Gender, Age of Migration, and Cognitive Life Expectancies Among Older Latinos: Evidence From the Health and Retirement Study

Abstract

Objectives

Migration and gender are important factors that differentiate the Latino immigrant experience in the United States. We investigate the association between nativity status, age of migration, and cognitive life expectancies among a nationally representative sample of Latino adults aged 50 and older to explore whether age of migration and gender influence cognitive aging across the life course.

Methods

This study used data from the Health and Retirement Study (1998–2016) to estimate Sullivan-based life tables of cognitive life expectancies by nativity, age of migration, and gender for older Latino adults. Cognitive status was based on the Langa–Weir algorithm. We test for both within-group (i.e., nativity and age of migration) and gender differences to explore the overall burden of disease among this rapidly growing population.

Results

Foreign-born Latinos, regardless of age of migration or gender, spend a greater number of years after age 50 with cognitive impairment/no dementia than U.S.-born Latinos. However, the number of years spent with dementia varied by subgroup with midlife immigrant men and late-life immigrant men and women exhibiting a significant disadvantage relative to the U.S.-born. Furthermore, we document a gender disadvantage for all Latino women, regardless of immigrant status.

Discussion

The robust relationship between nativity, age of migration, and cognitive aging suggests that older foreign-born Latinos experiencing cognitive decline may place serious burdens on families. Future research should target the needs of different subgroups of older Latinos who are entering their last decades of life to develop culturally appropriate long-term care programs.

Cognitive aging, Migration and the life course, Racial/ethnic data disaggregation

Alzheimer’s disease (AD) is a major public health concern that not only affects individuals with the disease and their families but also has a high economic impact on society. Currently, there are more than 6.2 million Americans aged 65 and older living with AD, and recent population estimates project this number to more than double by 2060 (Alzheimer’s Association, 2021). Extensive research shows that older U.S. Latinos experience an early onset of cognitive decline, and have a higher incidence and prevalence of AD compared to non-Latino Whites (Chen & Zissimopoulos, 2018; Matthews et al., 2019; Mayeda et al., 2016; Moon et al., 2019; Zhu et al., 2021). However, the Latino pan-ethnic categorization obscures important nativity differences, and among the foreign-born, migrations histories that are important to understanding the lifetime burden of AD. As a part of the gendered life course framework, these differences among Latinos interact with gender (i.e., reason for migration and health selectivity) in unique ways to shape late-life health and mortality (Reyes & Garcia, 2020). Thus, researchers and policymakers must devote additional attention to understanding differences in cognitive risk profiles by gender among older Latinos to develop effective policies aimed at optimizing health and reducing the burden of the disease among this rapidly aging and diverse population (Babulal et al., 2019; Brewster et al., 2019).

The higher incidence and prevalence of AD among Latinos have been attributed in part to multiple factors, including longer life expectancy (Arias et al., 2021; Garcia et al., 2019), lower levels of educational attainment (Flores et al., 2017; Garcia et al., 2021), higher rates of poverty (Creamer, 2020), and a higher prevalence of chronic conditions such as diabetes and cardiovascular disease (Brown, 2018; Garcia, Garcia, et al., 2018; Garcia, Garcia, et al., 2018), that disproportionately affect Latino populations compared to non-Latino Whites. Furthermore, where a person was born, and among the foreign-born, what stage of the life course they migrated to the United States may shape life experiences for key predictors of AD, such as education, and access to resources that may be protective of cognitive health.

Life course theory emphasizes the importance of the “timing” of life events, such as migration, and the accumulation of exposures that ultimately shape late-life health outcomes (Dannefer, 2003; Elder et al., 2003). Despite having lower levels of education, foreign-born Latinos have longer life expectancies than their U.S.-born counterparts (Arias et al., 2020), particularly those who migrated in midlife (Reyes & Garcia, 2020). This may in part reflect the positive selection of immigrants who migrate in midlife, although longer lives do not always mean healthier lives (Boen & Hummer, 2019; Gubernskaya, 2015; Reyes & Garcia, 2020). Life course timing of migration may also shape cognitive health among older Latinos in unique ways by serving as a proxy for exposure to varied educational institutions, environmental risks, and health care systems in diverse countries of origin. Previous research has found that foreign-born Latinos spend a greater proportion of their lives with dementia (Garcia et al., 2019). In addition, findings from a regional study of older Mexican-origin adults showed that midlife migrants exhibit an advantage relative to their U.S.-born and foreign-born counterparts in the proportion of remaining late-life years spent with cognitive impairment (Garcia, Saenz, et al., 2018).

Gender and age of migration may also intersect in unique ways to shape AD given gender differences in selectivity during the migration process (Donato, 2010; Hill et al., 2012). These gender differences in selectivity are more pronounced for those migrating in midlife, rather than as children or older adults. Women migrating to the United States are often following husbands or other family members in part due to shifts in immigration policy that favored family reunification (Donato, 2010). This pattern of male-led migration suggests that successful male migrants may be more positively selected on health and other attributes than women. The factors associated with positive health selection among migrants, better health and education, are also protective for AD.

Gender has also emerged as a key factor for understanding the cognitive decline and prevalence of AD among older adults. Women have been shown to experience faster cognitive decline and exhibit greater rates of AD than men (Chanti-Ketterl et al., 2021; Levine et al., 2021; Zhu et al., 2021). These gender differences are likely a result of both social and biological differences that influence the risk of AD. For instance, women on average have longer life expectancies than men (Arias & Xu, 2022) which puts them at a higher risk of AD. In addition, a gendered life course perspective argues gender differences across the life course in educational attainment, socioeconomic status, chronic health conditions, and health behaviors will shape pathways that ultimately lead to variation in AD.

However, AD research aimed at exploring gender differences by nativity status and timing of migration among a nationally representative sample of older Latinos is limited. Our study builds on previous research documenting differences in AD among the U.S. Latino population by making two important contributions: first, we assess differences by nativity and age of migration among Latinos using nationally representative data, and second, we test for differences by gender. Understanding the heterogeneity in cognitive risk profiles of older Latinos both by nativity status (i.e., U.S.-born vs foreign-born) and age of migration (i.e., early-, mid-, and late-life) is necessary to disentangle how these within-group variations differentially contribute to AD among older Latinos. Thus, the primary aim of this study is to use a gendered life course perspective to examine gender, nativity, and age of migration differences in cognitive life expectancies among older Latinos using a nationally representative population.

Drawing on the gendered life course perspective, we hypothesize that among Latino men, midlife migrants will have a greater proportion of cognitively healthy life years compared to U.S.-born Latinos, late-life migrants will have a lower proportion of cognitively healthy years compared to U.S.-born Latinos, and early-life migrants will have a similar proportion of cognitively health years compared to U.S.-born Latinos.

We further hypothesize that Latino women will have worse cognitive health than men, (i.e., a greater number of years spent in an impaired state, and a lower proportion of cognitively healthy years), and that midlife migrant women will not exhibit the same advantage in cognition over U.S.-born women that is expected among men.

Method

Data

This study used data from the Health and Retirement Study (HRS), a nationally representative study of individuals over age 50 and their spouses in the United States. The HRS collects information every 2 years starting in 1992; we pool waves from 1998 to 2016 to have a sufficient sample size to estimate differences by gender, nativity, and age of migration, for cognitive life expectancies (cognitively normal, cognitively impaired not dementia [CIND], and dementia) and death for adults 50 years and older. Although the HRS follows up with individuals who become institutionalized, we exclude the observations of respondents who have entered nursing homes at follow-up. The final analytic sample includes 4,321 unique individuals and 36,012 person-years of data. Supplementary Table 1 provides a detailed listing of sample size for Latino subgroups by HRS survey.

Cognitive Assessment and Classification

We used the Telephone Interview for Cognitive Status (TICS-M; Brandt et al., 1988) to measure participants’ cognitive function. The adopted 27-point TICS-M scale includes: two indices (0–20 points) for immediate and 5-min delayed recall of 10-non semantically linked words for memory performance; assessment of working memory through a serial sevens subtraction test (0–5 points); and a processing speed measure tested through counting backward (0–2 points). TICS-M was administered to participants responding directly but not to those responding through a proxy.

To classify individuals according to cognitive status, we adopted the Langa–Weir (LW) algorithm (Crimmins et al., 2011; Langa et al., 2008; Plassman et al., 2007; Rocca et al., 2011). LW classified self-reporting participants relative to the threshold extrapolated from and validated against details diagnostic assessment derived from the ADAMS, a detailed neurological and neuropsychological study of a subsample of HRS participants. For direct participant interviews, LW classified individuals as cognitively normal if scoring between 12 and 27 on the TICS-M, and CIND and dementia if their scores fall between 7–11 and 0–6, respectively.

For proxy respondents no direct assessment of cognitive function is available. However, the LW algorithm uses proxy reported information on respondents’ instrumental activities of daily living (range 0–5), and memory (0 = excellent to 4 = poor), as well as interviewer assessment of respondents’ cognitive status (0 = no cognitive impairment, 1 = possible impairment, and 2 = impaired; starting in the 2000 wave) to generate a proxy-based index (range 0–11; 0–9 in prior waves) of cognitive function with higher values indicating higher levels of cognitive impairment. Proxied participants are subsequently classified as cognitively normal if scoring between 0 and 2, CIND and dementia if scoring 3–5 (3–4 in 1998) and 6–11 (5–9 in 1998), respectively. These classifications are combined into a single measure allowing classification across both self and proxy responding participants. LW classifications have been validated in multiple studies and have been used extensively in the literature (Crimmins et al., 2011; Langa et al., 2008; Plassman et al., 2007; Rocca et al., 2011).

Stratification

Given the scope of this work, several HRS measures were used to group participants according to ethnicity, nativity, and age of migration. Latinos were divided into U.S.-born and foreign-born categories based on self-reported nativity status. Furthermore, we divided foreign-born Latinos into three subgroups based on the timing of migration to the United States: (1) individuals migrating in their formative years (i.e., before age 18), (2) individuals migrating in early adulthood, and postformative years (between the ages of 18 and 35), and (3) those migrating later in their adulthood (between the ages 36 and 49). The sampling design of the HRS recruits participants at age 51; therefore, we are unable to study foreign-born respondents who migrated in late life (i.e., after age 50). Similar groupings have been adopted in prior studies using the HRS (Gubernskaya, 2015). Given known gender differences in aging within and across immigrant groups, all analyses were conducted by stratifying on self-reported gender. Age and educational years were also included to characterize the target population by gender, nativity/age at migration, and cognitive status.

Cognitive Life Expectancies

We used data on age-specific CIND and dementia prevalence combined with mortality information to model total, cognitive healthy, CIND, and dementia life expectancies and generate Sullivan Life Tables stratified by gender, ethnicity, nativity, and age of migration.

Generating these estimates was based on a multistep process. First, we used logistic regression models to generate gender-specific weighted prevalence estimates of CIND and dementia, generalizable to noninstitutionalized community-dwelling adults 50 years and older in the ethnicity, nativity, and age of migration groups of interest. Second, we estimated weighted mortality rates using Gompertz hazard models and used these estimates to calculate total life expectancy. Subsequently, estimates of total life expectancy at age 50 were split into three subsets of health states (cognitive normal, CIND, and dementia) according to age-specific prevalence estimates for each of these cognitive classifications. In this context, status-specific cognitive life expectancies represent the average group-specific number of life span years at age 50 remaining in each cognitive state (Jagger et al., 2006). Third, standard errors and empirical confidence intervals for group-specific estimates, and differences in estimates (across ethnicity, nativity and age of migration groups, by gender), within each cognitive state were generated using bootstrap techniques. Specifically, we used 300 bootstrap samples to estimate sampling variability for the life table functions and life expectancies.

An estimate of “healthy” life expectancy captures the number of years an individual can expect to live given the absence of a prespecified single health condition or cluster of comorbidities (Dubois & Hebert, 2006). In the context of this study, healthy indicates the absence of CIND or dementia. Estimates of healthy life expectancy facilitate cross-population epidemiological comparisons of disease burden taking into consideration the variability across group age distributions (Jagger et al., 2006). Life expectancy states (healthy or their disease complements) can thus be interpreted as the average number of years spent in a specific health/disease state for a member of a group in a specific cohort assuming a shared, constant, risk of disease and mortality.

Descriptives

For all Latino subgroups, the distribution of cognitive functioning shifts to higher proportions with dementia across age groups (see Table 1). U.S.-born Latina women in the 50–59 age group have low levels of dementia compared to midlife and late-life migrants but by age 80 this advantage is no longer evident as a greater proportion of U.S.-born women over age 80 have dementia compared to all three immigrant groups. In the youngest age group (50–59), midlife and late-life migrant men have relatively low levels of dementia, but in the oldest age group, it is early-life migrant men who exhibit the lowest levels of dementia. These results suggest the timing and trajectory of cognitive decline vary across groups. Average age differed across cognitive status groups, with those with dementia being the oldest for all groups. We also found a gradient in the average education by cognitive status, whereby individuals with normal cognition reported more years of education compared to those with CIND, and dementia.

Table 1.

Descriptive Characteristics of U.S.-Born and Foreign-Born Age of Migration Latino Subgroups by Gender and Cognitive Classification

	USBL			ELL (0–17)			MLL (18–34)			LLL (35+)
	CN	CIND	Dementia	CN	CIND	Dementia	CN	CIND	Dementia	CN	CIND	Dementia
Panel A: females
Age groups (%):
50–59	79.0	18.7	2.3	77.1	21.3	1.7	76.9	19.2	3.8	77.0	19.6	3.5
60–69	72.8	21.3	5.8	61.4	34.1	4.5	69.5	24.0	6.4	62.3	29.3	8.4
70–79	54.7	31.8	13.5	40.2	50.3	9.5	52.4	32.0	15.6	52.7	35.6	11.7
80+	23.8	36.1	40.1	11.0	59.4	29.5	27.4	41.4	31.2	28.5	34.1	37.5
Age (mean)	62.7	67.8	76.8	59.3	63.6	69.8	62.1	66.5	71.1	64.8	70.1	78.1
Education (mean)	11.8	9.1	6.8	9.5	6.7	5.5	9.9	6.0	5.0	8.4	5.4	4.6
Panel B: males
Age groups (%):
50–59	79.8	17.1	3.1	70.9	24.5	4.6	76.3	21.3	2.3	88.6	10.0	1.5
60–69	74.4	20.5	5.1	71.6	23.5	4.9	68.4	27.9	3.6	71.4	25.3	3.3
70–79	57.0	32.9	10.0	57.5	30.3	12.2	60.3	31.7	8.0	61.9	29.3	8.8
80+	23.9	37.4	38.7	20.1	53.7	26.2	22.4	45.7	31.9	34.7	36.3	29.0
Age (mean)	62.2	66.4	72.3	59.8	61.3	63.6	61.2	64.2	71.2	65.5	72.3	80.7
Education (mean)	12.1	9.7	8.0	11.7	9.4	9.1	8.8	5.2	3.6	9.5	5.3	3.6

Notes: CN = cognitively normal; CIND = cognitively impaired not dementia. CIND and dementia classification are based on Langa–Weir algorithm. ELL = early-life migrating Latinos; LLL = later-life migrating Latinos; MLL = early adulthood/midlife migrating Latinos; USBL = U.S.-born Latinos.

Source: HRS 1998–2016; 4,321 unique observations contribute 18,891 age-specific observations.

Life expectancy estimates

At age 50, U.S.-born Latino males (27.6 years) had the lowest average total life expectancy, whereas late-life migrating females (36.4 years) had the highest expected average (Table 2; Figure 1). Midlife migrating females and late-life migrating males had the highest number of cognitively healthy years (20.9 years for females and 23.1 years for males); the equivalent of 59% and 67% of their life expectancy at age 50, respectively. We found important variations in average years spent in CIND and dementia by nativity status and age of migration as well as gender (see Figure 1).

Table 2.

Total and Cognitive Classification Specific Life Expectancies by Nativity and Age of Migration for Latino Males and Females

	Female
	Totala	Healthy	CIND	Dementia	Ratio healthy
USBL	33.9 (0.78)	20.4 (0.53)	8.6 (0.41)	4.9 (0.42)	0.60
ELL (0–17)	33.4 (1.79)	17.6 (1.05)	12.6 (1.44)	3.1 (0.77)	0.53
MLL (18–34)	35.6 (0.91)	20.9 (0.75)	9.9 (0.66)	4.8 (0.57)	0.59
LLL (35+)	36.4 (1.13)	20.2 (0.81)	10.6 (0.64)	5.6 (0.65)	0.56
Exposure	8687	1755	5805.5	4227.5

	Male
	Total	Healthy	CIND	Dementia	Ratio healthy
USBL	27.6 (0.73)	18.4 (0.63)	6.5 (0.4)	2.6 (0.3)	0.67
ELL (0–17)	30 (1.75)	19.2 (1.4)	8.4 (1.16)	2.5 (0.82)	0.64
MLL (18–34)	34.0 (1.42)	20.7 (0.86)	9.9 (0.86)	3.4 (0.68)	0.61
LLL (35+)	34.7 (1.05)	23.1 (0.89)	8.3 (0.70)	3.3 (0.55)	0.67
Exposure	6,640	1,363	4,777.5	2,756.5

Notes: CIND = cognitively impaired not dementia; ELL = early-life migrating Latinos; LLL = later-life migrating Latinos; MLL = early adulthood/midlife migrating Latinos; USBL = U.S.-born Latinos.

^aAverage life expectancy and (empirical intervals).

Figure 1.

Cognitive life expectancies at age 50 by gender among U.S.-born and foreign-born age of migration Latino subgroups. CIND = cognitively impaired not dementia; ELL = early-life migrating Latinos; LE = life expectancy; LLL = later-life migrating Latinos; MLL = early adulthood/midlife migrating Latinos; USBL = U.S.-born Latinos.

Variability in CIND and dementia years by age of migration

Foreign-born Latinos, regardless of age of migration or gender, spent a greater number of years after age 50 with CIND compared to U.S.-born Latinos (see Figure 2). Early-life female migrants had the highest average expected number of years spent in CIND (12.6 years), followed by late-life (10.6 years) and midlife migrants (9.9 years). The average life spent in CIND among foreign-born Latino males was largely similar. Late-life female migrants had the highest average number of years spent with dementia (5.6 years).

Figure 2.

Gender-specific differences in total and cognitive life expectancies at age 50 between U.S.-born and foreign-born age of migration Latino subgroups. CIND = cognitively impaired not dementia; ELL = early-life migrating Latinos; LE = life expectancy; LLL = later-life migrating Latinos; MLL = early adulthood/midlife migrating Latinos; USBL = U.S.-born Latinos.

Variability in cognitively healthy years by gender

Foreign-born females on average and independent of age of migration live longer than males. However, despite their shorter life span, males had consistently higher ratios of healthy life years compared to their female counterparts (see Figure 3).

Figure 3.

Differences in average total and cognitive life expectancy at age 50 between males and females within U.S.-born and foreign-born age of migration Latino subgroups. CIND = cognitively impaired not dementia; ELL = early-life migrating Latinos; LE = life expectancy; LLL = later-life migrating Latinos; MLL = early adulthood/midlife migrating Latinos; USBL = U.S.-born Latinos

Discussion

Promoting healthy aging and reducing risk factors for AD is a primary goal of the National Plan to Address Alzheimer’s Disease: 2021 Update (US Department of Health and Human Services, 2021), yet nationally representative studies examining the diversity within the older Latino adult population in the United States are lacking. Despite a growing literature examining cognitive outcomes among older Latinos residing in the United States, several critical questions regarding the cognitive aging of this diverse population remain unanswered. Much of the prior literature using nationally representative data have examined Latinos as a pan-ethnic group or has overlooked the demographic diversity within the older foreign-born Latino population. Using a gendered life course perspective and longitudinal nationally representative data from the HRS to examine the cognitive risk profiles of older Latinos by nativity status and age of migration we find significant variation in both the number and proportion of cognitively healthy years by gender and age of migration.

We find mixed support for our hypotheses. Midlife male migrants exhibit more healthy years compared to U.S.-born Latinos, but not a greater proportion of cognitively healthy years. In other words, they live longer but spend a similar proportion of later life with dementia. This suggests that positive selection for longer life expectancies does not translate to better cognitive health among older Latinos. This is also true for foreign-born Latina women who spend a greater number of years with CIND and have a similar proportion of cognitively healthy years compared to U.S.-born Latina women. We do find that late-life migrant men and women have a lower proportion of cognitively healthy years compared to U.S.-born men and women, respectively. Contrary to our hypothesis, we find that early-life migrants have worse cognitive health compared to their U.S.-born counterparts.

Heterogeneity among Latinos represents not only important demographic differences but also represents different life course trajectories that may have important implications for cognitive functioning in later life. The experiences and life course trajectories are shaped in large part by nativity status, as this determines access to and quality of educational attainment, a key determinant of cognitive functioning may shape life course trajectories. However, the interactions between educational attainment and nativity may be more complex because although immigrants who arrive as children may adapt more easily and have educational attainment levels similar to Latinos born in the United States, we find they have worse cognitive health outcomes.

Importantly, previous research has also suggested that immigrants who arrive in midlife are positively selected on health. (Angel et al., 2010; Gubernskaya, 2015). However, our results suggest this positive health selection does not extend to cognitive healthy years. These differences may represent the lower levels of education and acculturation stress experienced across the life course by many of these immigrants that are unaccounted for in our models (Boen & Hummer, 2019; Muñoz et al., 2019).

The above results shed light on how nativity and age of migration intersect with gender in important ways for older Latinos. Prior research has suggested midlife female migrants may be less health selected in the migration process and has found greater health declines after migration for women (Angel et al., 2015) and the immigrant mortality advantage to be not as strong for women as it is for men (Reyes & Garcia, 2020). Explanations for these differences have been attributed to the gendered life course that differentially shapes the social and economic outcomes of men and women. As men and women accumulate different skills and experiences over their life course, these may shape cognitive health in important ways.

A notable limitation of this study is the reliance on a threshold-based algorithm for dementia classification, namely LW that has not been validated among Latino populations. Recent work by Gianattasio and colleagues tested the sensitivity, specificity, and accuracy of different threshold-based algorithms for dementia classification (Gianattasio et al., 2019). While overall, LW provides appropriate characteristics for estimating generalizable dementia prevalence and incidence rates (i.e., high specificity and accuracy), this work finds critical differences in the performance of several algorithms over subpopulation stratas (including by gender, age, education, and race/ethnicity) that varied across training and validation sets. For Latinos in the training set (n = 76), LW had an 86% (67–100) sensitivity, 71% (59–83) specificity, and 75% (65–84) accuracy, which was higher than non-Latino Blacks, but with the exception of sensitivity, lower than non-Latino Whites. Although sensitivity was markedly lower in the validation set (n = 31), LW specificity was 77% (61–91), and accuracy was 66% (50%–80%). Notably, the sample of Latinos in the training and validation data used in that study was relatively small; the analyses urge caution in overinterpretation when subpopulation stratas are used, particularly in the context of racial/ethnic disparities. Thus, additional work should be conducted, and larger data sets generated to validate classification algorithms in understudied and underrepresented populations, including subgroups of Latinos from various country-of-origin backgrounds. Future studies should build on the results presented above by applying our methods to alternative algorithmic classifications as detailed in prior research (Gianattasio et al., 2020).

The robust relationship between nativity status, age of migration, and cognitive aging suggests that foreign-born Latino adults at risk for cognitive impairment may place serious burdens on families. Research focusing on modifiable risk factors for foreign-born adults, including social determinants of health earlier in the migration process, can help reduce the overall burden of AD in this population. This issue merits special attention in the development of community-based long-term care programs to appropriately target the specific needs of different subgroups of older Latinos who are entering their last decades of life.

Funding

We gratefully acknowledge the financial support for this research provided by the National Institute on Aging (P30AG043073).

Conflict of Interest

None declared.

Author Contributions

M. A. Garcia planned the study, secured funding, and wrote the initial draft of the manuscript; W. Tarraf and C.-T. Chiu performed the data analyses; A. M. Reyes helped plan the study and write the initial draft of the manuscript; all authors contributed to the interpretations of the results, revisions of the manuscript, and approval of the final submission.