Influence of Birthplace and Age at Migration on Cognitive Aging Among Hispanic/Latino Populations in the United States: Study of Latinos-Investigation of Neurocognitive Aging

Abstract

Background and Objectives

Although Hispanic/Latino populations in the United States are remarkably diverse in terms of birthplace and age at migration, we poorly understand how these factors are associated with cognitive aging. Our research seeks to operationalize a life course perspective of migration and health and contribute new understanding of Alzheimer’s disease/Alzheimer’s disease-related dementias among U.S.-based Hispanic/Latino older adults.

Research Design and Methods

Harnessing the Hispanic Community Health Study/Study of Latinos (n = 16,415) and the Study of Latinos-Investigation of Neurocognitive Aging (n = 6,377) data, we compare baseline cognition and 7-year cognitive change among U.S./mainland-born Hispanic/Latino adults relative to foreign/island-born immigrants by age of migration (4 groups: born in mainland United States, immigrated <16 years, 16–34 years, >34 years). Global cognition was calculated as a composite measure, and domain-specific measures were considered in secondary analyses. We employed linear regressions, ANOVA contrasts, and Blinder–Oaxaca decomposition techniques.

Results

All Hispanic/Latino immigrant adults, regardless of age at migration, have a cognitive health disadvantage (at each visit and over time) relative to U.S./mainland-born Hispanic/Latino individuals. Differences did not endure the inclusion of covariates and were explained predominantly by first socioeconomic and then acculturative factors, and far less by health and health behaviors. Acculturative factors are particularly important for individuals who migrated after childhood.

Discussion and Implications

Socioeconomic and acculturation factors have outsized roles in explaining gaps in cognitive aging among U.S.-born and migrant Hispanic/Latino adults. It is then vital to examine whether disrupting socioeconomic and acculturation inequalities closes such gaps in cognitive aging.

Background and Objectives

Hispanic/Latino communities in the United States have a disproportionately high incidence of Alzheimer’s disease and related dementias (ADRD) (Alzheimer’s Association, 2019b), but the reasons are not yet well understood. At the same time, while early and middle-life circumstances “set the stage” in aging (Jagust, 2016), research linking these life course determinants to ethnoracial disparities in ADRD remains limited (Glymour & Manly, 2008). In 2015, the National Institute on Aging put forth a health disparities research framework that included four areas of concentration—environmental, sociocultural, behavioral, and biological—and encouraged taking a life course perspective (Hill et al., 2015). Nearly 10 years later, several studies have identified factors in each of these areas that influence AD/ADRD risk (Livingston et al., 2024), but these factors are often investigated in isolation and/or in predominantly non-Latino White samples, which limits generalizability and can omit culturally relevant factors (Quiroz et al., 2022; Vila-Castelar et al., 2022). Migration is a major life course event with implications for socioeconomic status (SES), lifestyle modifications, and exposure to stressors—all of which may affect health in later life (Alcántara et al., 2014; Sangalang et al., 2019; Torres & Wallace, 2013), but it is quite underresearched with regards to ADRD. Focusing on birthplace and age at migration can help us better understand the high risk and prevalence of ADRD that Hispanic/Latino older adults living in the U.S. experience (Alzheimer’s Association, 2019a; Matthews et al., 2019).

The United States’ (U.S.’) Hispanic/Latino populations are remarkably diverse (see Author Note 1). Important sources of diversity include birthplace and age at migration: for example, about two-thirds were born in one of the 50 U.S. states/DC (Krogstad et al., 2022), whereas the remaining third was born elsewhere. Migration’s influence on health will vary according to the circumstances and stage of the life course in which people arrived (Abraído-Lanza et al., 2016; Torres & Wallace, 2013). However, it remains unclear how cognitive function among foreign-born older adults may differ from their U.S.-born (USB) co-ethnic counterparts, and how this may depend on the age at migration.

Focusing on age at migration (rather than years lived in the United States) allows us to operationalize a life course perspective of migration and health. The Healthy Migrant hypothesis—that healthy migrants are selected into the United States—is influential (Abraido-Lanza et al., 1999). Individuals who migrate during childhood usually accompany their parents and so are not themselves likely to be selected for good health, but may gain late-life cognitive and health benefits from more years of education in the United States. Education, however, may be less and/or not protective among Latino populations than in other groups (Avila et al., 2021; Kimbro et al., 2008; Mungas et al., 2018; Zahodne et al., 2019). Younger age at migration is also linked to greater accumulation of wealth over the life course (Wakabayashi, 2010), which then can influence cognitive function in later life (Haan et al., 2011). These factors, health and SES, reflect the behavioral, biological, and the environmental components of the National Institute on Aging health disparities research framework (Hill et al., 2015). Individuals who migrate during young adulthood are more likely to be positively selected for good health and may still benefit from additional training and immersion in the destination labor market (Landale et al., 2006; Riosmena et al., 2013, 2017; Thomson et al., 2013). Individuals who migrate during mid- and late adulthood may be less selected on health factors than the young adult migrants and less likely to gain additional training at their destinations relative to those migrating at younger ages. Previous research of Hispanic/Latino populations in the United States find that the risks of unfavorable health conditions like diabetes (Schneiderman et al., 2014) and obesity (Isasi et al., 2015) and unhealthy behaviors (Abraido-Lanza et al., 2005; Markides & Eschbach, 2011) increase with longer duration in the United States. Existing studies have linked earlier age at migration to higher acculturation (greater adoption of Anglo-American culture; Colón-López et al., 2009), and higher acculturation (e.g., language preferences, social exposures, preferences, etc.) is associated with better cognitive function among Hispanic/Latino adults (Lamar et al., 2020; Mendoza et al., 2022; Tan et al., 2021). Acculturation fits within the sociocultural component of the health disparities research framework (Hill et al., 2015).

Previous research on birthplace, age at migration, and cognitive aging among older Hispanic/Latino immigrants in the United States have shown mixed results. Compared to USB Mexican Americans, Mexican Americans who migrated to the United States during middle-life (aged 20–49) have higher cognitive test performance, and men who migrated at middle-life have slower rates of cognitive decline (Hill et al., 2012). Similarly, relative to USB Mexican Americans, middle-life immigrant Mexican men (migrating aged 20–49) have lower risk of cognitive impairment, whereas late-life immigrant women (after age 50) have higher risk (Garcia et al., 2017). Other studies do not detect such differences. For example, a study of the U.S.-representative Health and Retirement Study finds that age at migration differences among older foreign-born Hispanic/Latino adults in baseline cognitive function are fully explained by years of education. Relative to the foreign-born, USB Hispanic/Latino adults have lower baseline cognitive scores and similar cognitive decline trajectories (Garcia et al., 2020). Additional cross-sectional and longitudinal studies are needed to untangle birthplace and age at migration differences in cognitive aging and the potential drivers for these differences.

In this article, we sought to contribute novel evidence (including test-specific cognitive outcomes) and deepen understanding of the drivers of differences in later-life cognitive function among older Hispanic/Latino migrants in the United States. We examined drivers across the National Institute on Aging Health Disparities Research Framework (i.e., environmental, sociocultural, behavioral, and biological) in order to better understand the life course implications of age at migration on late-life cognitive function among diverse Hispanic/Latino adults. To do so, we employed data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) and SOL-Investigation of Neurocognitive Aging (SOL-INCA; González et al., 2019). We built on prior research to compare baseline cognitive scores and cognitive change (over an average 7-year period) by birthplace and age at migration. Specifically, we test the hypotheses that:

(H1) Foreign-born Hispanic/Latino adults at various ages at migration would have comparable or better global and test-specific cognitive outcomes compared to U.S.-born Hispanics/Latinos.

(H2) Birthplace or age at migration differences would be partially explained by 1. Demographic (e.g., Hispanic/Latino heritage), 2. Socio-economic (e.g., educational attainment), 3. Health conditions (e.g., depressive symptoms), 4. Health behaviors (e.g., diet), 5. Acculturation (e.g., language use) factors. Previous studies highlight the salience of socioeconomic factors and health conditions and behaviors (Abraido-Lanza et al., 2005; Markides & Eschbach, 2011; Vega et al., 2009; Velasco-Mondragon et al., 2016), so we expect these to be especially important.

Because the potential drivers of differences in later-life cognitive function among U.S.-born Hispanic/Latino older adults and those migrating at different ages are multiple, we use the Oaxaca–Blinder decomposition method to analyze the relative contributions of demographic, socioeconomics, health behaviors, health conditions, and acculturation factors in explaining cognitive testing differences among U.S.-born Hispanic/Latino older adults versus Hispanic/Latino older adults who migrated to the United States at different ages.

Research Design and Methods

Study Design

The HCHS/SOL is a population-based, prospective cohort study of diverse Latinos (Visit 1 2008–2011). The study designs and sampling procedures have been published (LaVange et al., 2010). HCHS/SOL used complex survey design and sampling procedures to obtain representative data of diverse Latinos. A total of N = 16,415 self-identified Latinos (ages 18–74) were sampled from four major U.S. metropolitan areas: Bronx, NY; Chicago, IL; Miami, FL; and San Diego, CA. The Study of Latinos-Investigation of Neurocognitive Aging (SOL-INCA), is an ancillary study of HCHS/SOL. At HCHS/SOL Visit 2, the Coordinating Center identified 7,420 potentially eligible participants for SOL-INCA. Inclusion criteria were: (1) Visit 1 neurocognitive testing completion, (2) Visit 2 completion, and (3) aged 50 years and older at Visit 2. Of these, 222 were determined to be ineligible (e.g., missing Visit 1 data), 569 were eligible but refused, and 6,377 were eligible and agreed to participate. To address possible biases (e.g., by sample attrition), study-specific probability weights that adjust for nonresponse (e.g., deaths) were generated to allow generalization of estimates to the HCHS/SOL target populations aged 50 and older. For the analysis, we excluded 134 individuals with mixed or missing Hispanic/Latino heritage information. We also excluded 23 individuals whose age at Visit 1 was less than 45 and did not participate in the neurocognitive module, and n = 573 individuals who had any missing covariate for a final unweighted analytic sample of n = 5,647. See Supplementary Table 1 for a simplified participant flow diagram. Individuals included in the analysis did not significantly differ in age (p = .613), sex (p = .4685), or education (p = .2563) compared to those excluded.

Primary Exposure

Birthplace/age at migration to the United States (50 states and Washington DC/mainland) were coded into four categories based on self-reported birthplace (USB vs foreign-born) and age at migration to the United States. Note that the mainland and island-born distinction pertains to Puerto Ricans with island-born individuals counted as migrants. We were interested in individuals who migrated prior to age 16 (school age); migrated between ages 16 and 34 (potentially selected for health/capacity to work); and migrated after age 35 (less likely to be selected for health and receive educational resources). Age at migration to the United States was calculated by subtracting the age of the respondents at interview from the reported number of years spent in the United States. In the overall HCHS/SOL sample, the vast majority (89%) of those who were USB were second generation, with only 5% being third generation and 6% being fourth generation or higher. Therefore, we did not split the USB group by generational status.

Cognitive Outcomes

All testing was administered in the participant’s preferred language. Visit 1 cognitive tests measured learning and memory (the Brief-Spanish English Verbal Learning Test Sum of Trials and Delayed Recall), processing speed (Digit-Symbol Substitution; DSS), and language (Word Fluency (WF)—phonemic fluency). SOL-INCA repeated this battery (henceforth Visit 2) on average 7 years later and added two additional tests of processing speed and executive functioning (Trails Making Test—A and B), which were reverse scored so that higher numbers equal faster (better) performance. Therefore, Trails A and B scores were only used in the Visit 2 cognitive function outcomes. Cognitive scores were standardized (z-scored) at each visit, and global composite scores (based on four scores for Visit 1 and six scores for Visit 2) were generated by averaging across these values. To examine cognitive change, change scores for repeated tests were calculated using regression-based techniques. First, cognitive performance at Visit 2 was modeled as a function of performance at Visit 1, adjusting for days lapsed between assessments (Duff, 2012). Second, global and test-specific standardized measures of change were calculated using (T2 − T2pred)/RMSEA where T2 was the respondent cognitive score at Visit 2, T2pred was their predicted score, and RMSEA was the standard error of the regression estimator (Arellano-Morales et al., 2015; Ramos et al., 2019). More detailed explanation can be found in Duff (Duff, 2012).

Covariates

Covariates, measured at Visit 1, were organized into domains. First, the Demographic domain included age (years), sex (male, female), marital status (single, married/partnered, separated/divorced/widowed), and heritage (Central American, Cuban, Dominican, Mexican, Puerto Rican, South American). Second, the Socioeconomic domain includes educational attainment (years); income (≤$10,000, $10,001−$20,000, $20,001−$40,000, $40,001−$75,000, more than $75,000, not reported/missing); occupation (nonskilled worker, service worker, skilled worker, professional/technical administrative/executive or office staff, other occupation, retired and not currently employed, not retired and not currently employed); and health insurance status. Third, the Health Behavior domain includes self-reported sleep duration (<6 hr, 6−9 hr, >9 hr); diet based on consumption of saturated fatty acids, dietary fiber, calcium, and potassium and dichotomized based on gender specific percentiles (<60th vs ≥60th, see Daviglus et al. 2012); exercise using activity level per 2008 U.S. physical activity guidelines (inactive, low activity, medium activity, high activity; U.S. Department of Health and Human Services, 2008). The fourth domain is Health, which includes self-reported stroke/transient ischemic attack (yes/no) and depressive symptoms (Center for Epidemiologic Studies-Depression scale, revised score, ranging 0 to 30). The final domain is the Acculturation and includes participants’ language preference (Spanish, English); and the Short Acculturation Scale for Hispanics (SASH)—Social and Language subscales (Ellison et al., 2011; Marin & Gamba, 1996). Items were rated on a 5-point scale, with lower scores indicating lower levels of acculturation. We used a 10-item modified version of the SASH that included two factors (the SASH-Language and SASH-Social subscales; Arellano-Morales et al., 2015). Items in the SASH-Language subscale items asked about exposure/speaking in various contexts with participants ranking items from “Only Spanish” to “Only English.” SASH-Social subscale items asked about exposure/preferences for spending time with groups of people in various contexts with participants ranking items from “All Hispanics/Latinos” to “All Americans.” All regression models also include study site (Bronx, Chicago, Miami, and San Diego).

Analytic Approach

First, we generated weighted descriptive statistics for the analytic subpopulation. Second, we performed survey generalized linear regressions to model the associations between migration status with Visit 1 cognitive function, Visit 2 cognitive function, and change in cognitive function as a function of (1) the primary exposure (crude), and (2) fully adjusting for covariates. Our primary analyses focused on global measures of Visits 1 and 2 cognitive function and cognitive change (Table 2). Test-specific scores were considered in secondary analyses (Table 3). Estimates for the regression coefficients and their standard errors are presented in Tables 2 and 3. In post hoc analyses, we calculate average marginal means, and conduct ANOVA-based contrasts, and plot these estimates and their 95% confidence intervals to highlight differences across the exposure groups. Third, we used Blinder–Oaxaca decomposition techniques (Firpo, 2017; Jann, 2008; Rahimi & Hashemi Nazari, 2021) to examine contributors to mean outcome differences in cognitive performance and change among the three migrant groups, independently, relative to USB. Briefly, the procedures allow us to separate group differentials (e.g., USB vs migrated prior to age 16) into parts explained by differences attributable to characteristics (the domains mentioned above) and unexplained (residual) parts attributable to potentially unmeasured characteristics. To do so, outcome differences are divided into “endowment effects” or group differences attributable to model covariates and coefficient differences. For example, when two groups are compared, the endowment component would quantify the change in the specific age at migration group average outcome if they had the same characteristics (i.e., predictor values) as the USB group. The coefficient component quantifies the expected change in migration group average outcome under the assumption that they have the same coefficient values for the covariates as the USB group. Detailed attribution of single (or sets of) covariates to explained and unexplained parts can be generated to evaluate the contribution of differences in domains to group differences.

Table 2.

Associations Between Age at Migration With Global Cognition at Visit 1, at Visit 2, and 7-Year Cognitive Change (Betas, 95% Confidence Intervals)

Born in US	Visit 1		Visit 2		Change
	Global cognition
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
	Reference	Reference	Reference	Reference	Reference	Reference
Migration age <16 years	−0.26*** [−0.39; −0.14]	−0.02 [−0.12; 0.07]	−0.36*** [−0.49; −0.22]	−0.06 [−0.17; 0.04]	−0.21* [−0.39; −0.04]	−0.09 [−0.26; 0.08]
Migration age 16–34 years	−0.30*** [−0.40; −0.20]	0.04 [−0.06; 0.14]	−0.40*** [−0.51; −0.29]	−0.01 [−0.13; 0.10]	−0.22** [−0.37; −0.08]	−0.11 [−0.28; 0.07]
Migration age >34 years	−0.34*** [−0.44; −0.24]	0.04 [−0.07; 0.14]	−0.39*** [−0.50; −0.28]	0.02 [−0.10; 0.15]	−0.15* [−0.29; −0.00]	−0.04 [−0.23; 0.15]

Note: Model 1: crude; Model 2 controlled for age, sex, marital status, heritage, education, income, occupation, health insurance status, sleep duration, diet, physical activity, self-reported stroke/transient ischemic attack, depressive symptoms, social acculturation, and language acculturation. ***p <.001; **p <.01; *p <.05.

Table 3.

Associations Between Age at Migration With Individual Tests at Visit 1, at Visit 2, and 7-Year Cognitive Change (Betas, 95% Confidence Intervals)

	Visit 1		Visit 2		Change
	Z-score SEVLT Sum
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Born in US	Reference	Reference	Reference	Reference	Reference	Reference
Migration age <16 years	−0.21** [−0.36; −0.06]	−0.02 [−0.15; 0.11]	−0.26** [−0.44; −0.09]	−0.05 [−0.20; 0.10]	−0.14 [−0.31; 0.02]	−0.04 [−0.20; 0.11]
Migration age 16–34 years	−0.11 [−0.23; 0.02]	0.10 [−0.04; 0.24]	−0.13 [−0.27; 0.00]	−0.01 [−0.17; 0.15]	−0.07 [−0.20; 0.06]	−0.10 [−0.28; 0.07]
Migration age >34 years	−0.16* [−0.29; −0.03]	0.09 [−0.06; 0.24]	−0.06 [−0.19; 0.07]	0.04 [−0.13; 0.22]	0.04 [−0.08; 0.17]	−0.04 [−0.22; 0.14]
	Z-Score SEVLT recall
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Born in US	Reference	Reference	Reference	Reference	Reference	Reference
Migration age <16 years	−0.18* [−0.35; −0.02]	−0.01 [−0.15; 0.13]	−0.25** [−0.42; −0.08]	−0.06 [−0.22; 0.10]	−0.16 [−0.35; 0.03]	−0.05 [−0.23; 0.13]
Migration age 16–34 years	−0.06 [−0.19; 0.06]	0.08 [−0.07; 0.22]	−0.12 [−0.27; 0.04]	−0.01 [−0.18; 0.17]	−0.09 [−0.25; 0.07]	−0.06 [−0.24; 0.12]
Migration age >34 years	−0.12 [−0.25; 0.00]	0.06 [−0.09; 0.20]	−0.09 [−0.24; 0.06]	0.03 [−0.16; 0.22]	−0.04 [−0.20; 0.12]	−0.01 [−0.21; 0.19]
	WF
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
Born in US	Reference	Reference	Reference	Reference	Reference	Reference
Migration age <16 years	−0.24** [−0.38; −0.09]	−0.01 [−0.14; 0.12]	−0.37*** [−0.55; −0.20]	−0.05 [−0.22; 0.11]	−0.23* [−0.42; −0.05]	−0.08 [−0.27; 0.11]
Migration age 16–34 years	−0.26*** [−0.38; −0.14]	0.13 [−0.01; 0.27]	−0.41*** [−0.55; −0.27]	0.14 [−0.02; 0.31]	−0.24** [−0.40; −0.09]	0.05 [−0.15; 0.26]
Migration age >34 years	−0.23*** [−0.35; −0.11]	0.18* [0.04; 0.33]	−0.38*** [−0.52; −0.24]	0.22* [0.04; 0.40]	−0.24** [−0.39; −0.09]	0.09 [−0.12; 0.31]
	DSS
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
	Reference	Reference	Reference	Reference	Reference	Reference
Born in US	−0.43*** [−0.57; −0.29]	−0.08 [−0.18; 0.02]	−0.53*** [−0.69; −0.36]	−0.10 [−0.21; 0.02]	−0.19* [−0.36;−0.01]	−0.08 [−0.26; 0.10]
Migration age <16 years	−0.79*** [−0.90; −0.69]	−0.17** [−0.28; −0.07]	−0.93*** [−1.05; −0.80]	−0.18** [−0.32; −0.05]	−0.21** [−0.36;−0.07]	−0.07 [−0.25; 0.11]
Migration age 16–34 years	−0.86*** [−0.97; −0.75]	−0.20*** [−0.31; −0.09]	−1.01*** [−1.15; −0.88]	−0.21** [−0.35; −0.07]	−0.22** [−0.36;−0.07]	−0.04 [−0.23; 0.16]
			Trails A (reversed)
Born in US			Reference	Reference
Migration age <16 years			−0.21*** [−0.30; −0.12]	0.10* [0.02; 0.18]
Migration age 16–34 years			−0.53*** [−0.62; −0.45]	0.02 [−0.08; 0.12]
Migration age >34 years			−0.71*** [−0.80; −0.61]	−0.09 [−0.21; 0.03]
			Trails B (reversed)
			Reference	Reference
Born in US			0.00 [0.00; 0.00]	0.00 [0.00; 0.00]
Migration age <16 years			−0.45*** [−0.61; −0.30]	−0.11 [−0.24; 0.02]
Migration age 16–34 years			−0.80*** [−0.92; −0.67]	−0.20** [−0.35; −0.05]
Migration age >34 years			−0.79*** [−0.91; −0.67]	−0.12 [−0.26; 0.03]

Notes: B-SEVLT = Brief-Spanish English Learning Test; DSS = digit symbol substitution; WF = word fluency.

Trails A and B were only administered at Visit 2. Model 1: crude; Model 2: controlled for age, sex, marital status, heritage, education, income, occupation, health insurance status, sleep duration, diet, physical activity, self-reported stroke/transient ischemic attack, depressive symptoms, social acculturation, and language acculturation. ***p <.001; **p <.01; *p <.05.

Results

Descriptive Statistics

Fifty-five percent of the target population were female (Table 1). The sample is diverse in terms of age at migration: 7.6% were born USB; 8.7% people migrated before age 16; 37.1% migrated between 16 and 34; and 46.6% migrated after age 34. Most USB participants were of Mexican and Puerto Rican descent, whereas foreign-born participants were more varied in heritage. The USB had, on average, higher income (less likely to report income lower than $40,000), higher educational attainment (more likely to have greater than high school education), and greater social and language acculturation (SASH scores). Relative to other groups, the USB were more likely to be single, report skilled employment or professional and technical jobs. Fifty-seven percent of individuals had health insurance with higher rates for USB and those who migrated earlier in age. Individuals migrating at ≥35 years old were more likely to be Cuban and less likely to be Puerto Rican, tested at the Miami site, report employment in the service sector, and complete their interview in Spanish. They were less likely to be insured and reported fewer depressive symptoms on average.

Table 1.

Descriptives by Age at Migration for the Study of Latinos-Investigation of Neurocognitive Aging

		Foreign/island-born
	U.S./mainland-born	Migrated <16 years	Migrated 16–34 years	Migrated >34 years	Total	p Value
Unweighted n	473	521	2430	2,223
Weighted %	7.6	8.7	39.3	44.4	100.0
	% (SE)	% (SE)	% (SE)	% (SE)	% (SE)
Sex
Female	51.7 (2.99)	54.0 (2.76)	52.9 (1.41)	57.3 (1.29)	54.9 (0.89)	p = .065
Male	48.3 (2.99)	46.0 (2.76)	47.1 (1.41)	42.7 (1.29)	45.1 (0.89)
Marital status
Single	30.4 (2.65)	29.6 (2.79)	15.5 (1.14)	12.0 (1.07)	16.3 (0.80)
Married/partnered	44.1 (3.22)	43.0 (2.95)	56.6 (1.71)	59.4 (1.86)	55.7 (1.27)
Separated/divorced/widowed	25.5 (2.85)	27.4 (2.48)	27.9 (1.48)	28.6 (1.63)	28.0 (1.06)
Latino/Hispanic heritage
Dominican	1.3 (0.54)	4.0 (1.16)	12.8 (1.25)	10.3 (1.13)	10.0 (0.83)	p < .001
Central American	0.5 (0.29)	0.7 (0.40)	9.8 (0.97)	7.8 (0.74)	7.4 (0.57)
Cuban	5.3 (1.86)	15.0 (2.54)	13.8 (1.55)	41.8 (2.96)	25.7 (1.90)
Mexican	40.0 (3.19)	27.6 (2.60)	44.2 (2.07)	28.8 (2.49)	35.6 (1.77)
Puerto Rican	52.2 (3.27)	51.8 (3.15)	14.0 (1.04)	4.5 (0.69)	16.0 (0.86)
South American	0.7 (0.31)	0.8 (0.44)	5.4 (0.64)	6.7 (0.70)	5.2 (0.41)
Study site
Bronx	45.8 (3.30)	47.5 (3.24)	29.7 (1.97)	17.3 (1.76)	27.0 (1.55)	p < .001
Chicago	12.0 (1.50)	13.9 (1.78)	18.2 (1.22)	8.7 (0.92)	13.1 (0.88)
Miami	5.2 (1.89)	16.2 (2.57)	22.7 (2.17)	54.3 (3.04)	34.8 (2.26)
San Diego	37.1 (3.09)	22.4 (2.43)	29.4 (2.03)	19.7 (2.33)	25.1 (1.76)
Educational attainment
Less than HS	24.7 (2.76)	38.5 (2.96)	45.1 (1.58)	35.9 (1.78)	38.9 (1.17)	p < .001
HS or Equivalent	20.2 (2.28)	22.4 (2.58)	23.2 (1.33)	20.2 (1.26)	21.6 (0.82)
Greater than HS or equivalent	55.0 (3.14)	39.1 (2.93)	31.7 (1.53)	43.9 (1.57)	39.5 (1.05)
Household income
≤$10,000	10.2 (1.47)	15.6 (2.07)	15.3 (1.23)	19.1 (1.35)	16.6 (0.88)	p < .001
$10,000–$20,000	22.1 (2.58)	27.5 (2.48)	30.2 (1.44)	32.2 (1.56)	30.2 (1.03)
$20,000–$40,000	27.3 (2.71)	30.2 (2.83)	31.3 (1.34)	26.0 (1.28)	28.5 (0.87)
$40,000–$75,000	22.3 (2.52)	12.0 (1.76)	14.6 (1.22)	7.4 (0.93)	11.8 (0.75)
More than $75,000	13.9 (2.05)	9.0 (2.06)	3.3 (0.54)	1.9 (0.56)	4.0 (0.43)
Not reported/missing	4.1 (1.29)	5.6 (1.69)	5.5 (0.64)	13.3 (1.31)	8.9 (0.68)
Occupation
Nonskilled worker	6.7 (1.48)	8.0 (1.36)	14.3 (1.10)	11.5 (0.98)	11.9 (0.63)	p < .001
Service worker	4.5 (0.92)	6.8 (1.49)	8.5 (0.79)	14.6 (1.11)	10.7 (0.63)
Skilled worker	16.6 (2.12)	11.1 (1.75)	11.0 (0.95)	7.5 (0.81)	9.9 (0.59)
Professional/technical administrative/executive or office staff	12.1 (1.92)	8.1 (1.67)	4.3 (0.56)	4.9 (0.85)	5.5 (0.54)
Other occupation	8.0 (1.88)	3.6 (1.09)	8.9 (0.79)	7.7 (0.81)	7.9 (0.50)
Retired and not currently employed	19.4 (2.38)	22.2 (2.57)	20.6 (1.51)	17.8 (1.34)	19.4 (0.92)
Not retired and not currently employed	32.7 (2.77)	40.3 (2.85)	32.4 (1.43)	36.1 (1.54)	34.7 (0.93)
Insurance
No current health insurance	24.0 (2.86)	25.2 (2.53)	39.4 (1.67)	52.3 (1.74)	42.7 (1.25)	p < .001
Currently have health insurance	76.0 (2.86)	74.8 (2.53)	60.6 (1.67)	47.7 (1.74)	57.3 (1.25)
Language preference
Spanish	22.7 (2.47)	51.4 (3.12)	95.8 (0.64)	99.1 (0.23)	87.9 (0.74)	p < .001
English	77.3 (2.47)	48.6 (3.12)	4.2 (0.64)	0.9 (0.23)	12.1 (0.74)
Sleep duration
<6 hr	9.8 (1.68)	9.5 (1.72)	7.2 (0.69)	5.5 (0.56)	6.8 (0.44)	p = .002
6–9 hr	72.7 (2.61)	70.3 (2.92)	78.5 (1.18)	80.9 (1.26)	78.4 (0.77)
>9 hr	17.5 (2.19)	20.2 (2.62)	14.3 (1.05)	13.6 (1.14)	14.8 (0.67)
Stroke/TIA
No	96.2 (1.05)	95.5 (1.23)	96.7 (0.69)	96.9 (0.61)	96.6 (0.41)	p = .704
Yes	3.8 (1.05)	4.5 (1.23)	3.3 (0.69)	3.1 (0.61)	3.4 (0.41)
Diet
<60%	54.9 (3.08)	57.7 (3.03)	42.0 (1.60)	41.8 (1.74)	44.2 (1.18)	p < .001
≥60%	45.1 (3.08)	42.3 (3.03)	58.0 (1.60)	58.2 (1.74)	55.8 (1.18)
Physical activity
Inactive	21.4 (2.45)	21.2 (2.10)	23.6 (1.27)	27.9 (1.36)	25.1 (0.89)	p = .019
Low activity	11.8 (1.70)	13.8 (2.03)	15.5 (1.28)	15.5 (1.05)	15.0 (0.75)
Medium activity	13.3 (2.23)	11.2 (1.79)	11.7 (0.88)	12.4 (1.07)	12.1 (0.65)
High activity	53.5 (2.85)	53.8 (2.85)	49.3 (1.59)	44.2 (1.36)	47.7 (0.96)
	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)
Age at baseline (years)	53.20 (7.04)	55.71 (8.01)	55.37 (8.43)	58.05 (7.81)	56.42 (8.20)	p < .001
SASH-Social subscale	2.65 (0.55)	2.51 (0.57)	2.17 (0.60)	1.92 (0.51)	2.12 (0.60)	p < .001
SASH-language subscale	3.64 (1.03)	2.92 (0.99)	1.57 (0.63)	1.24 (0.36)	1.70 (0.95)	p < .001
Depressive symptoms (CESD-10)	8.28 (6.50)	8.44 (7.30)	7.02 (6.34)	7.24 (5.95)	7.34 (6.30)	p = .006

Notes: CESD = Center for Epidemiological Studies-Depression scale; SASH = Short Acculturation Scale for Hispanics; SD = standard deviation; SE = standard error; TIA = transient ischemic attack; US = United States.

Diet percentile based on gender-specific profiles previously published in Daviglus et al. (2012).

Birthplace, age at migration, and cognitive functioning

Migrants, independent of age at migration, appeared to have lower global cognitive scores compared with USB in crude models, but this was fully explained through covariates adjustment (Table 2, Figure 1). We found similar associations for test-specific performance at Visit 1, although associations with WF and DSS were maintained after adjusting for covariates (Table 3). Notably, only in fully adjusted models, WF was better among individuals who migrated after 34 years compared to USB. In contrast, worse performance on the DSS was maintained in fully adjusted models for individuals who migrated between 16 and 34 and those who migrated after 34 years relative to the USB group. Similar patterns were also detected for global and test-specific cognitive performance at Visit 2. Given the wide range of covariates in our model, we then used Oaxaca–Blinder tests to determine which covariate domains best contribute to these differences.

Figure 1.

Associations between age at immigration with global cognition at Visit 1, global cognition at Visit 2, and 7-year cognitive change. Baseline = Visit 1; INCA = Visit 2. Model 1: crude. Model 2: adjusted for age, sex, marital status, heritage, education, income, occupation, health insurance status, sleep duration, diet, physical activity, self-reported stroke/transient ischemic attack, depressive symptoms, social acculturation, and language acculturation.

Effect decomposition of cognitive functioning

Model covariates explained 74% of the difference in Visit 1 global cognitive function between USB individuals and those migrating prior to 16 (Δ = −0.26; Supplementary Table 2; also see Figure 2). Of the total explained difference in estimates, 49% was attributed to socioeconomic domain and 39% to the acculturation domain. The model covariates also explained 92% of the difference in global cognitive function (Δ = −0.30) between the USB and those migrated between the ages of 16 and 34. Socioeconomic and acculturation domains were substantial and equal contributors to the explained difference (51% and 57%, respectively). Finally, when comparing the USB with migrants arriving at or after age 35, socioeconomic and acculturation domains were responsible for 32% and 69% of the explained differences (Δ = −0.34), respectively. Demographic, health condition, and health behavior domains each had smaller contributions to between-group differences across the USB—migrant group cognitive function at Visit 1 comparisons (i.e., ≥8%). When significant differences were uncovered in specific test scores, effect decompositions showed similar patterns to global cognitive function whereby acculturation and socioeconomic domains were the primary drivers of explained differences in Visit 1 performance (Supplementary Table 3).

Figure 2.

Results from the Oaxaca–Blinder tests comparing percent of global cognition at Visit 1, global cognition at Visit 2, and 7-year cognitive change scores explained, split by age at immigration group relative to U.S.-born group. Baseline = Visit 1; INCA = Visit 2. Groupings consisted of the following: demographics: age, sex, Hispanic/Latino background, and marital status. Health: depressive symptoms and stroke/transient ischemic attack. Socioeconomic: education, income, occupation, and insurance status. Acculturation: cognitive testing language preference, SASH-Language, and SASH-Social. Health Behaviors: sleep duration, physical activity (U.S. Department of Health and Human Services, 2008), diet (score > 60%; Daviglus et al., 2012). Accult = acculturation; Demo = demographics; Hbehavior = health behavior; SASH = Short Acculturation Scale for Hispanics; SES = socioeconomic status.

Similar patterns (i.e., the socioeconomic and acculturation domains explained the largest percent differences between USB and each migrant group) were observed for global cognitive function and individual test scores at Visit 2 (Supplementary Tables 2 and 4). At times, the demographic domain had larger nominal contributions to Visit 2 cognitive performance than in Visit 1.

Birthplace, age at migration, and 7-year cognitive change

In crude models, all migrant groups, regardless of age at migration, had greater adverse change in cognitive function compared to the USB group (Table 2). However, group differences were fully explained by adjustment to model covariates. Regarding specific cognitive tests, all migrant groups had greater adverse change in WF and DSS than the USB group, and this too was fully explained by covariates adjustment (Table 3).

Effect decomposition of average 7-year cognitive change

Model covariates explained nearly half (47%) of the difference in estimates between the USB and those who migrated prior to age 16 with most of the explained difference attributed to the socioeconomic domain (77%) followed by demographic domain (36%) (Supplementary Table 2, Figure 2). The acculturation domain was disadvantageous (i.e., contributed to more adverse cognitive change) among individuals who migrated before 16 years of age migrants (−12%). Model covariates explained 64% of the difference in global cognitive change between USB and those migrating between the ages of 16 and 34. Demographic and socioeconomic domains contributed 25% and 31% of the explained difference, respectively, whereas 52% was attributed to the acculturation domain. Similar patterns emerged for differences in global cognitive change between USB and those who migrated after age 35. That is, 34% of the explained difference was specific to the demographic domain, 51% to the socioeconomic domain, and 19% to the acculturation domain. As with cognitive performance at Visits 1 and 2, health conditions and health behaviors did not contribute substantively to differences between groups.

Test-specific differences in cognitive change were largely restricted to verbal fluency and processing speed (Supplementary Table 5). Demographic, socioeconomic, and acculturation domains explained nearly all the differences in change in WF between USB and those migrating prior to age 16 as well as those migrating after age 35, whereas only socioeconomic and acculturation domains contributed to explain differences between the USB group relative to those migrating between ages 16 and 34. Demographic and socioeconomic domains contributed to differences in change in DSS between USB and all-age-at migration groups. The acculturative domain was less critical in the explained differences in change in DSS among those migrating prior to 16 and between 16 and 34, compared to the USB, but absorbed 28.6% of the explained difference between USB and those migrating after the age of 34.

Discussion and Implications

Using two timepoints of data from diverse Hispanic/Latino populations in the United States, our study contributes to understanding cognitive aging in this understudied population by providing new evidence for how individuals’ birthplace and age at migration are associated with their cognitive function cross-sectionally, as well as over time. Guided by the National Institute on Aging’s health disparities research framework (Hill et al., 2015), we found a significant difference in cognitive score as well as cognitive change over time within this cohort depending upon important lived experience variables of interest. Furthermore, we decomposed the birthplace difference in cognitive scores in older adulthood using several pathways: demographic, socioeconomic, acculturation, health, and health behaviors. Our findings underscore socioeconomic and acculturation pathways.

In contrast to previous research of comparable or advantageous cognitive health for migrant Hispanics/Latino adults (Garcia et al., 2017, 2020; Hill et al., 2012) (Hypothesis H1), our study found evidence that Hispanics/Latino migrants had lower Visit 1 cognitive scores than USB individuals. The most likely explanation is that these lower scores reflect lower levels of educational attainment and associated test-taking skills among Hispanic/Latino migrant groups. Indeed, the inclusion of covariates—particularly related to socioeconomic status—completely tempers cognitive score differences as discussed below.

Further, unlike previous research that found some age-at-migration differences in cognitive health for foreign-born Hispanics/Latinos (Garcia et al., 2020), in our study, Hispanics/Latino individuals who migrated during young adulthood appeared to have similar Visit 1 cognitive scores as those who migrated during childhood or mid/late-adulthood. Study findings may differ due to sampling. For example, Garcia and colleagues (2020) reported on data from the Health and Retirement Study, which is nationally representative, whereas HCHS/SOL is representative of the four sites sampled (the Bronx, Chicago, Miami, and San Diego). Sampling from regions with high proportions of Hispanic/Latino adults, HCHS/SOL may increase the likelihood of participants (regardless of country of birth) residing in enclaves, which has been associated with lower prevalence of cognitive impairment (Weden et al., 2017). Additionally, HCHS/SOL accounted for Hispanic/Latino heritage. Although census region, which overlaps with heritage to a certain degree, was controlled for in Garcia and colleagues (2020) analysis, it cannot fully account for heritage. Furthermore, certain cultural factors that have been implicated in cognitive function (e.g., familism) may be more salient in specific heritage groups due to sociopolitical events and policies (Barbosa et al., 2023; Estrella et al., 2024). Additionally, certain heritage groups may be more inclined to acculturate given access to certain resources (e.g., citizenship for those of Puerto Rican heritage; Baldoz & Ayala, 2013). Future studies are needed to determine whether such cultural and acculturative factors influence cognitive function regardless of age at migration.

In terms of decomposing the differences, a large proportion of the total birthplace/age-at-migration difference in cognitive score—74% and more—was explained by the observed domains (Hypothesis H2). Overall, these results indicate the importance of socioeconomic and acculturation factors in these disparities, which is consistent with previous literature, but suggests that these domains are particularly relevant to cognitive outcomes for migrant Hispanic/Latino adults compared to their USB peers (Haan et al., 2011; Mendoza et al., 2022; Tan et al., 2021; Zeki Al Hazzouri et al., 2011). Although both domains significantly contributed to differences in global cognitive function and change in cognitive function over time across all-age-at migration groups, socioeconomic factors appeared to be most influential for those who migrated at younger ages, while acculturation factors appeared to have a greater role for individuals who migrated at older ages. In contrast and surprisingly, self-reported health and health behavior domains explained very little, signaling a more limited role for them in intra-Hispanic/Latino birthplace/age-at-migration disparities and the health selection and health behavior explanations they might represent. Similarly, Garcia and colleagues (2020) found that health and health behaviors did not account for differences in cognitive function between non-Hispanic/Latino White and Hispanic/Latino adults of any age at migration. Importantly, our results do not suggest that health or health behaviors are irrelevant to cognition, nor that health status is the same across Hispanic/Latino adults regardless of place of birth and age at migration. Rather, our results indicate that health and health behaviors may have a universal impact on cognitive function in this population. Taken together, these results require further study to understand how age at migration influences Hispanic/Latino migrants’ cognitive health.

Our study leveraged the unique properties of the HCHS/SOL and SOL-INCA to strengthen the existing literature on age at migration and cognition. Notably, our sample was majority migrant (>90%) compared to other studies (e.g., the Health and Retirement Study and Hispanic Established Populations for the Epidemiological Study of the Elderly) whose samples are less than 50% migrant (Downer et al., 2018; Garcia et al., 2017, 2020). Our oldest age at migration group (>34 years) was particularly large compared to previous work, which may have increased our ability to identify which domains were more relevant to cognitive function in this group versus the U.S.-born group. Thus, our results may better generalize to the Hispanic/Latino migrant population, which historically lacks representation in cognitive aging research.

Most studies of age at migration and later-life cognitive function have focused on mental status scores and proxies of dementia outcomes among older Mexican Americans (Downer et al., 2018; Garcia et al., 2017, 2020). We have expanded the existing literature to include test-specific outcomes among diverse Hispanic/Latino individuals who do not have a dementia diagnosis. In our study, age at migration was not associated with any learning and memory outcomes. We also found that individuals who migrated after age 34 had better word fluency than USB Hispanic/Latino individuals, possibly related to different experiences of bilingual (English–Spanish) language use throughout the life course. Work with SOL-INCA’s native Spanish speakers suggests that higher English use and proficiency are linked to better performance and maintenance of word fluency over time (Lamar et al., 2023). Migration to the United States between 16–34 and after 34 years of age was associated with worse performance on processing speed and speeded executive functioning tests (based on the results from the DSS at Visits 1 and 2 and Trails B, which was only administered at Visit 2). Similar, though less robust results were observed among Hispanic/Latino and Asian American older adults in the Kaiser Healthy Aging and Diverse Life Experiences study, possibly reflecting small effect sizes of age at migration on executive functioning and other cognitive domains (Meyer et al., 2023). We did not detect differences in 7-year change across any of the cognitive domains examined, suggesting that age-at-migration-based cross-sectional differences do not reflect risk for cognitive decline.

A primary innovation of our study is to focus fully on diverse Hispanic/Latino individuals of various heritages and their cognitive health, trajectories, and socioeconomic inequalities, without referencing other groups (e.g., non-Hispanic/Latino White individuals). Such within-population research is key for understanding and improving Hispanic/Latino cognitive aging. However, our study has some limitations. First, the age at migration cutoffs varies between studies, and this may have contributed to slight differences in findings between cohorts. Importantly, we selected our cutoffs to reflect the rationale behind migration (see above). Second, although our sample is representative of Hispanic/Latino adults from four metropolitan areas, it is not nationally representative. These four areas have high proportions of Hispanic/Latino individuals, suggesting increased access to others in that community and enclaves that may have specific benefits and disadvantages on overall health and well-being (Osypuk et al., 2009). Third, most of our participants were first- or second-generation migrants, and we could not examine generational status within the USB group. Fourth, we did not examine Hispanic/Latino heritage-specific (e.g., Puerto Rican, Mexican) relationships between age at migration with cognitive outcomes. Although omitting this increases our statistical power, it may overlook some unique relationships based on heritage. Importantly, reasons for migration (e.g., fleeing persecution, seeking job opportunities, and pursuing higher education) broadly vary by Hispanic/Latino heritage in addition to at the individual level. Future work should incorporate self-report measures of reasons for migration. Fifth, despite evidence that Hispanic/Latino older adults’ cognitive aging in the United States and Puerto Rico may be stratified by race and color (Liu, Crowe, et al., 2022; Liu, Telles, et al., 2022), we did not examine these, and future study should. Sixth, future study could also include other markers of SES (e.g., childhood socioeconomic position, home ownership) and change (e.g., income change at older ages) across the life course (Filigrana et al., 2023; Zeki Al Hazzouri et al., 2011). Seventh, our findings warrant further studies of the specific factors (e.g., bilingualism) that may connect acculturation to cognitive aging. Finally, our study focuses on age at migration and cannot, due to data limitations, identify duration of residence at the same time. Future studies should consider doing so (Van Hook et al., 2018).

Across a diverse sample of over 5,000 Hispanic/Latino middle-aged and older adults, relative cognitive advantages for USB relative to migrants at various ages at migration were attributable to sociodemographic factors. Further, socioeconomic factors appeared to account for a larger share of variance in cognitive outcomes for migrants (at any age at migration) than for USB individuals. Also compared to US-born individuals, acculturative factors were also associated with cognitive outcomes to a greater extent among migrants, especially among those who migrated in late adolescence through early adulthood and later adulthood. Our findings underscore the importance of examining whether disrupting socioeconomic inequalities closes gaps in cognitive aging among Hispanic/Latino adults.

Author Note

1. In this article, we use the term U.S.-born to refer to persons born in the United States. 50 states/DC and the term foreign-born refers to persons born in a foreign country or in a U.S. territory (e.g., Puerto Rico). We treat individuals born in U.S. territories as such because they encounter similar migration and acculturation experiences as the foreign-born (Immerwahr, 2019; Landale et al., 2006).

Funding

This work was supported by the National Institute of Aging [R01AG064778, R01AG048642, R56AG048642, RF1AG054548, RF1AG061022, R01AG075758, and R01AG068392]. Additional support includes K08AG075351, L30AG074401, and U54CA267789 to Dr. A. M. Stickel, P30AG012839 to Dr. W. H. Dow, R01AG062711 to Dr. M. Lamar; R01NS131225 to Dr. T. P. Garcia, P30AG062429 to Dr. H. M. González, and RF1AG077639 to Dr. C. R. Isasi. The Hispanic Community Health Study/Study of Latinos is a collaborative study supported by contracts from the National Heart, Lung, and Blood Institute (NHLBI) to the University of North Carolina [HHSN268201300001I/N01-HC-65233], University of Miami [HHSN268201300004I/N01-HC-65234], Albert Einstein College of Medicine [HHSN268201300002I/N01-HC-65235], University of Illinois at Chicago [HHSN268201300003I/N01- HC-65236 Northwestern University], and San Diego State University [HHSN268201300005I/N01-HC-65237]. The following Institutes/Centers/Offices have contributed to the HCHS/SOL through a transfer of funds to the NHLBI: National Institute on Minority Health and Health Disparities, National Institute on Deafness and Other Communication Disorders, National Institute of Dental and Craniofacial Research, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Neurological Disorders and Stroke, and NIH Institution-Office of Dietary Supplements.

Conflict of Interest

None.

Data Availability

Data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) and SOL-Investigation of Neurocognitive Aging (SOL-INCA) are available at https://biolincc.nhlbi.nih.gov/studies/hchssol/. This study was not preregistered.