Mapping the Trajectories of Social Relations for White, Black, and Hispanic/Latino Individuals Approaching Death With Dementia.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice J Gerontol B Psychol Sci Soc Sci . 2025 Apr 12;80(7):gbaf062. doi: 10.1093/geronb/gbaf062 Search in PMC Search in PubMed View in NLM Catalog Add to search Mapping the Trajectories of Social Relations for White, Black, and Hispanic/Latino Individuals Approaching Death With Dementia Zachary G Baker Zachary G Baker , PhD 1 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA Find articles by Zachary G Baker 1, ✉ , Andrew Alberth Andrew Alberth , MS, MPH 2 Department of Gerontology, University of Massachusetts Boston, Boston, Massachusetts, USA Find articles by Andrew Alberth 2 , M Aaron Guest M Aaron Guest , PhD 3 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA Find articles by M Aaron Guest 3 , Allie Peckham Allie Peckham , PhD 4 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA Find articles by Allie Peckham 4 , Joahana Segundo Joahana Segundo , MA 5 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA Find articles by Joahana Segundo 5 , Joseph Saenz Joseph Saenz , PhD 6 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA Find articles by Joseph Saenz 6 Editor: Marc A Garcia 7 Author information Article notes Copyright and License information 1 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA 2 Department of Gerontology, University of Massachusetts Boston, Boston, Massachusetts, USA 3 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA 4 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA 5 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA 6 Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA 7 (Social Sciences Section) ✉ Address correspondence to: Zachary G. Baker, PhD. E-mail: [email protected] Roles Marc A Garcia : PhD , Decision Editor Received 2024 Sep 25; Collection date 2025 Jul. © The Author(s) 2025. Published by Oxford University Press on behalf of the Gerontological Society of America. All rights reserved. For commercial re-use, please contact [email protected] for reprints and translation rights for reprints. All other permissions can be obtained through our RightsLink service via the Permissions link on the article page on our site—for further information please contact [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model ( https://academic.oup.com/pages/standard-publication-reuse-rights ) PMC Copyright notice PMCID: PMC12302131  PMID: 40220281 Abstract Objectives Larger social networks are associated with a lower risk of dementia, but little is known about how social networks shift as someone with dementia approaches death. We investigate these shifts while giving special attention to race and ethnicity, which are related to different dementia patterns, social network sizes, and social network makeup. Methods This study included 2,301 deceased people with dementia from the Health and Retirement Study (2004–2018; waves = 8). Multilevel models estimated associations between dementia, race/ethnicity, time, and close family and friend network size while controlling for several variables, including instrumental activities of daily living, age, proxy status, and disease count, using retrospective and proxy data. Results Social networks shrank linearly as death approached. A decrease in close friends primarily drove shrinkage. However, when race/ethnicity was crossed with time, Hispanic/Latino persons with dementia showed the opposite pattern. As Hispanic/Latino persons with dementia approached death, the number of close extended family members increased dramatically: one additional person every 4 years. Discussion Dementia risk, social networks, and patterns of social network shrinking are unequal across people of different races and ethnicities. Adding nuance to known patterns, network shrinkage may be a phenomenon of White persons with dementia. In contrast, patterns of stable or even increasing numbers of network members may better describe Black and Hispanic/Latino networks, respectively. These findings may reveal unique strengths of Black and Hispanic/Latino networks that could be leveraged to develop care and support for individuals with dementia and those they leave behind. Keywords: Alzheimer’s disease and related dementias, Bereavement, Death, Social networks Social relationships are critical in shaping individuals’ health and well-being, perhaps even protecting against their risk of developing dementia ( Khondoker et al., 2017 ; Kuiper et al., 2015 ). Robust social networks serve as a protective factor for unmet needs and a range of health outcomes, including heart disease, depression, and earlier mortality ( Holt-Lunstad et al., 2010 ; Miranda-Castillo et al., 2010 ). Strong and stable social relationships are especially important in older age as they can offer support that helps when navigating aging challenges ( Ajrouch et al., 2018 ; Miranda-Castillo et al., 2010 ). One challenge that becomes more common as people age is the development of Alzheimer’s disease or related dementias (hereafter “dementia”; Alzheimer’s Association, 2024 ). Nearly 7 million Americans aged 65 years and older live with dementia ( Alzheimer’s Association, 2024 ). Dementia clinical symptoms include challenges with memory, attention, communication and reasoning, judgment, and problem-solving ( CDC, 2019 ), which can hinder the ability to maintain existing, and form new, social relationships ( Alzheimer’s Association, 2024 ). The progression of dementia and associated cognitive and functional declines can significantly impact relationship quality and dynamics within an individual’s social network. For example, the onset of dementia often leads to shifts in power dynamics within relationships ( Wadham et al., 2016 ). These changes in social interactions can vary depending on the severity of dementia, the individual’s prior social relationships, and/or the strategies they use to maintain social connections ( Birt et al., 2020 ). Immediate family members, particularly spouses and children, remain the core of individuals’ social networks as they progress through dementia ( Merz & Huxhold, 2010 ; Wenger, 1994 ). However, these relationships often become strained as caregivers navigate their shifting roles within the relationship, leading to increased rates of stress and caregiver burden ( Blieszner & Roberto, 2010 ). While much of the emphasis in the literature has been placed on the shifting roles of family relationships in dementia disease progression, the role of nonfamily (e.g., friends, acquaintances) is also critical ( Ali et al., 2022 ; Genoe et al., 2022 ; Odzakovic et al., 2021 ). Perry et al. (2022) found that older adults without dementia had higher proportions of close friends and more extensive and diverse social networks than those with mild cognitive impairment. Relatedly, poor or limited social networks can raise dementia risk by as much as 60% ( Fratiglioni et al., 2000 ), highlighting the substantial associations between cognition and the development and maintenance of social ties. Socioemotional selectivity theory (SST; Carstensen, 1992 ) offers one framework for understanding the changes in social networks as one ages ( Carstensen, 2021 ; Löckenhoff & Carstensen, 2004 ). SST suggests that as time horizons become shorter—whether due to aging and/or cognitive decline—individuals increasingly prioritize emotionally meaningful goals, such as maintaining a sense of purpose, belonging, and emotional satisfaction, often leading to a narrowing of social networks to focus on closer, more emotionally fulfilling relationships. However, findings from Cornwell et al. (2014) complicate this narrative by demonstrating that older adults’ personal social networks are rarely stable. They often exhibit significant turnover within a few years. Cornwell and colleagues found that networks were more likely to expand than shrink, became less household-centric, and drew on weaker ties. Likewise, peripheral and weaker ties have been shown to be influential in making health decisions ( Fingerman, 2009 ; Fingerman et al., 2011 ). These findings suggest that while emotional goals emphasized by SST may influence some aspects of social network composition, other factors—such as life transitions, cultural expectations, or opportunities for social engagement—play a critical role in shaping these dynamics, highlighting the need for a more nuanced understanding of social network changes in later life. One way that robust social ties (i.e., those that demonstrate resilience and adaptability to changing situations) have been found to mitigate some of the effects of cognitive decline is by providing a sense of belonging, reducing loneliness, and preserving identity ( Cardona & Andrés, 2023 ; Holt-Lunstad, 2021 ). Robustness has been associated with closeness or the strength and frequency of interactions with individuals, and quality, the subjective value someone places on a particular network. As dementia progresses, cognitive ability declines, which can cause social networks to shrink ( Röhr et al., 2020 ). According to SST, this narrowing reflects a natural prioritization of close, emotionally fulfilling relationships as time horizons shorten. Individuals reflect on the quality of their connections and on the closeness these connections offer. In doing so, according to SST, they aim to retain robust networks while removing lower-quality or peripheral relationships. Yet, dementia introduces additional challenges to maintaining these core relationships, as cognitive decline and difficulties recognizing others can disrupt even the most meaningful connections. This disconnection not only thwarts the emotional goals emphasized by SST but also can contribute to increasing loneliness and isolation, particularly as individuals become less able to recognize those around them, contributing to further adverse health outcomes ( Moyle et al., 2011 ). The Importance of Race and Ethnicity Individuals with dementia have an average of 9.4 close family and friends at death, but this figure alone obfuscates the important picture that emerges when race/ethnicity are considered ( Baker et al., 2024 ). This summary figure is reasonably reflective of non-Hispanic/Latino White (hereafter “White”) people with dementia, who averaged 9.2 close friends and family at death. However, the networks for Hispanic/Latino people with dementia are almost an entire person larger (9.9 close family and friends). This difference continues to grow with non-Hispanic/Latino Black (hereafter “Black”) people with dementia averaging 10.8 close friends and family at death. As reviewed above, these social network differences are important for their implications for health broadly and conditions like dementia, specifically, but much past research gave limited consideration to race and ethnicity. Larger social networks have not ameliorated the disparity in which Hispanic/Latino older adults experience dementia at 1.5× the rate and Black older adults at 2× the rate of their White counterparts ( Alzheimer’s Association, 2024 ). Research on social engagement among community-dwelling Latino and White older adults has yielded mixed results, with some studies finding higher engagement among Latino older adults, others among White older adults, and some reporting no differences ( Tibiriçá et al., 2022 ; Wang et al. 2025 ). Additionally, evidence suggests that the negative health effects of loneliness—such as impaired daily functioning, depression, and memory decline—are stronger in collectivist cultures ( Beller & Wagner, 2020 ). This aligns with the revised sociocultural stress and coping model, which highlights how cultural values like familism—particularly obligation values—shape caregiving stress, coping, and health outcomes across ethnic groups ( Knight & Sayegh, 2010 ). Larger social networks among Hispanic/Latino and Black older adults may reflect adaptive practices that positively influence health through enhanced social support and effective coping styles. Recent research also demonstrates that the nature of social networks within different racial and ethnic groups differs substantially ( Baker et al., 2024 ). For instance, Hispanic/Latino people with dementia have more than twice as many extended family members in their networks at death as White people with dementia. Black people with dementia have 2.5 fewer extended family members than Hispanic/Latino people with dementia, but still 1.4 more than White people with dementia. Black people with dementia are also less likely to be married at the time of death than either Hispanic/Latino or White people with dementia. The Present Study If people with dementia in different racial and ethnic groups have differently sized social networks that decline at similar rates, it may suggest the importance of building greater social reserve prior to dementia. Conversely, if some groups exhibit networks that decline more slowly or increase in size, this may suggest intervention targets during the dementia disease process. The goal of this study is to evaluate the longitudinal patterns of social networks in individuals of different racial/ethnic groups living with dementia as they approach death. We will also assess how these shifts in network size are driven by specific types of relationships (e.g., family vs. friends) and how these patterns are modified by race/ethnicity. Method Data In this study, we used eight waves of data between 2004 and 2018 from the Health and Retirement Study (HRS). A local Institutional Review Board determined this secondary analysis was not human subjects research. The HRS is a nationally representative panel study that surveys older adults every 2 years on a wide range of topics ( Sonnega et al., 2014 ). The HRS started in 1992, and a new cohort of study participants is enrolled every 6 years. New study participants are 51–56 years old, not institutionalized at baseline, and include spouses of any age living in the continental United States of America ( Sonnega et al., 2014 ). The HRS continues to follow study respondents who become institutionalized and allows proxy respondents ( Fisher & Ryan, 2018 ; Michaud et al., 2011 ), strengthening the utility of the HRS for dementia research. The HRS uses face-to-face interviews, telephone interviews, and mail-back psychosocial questionnaires, or Leave-Behind Questionnaires (LBQs). The present study used data from the 2004–2018 LBQ that collected respondents’ information about their social network size ( Fisher & Ryan, 2018 ). Data from the RAND HRS Longitudinal File 2018 (V1; HRS, 2023 ) and the Gianattasio-Power Predicted Dementia Probability Scores and Dementia Classifications version 2.0 data file ( Gianattasio et al., 2020 ) were also merged to create the master analytic file. Analytic Sample Between 2004 and 2018, the number of respondents within each wave ranged from 19,459 to 27,198 ( HRS Staff, 2023 ). The analytic sample for this study included individuals who died between 2004 and 2018 and who were classified with dementia at the time of death. Mortality information was obtained from the HRS “Cross-Wave Tracker File” which documents each participant’s vital status across HRS waves. This study excluded individuals who did not respond to the LBQ network characteristics instruments, individuals who did not die, and individuals who died without having dementia. Individuals missing values on control variables were also dropped from the analytic sample. The final analytic sample size was 2,301 descendants with 3,986 observations across the study timeframe. Proxy responses were included. Measures Dementia We based our dementia classification on Expert model algorithms for dementia classification ( Gianattasio et al., 2020 ). The Expert model is a validated dementia algorithm based on a logistic model that predicts dementia using sociodemographic, physical health, and cognitive data, as well as relevant interactions between dementia predictors from the HRS core interviews. The Expert model was designed with racial/ethnic disparities research in mind to have comparable out-of-sample sensitivity and specificity values regardless of race/ethnicity. More details on the Expert classification algorithm, including the specific variables used in the Expert model algorithm off which we base our analyses, are available in other work ( Gianattasio et al., 2020 ; Power et al., 2021 ) and online: https://hrs.isr.umich.edu/data-products/cognition-data Network size The LBQ began in 2004 and is applied to an alternating half of the HRS sample ( Smith et al., 2017 ). After completing the HRS interview, respondents are given the LBQ to complete on their own time and then return by mail. We used responses to questions assessing the number of close family members and friends to calculate network size. First, we calculated whether the respondents had a close romantic partner/spouse based on their report of feeling “very close” or “quite close” to their partner. Respondents who reported not having a husband, wife, or partner with whom they lived or reported a “not very close” or “not at all close” relationship with the spouse/partner were considered not to have a close romantic partner. Second, we considered the number of close children reported using the item, “How many of your children would you say you have a close relationship with?” Third, respondents reported the number of other immediate family members (brothers or sisters, parents, cousins, or grandchildren) with whom they had a close relationship. Fourth, respondents were asked how many friends they would say they have a close relationship with. We calculated network size as the sum of these four components for our main analyses. We also conducted analyses to evaluate whether changes in network size differed by relationship type. Time till death The time variable was generated by subtracting the last year respondents were alive from each interview year. The last year alive was drawn from the HRS tracker file and had been determined by the decedent’s most recent core interview date, the core interview date of the decedent’s spouse/partner where the decedent was reported alive, or the last date during field activity where direct contact with the respondent or someone with knowledge of the respondent was had. Core interview dates were recorded for each observation. Race/ethnicity Race and ethnicity were self-reported. Race/ethnicity were re-coded from the RAND Longitudinal File’s “race” variable, which included categories “White/Caucasian,” “Black/African American,” and “Other,” as well as “ethnicity,” which was binarily assessed as “Hispanic/Latino” or “non-Hispanic/Latino.” The re-coded analytic variable included the categories White, Black, and Hispanic/Latino, with the White category as the reference group. Analytic Strategy We conducted descriptive analyses outlining the distribution, frequency, mean, and proportion of the sample characteristics. These characteristics were also stratified by race/ethnicity, and bivariate ANOVA was used to assess differences between race/ethnicity and each variable. Descriptive analyses were based on the characteristics of decedents in the first wave, in which they provided LBQ data on network size, which we defined as baseline. Growth curve models assessed changes in network size and its constituent components over time. Unadjusted models tested associations between race/ethnicity and time with network size, close spousal relationships, number of close children relationships, number of close friendships, and number of close extended family relationships. Adjusted models controlled for the number of diseases, age, proxy status, census region where the respondent lived, number of IADLs, nursing home residence, marital status, sex, and household wealth (selected for their likely importance to both predictors and outcomes of the models). For example, cognitive, functional, and mobility impairments are likely important to focal variables, and several of these control variables are expected to be associated with those impairments (e.g., IADLs, age, proxy status, and disease count). Further description of these variables is provided in Supplementary Material . We also tested a race/ethnicity by time till death interaction term. Significant interactions helped to identify whether members of different racial/ethnic groups experienced substantial differences in network size as they approached death. All models were estimated using random effect estimates and random effect standard errors with ɑ = .05. All data were maintained and analyzed using Stata 17.0 ( StataCorp, 2023 ). Results Participants The average network size at death was 10.28 people ( SD = 11.57), with friends and extended family members contributing the most to the total network size (3.92 people and 3.89 people, respectively). The demographic characteristics ( Table 1 ) of the sample were mostly female (59.10%), married (48.89%), and White (80.66%). Approximately 40.33% lived in the Southern region of the United States. The average education was 11.42 years ( SD = 3.38 years), and the average wealth decile was 4.83 ( SD = 2.75). Table 1. Unweighted Descriptive Statistics of Variables by Race/Ethnicity Variable Total sample ( N = 2,301) White ( n = 1,856) Black ( n = 333) Hispanic/Latino ( n = 112) p Network size at baseline, M ( SD ) Total network size 11.01 11.86 10.96 11.56 11.76 13.46 9.64 11.51 Children 2.72 3.46 2.61 3.29 3.22 4.40 3.04 3.03 ** Friends 4.46 6.32 4.58 6.25 4.29 7.11 2.93 4.60 * Extended family 4.11 6.29 3.93 5.67 5.07 8.21 4.25 9.21 * Spouses 0.88 0.32 0.90 0.29 0.81 0.40 0.72 0.45 *** Network size at endpoint, M ( SD ) Total network size 10.28 11.57 10.10 11.37 11.62 12.73 9.40 11.08 Children 2.81 4.16 2.72 4.14 3.28 4.56 3.04 3.00 Friends 3.92 5.87 3.98 5.73 4.07 7.20 2.43 2.58 * Extended family 3.89 6.19 3.71 5.88 4.79 6.50 4.37 9.38 * Spouses 0.89 0.32 0.90 0.30 0.82 0.38 0.84 0.37 * Demographics Age at baseline, M ( SD ) 80.45 7.35 80.72 7.21 78.59 7.77 81.47 7.53 *** Age at endpoint, M ( SD ) 83.36 7.04 83.79 6.87 80.96 7.47 83.38 7.27 *** Female, n (%) 1,360 59.10 1,076 57.97 211 63.36 73 65.18 Region ab , n (%) Northeast 380 16.51 330 17.78 37 11.11 13 11.61 ** Midwest 627 27.25 545 29.36 80 24.02 2 1.79 *** South 928 40.33 678 36.53 202 60.66 48 42.86 *** West 366 15.91 303 16.33 14 4.20 49 43.75 *** Marital status b , n (%) Married/partnered 1,125 48.89 946 50.97 125 37.54 54 48.21 *** Widowed 988 42.94 793 42.73 150 45.05 45 40.18 Divorced/separated/never married 188 8.17 117 6.30 58 17.42 13 11.61 *** Socioeconomic status Years of education, M ( SD ) 11.42 3.38 11.96 2.93 9.93 3.68 6.81 4.50 *** Wealth decile b , M ( SD ) 4.83 2.75 5.37 2.63 2.55 2.02 2.61 1.84 *** Functionality IADL count b , M (SD) 0.80 1.39 0.73 1.32 0.96 1.44 1.57 1.90 *** Disease count bc , M ( SD ) 2.56 1.39 2.55 1.40 2.66 1.31 2.45 1.45 Utilizes a proxy b , n (%) 179 7.78 123 6.63 32 9.61 24 21.43 *** Nursing home resident d , n (%) 69 3.00 58 3.13 4 1.20 7 6.25 * Open in a new tab Notes : IADL = instrumental activities of daily living; SD = standard deviation. p -Values are derived from analyses of variance and chi-squares. a Excluding noncontinental regions. b Baseline data were used. c Disease count includes whether respondents were diagnosed with heart disease, hypertension, stroke, diabetes, arthritis, lung disease, or cancer (excluding skin cancer) (range: 0–7. d Values reflect the number of respondents who reported living in nursing home care while enrolled in the Health and Retirement Study. * p < .05. ** p < .01. *** p < .001. The sample differed by race/ethnicity. White decedents reported fewer close relationships with children at baseline than Black and Hispanic/Latino decedents. Black and Hispanic/Latino decedents had significantly fewer close spousal relationships at baseline than White decedents. At death, these associations were similar; however, Black and Hispanic/Latino decedents reported significantly more close extended family relationships than White decedents. Black and Hispanic/Latino decedents reported living in the Southern region of the United States in greater proportions than White decedents, and Black and Hispanic/Latino decedents were significantly more likely to be divorced and female than White decedents. White decedents were more educated and had higher wealth deciles than Black and Hispanic/Latino decedents. White decedents had significantly fewer IADL challenges than Black and Hispanic/Latino decedents, and Hispanic/Latino decedents used a proxy in a greater proportion than Black and White respondents. Models Unadjusted growth curve models Table 2 presents unadjusted growth curve models estimating changes in overall network size and each component of network size (i.e., close spousal, child, friendships, and extended family relationships) by race, ethnicity, and time till death. Note that positive parameter estimates for time till death indicate a shrinking network size (i.e., as death approaches, the network size decreases), whereas negative values represent a growing network (i.e., as death approaches, the network size increases). Overall network size decreased as respondents grew closer to death ( b = 0.27, SE = 0.05, p < .001). Black decedents reported larger network sizes ( b = 1.21, SE = 0.58, p = .038) than White decedents. No changes were observed in spousal relationships as respondents approached death. However, Black decedents ( b = −0.11, SE = 0.03, p < .001) and Hispanic/Latino decedents ( b = −0.15, SE = 0.04, p < .001) reported fewer close spousal relationships compared to White decedents. The number of close relationships with children did not differ as respondents approached death. However, Black decedents reported more close relationships with children ( b = 0.50, SE = 0.20, p = .014) than White decedents. As respondents approached death, the number of close friendships diminished ( b = 0.18, SE = 0.03, p < .001). Hispanic/Latino decedents reported fewer close friendships ( b = −1.52, SE = 0.53, p = .004) than White decedents. The number of close, extended family relationships diminished as respondents approached death ( b = 0.08, SE = 0.03, p = .003). However, Black decedents reported substantially more extended family relationships ( b = 1.29, SE = 0.32, p < .001) than White decedents. Table 2. Unadjusted Growth Curve Models of Changes in Network Size of People With Dementia as They Approach Death Variable Network size Spouses Children Friends Extended family Estimate SE Estimate SE Estimate SE Estimate SE Estimate SE Time till death 0.27 *** 0.05 −0.00 0.00 −0.00 0.02 0.18 *** 0.03 0.08 ** 0.03 Race/ethnicity (ref. = White) Black 1.21 * 0.58 −0.11 *** 0.03 0.50 * 0.20 −0.03 0.31 1.29 *** 0.32 Hispanic/Latino −0.86 0.98 −0.15 *** 0.04 0.26 0.34 −1.52 ** 0.53 0.64 0.53 Intercept 9.36 *** 0.32 0.91 *** 0.01 2.70 *** 0.12 3.49 *** 0.17 3.40 *** 0.17 Random effects (σ 2 ) Time till death 0.13 0.07 0.00 0.00 0.00 0.00 0.15 0.02 0.00 0.00 Intercept 32.05 4.07 0.06 0.00 2.19 0.37 5.11 1.16 11.29 1.08 Residual 91.45 3.31 0.04 0.00 13.63 0.44 24.65 1.00 22.70 0.90 Wald χ 2 31.66 *** 30.78 *** 6.41 42.68 *** 25.91 *** Open in a new tab Notes : N = 2,301. Data are from the 2004–2018 waves of the Health and Retirement Study. Network size = overall network size ( n = 2,301); Spouses = close spousal relationships ( n = 1,247); Children = close child relationships ( n = 2,257); Friends = close friendships ( n = 2,196); Extended family = close extended family relationships ( n = 2,204). Positive parameter estimates for time till death indicate a shrinking network size (i.e., as death approaches, the network size decreases), whereas negative values represent a growing network (i.e., as death approaches, the network size increases). ref = reference; SE = standard error. * p < .05. ** p < .01. *** p < .001. Adjusted growth curve models Table 3 presents adjusted growth curve models estimating changes in overall network size and its subsidiary components. Overall network size diminished as respondents approached death ( b = 0.23, SE = 0.06, p < .001). More education was associated with smaller network sizes ( b = −0.23, SE = 0.07, p = .001). Midwestern decedents ( b = 2.28, SE = 0.62, p < .001) and Southern decedents ( b = 1.80, SE = 0.59, p = .002) reported larger network sizes than Northeastern decedents. Unmarried decedents (i.e., divorced, separated, or never married) had smaller network sizes ( b = −2.45, SE = 0.76, p = .001) than married decedents. Table 3. Adjusted Growth Curve Models of Changes in Network Size of People With Dementia as They Approach Death Variable Network size Spouses Children Friends Extended family Estimate SE Estimate SE Estimate SE Estimate SE Estimate SE Time till death 0.23 *** 0.06 −0.00 0.00 −0.02 0.02 0.17 *** 0.03 0.09 ** 0.03 Race/ethnicity (ref. = White) Black 1.06 0.63 −0.01 0.03 0.50 * 0.22 0.04 0.33 0.99 ** 0.35 Hispanic/Latino −1.31 1.07 −0.10 * 0.04 0.23 0.37 −1.41 * 0.57 −0.02 0.58 Disease count 0.10 0.14 0.00 0.01 0.04 0.05 0.06 0.07 0.06 0.08 Age −0.01 0.03 0.00 0.00 −0.02 0.01 0.02 0.02 0.02 0.02 Education −0.23 ** 0.07 0.00 0.00 −0.05 * 0.02 −0.05 0.04 −0.15 *** 0.04 Uses proxy (ref. = no proxy) −0.23 0.74 −0.06 * 0.03 0.12 0.27 −0.61 0.38 0.14 0.39 Census region (ref. = Northeast) Midwest 2.28 *** 0.62 −0.03 0.03 0.41 0.21 1.03 ** 0.33 0.78 * 0.34 South 1.80 ** 0.59 −0.00 0.02 0.11 0.20 1.12 *** 0.31 0.58 0.32 West 0.74 0.71 0.04 0.03 −0.06 0.24 0.69 0.38 0.11 0.38 Instrumental activities of daily living −0.17 0.15 −0.00 0.01 −0.06 0.05 −0.14 0.08 −0.05 0.08 Nursing home resident (ref. = community resident) −1.32 0.89 −0.07 0.05 −0.61 0.33 −0.09 0.46 −0.65 0.47 Marital status (ref. = married/partnered) Widowed −0.81 0.47 −0.24 *** 0.03 −0.12 0.17 −0.10 0.25 −0.14 0.25 Unmarried a −2.45 ** 0.76 −0.44 *** 0.05 −0.89 ** 0.26 −0.63 0.40 −0.41 0.41 Female (ref. = male) 0.16 0.45 −0.02 0.02 −0.03 0.15 −0.20 0.24 0.71 ** 0.24 Wealth decile 0.08 0.08 0.00 0.00 0.01 0.03 0.04 0.04 0.02 0.04 Intercept 10.83 *** 1.08 0.88 *** 0.04 3.30 *** 0.37 3.25 *** 0.57 4.25 *** 0.58 Random effects (σ 2 ) Time till death 0.15 0.07 0.00 0.00 0.00 0.00 0.15 0.02 0.00 0.00 Intercept 28.78 4.06 0.05 0.00 1.98 0.36 4.69 1.15 10.69 1.07 Residuals 92.13 3.34 0.04 0.00 13.68 0.44 24.69 1.00 22.80 0.91 Wald χ 2 82.77 *** 241.97 *** 41.49 *** 76.58 *** 68.31 *** Open in a new tab Notes : N = 2,301. Data are from the 2004–2018 waves of the Health and Retirement Study. Network size = overall network size ( n = 2,299); Spouses = close spousal relationships ( n = 1,245); Children = close child relationships ( n = 2,255); Friends = close friendships ( n = 2,195); Extended family = close extended family relationship ( n = 2,202). Positive parameter estimates for time till death indicate a shrinking network size (i.e., as death approaches, the network size decreases), whereas negative values represent a growing network (i.e., as death approaches, the network size increases). a Unmarried includes divorced, separated, and never married. ref = reference; SE = standard error. * p < .05. ** p < .01. *** p < .001. Close spousal relationships did not change as respondents approached death in the adjusted model. However, Hispanic/Latino decedents reported fewer close spousal relationships ( b = −0.10, SE = 0.04, p = .016) compared to White decedents, and decedents who used a proxy had fewer close spousal relationships ( b = −0.06, SE = 0.03, p = .017). Intuitively, decedents had fewer close spousal relationships if they were widowed ( b = −0.24, SE = 0.03, p < .001) or unmarried ( b = −0.44, SE = 0.05, p < .001) compared to married decedents. Close relationships with children did not differ as respondents grew closer to death. Black decedents reported more close relationships with children ( b = 0.50, SE = 0.22, p = .021) than White decedents. More education was associated with fewer close relationships with children ( b = −0.05, SE = 0.02, p = .029). Similarly, unmarried decedents reported fewer close relationships with children ( b = −0.89, SE = 0.26, p = .001) than married decedents. The number of close friendships diminished ( b = 0.17, SE = 0.03, p < .001) as respondents with dementia approached death. Hispanic/Latino decedents reported fewer close friendships ( b = −1.41, SE = 0.57, p = .014) than White decedents. Decedents in the Midwest ( b = 1.03, SE = 0.33, p = .002) and South ( b = 1.12, SE = 0.31, p < .001) reported more close friendships than decedents in the Northeast. The number of close extended family members diminished as respondents with dementia approached death ( b = 0.09, SE = 0.03, p = .006). However, Black decedents had more close extended family relationships ( b = 0.99, SE = 0.35, p = .004) than White decedents. More education was associated with fewer close extended family relationships ( b = −0.15, SE = .04, p < .001). Compared to the northeastern region, decedents in the Midwestern region reported more close extended family relationships ( b = 0.78, SE = 0.34, p = .020). Female decedents reported more close extended family relationships ( b = 0.71, SE = 0.24, p = .003) than male decedents. Adjusted growth curve models interacting race/ethnicity with time Table 4 and Figure 1 present the interaction effects between race/ethnicity and time till death. In the models where overall network size and immediate family and friend relationships were the outcomes, the interaction terms did not differ for Black and Hispanic/Latino decedents compared to White decedents. However, when close extended family relationships were the outcome, Hispanic/Latino respondents with dementia indicated more close family relationships ( b = −0.34, SE = 0.14, p = .015) compared to White respondents with dementia as death approached. Table 4. Adjusted Growth Curve Models of Changes in Network Size of People With Dementia Interacting Race and Ethnicity With Time Till Death Network size Spouses Children Friends Extended family Variable Estimate SE Estimate SE Estimate SE Estimate SE Estimate SE Time till death 0.28 *** 0.07 −0.00 0.00 −0.02 0.02 0.18 *** 0.04 0.10 ** 0.03 Race/ethnicity (ref. = White) Black 2.12 * 0.92 −0.01 0.04 0.48 0.34 0.57 0.48 1.11 * 0.50 Hispanic/Latino 0.75 1.57 −0.07 0.06 0.53 0.58 −1.63 0.85 1.49 0.85 Disease count 0.10 0.14 0.00 0.01 0.04 0.05 0.06 0.07 0.06 0.08 Age −0.01 0.03 0.00 0.00 −0.02 0.01 0.02 0.02 0.02 0.02 Education −0.22 ** 0.07 0.00 0.00 −0.05 * 0.02 −0.05 0.04 −0.15 *** 0.04 Uses proxy (ref. = no proxy) −0.30 0.74 −0.06 * 0.03 0.12 0.27 −0.63 0.39 0.11 0.39 Census region (ref. = Northeast) Midwest 2.25 *** 0.62 −0.03 0.03 0.41 0.21 1.02 ** 0.33 0.77 * 0.33 South 1.76 ** 0.59 −0.00 0.02 0.11 0.20 1.11 *** 0.31 0.57 0.32 West 0.75 0.71 0.04 0.03 −0.06 0.24 0.67 0.38 0.13 0.38 Instrumental activities of daily living −0.19 0.15 −0.00 0.01 −0.06 0.05 −0.14 0.08 −0.06 0.08 Nursing home resident (ref. = community resident) −1.25 0.89 −0.07 0.05 −0.61 0.33 −0.05 0.46 −0.64 0.47 Marital status (ref. = married/partnered) Widowed −0.81 0.47 −0.24 *** 0.03 −0.13 0.17 −0.10 0.25 −0.14 0.25 Unmarried a −2.48 ** 0.76 −0.44 *** 0.05 −0.89 ** 0.26 −0.63 0.40 −0.43 0.41 Female (ref. = male) 0.16 0.45 −0.02 0.02 −0.03 0.15 −0.21 0.24 0.71 ** 0.24 Wealth decile 0.07 0.08 0.00 0.00 0.01 0.03 0.03 0.04 0.02 0.04 Race/ethnicity * Time till death (ref. = White * Time till death) Black * Time till death −0.25 0.16 −0.00 0.01 0.01 0.06 −0.14 0.09 −0.02 0.08 Hispanic/Latino * Time till death −0.47 0.26 −0.01 0.01 −0.06 0.10 0.06 0.16 −0.34 * 0.14 Intercept 10.61 *** 1.09 0.88 *** 0.04 3.29 *** 0.38 3.19 *** 0.57 4.18 *** 0.59 Random effects (σ 2 ) Time till death 0.14 0.07 0.00 0.00 0.00 0.00 0.15 0.02 0.00 0.00 Intercept 28.69 4.05 0.05 0.00 1.97 0.36 4.69 1.15 10.58 1.07 Residuals 92.11 3.34 0.04 0.00 13.68 0.44 24.68 1.00 22.82 0.91 Wald χ 2 88.15 *** 242.36 *** 41.97 ** 79.29 *** 74.48 *** Open in a new tab Notes : N = 2,301. Data are from the 2004–2018 waves of the Health and Retirement Study. Network size = overall network size ( n = 2,299); Spouses = close spousal relationships ( n = 1,245); Children = close child relationships ( n = 2,255); Friends = close friendships ( n = 2,195); Extended family = close extended family relationship ( n = 2,202). Positive parameter estimates for time till death indicate a shrinking network size (i.e., as death approaches, the network size decreases), whereas negative values represent a growing network (i.e., as death approaches, the network size increases). a Unmarried includes divorced, separated, and never married. ref = reference; SE = standard error. * p < .05. ** p < .01. *** p < .001. Figure 1. Open in a new tab Social network trajectories and composition in the last decade of life of people with dementia by race/ethnicity. Trajectories are based on relevant parameter estimates (i.e., intercepts, slopes, main effects of race/ethnicity, and slope by race/ethnicity interactions) from adjusted models in Table 4 . Discussion This work explored how the size of people with dementia’s social networks changes as death approaches, using data from the nationally representative HRS. For people with dementia, social networks generally shrank as they approached death. On average, their network size decreased by one person every 3.7 years. The size of immediate family networks remained relatively stable, whereas friendship and extended family networks shrank by one person every 5.56 years and 12.5 years, respectively. These social network size trajectories were not always equal when examined through the lens of race and ethnicity. Race and Ethnicity Results revealed significant differences in social network size among different racial and ethnic groups as people with dementia approached death, replicating past work ( Ajrouch et al., 2001 ; Baker et al., 2024 ). In unadjusted models, Black people with dementia had a broader social circle, as evidenced by larger overall networks than their White counterparts at the time of death. Similarly, in both unadjusted and adjusted models they had significantly larger networks of close children and extended family at the time of death. As for Hispanic/Latino people with dementia, unadjusted models showed they had significantly smaller networks of close spouses and friends at the time of death, compared to their White counterparts. Similarly, in adjusted models, their close spousal and friendship networks remained significantly smaller than those of White people with dementia. We also investigated how the social networks of different people with dementia from different racial/ethnic groups changed over time as they approached death. We found racial/ethnic differences where there was an increase in extended family size as Hispanic/Latino respondents approached death. However, when observing figures ( Figure 1 ) created with these data, it is notable that the characteristic trend of social networks shrinking as dementia death approaches may be a trend that is most characteristic of White respondents. For instance, while overall network size decreased among White respondents, components and overall network size stayed steady for Black respondents, suggesting that dementia’s progression might not extensively impact Black social network size. Still, the differences in those coefficients did not reach statistical significance, suggesting considerable caution in interpretations is warranted. Comparatively, Hispanic/Latino respondents exhibited a pattern of decrease in close friends as death approached, but their overall network size and number of close extended family and close children appeared to increase. This finding for close children was also not statistically significant, and while interesting, we cannot determine with these data whether that failure to find statistical significance is because there is not a true difference between these populations or because of Type II error. While not specifically concerned with racial/ethnic differences, the findings of differential changes in dementia networks over time in different contexts do align with past work ( Spillman et al., 2020 ). One explanation for why Black and Hispanic/Latino respondents maintained or expanded their social networks as death approached could be cultural values that prioritize family connections and collective care. Collectivist cultural orientations, which are prevalent in Black and Hispanic/Latino communities, prioritize maintaining close relationships with family and extended networks ( Falzarano et al., 2022 ; Ruiz & Ransford, 2012 ). In line with this, the revised sociocultural stress and coping model ( Knight & Sayegh, 2010 ) provides a useful lens to interpret this finding, emphasizing how cultural values can positively influence social support and coping styles and, in turn, improve health outcomes. For these groups, larger or sustained social networks may reflect a proactive strategy to mobilize resources and strengthen support systems during challenging times. This underscores the importance of considering cultural contexts in understanding caregiving and the social dynamics of people with dementia near the end of life. These findings are especially interesting for a few reasons. First, social networks can serve as considerable protective factors for a wide range of health outcomes ( Holt-Lunstad et al., 2010 ; Miranda-Castillo et al., 2010 ). This is particularly important in the context of dementia, which afflicts Black and Hispanic/Latino older adults at substantially higher rates than their White counterparts ( Alzheimer’s Association, 2024 ). If future research confirms these patterns, exploring the benefits conferred by maintaining and (in some cases) growing social networks for individuals with dementia from these groups may be a fruitful area of study. The implications of these social network patterns are also substantial for those surrounding people with dementia (e.g., for caregivers who have previously exhibited social network loss as a result of providing dementia care; Liu et al., 2021 ). Indeed, reviews consistently find that bereaved dementia caregivers are challenged by social isolation and loneliness ( Arruda & Paun, 2017 ) and that social support is a meaningful factor for post-death grief experiences ( Crawley et al., 2023 ). One newly proposed model even incorporates relatedness as a core component for understanding which bereaved dementia caregivers might need the most aid to reduce the consequences of their bereavement ( Baker et al., 2024 ). Most relevant to the present study, bereaved dementia caregivers with fewer people in their social networks may experience more grief ( Bergman et al., 2011 ). Second, through a strengths-based lens, this research might suggest that other racial and ethnic groups could learn from older adults who are Black or Hispanic/Latino. Given the importance of social networks, especially during challenging times, and their tendency to shrink closer to death, if individuals could adopt the effective strategies used by Black and Hispanic/Latino older adults to maintain and grow their social networks, they might experience similar benefits of robust social networks. This would be an incredible boon to those other groups, given the massive benefits of social networks in the context of dementia ( Fratiglioni et al., 2000 ). This study makes several significant contributions to the literature on dementia and social networks. First, it contributes to our understanding of how social networks change as individuals with dementia approach death by using longitudinal data from a nationally representative sample. Unlike cross-sectional studies, these findings capture the dynamic nature of social networks by highlighting trends such as the shrinking of friendship networks and the stability or growth of family networks among certain racial and ethnic groups. Second, it provides a better understanding of Hispanic/Latino and Black older adult social networks, which could be instructive in developing interventions based on the strengths within these groups. Indeed, social isolation has been identified to be the single largest potentially modifiable risk factor for dementia in late life ( Livingston et al., 2024 ). These interventions may work to maintain or even increase social network size so that more people can benefit from social networks. Limitations and Future Directions First, several of the observed patterns were not statistically significant, and the small Black and Hispanic/Latino samples resulted in large standard errors. These trends are not definitive and warrant replication attempts. Second, data presented were predominantly self-reported. One alternative strategy is asking both the person with dementia and their family member to report relationship closeness. This might offer a more accurate depiction of both members’ perceptions of their close relationship. Moreover, there is a risk of inaccuracies or biases in how social network sizes and characteristics are reported. Third, our network size calculation method provides a quantifiable measure of social networks. However, it may not fully capture the changes in social connections of these relationships and network support as dementia progresses. More nuanced methodologies may help better account for their social network’s quality, frequency of social interaction, and variability. For instance, qualitative research methods, including in-depth interviews, could provide richer insights into the nature of social support and changes in relationship dynamics over time ( Fuhse & Mützel, 2011 ). Future research implementing a more detailed investigation of persons with dementia’s social networks could improve the accuracy of findings and contribute to a deeper understanding of network changes as people navigate dementia’s challenges. Likewise, while the number of people in a social network and what categories those people might fall into are important, they fail to include other important social network characterizations. Relationship quality, frequency of interaction, and perceptions of support received from social networks are essential elements to be explored in future directions of this line of research. There must also be consideration of peripheral network members whose ties are generally weaker ( Fingerman, 2009 ). As these social ties can often outnumber close social ties, but are increasingly ( Fingerman et al., 2024 ) viewed as providing a critical function to individuals’ well-being. Finally, future research should explore how culture shapes persons with dementia’s social networks. Cultural values can influence social support and expectations of responsibility for aging family members’ care ( Knight & Sayegh, 2010 ), thus impacting social network size. For instance, collectivist cultures prioritize intergenerational caregiving within families ( Shi et al., 2024 ), and extended family and community members may play a more active role in providing social support ( Knight & Sayegh, 2010 ), perhaps resulting in larger and more interconnected social networks. Whereas individualistic cultures emphasize nuclear family relationships, autonomy, and independence ( Willis, 2012 ), which may contribute to smaller social networks for people with dementia. In turn, this will influence the positioning of ties as either close or peripheral in one’s network. Conclusion Results revealed that social networks shrink linearly as death with dementia approaches, but that this may be a finding most applicable to White people. Findings for Black and Hispanic/Latino respondents appeared more desirable: social networks either stayed a similar size or grew as death approached. If replicated, these patterns may be the result of protective factors for these Black and Hispanic/Latino people with dementia. The strategies employed by these groups to maintain their networks might also provide a useful avenue for future interventions to help slow the diminishment of social networks as death with dementia approaches for other racial/ethnic groups. Supplementary Material gbaf062_suppl_Supplementary_Materials gbaf062_suppl_supplementary_materials.docx (16.5KB, docx) Acknowledgments This study uses data from the Health and Retirement Study (HRS). The HRS is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. We also wish to thank Yingyan Huang for assisting in quality control of our statistical reporting. Contributor Information Zachary G Baker, Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA. Andrew Alberth, Department of Gerontology, University of Massachusetts Boston, Boston, Massachusetts, USA. M Aaron Guest, Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA. Allie Peckham, Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA. Joahana Segundo, Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA. Joseph Saenz, Edson College of Nursing and Health Innovation, Arizona State University, Phoenix, Arizona, USA. Marc A Garcia, (Social Sciences Section). Funding Research reported in this publication was supported by the National Institute on Aging of the National Institutes of Health (R00AG073463, R00AG058799). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Conflict of Interest None. Data Availability Data for this study are made publicly available by the Health and Retirement Study on their individual website ( https://hrs.isr.umich.edu/ ). The present analysis of these data was not preregistered. Author Contributions Z.G. Baker contributed to the conceptualization of the study, helped with the initial draft, revised the paper, supervised the project, ensured the validation of the research findings, managed project administration, acquired funding, and provided resources. A. Alberth handled data curation, performed formal analysis, contributed to the methodology and the visualization of tables, helped with the initial draft, and revised the paper. M.A. Guest contributed to writing the initial draft and revised the paper. A. Peckham assisted in writing the initial draft and revised the paper. J. Segundo ensured the validation of research findings, visualized tables, helped with the initial draft, revised the paper, and managed project administration. J. Saenz performed formal analysis, contributed to the methodology, visualized figure, helped with the initial draft, revised the paper, and was involved in funding acquisition. References Ajrouch, K. J., Antonucci, T. C., & Janevic, M. R. (2001). Social networks among Blacks and Whites: The interaction between race and age. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 56(2), S112–S118. https://doi.org/ 10.1093/geronb/56.2.s112 [ DOI ] [ PubMed ] [ Google Scholar ] Ajrouch, K. J., Fuller, H. R., Akiyama, H., & Antonucci, T. C. (2018). Convoys of social relations in cross-national context. The Gerontologist, 58(3), 488–499. https://doi.org/ 10.1093/geront/gnw204 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ali, T., Elliott, M. R., Antonucci, T. C., Needham, B. L., Zelner, J., & Mendes de Leon, C. F. (2022). Multidimensional social network types and their correlates in older Americans. Innovation in Aging, 6(1), 1–16. https://doi.org/ 10.1093/geroni/igab053 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Alzheimer’s Association. (2024). 2024 Alzheimer’s disease facts and figures. Alzheimer’s & Dementia, 20(5), 3708–3821. https://doi.org/ 10.1002/alz.13809 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Arruda, E. H., & Paun, O. (2017). Dementia caregiver grief and bereavement: An integrative review. Western Journal of Nursing Research, 39(6), 825–851. https://doi.org/ 10.1177/0193945916658881 [ DOI ] [ PubMed ] [ Google Scholar ] Baker, Z. G., Han, S., Sefcik, J. S., Petrovsky, D. V., Ma, K. P. K., Smith, M. L., Bacsu, J. R., Rahemi, Z., & Saenz, J. (2024). Mapping the landscape of those left behind when a person with dementia dies: Roles of race and ethnicity. Alzheimer Disease and Associated Disorders, 38(4), 332–337. https://doi.org/ 10.1097/WAD.0000000000000645 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Baker, Z. G., O’Donnell, M. G., Garcia-Arias, S., Huang, Y., Segundo, J., Millenbah, A. N., Neubert, O. M., & Huerta, I. (2024). Protocol for a telephonic mixed methods study to understand needs and find solutions for bereaved dementia caregivers. BMJ Open, 14(7), e086559. https://doi.org/ 10.1136/bmjopen-2024-086559 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Beller, J., & Wagner, A. (2020). Loneliness and health: The moderating effect of cross-cultural individualism/collectivism. Journal of Aging and Health, 32(10), 1516–1527. https://doi.org/ 10.1177/0898264320943336 [ DOI ] [ PubMed ] [ Google Scholar ] Bergman, E. J., Haley, W. E., & Small, B. J. (2011). Who uses bereavement services? An examination of service use by bereaved dementia caregivers. Aging & Mental Health, 15(4), 531–540. https://doi.org/ 10.1080/13607863.2010.543661 [ DOI ] [ PubMed ] [ Google Scholar ] Birt, L., Griffiths, R., Charlesworth, G., Higgs, P., Orrell, M., Leung, P., & Poland, F. (2020). Maintaining social connections in dementia: A qualitative synthesis. Qualitative Health Research, 30(1), 23–42. https://doi.org/ 10.1177/1049732319874782 [ DOI ] [ PubMed ] [ Google Scholar ] Blieszner, R., & Roberto, K. A. (2010). Care partner responses to the onset of mild cognitive impairment. The Gerontologist, 50(1), 11–22. https://doi.org/ 10.1093/geront/gnp068 [ DOI ] [ PubMed ] [ Google Scholar ] Cardona, M., & Andrés, P. (2023). Are social isolation and loneliness associated with cognitive decline in ageing? Frontiers in Aging Neuroscience, 15, 1075563. https://doi.org/ 10.3389/fnagi.2023.1075563 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Carstensen, L. L. (1992). Social and emotional patterns in adulthood: Support for socioemotional selectivity theory. Psychology and Aging, 7(3), 331–338. https://doi.org/ 10.1037//0882-7974.7.3.331 [ DOI ] [ PubMed ] [ Google Scholar ] Carstensen, L. L. (2021). Socioemotional selectivity theory: The role of perceived endings in human motivation. The Gerontologist, 61(8), 1188–1196. https://doi.org/ 10.1093/geront/gnab116 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] CDC. (2019, April 5). What is dementia? https://www.cdc.gov/alzheimers-dementia/about/index.html [ Google Scholar ] Cornwell, B., Schumm, L. P., Laumann, E. O., Kim, J., & Kim, Y.-J. (2014). Assessment of social network change in a national longitudinal survey. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69(Suppl 2), S75–S82. https://doi.org/ 10.1093/geronb/gbu037 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Crawley, S., Sampson, E. L., Moore, K. J., Kupeli, N., & West, E. (2023). Grief in family carers of people living with dementia: A systematic review. International Psychogeriatrics, 35(9), 477–508. https://doi.org/ 10.1017/S1041610221002787 [ DOI ] [ PubMed ] [ Google Scholar ] Falzarano, F., Moxley, J., Pillemer, K., & Czaja, S. J. (2022). Family matters: Cross-cultural differences in familism and caregiving outcomes. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 77(7), 1269–1279. https://doi.org/ 10.1093/geronb/gbab160 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Fingerman, K. L. (2009). Consequential strangers and peripheral ties: The importance of unimportant relationships. Journal of Family Theory & Review, 1(2), 69–86. https://doi.org/ 10.1111/j.1756-2589.2009.00010.x [ DOI ] [ Google Scholar ] Fingerman, K. L., Brown, B. B., & Blieszner, R. (2011). Informal ties across the life span: Peers, consequential strangers, and people we encounter in daily life. In Fingerman K. L., Berg C. A., Smith J., & Antonucci T. C. (Eds.), Handbook of life-span development (pp. 487–511). Springer Publishing Company. [ Google Scholar ] Fingerman, K. L., Zhou, Z., Huo, M., Luong, G., & Birditt, K. S. (2024). Enduring bonds: Duration and contact in close relationships in late life. The Gerontologist, 64(4), gnad091. https://doi.org/ 10.1093/geront/gnad091 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Fisher, G. G., & Ryan, L. H. (2018). Overview of the Health and Retirement Study and introduction to the special issue. Work, Aging and Retirement, 4(1), 1–9. https://doi.org/ 10.1093/workar/wax032 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Fratiglioni, L., Wang, H.-X., Ericsson, K., Maytan, M., & Winblad, B. (2000). Influence of social network on occurrence of dementia: A community-based longitudinal study. Lancet (London, England), 355(9212), 1315–1319. https://doi.org/ 10.1016/S0140-6736(00)02113-9 [ DOI ] [ PubMed ] [ Google Scholar ] Fuhse, J., & Mützel, S. (2011). Tackling connections, structure, and meaning in networks: Quantitative and qualitative methods in sociological network research. Quality & Quantity, 45(5), 1067–1089. https://doi.org/ 10.1007/s11135-011-9492-3 [ DOI ] [ Google Scholar ] Genoe, M. R., Fortune, D., & Whyte, C. (2022). Strategies for maintaining friendship in dementia. Canadian Journal on Aging = La revue Canadienne du vieillissement, 41(3), 431–442. https://doi.org/ 10.1017/S0714980821000301 [ DOI ] [ PubMed ] [ Google Scholar ] Gianattasio, K. Z., Ciarleglio, A., & Power, M. C. (2020). Development of algorithmic dementia ascertainment for racial/ethnic disparities research in the US Health and Retirement Study. Epidemiology, 31(1), 126–133. https://doi.org/ 10.1097/EDE.0000000000001101 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Health and Retirement Study (HRS). (2023). Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740) [Dataset]. Holt-Lunstad, J. (2021). Loneliness and social isolation as risk factors: The power of social connection in prevention. American Journal of Lifestyle Medicine, 15(5), 567–573. https://doi.org/ 10.1177/15598276211009454 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality risk: A meta-analytic review. PLoS Medicine, 7(7), e1000316. https://doi.org/ 10.1371/journal.pmed.1000316 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] HRS Staff. (2023). HRS core interview sample sizes and response rates . https://hrs.isr.umich.edu/documentation/survey-design/response-rates [ Google Scholar ] Khondoker, M., Rafnsson, S. B., Morris, S., Orrell, M., & Steptoe, A. (2017). Positive and negative experiences of social support and risk of dementia in later life: An investigation using the English Longitudinal Study of Ageing. Journal of Alzheimer's Disease: JAD, 58(1), 99–108. https://doi.org/ 10.3233/JAD-161160 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Knight, B. G., & Sayegh, P. (2010). Cultural values and caregiving: The updated sociocultural stress and coping model. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 65B(1), 5–13. https://doi.org/ 10.1093/geronb/gbp096 [ DOI ] [ PubMed ] [ Google Scholar ] Kuiper, J. S., Zuidersma, M., Oude Voshaar, R. C., Zuidema, S. U., van den Heuvel, E. R., Stolk, R. P., & Smidt, N. (2015). Social relationships and risk of dementia: A systematic review and meta-analysis of longitudinal cohort studies. Ageing Research Reviews, 22, 39–57. https://doi.org/ 10.1016/j.arr.2015.04.006 [ DOI ] [ PubMed ] [ Google Scholar ] Liu, C., Fabius, C. D., Howard, V. J., Haley, W. E., & Roth, D. L. (2021). Change in social engagement among incident caregivers and controls: Findings from the Caregiving Transitions Study. Journal of Aging and Health, 33(1–2), 114–124. https://doi.org/ 10.1177/0898264320961946 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Livingston, G., Huntley, J., Liu, K. Y., Costafreda, S. G., Selbæk, G., Alladi, S., Ames, D., Banerjee, S., Burns, A., Brayne, C., Fox, N. C., Ferri, C. P., Gitlin, L. N., Howard, R., Kales, H. C., Kivimäki, M., Larson, E. B., Nakasujja, N., Rockwood, K., … Mukadam, N. (2024). Dementia prevention, intervention, and care: 2024 report of the Lancet Standing Commission. Lancet (London, England), 404(10452), 572–628. https://doi.org/ 10.1016/S0140-6736(24)01296-0 [ DOI ] [ PubMed ] [ Google Scholar ] Löckenhoff, C. E., & Carstensen, L. L. (2004). Socioemotional selectivity theory, aging, and health: The increasingly delicate balance between regulating emotions and making tough choices. Journal of Personality, 72(6), 1395–1424. https://doi.org/ 10.1111/j.1467-6494.2004.00301.x [ DOI ] [ PubMed ] [ Google Scholar ] Merz, E.-M., & Huxhold, O. (2010). Wellbeing depends on social relationship characteristics: Comparing different types and providers of support to older adults. Ageing and Society, 30(5), 843–857. https://doi.org/ 10.1017/s0144686x10000061 [ DOI ] [ Google Scholar ] Michaud, P.-C., Kapteyn, A., Smith, J. P., & van Soest, A. (2011). Temporary and permanent unit non-response in follow-up interviews of the Health and Retirement Study. Longitudinal and Life Course Studies, 2(2), Article 2. https://doi.org/ 10.14301/llcs.v2i2.114 [ DOI ] [ Google Scholar ] Miranda-Castillo, C., Woods, B., Galboda, K., Oomman, S., Olojugba, C., & Orrell, M. (2010). Unmet needs, quality of life and support networks of people with dementia living at home. Health and Quality of Life Outcomes, 8(1), 132. https://doi.org/ 10.1186/1477-7525-8-132 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Moyle, W., Kellett, U., Ballantyne, A., & Gracia, N. (2011). Dementia and loneliness: An Australian perspective. Journal of Clinical Nursing, 20(9–10), 1445–1453. https://doi.org/ 10.1111/j.1365-2702.2010.03549.x [ DOI ] [ PubMed ] [ Google Scholar ] Odzakovic, E., Kullberg, A., Hellström, I., Clark, A., Campbell, S., Manji, K., Rummery, K., Keady, J., & Ward, R. (2021). “It’s our pleasure, we count cars here”: An exploration of the “neighbourhood-based connections” for people living alone with dementia. Ageing & Society, 41(3), 645–670. https://doi.org/ 10.1017/S0144686X19001259 [ DOI ] [ Google Scholar ] Perry, B. L., McConnell, W. R., Peng, S., Roth, A. R., Coleman, M., Manchella, M., Roessler, M., Francis, H., Sheean, H., & Apostolova, L. A. (2022). Social networks and cognitive function: An evaluation of social bridging and bonding mechanisms. The Gerontologist, 62(6), 865–875. https://doi.org/ 10.1093/geront/gnab112 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Power, M. C., Bennett, E. E., Turner, R. W., Dowling, N. M., Ciarleglio, A., Glymour, M. M., & Gianattasio, K. Z. (2021). Trends in relative incidence and prevalence of dementia across non-Hispanic Black and White individuals in the United States, 2000-2016. JAMA Neurology, 78(3), 275–210. https://doi.org/ 10.1001/jamaneurol.2020.4471 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Röhr, S., Löbner, M., Gühne, U., Heser, K., Kleineidam, L., Pentzek, M., Fuchs, A., Eisele, M., Kaduszkiewicz, H., König, H.-H., Brettschneider, C., Wiese, B., Mamone, S., Weyerer, S., Werle, J., Bickel, H., Weeg, D., Maier, W., Scherer, M., … Riedel-Heller, S. G. (2020). Changes in social network size are associated with cognitive changes in the oldest-old. Frontiers in Psychiatry, 11, 330. https://doi.org/ 10.3389/fpsyt.2020.00330 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ruiz, M. E., & Ransford, H. E. (2012). Latino elders reframing familismo: Implications for health and caregiving support. Journal of Cultural Diversity, 19(2), 50–57. [ PubMed ] [ Google Scholar ] Shi, J., Zhang, D., & Liu, X. (2024). Intergenerational caregiving patterns and cognitive health among the sandwich generation within four-generation families. International Journal of Aging & Human Development, 99(3), 307–337. https://doi.org/ 10.1177/00914150241235088 [ DOI ] [ PubMed ] [ Google Scholar ] Smith, J., Ryan, L. H., Fisher, G. G., Sonnega, A., & Weir, D. R. (2017). HRS Psychosocial and Lifestyle Questionnaire 2006–2016. Survey Research Center, Institute for Social Research, University of Michigan. [ Google Scholar ] Sonnega, A., Faul, J. D., Ofstedal, M. B., Langa, K. M., Phillips, J. W., & Weir, D. R. (2014). Cohort profile: The Health and Retirement Study (HRS). International Journal of Epidemiology, 43(2), 576–585. https://doi.org/ 10.1093/ije/dyu067 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Spillman, B. C., Freedman, V. A., Kasper, J. D., & Wolff, J. L. (2020). Change over time in caregiving networks for older adults with and without dementia. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(7), 1563–1572. https://doi.org/ 10.1093/geronb/gbz065 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] StataCorp. (2023). Stata Statistical Software: Release 18 [Computer software]. StataCorp LLC. [ Google Scholar ] Tibiriçá, L., Jester, D. J., & Jeste, D. V. (2022). A systematic review of loneliness and social isolation among Hispanic/Latinx older adults in the United States. Psychiatry Research, 313, 114568. https://doi.org/ 10.1016/j.psychres.2022.114568 [ DOI ] [ PubMed ] [ Google Scholar ] Wadham, O., Simpson, J., Rust, J., & Murray, C. (2016). Couples’ shared experiences of dementia: A meta-synthesis of the impact upon relationships and couplehood. Aging & Mental Health, 20(5), 463–473. https://doi.org/ 10.1080/13607863.2015.1023769 [ DOI ] [ PubMed ] [ Google Scholar ] Wang, C. X., Hussain, M. A., Tibiriçá, L., Castellanos-Ponce, V., Cherner, M., Palmer, B. W., Kamalyan, L., Yassai-Gonzalez, D., Umlauf, A., Barnes, L. L., Heaton, R. K., Ellis, R. J., & Marquine, M. J. (2025). Social engagement and neurocognition in Latino and non-Latino White older persons with HIV. AIDS and Behavior, 29, 1096–1108. https://doi.org/ 10.1007/s10461-024-04586-7 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wenger, G. C. (1994). Support networks of older people: A guide for practitioners. Routledge. [ Google Scholar ] Willis, R. (2012). Individualism, collectivism and ethnic identity: Cultural assumptions in accounting for caregiving behaviour in Britain. Journal of Cross-Cultural Gerontology, 27(3), 201–216. https://doi.org/ 10.1007/s10823-012-9175-0 [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials gbaf062_suppl_Supplementary_Materials gbaf062_suppl_supplementary_materials.docx (16.5KB, docx) Data Availability Statement Data for this study are made publicly available by the Health and Retirement Study on their individual website ( https://hrs.isr.umich.edu/ ). The present analysis of these data was not preregistered. Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press ACTIONS View on publisher site PDF (1.1 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top