mediate 37% (95% CI: 20.1, 54.7) of the association with SCD.

Figure 2.

Mediation analyses: depressive symptoms as potential mediators and subjective cognitive decline as the outcome.

Note. SES = socioeconomic status. Natural direct effects, natural indirect effects, and percentage mediated were estimated separately for each lifecourse SES factor. Subjective cognitive decline was measured by the Everyday Cognition (ECog-12). The corresponding total effect for each SES factor is presented in Models 1–6 of Table 3.

The natural direct effect of childhood social capital on SCD was statistically nonsignificant, whereas the natural indirect effect was significant, suggesting that depressive symptoms mediate the association between childhood social capital and SCD (Figure 2; Supplementary Tables S2). Approximately 64.4% (95% CI: 27.0, 101.8) of the association between childhood social capital and SCD was mediated by late-life depressive symptoms. There was no evidence of a natural direct effect of late-life social capital on SCD while the natural indirect effect was significant (Figure 2; Supplementary Tables S2). An estimated 51.8% (95% CI: 10.0, 93.5) of the association between late-life social capital and SCD was mediated by late-life depressive symptoms.

There was no evidence that depressive symptoms mediated the associations between lifecourse cultural capital and SCD as the natural indirect effects and proportion mediated in the childhood and late-life cultural capital models were nonsignificant.

Discussion

This study examined multiple forms of socioeconomic capital (financial, social, and cultural) across two critical life stages (childhood and late-life), providing a lifecourse perspective on SCD in a large, diverse sample. Our findings demonstrate that, with the exception of childhood cultural capital, each type of socioeconomic capital at both life stages was independently associated with the degree of SCD in late-life. Specifically, individuals with low childhood and late-life financial and social capital, as well as low late-life cultural capital, were more likely to endorse greater SCD. When concurrently examining the three forms of SES at both lifecourse stages, we observed similar associations between each lifecourse SES measure and SCD, indicating that both early- and late-life SES disadvantage uniquely contributed to greater SCD. We also observed several notable patterns of moderation in the associations between lifecourse SES and SCD. The adverse effect of low late-life financial capital on SCD was significantly stronger among Black participants, and low childhood social capital had a greater impact among Asian participants than among White participants. Sex/gender modified the association between childhood cultural capital and subjective cognitive ratings, with a stronger negative effect observed among women. Additionally, low late-life cultural capital was more strongly associated with SCD among Black participants, but not among White participants.

Given that SCD can be part of the preclinical stage of AD and a risk for cognitive decline and dementia (Jessen et al., 2020; Mitchell et al., 2014), it is important to understand the factors that contribute to SCD in late life. Lower SES across the lifecourse may limit access to cognitively enriching resources and social support (Office of Disease Prevention and Health Promotion, n.d.), thereby influencing both perceptions of cognitive decline and future risk of dementia (Marden et al., 2017). These results highlight the need for research on the pathways and mechanisms through which socioeconomic disadvantage and other structural factors shape subjective cognitive experiences and late-life cognitive outcomes.

Our findings further highlight the detrimental impact of socioeconomic disadvantage across the lifespan on cognitive health. They emphasize the importance of adopting a lifecourse perspective, which considers how exposures at different periods of life may uniquely and cumulatively influence late-life cognitive outcomes. This approach aligns with two theoretical frameworks: (1) the sensitive period model, which posits that exposures during sensitive developmental windows (e.g., childhood) may have lasting effects, and (2) the accumulation model, which suggests that socioeconomic disadvantages across multiple life stages compound over time to influence cognitive aging trajectories (Wagner et al., 2024). Using the KHANDLE cohort, Peterson et al. (2021a) reported that late-life cultural capital had a more pronounced impact on objective cognitive outcomes compared to childhood cultural capital. Similarly, we found that the negative effects of lower cultural capital on SCD were stronger in late life compared to childhood. One possible explanation is that late-life cultural capital (i.e., education) may influence subjective cognitive decline through several pathways. For example, more education may directly influence how people perceive cognitive changes, possibly making them more sensitive to these changes as they age. Additionally, higher education may create greater opportunities for cognitive stimulation that could impact the risk of cognitive changes.

Our findings add to the growing literature demonstrating that SES indicators across the lifespan are important determinants of late-life SCD. While prior studies have typically examined childhood and late-life SES separately (Peterson et al., 2019; Simons et al., 2023), our study found that both timepoints contribute to greater SCD independently and cumulatively, indicating the need to consider lifecourse socioeconomic disadvantage on SCD. A Behavioral Risk Factor Surveillance System (BRFSS) and U.S. Census-based study observed that adults aged 45+ with lower income (financial capital) or lower educational attainment (cultural capital) had higher odds of SCD (Peterson et al., 2019). We also found that lower financial capital and lower cultural capital in late life were independently associated with greater SCD. However, our study additionally found that low childhood financial capital also contributes to late-life SCD. Importantly, we also extended this work by examining a third SES factor, social capital in late-life and childhood. We found that low social capital in both timepoints was associated with greater SCD, suggesting that social relationships and emotional support are also unique contributors to SCD beyond traditional financial and educational measures. Prior research has linked poor social support to worse subjective cognition, including a study of adults aged 18–99 from the Virginia Cognitive Aging Project, which found an association between negative social support and worse subjective cognitive functioning (Mueller et al., 2023). Here, our inclusion of childhood social capital furthers this work by indicating that early-life social disadvantage may have long-term consequences for how individuals perceive their cognitive health in later life.

Our findings add new evidence that the impact of lifecourse SES factors on SCD differs across self-reported sex/gender, an area where prior research has been limited. In the current study, we found that childhood cultural capital showed sex/gender-specific associations, with women exhibiting stronger negative impacts on SCD than men. Women have been found to report more SCD than men, which may reflect sex/gender differences in awareness, emotional processing, or health-seeking behavior (Heser et al., 2019). Little research has directly examined parental education and SCD by sex/gender. However, a relevant study using the Health and Retirement Study data found that women with low maternal education had worse memory, but did not find this for men (Lyu, 2015). Our findings contribute novel evidence that the association between childhood cultural capital and SCD may operate differently for women and men.

Previous work examining SCD across different racial and ethnic groups is limited, but some studies suggest differences in both its prevalence (Wooten et al., 2023) and its associations with objective cognitive performance (Katz et al., 2023; Tomaszewski Farias et al., 2025). These differences highlight the importance of examining whether socioeconomic factors relate to SCD similarly across diverse populations. To address this gap, we stratified our analyses by racial and ethnic group. We found that low late-life cultural capital had stronger adverse effects on SCD among Black participants compared to other racial and ethnic groups. Our findings align with prior literature that show differential returns of lifecourse SES across racial groups. For example, one study found that the association between higher adulthood SES (composite of education, occupation, income, and wealth) and better objective memory performance was stronger among White participants compared to Black participants, indicating diminished cognitive benefits of SES among Black participants (Zeng et al., 2022). Other research suggests that even when lifecourse SES levels are similar across groups, the cognitive benefits associated with higher SES may be diminished for racially minoritized individuals due to structural inequities and accumulated disadvantage, which may not be fully captured in our analyses (Glymour & Manly, 2008; Zahodne et al., 2021). These structural barriers, such as systemic discrimination in education, can limit the extent to which higher SES translates into cognitive health benefits for racially minoritized groups. Overall, these findings emphasize the need to consider both psychosocial and structural factors that influence subjective cognitive experiences across racial and ethnic groups.

Finally, previous work has consistently demonstrated a relationship between SCD and depression. As such, it is important to understand the extent to which depressive symptoms mediate the association between lifecourse SES disadvantage and SCD. Although depression has been examined as a mediator of the relationship between composite measures of SES and SCD (Brown et al., 2022; Simons et al., 2023), our study differentiates between distinct forms of lifecourse capital (financial, social, and cultural). We found that depressive symptoms partially mediated the associations between both social and financial capital and SCD. Notably, when accounting for depressive symptoms, the natural direct effects of childhood and late-life social capital were attenuated and became non-significant, suggesting that depressed mood may largely mediate this association. Low social capital may lead to increased feelings of loneliness, reduced emotional support, and a diminished sense of belonging, all of which are all factors strongly associated with depression (Holt-Lunstad, 2021; Wickramaratne et al., 2022). The mediating role of depressive symptoms in the association between both childhood and late-life financial capital and SCD may reflect the psychological burden of economic strain. Financial insecurity is associated with chronic stress, reduced access to health-promoting resources, and heightened emotional distress, which can contribute to depressive symptoms (Guan et al., 2022; Ryu & Fan, 2023). Together, these findings highlight mental health and emotional well-being as key mechanisms through which lifecourse socioeconomic disadvantage may shape late-life SCD.

Several limitations should be considered. First, the use of dichotomous measures for capital simplifies complex socioeconomic gradients, and continuous or ordinal measures may better capture subtle differences in exposure. Second, unmeasured confounding remains possible, particularly from factors such as childhood health or education quality, emphasizing the need for richer datasets that include more detailed early-life conditions. Third, although we assess childhood and late-life SES separately, we did not examine socioeconomic mobility and trajectories (e.g., consistently low, consistently high, upward mobility, or downward mobility) (e.g., Peterson et al., 2021a). Such patterns may have unique implications for cognitive aging and subjective cognitive perceptions and represent an important area for future research. Fourth, participants reported lifecourse SES retrospectively, which may introduce recall bias. Prospective data collection across the lifecourse would help validate these retrospective reports and reduce measurement error. Fifth, participants in this sample reported lower depressive symptoms than the average U.S. adult, which may limit the generalizability of the findings to populations with higher levels of depression. Lastly, the cross-sectional design limits causal inference, and mediation analyses assume that late-life financial and social capital precede depressive symptoms and that depressive symptoms precede SCD. Future research using longitudinal data is needed to confirm these pathways and better establish temporality. Addressing these limitations in future studies will help strengthen the evidence base and improve our understanding of how socioeconomic conditions shape subjective cognitive aging.

Conclusion

This study uses a lifecourse approach to examine how financial, social, and cultural capital in both childhood and late-life influence SCD, a known risk factor for cognitive impairment and dementia. By assessing multiple forms of capital, we capture both material (e.g., wealth) and non-material (e.g., social networks, educational attainment) resources across the lifecourse. Our findings show that lower childhood and late-life financial capital, lower childhood and late-life social capital, and lower late-life cultural capital are associated with greater SCD. We found that these associations were not completely uniform across social groups, with moderation by race/ethnicity and sex/gender highlighting the importance of considering intersectional social identities. Additionally, depressive symptoms partially mediated the associations between SES and SCD, emphasizing mental health as a key pathway through which socioeconomic disadvantage influences cognitive self-perception. Together, these findings highlight potential modifiable targets for intervention and suggest that policies and programs aimed at improving financial stability, promoting social engagement, expanding access to education, and supporting mental health throughout the lifecourse may help mitigate SCD and reduce disparities in cognitive aging.

Supplementary material

Supplementary material is available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online.

Funding

This work was supported by the National Institutes of Health National Institute on Aging [R01AG052132 (KHANDLE), R01AG050782 (STAR), and R01AG031563].

Conflicts of interest

None declared.

Data availability

Qualified researchers interested in accessing data utilized in these analyses should apply through the Research Collaboration Portal (https://divisionofresearch.kaiserpermanente.org/research/research-collaboration/). Access is contingent upon project approval, IRB approval, and execution of a Materials and Data Transfer Agreement. This study was not preregistered.

Author contributions

Nancy X. Chen performed all statistical analyses and wrote the paper. Sarah Tomaszewski Farias planned the study and supervised the data analysis. Rachel A. Whitmer and Paola Gilsanz were responsible for the acquisition of the data. All authors had access to the included data, critically reviewed and edited the paper for important intellectual content, approved the final version for publication, and agreed to be accountable for all aspects of the work.