Dual Trajectories of Dementia and Social Support in the Mexican-Origin Population

Abstract

Background and Objectives

In the next few decades, the number of Mexican American older adults with Alzheimer’s disease and related disorders will increase dramatically. Given that this population underutilizes formal care services, the degree of care responsibilities in Mexican American families is likely to increase at the same time. However, little is known about the changing need for assistance with instrumental day-to-day activities and emotional support by long-term patterns of cognitive impairment.

Research Design and Methods

We use 7 waves of the Hispanic Established Populations for the Epidemiologic Studies of the Elderly (1992/1993–2010/2011) and trajectory modeling to describe long-term patterns of perceived emotional and instrumental support, and dementia.

Results

Results revealed 2 latent classes of both emotional and instrumental support trajectories: low and high support. Specifically, those living alone were more likely to belong to the group with low support than to that with high support. Three latent classes for likely dementia were also revealed: likely dementia, increasing impairment, and no impairment. Those living alone were more likely to belong to the increasing impairment and likely dementia groups. The dual trajectory of emotional and instrumental support with likely dementia revealed that the probability of belonging to the low-support group was highest for those with increasing impairment.

Discussion and Implications

These findings highlight the risk and vulnerability of those who live alone concerning perceived social support and dementia. Implications of the findings for the potential dependency burden on Latino caregivers are discussed.

In the United States, the number of dementia cases in older Latinos is expected to dramatically increase (Llanque & Enriquez, 2012). This is an important problem because the life expectancy for Latinos currently outpaces other ethnic groups by 2.5 years (Arias, 2011). Extended longevity coupled with a disproportionate share of the burden of dementia means Latino older adults are spending a greater portion of their lives with dementia and functional disability and, therefore, need for assistance (Angel et al., 2016; Hayward et al., 2014). The Mexican-origin population represents about 65% of the Latino population in the United States (Malavé & Giordani, 2015) and as one of the fastest growing segments of the aging population is at especially high risk for cognitive impairment and dementia (Downer et al., 2018).

In general, Latino families, particularly those of Mexican origin, often benefit from aspects of their culture that have proven to be socially protective, especially their strong family orientation and systems of informal support (Angel & Angel, 2009). However, Latino families—like other families—are undergoing changes that will affect their ability to continue to care for their elders. The more prevalent role of women in the labor force, geographic dispersion of families, and increased dementia make it difficult for Latino families to maintain the cultural tradition of nearly exclusive informal care of elders. The Mexican American population experiences this hardship especially profoundly, as they are more economically disadvantaged than non-Latino whites and African Americans (Macartney et al., 2013). As of yet, we know very little about the need for emotional and instrumental support as dementia progresses in the Mexican-origin population.

We contribute to existing studies on dementia-related support needs by focusing on the largest subgroup of Latinos in the United States, the Mexican-origin population, and by describing the role of dementia for both emotional and instrumental support. In the current study, we use data from 2,880 Mexican Americans aged 65 and older from seven waves of the Hispanic Established Populations for the Epidemiologic Studies of the Elderly (H-EPESE, 1993/1994–2010/2011). By understanding how social support needs change at home as Mexican American older adults’ cognitive impairment worsens, we can make better predictions about the number of older adults and their family members at risk of needing additional support.

Latinos are more likely to develop mild cognitive impairment and Alzheimer’s disease than non-Latino whites and suffer from cognitive impairment for a longer time with more severe symptoms (Garcia et al., 2019; Hinton et al., 2006; Howrey et al., 2015; Raji et al., 2010). Latinos with mild cognitive impairments are more likely to progress into full Alzheimer’s than non-Latino whites (Manly et al., 2008). Their cognitive decline has been linked to increased fragility (Raji et al., 2010). Evidence from the H-EPESE shows that rates of dementia and possible dementia in the Mexican-origin population are higher than in the general U.S. population (Haan et al., 2003), and there is an indication of high levels of dementia-related behavioral symptoms (Rote et al., 2015; Salazar et al., 2017). Dementia is a key factor in the increase in disability over time for this population (SamperTernent et al., 2008), and cognitive functioning is a key component of the disablement process for Mexican Americans (Peek et al., 2005). Less information is available on the role of dementia in the need for emotional and instrumental assistance among aging Latinos.

Theoretical models of cognitive and physical disability emphasize person-in-environment. The disablement model developed by Verbrugge and Jette (1994) distinguishes between intrinsic and actual disability. Intrinsic disability reflects the individual’s ability to perform tasks independently. As memory and the ability to perform mental tasks becomes more taxing, the person living with dementia begins to rely more and more on family members or other convoys of support to meet their needs. Actual disability is concerned with the role of environmental modifications or social support from others and the degree to which an individual can overcome intrinsic disability with the assistance of people or assistive devices. Social support consists of instrumental actions and emotional support that lead individuals to feel valued and supported (Cobb, 1976).

Dementia requires individuals to call upon or mobilize network members to aid in day-to-day tasks and care; however, social stressors, such as health deterioration or cognitive decline, can weaken social resources (Turner et al., 1995). In addition, individuals may voluntarily isolate themselves from distant others, particularly as problematic behaviors such as memory loss or paranoia manifest, and thus make interactions with others more challenging (Hinton et al., 2006; Manly et al., 2008). Without adequate social support, older adults are not able to address intrinsic disability, may experience a rapid increase in dementia symptoms, and are at elevated risk for institutionalization or hospitalization.

But perhaps most importantly, researchers should consider additional risk factors in a more comprehensive approach to understanding dementia (Schoeni et al., 2018). Studies reveal that living arrangements, especially living alone, are connected to low support availability and dementia. People living alone are more isolated from family and experience greater emotional loneliness, and loneliness has been associated with increased risk of dementia (Evans et al., 2019; Sutin et al., 2018). Furthermore, people who have dementia and live alone at home were found to face challenges in attending to hygiene, maintaining adequate nutrition, keeping safe, managing money, and coping with technology (Evans et al., 2016). Similarly, older adults living alone with cognitive impairment often experience a sense of precarity (Portacolone et al., 2019).

In addition to living arrangements, gender and immigrant status can also affect risk for dementia with studies indicating that immigrant Latinos and women are at greater risk for impairment than their U.S.-born counterparts (Garcia et al., 2019). Mexican American women and immigrant older adults may be at especially high risk for cognitive impairment in late life (Garcia et al., 2018) due to social and economic inequalities that disproportionately affect these groups. There is also some evidence that immigrant older Mexican Americans have smaller support networks to rely on when faced with cognitive and physical impairment (Angel et al., 2014) and that women may have larger support networks to draw from when faced with impairment (Monserud, 2019). Yet, there is a lack of information on the dynamics of support need and dementia by these factors (e.g., living alone, gender, and immigrant status) for the Mexican-origin population.

In the current study, we examine trajectories in emotional and instrumental support and likely dementia in a cohort of older Mexican Americans between 1993/1994 and 2010/2011. We then assess how likely dementia is implicated in self-reported need in two domains of emotional and instrumental support using dual trajectory modeling. Given the above review, we hypothesize that (H₁) there will be distinct trajectories of both emotional and instrumental support needs. We also expect to find that (H₂) older adults living alone, immigrant older adults, and men will be at greater risk for low support. For likely dementia, we hypothesize that there will be (H₃) distinct trajectories of dementia, and considering previous research (H₄) we expect to find that women, immigrants, and older adults will be at greater risk for dementia in late life. Finally, we expect (H₅) to find that those with increasing dementia risk will be at greatest risk for low support.

Data

We use data from the H-EPESE, a national study of Mexican American 65 years and older residing in the Southwestern United States (Texas, New Mexico, Colorado, Arizona, and California; Black et al., 1998). In 1993/1994, 3,050 respondents were interviewed in their home in English or Spanish and follow-up interviews were conducted every 2 years. We use data from seven waves (H-EPESE, 1993/1994–2010/2011) spanning more than 15 years. The study sample includes all respondents who had complete data for all baseline variables (N = 2,880).

Measures

Dependent Variables

Emotional Support is based on a response to the question: “Can you talk about your deepest problems with at least some of your family or friends?” with response options of most of the time (1), some of the time (2), or none of the time (3). If a respondent reported none of the time, they were coded as having low emotional support.

Instrumental Support is based on a response to the question: “In times of trouble can you count on some of your family?” with response options of most of the time (1), some of the time (2), or none of the time (3). If a respondent reported none of the time, they were coded as having low emotional support.

Likely Dementia was created using both the Mini-Mental Status Examination (MMSE), a brief, standardized method used to grade cognitive status (Folstein et al., 1975; Nguyen et al., 2003), and Instrumental Activities of Daily Living (IADL). This specification of likely dementia has been validated and supported by previous research on Spanish-speaking older adults (Mejia et al., 2004; Mejia-Arango & Gutierrez, 2011). The MMSE assesses memory and reasoning abilities, with special emphasis on orientation, attention, immediate and short-term memory recall, language, and the ability to follow simple verbal and written commands. If the respondents’ MMSE was in the range of 24–30, and reported 0 IADL, they were coded as (1), “No Impairment.” If the MMSE was in the range of 0–23, and the respondent reported 0 IADL, they were coded as (2), “Cognitive Impairment—No IADL.” If the MMSE was in the range of 0–23, and the respondent reported 1 or more IADL, they were coded as (3), “Likely Dementia.”

Key Independent Variables

Living Arrangements is measured using the number of people living in the house and marital status at Wave 1. If the number of people living in the house was one and the respondent’s marital status was unmarried, it was coded as (1), “living alone.” If the number of people living in the house was two and the respondent’s marital status was married, it was coded as (2), “Spouse Only.” If the number of people living in the house was greater than two, regardless of the respondent’s marital status, it was coded as (3), “Living with Others” (reference category).

Control Variables

Education is based on years of formal education (range: 0–17). We also control for respondent’s age and gender. Medicaid is an indicator of access to health care and safety net for low-income adults and is a dichotomous variable (1 = yes, 0 = no). Nativity is coded Mexican born = 1 and U.S. born = 0.

Analytic Strategy

Our analyses are based on the complete cohort and conducted in three steps. First, we use a latent class growth model in Mplus 7.0 to develop trajectories of instrumental, emotional social support needs and likely dementia over time. Contrary to variable-centered approaches such as regression that focuses on describing the relationships among variables, growth mixture modeling takes on a person-centered approach that focuses on describing the relationship among individuals (Muthén & Muthén, 2000). By classifying individuals with similar response patterns into distinct categories, growth mixture modeling allows for identification of unobserved, homogeneous subpopulations within a larger heterogeneous population, and description of how and why these subpopulations differ in their longitudinal patterns of change (Muthén & Muthén, 2000). We use the Bayesian Information Criterion (BIC) value and the sample-size adjusted Bayesian Information Criterion (SABIC) to select an appropriate and distinguishable number of trajectories. In addition, means of the estimated posterior probabilities assigned to specific trajectories are used to evaluate the precision of the identification of the selected number of trajectories. Second, multinomial logistic regression analyses are performed to determine which predictors vary significantly by trajectory group. Finally, we used dual trajectory modeling to identify the concurrent change in social support and likely dementia over time.

Results

Sample Characteristics

Table 1 describes the analytic sample at baseline. More than half of the sample is Mexican American women (58%) and currently married (56%). For living arrangements, 21% lived alone, 33% lived only with their spouse, and about 46% lived in extended households. The average age at baseline was 74 years old and the average years of formal education was 5. In terms of health and functioning, the means score on the MMSE was about 25 and the average number of IADLs reported was 2. About 34% of older Mexican Americans had “Likely Dementia,” which means scoring 23 or less on the MMSE and reporting one or more IADL disability. Less than half (43%) reported receiving Medicaid. At baseline, more than 25% of respondents report that they could not rely on others for emotional or instrumental support.

Table 1.

Baseline Descriptive Statistics (N = 2,880)

Characteristics	Range	Percentage/Mean
Sociodemographics
Female (%)		57.7
Married (%)		55.5
Living alone (%)		21.0
Spouse only (%)		33.2
Living with others (%)		45.8
Age (mean)	65–108	73.6
Education (mean)	0–17	4.84
Health and functioning
Cognitive impairment (MMSE) (mean)	0–30	24.7
IADL (mean)	0–10	2.06
Likely dementia	1–3	1.86
No impairment (%)		48
Cognitive impairment no IADL (%)		18
Likely dementia (%)		34
Financial resources
Medicaid (%)		43.1
Perceived support (%)
Low instrumental support		25.5
Low emotional support		27.2

Note: IADL = instrumental activities of daily living; MMSE = Mini-Mental State Examination.

Hypothesis 1: Perceived Social Support Trajectories

To evaluate our first hypothesis (H1), we first identified trajectories of perceived emotional and instrumental support. To determine the appropriate number of emotional support trajectory classes, we fit three consecutive models from the two- to four-class models. Comparisons of BIC, SABIC estimates, entropy, and interpretability of the classes indicated that the two-class model best fit the data. The first class shows a persistent need for emotional support throughout the study and includes 9.3% of the sample (N = 249). The second class shows an overall low need for emotional support, indicating that their needs have been met from family and friends (90.7%, N = 2,631). The gap between these groups is widest in Wave 1, but we see a trend in a narrowing of the gap over time (Supplementary Figure 1).

To determine the appropriate number of instrumental support trajectories, we followed a similar process as we did for emotional support. Based on model fit criteria, the best fit for instrumental support was also the two-class model. Class 1 has low instrumental support need throughout the waves, indicating that these older adults have their support needs met some or most of the time by family and friends. Class 1 includes most of the sample (89.9%, N = 2,623). Class 2 shows an overall high instrumental support need and includes 10.1% of the sample (N = 257). Like the results for emotional support, the gap between these groups is widest in Wave 1 but we see a trend of a narrowing over time (Supplementary Figure 2).

Hypothesis 2: Logistic Regression of Perceived Social Support Trajectories

Table 2 addresses our second hypothesis (H2) by presenting results of binomial logistic regression of perceived emotional and instrumental support by key background factors. The results reveal that living alone was significantly associated with risk for low levels of both emotional and instrumental support through the study. For example, older Mexican Americans living alone had 3.52 times higher odds of belonging to the low-emotional support group (vs. high-support group) and 4.25 times higher odds of belonging to low-instrumental support need group (vs. high-support group) compared to their counterparts living in an extended household. In addition, while there were no significant differences in emotional support between those who were only living with a spouse and those living alone, Mexican American older adults living with their spouses only also had a 1.66 higher odds of low instrumental support availability compared to those living with others.

Table 2.

Logistic Regression of Support Trajectory Group Membership

Low support (vs. High support)	Emotional support	Instrumental support
	OR (95% CI)	OR (95% CI)
Living alone (ref. living with others)	3.52 (2.10–5.91)***	4.25 (2.16– 6.92)***
Only spouse (ref. living with others)	1.25 (0.74–2.11)	1.66 (1.01–2.73)*
Age	1.00 (0.97–1.03)	1.00 (0.97–1.02)
Female	0.49 (0.32–0.76)**	0.63 (0.43–0.93)*
Education	0.92 (0.86–0.99)*	0.94 (0.88–1.00)*
Medicaid	0.55 (0.36–0.86)*	0.70 (0.46–1.06)
Foreign born	1.40 (0.92–2.12)	1.87 (1.27–2.75)**
Likely dementia (ref. no impairment)	0.75 (0.47–1.21)	0.69 (0.45–1.08)

Note: CI = confidence interval; OR = odds ratio.

*p < .05, **p < .01, ***p < .001.

Results also reveal that Mexican American women also had significantly lower odds of low emotional and instrumental support, indicating the men are at greater risk for having a lack of availability of both types of support. For socioeconomic factors, surprisingly, Mexican American older adults who received Medicaid were less likely to belong to the low-emotional support group than their counterparts not receiving Medicaid. However, years of formal education was related to a significantly lower risk for low emotional and instrumental support. For nativity, differences are only observed for instrumental support. Mexican-born older adults had 1.87 times greater odds of low instrumental support compared to older adults born in the United States. We also found that baseline likely dementia is unrelated to support need trajectories; however, given the dynamic nature between these two factors, we are interested in trajectories in likely dementia over time.

Hypothesis 3: Likely Dementia Trajectories

To determine the appropriate number of likely dementia trajectory classes, three consecutive models were fit from the two- to four-class models. Comparisons of BIC, SABIC estimate, entropy, and interpretability of the classes indicated that the three-class model best fit the data (Figure 1). The first class starts off high and remains high throughout the waves and consists of 43.8% of the sample (N = 1,235). This class includes those who are likely to have dementia. The second class shows an overall low trajectory, with 18.3% of the sample (N = 504). This trajectory represents those with normal cognition or no impairment. The final class that consists of 36.9% of the sample (N = 1,018) starts off low but gradually increases over time. Those in the third class have little or no impairment at Wave 1 but suffer from increased impairment toward Wave 7.

Figure 1.

Trajectories of likely dementia.

Hypothesis 4: Multinomial Logistic Regression of Likely Dementia Trajectories

Table 3 presents the results of the multinomial logistic regression of likely dementia by living arrangements and other background factors. For living arrangements, there was only one significant finding. Older Mexican Americans living alone had 1.54 higher odds of relative risk for increasing impairment in late life (vs. likely dementia throughout late life) compared to those living in extended households. In terms of background factors, advanced age, less education, and Medicaid receipt were all risk factors for both likely dementia and increasing dementia (vs. no impairment). Women were also at greater relative risk than men for likely dementia throughout late life versus increasing impairment.

Table 3.

Multinomial Logistic Regression of Likely Dementia

Comparison group vs. Referent group	RRR (95% CI)	p Value
Likely dementia vs. No impairment
Living alone (ref. living with others)	0.90 (0.71–1.14)	.661
Only spouse (ref. living with others)	0.89 (0.74–1.07)	.521
Age	1.22 (1.20–1.24)***	.000
Female	2.06 (1.74–2.44)***	.000
Education	0.73 (0.72–0.75)***	.000
Medicaid	3.03(2.50–3.65)***	.000
Foreign born	1.02 (0.85–1.22)	.927
Increasing impairment vs. No impairment
Living alone (ref. living with others)	1.38 (1.07–1.79)	.202
Only spouse (ref. living with others)	0.99 (0.82–1.20)	.956
Age	1.08 (1.06–1.10)***	.000
Female	0.91 (0.77–1.09)	.610
Education	0.86 (0.84–0.89)***	.000
Medicaid	1.77 (1.44–2.17)**	.005
Foreign born	0.90 (0.74–1.10)	.593
Increasing impairment vs. Likely dementia
Living alone (ref. living with others)	1.54 (1.32–1.79)**	.005
Only spouse (ref. living with others)	1.11 (0.97–1.79)	.450
Age	0.88 (0.87–0.89)***	.000
Female	0.44 (0.39–0.50)***	.000
Education	1.18 (1.15–1.20)***	.000
Medicaid	0.58 (0.52–0.66)***	.000
Foreign born	0.89 (0.79–1.00)	.319

Note: CI = confidence interval; RRR = relative risk ratio.

*p < .05, **p < .01, ***p < .001.

Hypothesis 4: Dual Trajectories of Social Support With Dementia

Next, we analyzed dual trajectories of perceived social support and dementia (Tables 4–6). The results for emotional support are listed in the table with the results for instrumental support listed in the parentheses. Results from these tables show there is a similar pattern for both types of support with dementia. Table 4 presents the probability of membership in each of the dementia trajectory groups based on support, wherein the probabilities for low and high support each add up to 100%. From the table, the conditional probability of the increasing dementia risk group belonging to low-emotional (49.1%) and low-instrumental (48.4%) support groups is relatively higher than the groups with likely dementia and those with no impairment.

Table 4.

Dual Trajectory Model for Emotional (Instrumental) Support and Likely Dementia: Probability of Likely Dementia conditional on Emotional (Instrumental) Support

Support Need	Likely Dementia	Conditional Probability
Low Support	Likely Dementia	40.6% (40.0%)
	Increasing	49.1% (48.4%)
	No Impairment	10.3% (11.6%)
High Support	Likely Dementia	43.6% (43.7%)
	Increasing	36.9% (36.8%)
	No Impairment	19.5% (19.5%)

Table 5.

Dual Trajectory Model for Emotional (Instrumental) Support and Likely Dementia: Probability of Emotional (Instrumental) Support conditional on Likely Dementia

Likely Dementia	Support Need	Conditional Probability
Likely Dementia	Low Support	8.2% (9.1%)
	High Support	91.8% (90.9%)
Increasing	Low Support	11.3% (12.6%)
	High Support	88.7% (87.4%)
No Impairment	Low Support	4.8% (6.1%)
	High Support	95.2% (93.9%)

Table 6.

Dual Trajectory Model for Emotional (Instrumental) Support and Likely Dementia: Joint Probability of Emotional (Instrumental) Support and Likely Dementia

Support Need	Likely Dementia	Increasing	No Impairment
Low Support	3.6% (4.0%)	4.3% (4.8%)	0.9% (1.1%)
High Support	39.8% (39.4%)	33.6% (33.1%)	17.8% (17.6%)

Table 5 presents the reverse set of conditional probabilities. Here, the probabilities for likely dementia, increasing dementia, and no impairment each add up to 100%. We observe a similar pattern with the conditional probability of low emotional support 8.2% for the likely dementia group, 11.3% for the increasing dementia group, and 4.8% for the no impairment group. A similar pattern was observed for instrumental support. Table 6 shows the joint probability of membership in each of the social support and likely dementia trajectories, where the six different probabilities all add up to 100%. This table shows the long-term and interrelated patterns between support and dementia over time. These results indicate that there was a symbiotic relationship between support needs and dementia. There was a very low probability of the sample without impairment reporting low support (0.9% for emotional support and 1.1% of instrumental support). This is followed by the group with low emotional support and likely dementia, which was at 3.6%. Finally, the increasing impairment group is in most need of both emotional and instrumental support.

Discussion

Overall, our findings demonstrate trajectories of long-term patterns in perceived emotional and instrumental support and “likely dementia” for older Mexican Americans. Our research supports prior studies that indicate most older adults and particularly older Mexican Americans tend to report consistent, adequate social support (Hill et al., 2016; Wilmoth & Silverstein, 2017). However, close to 10% of older Mexican Americans experience late life without emotional and instrumental support. Lack of support increases the risk for frailty (Peek et al., 2012), depression (Rote et al., 2015), hospitalization (LaPlante et al., 2004), and even premature mortality (Hill et al., 2016). Therefore, intervention-related strategies that increase access to social resources and allow for support exchanges should target older Mexican American men, immigrants, and those not living in extended households. Future research should also determine whether a lack of support is more consequential for dementia onset for these at-risk subgroups.

Supporting prior research (Downer et al., 2018; Garcia et al., 2019; Mayeda et al., 2016; Wu et al., 2019) we also find that a large percentage of older Mexican Americans arrive in late life with dementia (44%) or experience rapid dementia onset during this time (40%). The dual trajectory model provides additional context given that the initial dementia levels reveal less than the subsequent pathways that dementia and support availability take over time. Almost all who belong to the trajectory of low support belong to the likely dementia or increasing dementia risk group, accentuating the need for support interventions and the detection of symptoms earlier in the disease trajectory. Older Mexican Americans tend to live in the community and rely on family members rather than formal services for dementia care (National Academies of Sciences, Engineering, and Medicine, 2016). Our findings support the importance of promoting family cohesion and dementia care coordination for older Mexican Americans to improve the emotional and instrumental support received, which also has shown to alleviate Latino dementia caregiver stress (Llanque & Enriquez, 2012; Rote et al., 2019).

The results from our findings also underscore how living alone is a new marker of inadequate support. While living alone may be a sign of independence, increasingly living alone is often accompanied by poor health (Cudjoe et al., 2020). Given the vulnerabilities associated with the fourth age, the period in the life course which is characterized as a span of years of biological and functional decline, we were surprised to discover a larger fraction of the cohort with likely dementia living alone. Despite the perception of independence, these older adults with likely dementia are paradoxically at a particularly high risk of possible dependence, unnecessary emergency room visits, premature nursing home admission, and hospitalization (Veazie et al., 2019). Given extended longevity and the risk of dementia in this cohort, especially among immigrants, such diminished health and loss of autonomy and social isolation are the newest realities of age-related health and long-term care needs in the community context. Municipal governments that promote aging in place policies, staying in one’s home for as long as possible, will need to consider added health threats associated with built environments that lack adequate resources for social interaction and support (Buffel et al., 2012).

One of the limitations of the current study is the use of a single-item measure for emotional and instrumental support. Even though these measures are validated social support indices in the EPESE sister studies (Mendes de Leon, 2007) and the gold standard measurement of key domains of aging-related support, we recognize the far more complex construct of social support that limits our interpretation. For example, our measure of emotional support includes potential support from both family and friends while instrumental support is limited to family members. Additionally, a more in-depth approach to understanding support needs would focus on whether low support availability indicates a need for more support. Clearly, a qualitative research approach would shed meaningful light on the trajectory of the availability and perceived need of support and sources of potential support both within and outside the family. Second, our method to assess cognitive function is limited. The MMSE is the only marker of cognitive function available in the H-EPESE. The MMSE lacks adequate sensitivity and specificity available in more common tests of cognitive aging (Barnes et al., 2014). Third, future studies should consider to what extent impairment affects the validity of reports of support by utilizing reported support from family caregivers. Nonetheless, our findings provide robust models that pinpoint risk factors for emotional and instrumental support among the most vulnerable population of older adults and can be replicated with more detailed markers of cognition in future studies.

Our study also speaks to strategies outlined in the 2013 National Alzheimer’s Project Act, in which leaders in the field and the public recognize the importance of education and support for persons with Alzheimer’s disease and their families. This is achieved particularly through enhanced coordinated care with long-term support services and improved transition between various care settings and the home across the trajectory of dementia (Gallagher-Thompson et al., 2003, 2020). In addition, experts acknowledge the importance of tailoring these strategies for a culturally diverse older adult population and their caregivers (Gallagher-Thompson et al., 2003). In the present study, household extension represents a potential source of support. Even so, the mobilization of household members is fraught with difficulty. As dementia progresses, changing needs for assistance arise and can overwhelm caregivers when providing support. Improving access to culturally and linguistically appropriate formal dementia care options to supplement family support for older Mexican Americans will become increasingly important. Finally, future research should replicate these findings on other rapidly growing older vulnerable populations—including different racial and ethnic populations, such as Chinese, Korean, Vietnamese, and other groups in the United States as well as other subpopulations of Latinos with large family networks and low use of formal dementia care services. These estimates can help service providers, governments, and caregivers plan for increasing diversity in dementia care needs.

Funding

This research was supported by the National Institute on Aging grants [1R03AG059107-01 and 1R03AG063183-01].

Conflict of Interest

None declared.