Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2022 May 4;77(10):1916–1927. doi: 10.1093/geronb/gbac069

Marital Loss and Cognitive Function: Does Timing Matter?

Zhenmei Zhang 1,✉, Hui Liu 2, Yan Zhang 3
Editor: Jessica Kelley
PMCID: PMC9535775  PMID: 35512289

Abstract

Objectives

This study examines the association between age at marital loss (i.e., divorce or widowhood) and cognitive function in later life and whether the association differs by gender.

Methods

We used mixed-effects models, drawing on longitudinal data from the Health and Retirement Study (1998–2016). The analytical samples included adults aged 51 and older who had ever been widowed (N = 5,639 with 25,537 person-waves) or divorced (N = 10,685 with 50,689 person-waves).

Results

We find that those who were widowed at younger ages had lower cognitive function than their counterparts who were widowed at older ages, for both men and women, after controlling for covariates. Household income and health-related factors partially accounted for the positive association between age at widowhood and cognitive function. Those who divorced at younger ages also had lower cognitive function than their counterparts who divorced at older ages, but this association was only present among men, not women. Health-related factors partially accounted for the association between age at divorce and cognitive function among men.

Discussion

Findings highlight the importance of considering the role of timing of marital loss in cognitive health among older adults.

Keywords: Cognition, Divorce, Gender, Timing of life course events, Widowhood

Marital loss (i.e., divorce or widowhood) has long topped the list of the most stressful life transitions an individual can experience. Historically, most marriages were dissolved through the death of a spouse. However, divorce rates have increased significantly since the 1970s, declining in recent years but remaining high in the United States (Carr & Utz, 2020; Raley & Sweeney, 2020). Recent studies have documented a dramatic increase in gray divorce, with 1 in 4 divorces involving individuals aged 50 and older (Brown & Lin, 2012). Marital loss can result in significant changes in an individual’s standard of living, social networks, and health behaviors, exerting short-term and long-term influences on their health (Leopold, 2018; Zhang et al., 2016). Despite a large literature on the health consequences of divorce and widowhood, few studies have focused on cognitive outcomes, and even fewer have examined whether age at marital loss can influence cognitive function in later life (Bertogg & Leist 2021; Brown et al., 2021). Cognitive function encompasses mental abilities to acquire and process information, think and reason, and plan and manage the demands of daily living. Maintaining cognitive function is essential for leading a productive and independent life (Langa et al., 2001).

A growing body of research has found that both divorce and widowhood are associated with lower cognitive function and higher risks of cognitive impairment in later life (Brown et al., 2021; Sundström et al., 2016). However, we know little about whether the timing of marital loss matters for later-life cognition. According to the life course perspective, the timing of an event can magnify or reduce its influence on different domains of life, including economic conditions and health (Lin et al., 2017; Liu, 2012). Previous research has suggested that the health consequences of marital loss depend on the age at which it occurs (Liu, 2012; Williams & Umberson, 2004). Moreover, the timing of marital loss may interact with gender to influence health outcomes. For example, Williams and Umberson (2004) found that while the transition to divorce or widowhood did not affect women’s self-assessed health at any age, the negative health effects of exiting marriage increased with age for men. Yet few studies have examined the potential mechanisms underlying the interplay among the timing of marital loss, gender, and health outcomes.

Our study aims to extend the previous literature by examining whether age at divorce and age at widowhood influence later-life cognitive function, measured by an adapted version of the Telephone Interview for Cognitive Status (TICS). As far as we know, our study is one of the first to examine the associations between the timing of marital loss and cognitive function among older Americans. Drawing on 18 years of data (1998–2016) from the Health and Retirement Study (HRS), we aim to answer the following questions:

  • (1) Is age at marital loss associated with cognitive function?

  • (2) To what extent is the association accounted for by economic resources and health-related factors?

  • (3) Are there gender differences in the association between age at marital loss and cognitive function?

Background

Marital Loss and Cognitive Health

The literature on the health consequences of divorce and widowhood suggests that marital loss is highly stressful and may have negative effects on health (Raley & Sweeney, 2020; Zhang et al., 2016). Recent research shows much heterogeneity in the association between marital loss and health, depending on age, prior marital quality, and parenthood status (Sbarra, 2015). For example, leaving a bad marriage may bring relief and mitigate the negative health consequences of divorce (Kalmijn & Monden, 2006; Sbarra, 2015). As for cognitive outcomes, the preponderance of evidence has shown negative effects of marital loss on cognitive health (Liu et al., 2020; Sommerlad et al., 2018), although some studies in European countries have found no significant effects (Vidarsdottir et al., 2014; Wörn et al., 2020).

Marital loss can affect cognitive health via biopsychosocial pathways. According to the marital resource model, the dissolution of marriage entails the loss of multiple economic and psychosocial resources that are protective of cognitive function (Liu et al., 2020; Zhang et al., 2016). In terms of economic resources, divorce and widowhood often lead to a lower standard of living, and some divorced and widowed adults may slip into poverty (Leopold, 2018; Lin et al., 2017). Previous research has shown that household income is positively associated with cognitive health (Scazufca et al., 2010; Yaffe et al., 2013). Marital dissolution also leads to the loss of daily mental stimulation, emotional support, and health monitoring from a spouse as well as losses in social networks (e.g., mutual friends and kin of spouse), which help to maintain the cognitive health of older adults (Fratiglioni et al., 2004; Leopold, 2018; Umberson, 1992).

According to the stress model, marital loss may exert stress on the divorced and the widowed (McFarland et al., 2013; Zhang & Hayward, 2006). The death of a spouse often has stronger emotional consequences (e.g., psychological distress) and triggers more negative effects on health than divorce (Pudrovska & Carr, 2008). Mounting research has shown that chronic stress can affect cognitive health through biochemical and behavioral mechanisms (Greenberg et al., 2014; Yan et al., 2018). Stress is also associated with chronic diseases such as cardiovascular disease and stroke, which are risk factors for dementia (Morley, 2017; Ramirez-Moreno et al., 2020). People may also adopt unhealthy behaviors (e.g., smoking and inactivity) to cope with the stress of marital loss. These unhealthy behaviors can significantly increase dementia risk (Zhou et al., 2014).

Many individuals remarry after experiencing marital loss. Remarriage is likely to improve economic conditions, provide social support and mental stimulation, and enlarge social networks, all of which may mitigate the negative effects of marital loss and enhance cognitive health. Recent research showed that the repartnered in midlife were similar to the stably married in the risk of cognitive impairment (Brown et al., 2021). However, it is unknown whether age at marital loss matters for cognitive function, net of current marital status.

The Role of Age at Marital Loss

The life course theory has long argued that “depending on when it occurs in the life course, the meaning of a transition differs and affects an individual differently” (Almeida & Wong, 2009, p. 142). However, research theorizing and empirically testing the association between the timing of marital loss and health has been limited. In this section, we develop our research hypotheses based on life course theories and previous research.

The life course perspective suggests that the timing of a life transition can influence (a) the normative status of the new role and (b) the challenges and the resources/skills needed to adjust to the new role (Williams & Umberson, 2004). It is suggested that an off-time event can be more stressful and have more negative consequences than an on-time event (Koropeckyj-Cox et al., 2007). For instance, widowhood is more common at older ages: the median age of widowhood from first marriage for Americans in 2009 was about 61 for men and 59 for women (Kreider & Ellis, 2011; Lin et al., 2017). Widowhood at younger ages is more likely to be unexpected and sudden, which can lead to greater difficulty in adjustment and higher levels of depressive symptoms (Carnelley et al., 1999; Carr & Utz, 2020; Sasson & Umberson, 2014). In terms of economic resources, adults who lose their spouses at younger ages may not have had time to build enough wealth to help them weather the financial losses that come after losing a spouse, nor are they eligible to receive social security benefits (Lin et al., 2017). Several studies have found that adults who became widowed at younger ages are more likely to live in poverty than their counterparts who lost their spouses at older ages (Holden et al., 1988; Lin et al., 2017). In terms of coping skills, according to the “age-as-maturity” perspective, as people get older, they have accumulated more life experiences and more coping skills to deal with hardships compared to those who are young and inexperienced (Liu et al., 2012; Mirowsky & Ross, 2001). Because depression and low income are associated with cognitive impairment (Chiao et al., 2014; James & Bennett, 2019), we hypothesize that widowhood at younger ages is associated with lower cognitive function in later life (Hypothesis 1a). On the other hand, it is also possible that people who are widowed at younger ages may recover more quickly than those at older ages because (a) at older ages additional stressors (e.g., chronic diseases, declining income, and losing family members and friends) may pile up and make it harder to adjust to the loss of a spouse (Mirowsky & Ross, 2001; Williams & Umberson, 2004), and (b) recency of spousal loss is associated with acute grief and more depression (Lin & Brown, 2020). We thus form the alternative hypothesis that widowhood at older ages is associated with lower cognitive function (Hypothesis 1b).

Far less is known about how the timing of divorce is related to later health. Due to demographic and cultural changes (e.g., longer life expectancy, more acceptance of divorce, and rising women’s employment), divorce is less tied to a specific stage of the life course (i.e., young adulthood) than it was in earlier decades (Brown & Lin, 2012). Previous literature has suggested two competing perspectives. According to the first perspective, although divorce at younger ages is more common, oftentimes the children of those marriages are still young, and divorce could lead to chronic stress due to either single-parenthood or coparenting and the associated conflicts with ex-spouses and their new partners (Dahl et al., 2015; Kamp Dush, 2013). In contrast, for those who divorce at older ages, their most intensive parenting days may be over, and the associated parenting stress and conflicts may be lower. Moreover, as we discussed before, with increasing age, people may also have more coping skills to resolve their conflicts with ex-spouses. Empirically, Liu (2012) found that, for the 1950s birth cohort in the United States, the negative effect of divorce on self-reported health was stronger for those who divorced at younger ages (35–42) than those who divorced at older ages (44–51). In another study in Germany, older age at divorce was associated with a slower decline in general well-being after divorce (Leopold & Kalmijn, 2016). Taking these findings together, we hypothesize that younger age at divorce is associated with lower cognitive function (Hypothesis 2a).

On the other hand, despite the rise in divorce among older couples in recent decades, divorce is still more normative at younger ages (Lin et al., 2017). In addition, divorce at older ages has been associated with greater financial losses and can leave fewer years for the divorced to rebuild their social support networks (Lin & Brown, 2020). For example, Lin et al. (2017) found that among divorced women aged 63 and older, those who had divorced before age 50 had a 30% lower poverty rate than their counterparts who had divorced after age 50. In comparison, divorce timing had a much smaller effect on the poverty rate among divorced men: Those who had divorced before age 50 had a 6% lower poverty rate than those who had divorced after 50. Thus, we hypothesize that older age at divorce is associated with lower cognitive function (Hypothesis 2b).

The Role of Gender

The literature about gender variation in the link between marital loss and health is mixed. A majority of studies suggest that divorce and widowhood exert stronger effects on men’s health than on women’s, whereas some have found no gender differences or stronger effects of divorce on women’s cardiovascular disease (Dupre et al., 2015; Dykstra & Fokkema, 2007; Sasson & Umberson, 2014). In a traditional marriage, the wife is the kinkeeper and caregiver, providing social support and social monitoring of the health behaviors of her husband. Thus, men may lose more when their marriage dissolves compared to women (Liu et al., 2020). On the other hand, marital dissolution brings greater financial loss to women than to men (Leopold & Kalmijn, 2016). Recent studies on marital status and cognitive outcomes in later life have also reported mixed results. For example, studies in the United States reported a stronger impact of divorce on the risk of dementia for men than for women (Liu et al., 2020; Zhang et al., 2021), while a study in Sweden did not find significant gender differences in the association between divorce and dementia risk after confounders were adjusted (Sundström et al., 2016). As for widowhood, recent articles have found greater negative effects on cognition for men than for women (Brown et al., 2021; Liu et al., 2020).

Whether gender moderates the association between the timing of marital loss and late-life health is unknown. On the one hand, in the past, divorce brought stigma, especially for women. Many divorced women in our sample who were divorced at younger ages had to live through a period of high societal disapproval of divorce (before the 1970s), which was especially challenging given women’s lack of access to economic opportunities and independence during that time (Amato & Irving, 2006; Gerstel, 1987). In this sense, age at marital loss may matter more for women than for men. On the other hand, because men benefit more from marriage than women, older age at first marital dissolution suggests that they may have stayed in the marriage longer, and thus may have reaped more benefits (e.g., emotional support and healthy lifestyles) from marriage than women did. Therefore, the timing of marital loss may matter more for men than for women. Given the limited research on this topic, we will explore whether the association between age at marital loss and late-life cognition varies by gender.

Data and Methods

Data

We used 10 waves of data from the HRS (1998–2016). The original HRS cohort consists of respondents aged 51–61 in 1992 and their spouses. In 1998, new cohorts were added to make the HRS a nationally representative sample of adults aged 50 and older. Since then, the HRS has refreshed its panel every 6 years (i.e., 2004, 2010, and 2016), adding new participants aged 51–56. The HRS oversamples Blacks and Hispanics and collects detailed information on cognitive, physical, economic, work, and family conditions as well as health behaviors by either telephone or in-person interview. Some of the covariates (e.g., household income) come from the RAND HRS DATA FILES. Our analytic samples were restricted to respondents aged 51 and older who experienced marital losses, had no missing values on age at a first marital loss and other covariates, and participated in the cognitive tests. Based on these criteria, we constructed two analytic samples: The ever-divorced sample included 10,685 respondents who had experienced divorce up to 2016 (N = 50,689 person-waves), and the ever-widowed sample included 5,639 respondents who had lost their spouses (N = 25,537 person-waves). The number of observations per person ranged from 1 to 10.

Measures

Dependent variable

Cognitive function

For respondents who agreed to participate in cognitive tests, the HRS assessed their cognitive function using the modified version of TICS. Respondents’ total cognitive scores are based on the tests of immediate and delayed recall of a list of 10 words (20 points), five trials of serial 7s (5 points), and backward counting (2 points). The summary score ranges from 0 (severely impaired) to 27 (high functioning). The TICS measure has good internal consistency and consists of two related factors (i.e., a memory factor and a mental state factor; Ofstedal et al., 2005). The HRS has developed a multiple imputation strategy that imputes missing cognitive variables for all waves, which we used (Fisher et al., 2017).

Independent variables

Age at first divorce (hereafter, age at divorce) and age at first widowhood (hereafter, age at widowhood). The HRS collected information on marital history when respondents first joined the study. Ever-married respondents were asked about the beginning and ending years of their previous three marriages and their most recent marriage in all HRS waves except the 1994 wave (which asked about the first marriage only) and the 1996 wave (which asked about the first and the most recent marriage). At each follow-up, respondents were asked whether there were any marital transitions since the last interview as well as the years of these transitions. Based on the retrospective report of marital history and the prospective follow-up interviews, we constructed the two variables that indicate the timing of marital loss (i.e., age at divorce and age at widowhood) for each respondent who experienced marital loss up to 2016.

Potential mediators

We tested two types of potential mediators (all time-varying): household income and health-related factors. Household income was measured by the total household income in the year before each interview wave. To adjust for the skewness of the income distribution and differences in household sizes, we divided the original household income by the square root of household size, adding one to eliminate zero incomes, and applying the logarithm (Glymour et al., 2008).

Health-related factors included four measures: smoking, chronic conditions, depressive symptoms, and psychiatric problems. Smoking was coded into a binary variable, either nonsmoker (reference) or current smoker. Current smoking is associated with a higher risk of dementia (Peters et al., 2008). Chronic conditions were a comorbidity index ranging from 0 to 5, representing the presence of five major physical diseases that are risk factors for cognitive impairment: stroke, diabetes, heart disease, high blood pressure, and chronic lung disease. Scores were based on respondents’ self-report of a physician’s diagnosis. Depressive symptoms were measured from a subset of the Center for Epidemiologic Studies Depression Scale. Participants were asked how often during the past week they felt depressed, lonely, sad, and happy, as well as how often they enjoyed life, had a restless sleep, felt everything was an effort, and could not get going. The sum of the scores ranged from 0 to 8, with higher scores indicating more depressive symptoms. Psychiatric problems, an independent risk factor for lower cognition and steeper cognitive decline (Brown, 2010), are coded as a dummy variable (1 = ever having emotional, nervous, or psychiatric problems).

Other Covariates

Covariates include age, age squared, gender, race/ethnicity, education, current marital status, employment status, number of marriages, and cohort. Age was measured as a continuous variable and centered. Age squared was included to model the nonlinear relationship between age and cognitive function. Gender was a dummy variable (0 = male, 1 = female). Race-ethnicity was coded into four categories, including non-Hispanic Whites (reference), non-Hispanic Blacks, Hispanics, and others. Education was categorized into four groups, including less than high school (reference), high school graduate, some college, and college graduate or above. Employment status was a dummy variable (1 = working for pay). Current marital status represented participants’ marital status at each wave, including remarried (reference), cohabiting, divorced/separated, and widowed. Number of marriages had three categories: 1 (reference), 2, 3, and more. Cohort was categorized into six groups based on birth year: early children of depression (1924–1930), late children of depression (1931–1941), war babies (1942–1947), early-baby boomers (1948–1953), mid-baby boomers (1954–1959), and late-baby boomers (1960–1965). All analytic covariates are time-varying except for gender, race/ethnicity, education, and cohort.

Analytic Strategy

We used mixed-effects models to examine age trajectories of cognitive function in relation to age at divorce and age at widowhood. Our independent variables included both time-invariant variables (i.e., age at marital loss, gender, race/ethnicity, education, and cohort) and time-varying variables (e.g., age, age squared, household income, and health-related variables). The basic equation takes the following form:

Level 1:

Yij=π0i+π1iAgeij+π2iAgeij2+Z′ijTVCij+εij

Level 2:

π0i=γ00+X′A0+ζ0i   
π1i=γ10+ζ1i
π2i= γ20

where Yij refers to cognitive function for individual i at time j and is modeled as a function of linear and quadratic terms of age for individual i at time j as well as time-varying covariates (TVCij). In level 2, π 0i is the ith individual’s intercept, and X΄ is a vector of time-invariant variables. π 1i and π 2i are the ith individual’s linear and quadratic age slopes, respectively. Because results from preliminary analysis (available upon request) suggested no significant effect of age at divorce or age at widowhood on the age slopes of cognitive function, we did not include covariates in the equations to predict age slopes. Both the intercept, π 0i, and the linear age slope, π 1i, were estimated as random effects, and the quadratic age slope, π 2i, was estimated as a fixed effect for model parsimony.

We estimated a series of nested mixed-effects models to understand how age at marital loss was associated with cognitive function. Model 1 included age at marital loss and basic demographic controls. Models 2 and 3 added household income and health-related factors separately to examine whether the association between age at marital loss and cognitive function would be explained by these factors. Model 4 added all covariates. In Model 5 we added an interaction term between age at marital loss and gender with all covariates controlled. We estimated all models using Stata 15 (StataCorp, 2017).

Results

Descriptive Statistics

Table 1 presents the descriptive statistics of our analytic sample. For the ever-widowed sample, the respondents’ average cognitive score was 14.45, on a scale of 0 to 27. The average age at first widowhood was about 56.07 years old, and the current age of ever-widowed respondents was about 70.31 years old. Women accounted for 75.28% of the sample. The majority were currently widowed (73.52%), 18.57% were remarried, 3.54% were divorced/separated, and 4.36% were cohabiting. About 31.71% had married twice, and 10.82% had married three or more times. For the ever-divorced sample, the respondents’ average cognitive score was 15.68. The average age at first divorce was about 34.36 years old, and the current age of respondents in the divorced sample was about 63.82 years old. Women accounted for 57.70% of the ever-divorced sample. Nearly half were currently remarried (49.08%), 33.75% were divorced/separated, 10.30% were widowed, and 6.87% were cohabiting. About 54.64% had married twice, and 21.71% had married three or more times.

Table 1.

Descriptive Statistics of All Analytical Variables, HRS 1998–2016, Mean (SD)/%

Variables Ever widowed Ever divorced Ever widowed Ever divorced
Cognition (0–27) 14.45 (4.62) 15.68 (4.33) Number of marriages (%)
Age at first widowhood (15–92) 56.07 (14.72)  1 57.47 23.65
Age at first divorce (15–86) 34.36 (10.57)  2 31.71 54.64
Gender (%)  3 and more 10.82 21.71
 Male 24.72 42.30 Cohort (%)
 Female 75.28 57.70  Early children of depression 17.94 5.43
Age in widowed sample (51–99) 70.31 (8.68)  Late children of depression 63.6 46.99
Age in divorced sample (51–102) 63.82 (8.18)  War babies 7.16 13.99
Race-ethnicity (%)  Early-baby boomers 6.86 18.25
 Non-Hispanic Whites 70.31 69.20  Mid-baby boomers 3.74 12.42
 Non-Hispanic Blacks 18.97 18.84  Late-baby boomers 0.70 2.91
 Hispanics 5.79 5.80 Employment status (%)
 Other 4.93 6.16  Not working for pay 74.27 53.01
Education (%)  Working for pay 25.73 46.99
 Less than high school 28.52 19.15 Household income (unit: 10k) 4.09 (12.60) 6.52 (29.74)
 High school graduate 36.91 32.05 Smoking (%)
 Some college 27.69 37.61  nonsmoker 84.46 79.12
 College graduate or above 6.89 11.20  Current smoker 15.54 20.88
Current marital status (%) Chronic conditions (0–5) 1.35 (1.10) 1.17 (1.09)
 Remarried 18.57 49.08 Depressive symptoms (0–8) 1.81 (2.13) 1.65 (2.13)
 Cohabiting 4.36 6.87 Psychiatric problems (%)
 Divorce/separated 3.54 33.75  No 82.89 80.47
 Widowed 73.52 10.30  Yes 17.11 19.53
Open in a new tab

Notes: HRS = Health and Retirement Study; SD = standard deviation. Number of ever-widowed person-waves = 25,537; number of ever-divorced person-waves = 50,689.

Age at Widowhood and Cognitive Function

Table 2 displays how age at widowhood was associated with cognitive function. The results from Model 1 of Table 2 indicate that among those who were ever widowed, there was a positive association between age at widowhood and cognitive function in later life, controlling for basic demographic covariates. In other words, younger age at widowhood was associated with lower cognitive function. Model 2 adjusted for household income. We found that the coefficient of age at widowhood was reduced slightly but remained statistically significant, and household income was positively associated with cognitive function. In Model 3, we tested whether health-related factors accounted for the association. The coefficient of age at widowhood was also reduced slightly after health-related factors were added but remained statistically significant; and chronic conditions, depressive symptoms, and psychiatric problems are all related to lower cognitive function. We then added all the covariates in Model 4, and the coefficient of age at widowhood remained statistically significant. In an additional analysis (available upon request), we found that younger age at widowhood was significantly associated with lower household income, more chronic disease, and current smoking. However, younger age at widowhood was also associated with a lower risk of having psychiatric problems. These results suggest household income and health-related factors (i.e., chronic conditions and smoking) are potential key mediators in the association between younger age at widowhood and lower cognition in later life. Finally, to test whether the association between age at widowhood and cognitive function was moderated by gender, we added Female × Age at widowhood in Model 5, but the interaction term was not statistically significant.

Table 2.

Mixed-Effects Models, Age at First Widowhood and Cognition, HRS 1998–2016

M1 M2 M3 M4 M5
Variables Coefficient SE Coefficient SE Coefficient SE Coefficient SE Coefficient SE
Age at first widowhood (centered) 0.018*** (0.004) 0.017*** (0.004) 0.017*** (0.003) 0.016*** (0.003) 0.020*** (0.006)
Gender (ref: male) 1.092*** (0.101) 1.108*** (0.100) 1.127*** (0.099) 1.141*** (0.098) 1.157*** (0.100)
Age (centered) −0.200*** (0.005) −0.202*** (0.005) −0.189*** (0.005) −0.191*** (0.005) −0.191*** (0.005)
Age squared (centered) −0.006*** (0.000) −0.006*** (0.000) −0.006*** (0.000) −0.006*** (0.000) −0.006*** (0.000)
Race/ethnicity (ref: non-Hispanic Whites)
 Non-Hispanic Blacks −2.741*** (0.116) −2.682*** (0.116) −2.731*** (0.114) −2.676*** (0.114) −2.675*** (0.114)
 Hispanics −2.086*** (0.185) −1.977*** (0.185) −2.068*** (0.181) −1.966*** (0.181) −1.963*** (0.181)
 Other −1.702*** (0.192) −1.612*** (0.191) −1.668*** (0.187) −1.584*** (0.187) −1.581*** (0.187)
Marital status (ref: remarried)
 Cohabiting −0.412** (0.155) −0.426** (0.155) −0.304* (0.154) −0.318* (0.154) −0.322* (0.154)
 Divorced/separated −0.284 (0.158) −0.213 (0.158) −0.096 (0.157) −0.033 (0.157) −0.037 (0.157)
 Widowed −0.340** (0.105) −0.280** (0.105) −0.157 (0.105) −0.104 (0.105) −0.109 (0.105)
Education (ref: <high school)
 High school 2.421*** (0.111) 2.353*** (0.110) 2.271*** (0.109) 2.211*** (0.108) 2.214*** (0.108)
 Some college 3.659*** (0.117) 3.545*** (0.118) 3.428*** (0.116) 3.328*** (0.116) 3.330*** (0.116)
 College or above 5.105*** (0.188) 4.934*** (0.188) 4.778*** (0.185) 4.627*** (0.185) 4.627*** (0.185)
Number of marriages (ref: once)
 Married twice 0.026 (0.101) 0.021 (0.100) 0.104 (0.099) 0.098 (0.099) 0.099 (0.099)
 Married three times and more −0.164 (0.149) −0.174 (0.148) 0.047 (0.147) 0.034 (0.146) 0.036 (0.146)
Birth cohorts (ref: early children of depression)
 Late children of depression −0.179 (0.126) −0.186 (0.125) −0.035 (0.123) −0.044 (0.123) −0.043 (0.123)
 War babies −0.403* (0.201) −0.440* (0.200) −0.202 (0.197) −0.240 (0.196) −0.238 (0.196)
 Early-baby boomers −0.826*** (0.201) −0.836*** (0.200) −0.542** (0.199) −0.557** (0.198) −0.555** (0.198)
 Mid-baby boomers −0.741** (0.242) −0.754** (0.241) −0.439 (0.239) −0.456 (0.238) −0.449 (0.238)
 Late-baby boomers −1.414*** (0.345) −1.396*** (0.344) −1.105** (0.342) −1.094** (0.341) −1.083** (0.341)
Employment status (ref: not working) 0.514*** (0.063) 0.452*** (0.064) 0.393*** (0.063) 0.336*** (0.064) 0.336*** (0.064)
Household income (logged) 0.136*** (0.019) 0.126*** (0.019) 0.126*** (0.019)
Current smoker (ref: nonsmoker) −0.032 (0.086) −0.020 (0.086) −0.020 (0.086)
Chronic conditions (0–5) −0.260*** (0.030) −0.256*** (0.030) −0.257*** (0.030)
Depressive symptoms (0–8) −0.120*** (0.012) −0.118*** (0.012) −0.118*** (0.012)
Psychiatric problems (ref: no) −0.658*** (0.093) −0.648*** (0.092) −0.647*** (0.092)
Female × Age at 1st widowhood −0.005 (0.006)
Random effects
 Within-person (level 1) 7.353* (0.079) 7.364* (0.079) 7.348* (0.078) 7.357* (0.079) 7.357* (0.079)
 In intercept (level 2) 7.236* (0.200) 7.090* (0.198) 6.761* (0.190) 6.643* (0.189) 6.644* (0.189)
 In linear growth rate (level 2) 0.016* (0.001) 0.015* (0.001) 0.015* (0.001) 0.015* (0.001) 0.015* (0.001)
 Covariance between random intercept and slope 0.068* (0.011) 0.066* (0.011) 0.063* (0.011) 0.061* (0.011) 0.061* (0.011)
Goodness of fit (BIC) 133646.8 133604.0 133404.3 133368.5 133378.0
Open in a new tab

Notes: BIC = Bayesian information criterion; HRS = Health and Retirement Study; SE = standard error. Random effects are significant at 95% level. Standard errors in parentheses. ***p < .001; **p < .01; *p < .05. Number of respondents = 5,639, number of person-periods = 25,537.

Age at Divorce and Cognitive Function

Next, we ran the same sets of analyses for those who had gone through a divorce. The results from Model 1 of Table 3 indicate that age at divorce was positively associated with cognitive function in later life, controlling for demographic variables, meaning that younger age at divorce was associated with lower cognitive function. When household income was added in Model 2, the coefficient of age at divorce barely changed. In Model 3, we added health-related factors to Model 1, and the coefficient of age at divorce was reduced but remained statistically significant. In our full model (Model 4), we included all the covariates, and the association between age at divorce and cognitive function remained robust.

Table 3.

Mixed-Effects Models, Age at First Divorce and Cognition, HRS 1998–2016

M1 M2 M3 M4 M5
Variables Coefficient SE Coefficient SE Coefficient SE Coefficient SE Coefficient SE
Age at first divorce (centered) 0.011*** (0.003) 0.011*** (0.003) 0.008** (0.003) 0.008** (0.003) 0.016*** (0.004)
Gender (ref: male) 0.819*** (0.060) 0.831*** (0.060) 0.869*** (0.059) 0.880*** (0.059) 0.881*** (0.059)
Age (centered) −0.124*** (0.003) −0.126*** (0.003) −0.118*** (0.003) −0.120*** (0.003) −0.120*** (0.003)
Age squared (centered) −0.006*** (0.000) −0.006*** (0.000) −0.005*** (0.000) −0.005*** (0.000) −0.005*** (0.000)
Race and ethnicity (ref: non-Hispanic Whites)
 Non-Hispanic Blacks −2.218*** (0.076) −2.170*** (0.075) −2.187*** (0.075) −2.144*** (0.074) −2.143*** (0.074)
 Hispanics −1.381*** (0.122) −1.293*** (0.122) −1.389*** (0.120) −1.308*** (0.120) −1.304*** (0.120)
 Other −1.716*** (0.112) −1.651*** (0.111) −1.668*** (0.110) −1.609*** (0.110) −1.605*** (0.110)
Marital status (ref: remarried)
 Cohabiting −0.183* (0.082) −0.175* (0.081) −0.112 (0.081) −0.106 (0.081) −0.111 (0.081)
 Divorced/separated −0.266*** (0.062) −0.175** (0.062) −0.138* (0.062) −0.058 (0.062) −0.061 (0.062)
 Widowed −0.225** (0.070) −0.157* (0.071) −0.091 (0.070) −0.033 (0.070) −0.036 (0.070)
Education (ref: <high school)
 High school 1.951*** (0.086) 1.892*** (0.086) 1.815*** (0.085) 1.766*** (0.085) 1.770*** (0.084)
 Some college 3.238*** (0.084) 3.137*** (0.084) 3.045*** (0.083) 2.960*** (0.083) 2.960*** (0.083)
 College or above 4.756*** (0.115) 4.592*** (0.116) 4.477*** (0.114) 4.338*** (0.115) 4.337*** (0.115)
Number of marriages (ref: once)
 Married twice 0.105 (0.080) 0.097 (0.080) 0.122 (0.079) 0.114 (0.079) 0.101 (0.079)
 Married three times and more −0.033 (0.101) −0.033 (0.100) 0.059 (0.100) 0.057 (0.099) 0.044 (0.099)
Birth cohorts (ref: early children of depression)
Late children of depression −0.034 (0.153) −0.048 (0.152) 0.069 (0.150) 0.053 (0.149) 0.060 (0.149)
War babies −0.221 (0.172) −0.256 (0.171) −0.059 (0.169) −0.094 (0.168) −0.089 (0.168)
Early-baby boomers −0.678*** (0.163) −0.700*** (0.163) −0.478** (0.161) −0.502** (0.160) −0.495** (0.160)
Mid-baby boomers −0.978*** (0.167) −1.001*** (0.166) −0.765*** (0.165) −0.790*** (0.164) −0.780*** (0.164)
Late-baby boomers −0.967*** (0.184) −0.998*** (0.183) −0.748*** (0.182) −0.781*** (0.181) −0.777*** (0.181)
Employment status (ref: not working for pay) 0.477*** (0.038) 0.414*** (0.038) 0.338*** (0.038) 0.283*** (0.039) 0.284*** (0.039)
Household income (logged) 0.128*** (0.011) 0.116*** (0.011) 0.116*** (0.011)
Current smoker (ref: nonsmoker) −0.142** (0.052) −0.128* (0.052) −0.129* (0.052)
Chronic conditions (0–5) −0.204*** (0.021) −0.199*** (0.021) −0.199*** (0.021)
Depressive symptoms (0–8) −0.117*** (0.008) −0.114*** (0.008) −0.114*** (0.008)
Psychiatric problems (ref: no) −0.433*** (0.061) −0.427*** (0.060) −0.428*** (0.060)
Female × Age at 1st divorce −0.015** (0.005)
Random effects
 Within-person (level 1) 6.988* (0.053) 6.994* (0.053) 6.983* (0.052) 6.986* (0.052) 6.986* (0.052)
 In intercept (level 2) 6.415* (0.129) 6.302* (0.127) 6.078* (0.123) 5.992* (0.122) 5.985* (0.122)
 In linear growth rate (level 2) 0.012* (0.001) 0.012* (0.001) 0.012* (0.001) 0.012* (0.001) 0.012* (0.001)
 Covariance between random intercept and slope 0.042* (0.007) 0.042* (0.007) 0.036* (0.007) 0.036* (0.007) 0.036* (0.007)
Goodness of fit (BIC) 260769.8 260654.6 260369.1 260275.0 260278.2
Open in a new tab

Notes: BIC = Bayesian information criterion; HRS = Health and Retirement Study; SE = standard error. Random effects are significant at 95% level. Standard errors in parentheses. ***p < .001; **p < .01; *p < .05. Number of respondents = 10,685, number of person-periods = 50,689.

To test whether gender moderates the association between age at divorce and cognitive function, we added the interaction term (Female × Age at divorce) in Model 5. The statistically significant interaction (b = −0.015, p < .01) indicates that the link between age at divorce and cognition varies by gender. We calculated the predicted cognitive scores based on the results in Model 5 of Table 3, and graphically presented them in Figure 1. It shows that the association between age at divorce and cognitive function was stronger for men than for women. Additional analyses show that the association between age at divorce and cognition was only statistically significant among men but not among women. We also found that age at divorce was significantly associated with three health-related factors (i.e., depressive symptoms, chronic disease, and smoking) but not with household income among men, suggesting that health-related factors are an important pathway linking age at divorce and cognition in men’s later life (results available upon request).

Figure 1.

Figure 1.

Open in a new tab

Age at first divorce and cognition by gender. Notes: We used the marginsplot command in STATA to generate a graph showing how the predicted value of cognition changes by age at first divorce for both men and women. Covariates are set at their means.

At first glance, the coefficients of age at widowhood and age at divorce appear small (<0.02) in both Tables 2 and 3, which indicates the average difference in cognition for a one-year increase in age at widowhood and divorce, respectively. To better understand the magnitude of the associations, we compare the cognitive scores of those who were widowed at age 70 with those who were widowed at age 40, all else equal. The latter group would have 0.48 (0.016 × 30) lower cognitive scores. In comparison, the coefficient of chronic conditions is −0.256 (Model 4 of Table 2).

We also note that although the remarried in general had higher cognition than other unmarried groups with the basic controls (Model 1) in both Tables 2 and 3, in our full model, the remarried were no longer different from the widowed or the divorced except that they still had higher cognition than the cohabitors in the ever-widowed sample. These results suggest that the cognitive differences between the currently widowed/divorced and the remarried are largely explained by their differences in economic resources and health behaviors.

Discussion

Marital loss through either divorce or widowhood has long been documented to have negative consequences on individuals’ mental and physical health (Sasson & Umberson, 2014; Zhang & Hayward, 2006). Few studies, however, have examined the relationship between the timing of marital loss and cognitive health in later life. This study is one of the first to use nationally representative data to examine how the timing of marital loss is associated with cognitive function among older adults in the United States, with close attention to potential gender differences. We have further advanced this line of research by testing whether two key potential mechanisms—economic resources and health-related factors—explain the association between the timing of marital loss and cognitive function.

We found that both divorce and widowhood at younger ages were associated with more negative consequences for cognitive function than those events at older ages. This finding suggests that divorce and widowhood at younger ages were especially difficult for the cohorts in the HRS. Some of them may have gotten divorced before the no-fault divorce law was passed and faced strong public disapproval. From a life-course perspective, divorce and widowhood at younger ages also mean that an individual may become the major caregiver for dependent children, which can exacerbate the negative effects of divorce and widowhood. Moreover, in younger and middle adulthood the marital role is stronger and more prevalent (Gove, 1973). This may make widowhood at younger ages especially difficult, as it is very nonnormative and traumatic (Kreider & Ellis, 2011). At the same time, compared to older individuals, younger individuals may have fewer coping skills to deal with the stress from either divorce or widowhood, and thus suffer more from the adverse consequences of marital loss. In contrast, older individuals may have gained more experience and skills for surviving life’s tribulations (Mirowsky & Ross, 2001).

We further went beyond previous literature to examine whether two potential mechanisms—economic resources and health-related factors—can explain the association between the timing of marital loss and cognitive function. The findings are mixed. We found that household income partly accounted for the association between age at widowhood and cognition. This finding is consistent with the broader health literature suggesting that economic resources play a key role in accounting for the widowhood disadvantage in health (Liu et al., 2020). Notably, the association between age at divorce and cognitive function was only significant among men, not women. As divorce may have less impact on men’s income than on women’s (Leopold, 2018), it is not surprising that income was not a key factor contributing to the association between age at divorce and men’s cognitive function.

Our results further show that health-related factors play some part in explaining the relationship between the timing of marital loss and cognitive function in later life. Those who divorced at younger ages had more depressive symptoms and more chronic health conditions and were more likely to smoke than those who divorced at later ages, factors that negatively affect cognitive function. Similarly, younger age at widowhood was also associated with more chronic conditions and a higher risk of smoking. Future studies should explore other factors, in particular sociopsychological factors such as reduced social support and network connection, that may explain the association between the timing of marital loss and cognitive function.

Finally, we found a significant gender difference in the association between age at divorce (but not widowhood) and cognitive function: It is significant for men but not for women. This finding is consistent with recent studies showing that divorce has a stronger impact on men’s risk of dementia than on women’s (Zhang et al., 2021). Divorced men, especially at younger ages, may be more socially isolated and tend to have unhealthy behaviors, which can produce long-term negative consequences for their cognitive health. Our findings highlight that men who experience divorce at younger ages may be particularly vulnerable to poor cognitive health in late life.

This study has several limitations. First, although we work from a causal framework to develop our research hypotheses, our analysis is intended primarily to document general associations rather than to determine causality. Indeed, we could not fully tease out the reversal influence—that is, that cognitive function may also affect the likelihood and timing of divorce and widowhood; some of our mediating factors such as health and income may also influence age at marital loss. Second, although we found that health-related factors and household income partially explained the association between timing of marital loss and cognitive function, this relationship remained statistically significant in the full models after controlling for all covariates. Future studies should investigate additional factors such as social isolation and social support that may help explain the association between the timing of marital loss and cognitive function. Third, we did not control for divorce/widowhood duration due to data limitations. One recent study has shown that time spent unmarried after the first marriage matters for cognition (Zaheed et al., 2021). Fourth, although we included birth cohort as a control in the analysis, future studies should also explore cohort as a potential moderator in the relationship between age at marital loss and cognition given cohort differences in the experience of marital loss. Older generations may have suffered more from the negative impacts of divorce due to stigma, whereas younger generations may have greater access to resources (e.g., divorce and grief counseling) to deal with marital loss (Liu, 2012). Moreover, the association between age at marital loss and cognitive function may vary by life course contexts such as repartnering, parenthood, grandparenthood, and spousal caregiving. Future studies should consider how these life course contextual factors may modify the impact of marital loss on cognitive health. Finally, because our sample included those who had experienced divorce or widowhood and survived to older ages, mortality selection may have caused our sample to include more of those who are robust and resilient after marital loss. Thus, our results should be interpreted with caution.

Conclusion

Despite these limitations, the current study makes important contributions to the literature on marital loss and health by examining the link between the timing of marital loss and cognitive function in later life. The results, which are based on longitudinal data drawn from a nationally representative sample of U.S. older adults, suggest that the timing of marital loss is associated with cognitive health. Specifically, both divorce and widowhood at younger ages are associated with poorer cognitive function than those that happen at older ages. We also found that the association between age at divorce and cognitive function was stronger for older men than for older women, suggesting that men who experience divorce at younger ages may be particularly vulnerable to poor cognitive health. Future research should explore a fuller picture of marital loss (e.g., duration spent unmarried, the context of loss such as relationship quality and caregiving) as well as the underlying mechanisms that may contribute to poor cognitive function in late life. This knowledge will aid in the design and implementation of effective interventions that support healthy cognitive aging, in particular for those who have experienced marital loss, especially at younger ages.

Acknowledgments

The article was presented virtually at the 2021 annual meeting of the Population Association of America. The authors thank Dr. Miles Taylor, the discussant, and participants in the session “Social and Behavioral Determinants of Cognitive Aging Across the Life Course” for their helpful feedback.

Contributor Information

Zhenmei Zhang, Department of Sociology, Michigan State University, East Lansing, Michigan, USA.

Hui Liu, Department of Sociology, Michigan State University, East Lansing, Michigan, USA.

Yan Zhang, Center for Demography of Health and Aging, University of Wisconsin-Madison, Madison, Wisconsin, USA.

Funding

This work was supported in part by the National Institute on Aging (Grants R03 AG062936 and R01 AG061118); the Center for Family and Demographic Research, Bowling Green State University, which has core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD050959).

Conflict of Interest

None declared.

Author Contributions

Z. Zhang conceptualized the study, supervised the data analysis, and wrote the article with input from all authors. H. Liu provided critical feedback on statistical analyses and contributed to writing the article. Y. Zhang performed the statistical analysis and contributed to writing the article.

References

  1. Almeida, D. M., & Wong, J. D. (2009). Life transitions and daily stress processes. In G. H.Elder, Jr. & Giele J. Z. (Eds.), The craft of life course research (pp. 141–162). The Guilford Press. [Google Scholar]
  2. Amato, P. R., & Irving, S. (2006). Historical trends in divorce in the United States. In Mark A. F., & Harvey J. H. (Eds.), Handbook of divorce and relationship dissolution (pp. 41–57). Psychology Press. [Google Scholar]
  3. Bertogg, A., & Leist, A. K. (2021). Partnership and cognitive aging in Europe: Mediating factors and social stratification. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 76(6), 1173–1185. doi: 10.1093/geronb/gbab020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Brown, M. T. (2010). Early-life characteristics, psychiatric history, and cognition trajectories in later life. The Gerontologist, 50(5), 646–656. doi: 10.1093/geront/gnq049 [DOI] [PubMed] [Google Scholar]
  5. Brown, S. L., & Lin, I. F. (2012). The gray divorce revolution: Rising divorce among middle-aged and older adults, 1990–2010. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 67(6), 731–741. doi: 10.1093/geronb/gbs089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brown, S. L., Lin, I. F., Vielee, A., & Mellencamp, K. A. (2021). Midlife marital dissolution and the onset of cognitive impairment. The Gerontologist, 61(7), 1085–1094. doi: 10.1093/geront/gnaa193 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Carnelley, K. B., Wortman, C. B., & Kessler, R. C. (1999). The impact of widowhood on depression: Findings from a prospective survey. Psychological Medicine, 29(5), 1111–1123. doi: 10.1017/S0033291799008971 [DOI] [PubMed] [Google Scholar]
  8. Carr, D., & Utz, R. L. (2020). Families in later life: A decade in review. Journal of Marriage and Family, 82(1), 346–363. doi: 10.1111/jomf.12609 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chiao, C., Botticello, A., & Fuh, J. L. (2014). Life-course socio-economic disadvantage and late-life cognitive functioning in Taiwan: Results from a national cohort study. International Health, 6(4), 322–330. doi: 10.1093/inthealth/ihu046 [DOI] [PubMed] [Google Scholar]
  10. Dahl, S. A., Hansen, H. T., & Vignes, B. (2015). His, her, or their divorce? Marital dissolution and sickness absence in Norway. Journal of Marriage and Family, 77(2), 461–479. doi: 10.1111/jomf.12166 [DOI] [Google Scholar]
  11. Dupre, M. E., George, L. K., Liu, G., & Peterson, E. D. (2015). Association between divorce and risks for acute myocardial infarction. Circulation: Cardiovascular Quality and Outcomes, 8(3), 244–251. doi: 10.1161/CIRCOUTCOMES.114.001291 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dykstra, P. A., & Fokkema, T. (2007). Social and emotional loneliness among divorced and married men and women: Comparing the deficit and cognitive perspectives. Basic and Applied Social Psychology, 29(1), 1–12. doi: 10.1080/01973530701330843 [DOI] [Google Scholar]
  13. Fisher, G. G., Hassan, H., Faul, J. D., Rogers, W., & Weir, D. R. (2017). Health and retirement study imputation of cognitive functioning measures: 1992–2014. Ann Arbor, MI: Survey Research Center, University of Michigan. [Google Scholar]
  14. Fratiglioni, L., Paillard-Borg, S., & Winblad, B. (2004). An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurology, 3(6), 343–353. doi: 10.1016/S1474-4422(04)00767-7 [DOI] [PubMed] [Google Scholar]
  15. Gerstel, N. (1987). Divorce and stigma. Social Problems, 34(2), 172–186. doi: 10.2307/800714 [DOI] [Google Scholar]
  16. Glymour, M. M., Avendaño, M., Haas, S., & Berkman, L. F. (2008). Lifecourse social conditions and racial disparities in incidence of first stroke. Annals of Epidemiology, 18(12), 904–912. doi: 10.1016/j.annepidem.2008.09.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gove, W. R. (1973). Sex, marital status, and mortality. American Journal of Sociology, 79(1), 45–67. doi: 10.1086/225505 [DOI] [PubMed] [Google Scholar]
  18. Greenberg, M. S., Tanev, K., Marin, M. F., & Pitman, R. K. (2014). Stress, PTSD, and dementia. Alzheimer’s and Dementia, 10(Suppl 3), S155–S 165. doi: 10.1016/j.jalz.2014.04.008 [DOI] [PubMed] [Google Scholar]
  19. Holden, K. C., Burkhauser, R. V., & Feaster, D. J. (1988). The timing of falls into poverty after retirement and widowhood. Demography, 25(3), 405–4 14. doi: 10.2307/2061540 [DOI] [PubMed] [Google Scholar]
  20. James, B. D., & Bennett, D. A. (2019). Causes and patterns of dementia: An update in the era of redefining Alzheimer’s disease. Annual Review of Public Health, 40, 65–84. doi: 10.1146/annurev-publhealth-040218-043758 [DOI] [PubMed] [Google Scholar]
  21. Kalmijn, M., & Monden, C. (2006). Are the negative effects of divorce on well-being dependent on marital quality? Journal of Marriage and Family, 68(5), 1197–1213. doi: 10.1111/j.1741-3737.2006.00323.x [DOI] [Google Scholar]
  22. Kamp Dush, C. M. (2013). Marital and cohabitation dissolution and parental depressive symptoms in fragile families. Journal of Marriage and Family, 75(1), 91–109. doi: 10.1111/j.1741-3737.2012.01020.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Koropeckyj-Cox, T., Pienta, A. M., & Brown, T. H. (2007). Women of the 1950s and the “normative” life course: The implications of childlessness, fertility timing, and marital status for psychosocial well-being in late midlife. International Journal of Aging and Human Development, 64, 299–330. doi: 10.2190/8PTL-P745-58U1-3330 [DOI] [PubMed] [Google Scholar]
  24. Kreider, R. M., & Ellis, R. (2011). Number, timing, and duration of marriages and divorces: 2009. Current Population Reports (pp. 70–125). Washington, DC: U.S. Census Bureau. https://www2.census.gov/library/publications/2011/demo/p70-125.pdf [Google Scholar]
  25. Langa, K. M., Chernew, M. E., Kabeto, M. U., Herzog, A. R., Ofstedal, M. B., Willis, R. J., & Fendrick, A. M. (2001). National estimates of the quantity and cost of informal caregiving for the elderly with dementia. Journal of General Internal Medicine, 16(11), 770–778. doi: 10.1111/j.1525-1497.2001.10123.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Leopold, T. (2018). Gender differences in the consequences of divorce: A study of multiple outcomes. Demography, 55(3), 769–797. doi: 10.1007/s13524-018-0667-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Leopold, T., & Kalmijn, M. (2016). Is divorce more painful when couples have children? Evidence from long-term panel data on multiple domains of well-being. Demography, 53(6), 1717–1742. doi: 10.1007/s13524-016-0518-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lin, I. F., Brown, S. L., & Hammersmith, A. M. (2017). Marital biography, social security receipt, and poverty. Research on Aging, 39(1), 86–110. doi: 10.1177/0164027516656139 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lin, I. F., & Brown, S. L. (2020). Consequences of later-life divorce and widowhood for adult well-being: A call for the convalescence model. Journal of Family Theory & Review, 12, 264–277. doi: 10.1002/JFTR.12366 [DOI] [Google Scholar]
  30. Liu, H. (2012). Marital dissolution and self-rated health: Age trajectories and birth cohort variations. Social Science and Medicine, 74(7), 1107–1116. doi: 10.1016/j.socscimed.2011.11.037 [DOI] [PubMed] [Google Scholar]
  31. Liu, H., Zhang, Z., Choi, S. W., & Langa, K. M. (2020). Marital status and dementia: Evidence from the Health and Retirement Study. The Journals of Gerontology, Series B: Psychological Science and Social Sciences, 75(8), 1783–1795. doi: 10.1093/geronb/gbz087 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. McFarland, M. J., Hayward, M. D., & Brown, D. (2013). I’ve got you under my skin: Marital biography and biological risk. Journal of Marriage and Family, 75, 363–380. doi: 10.1111/jomf.12015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mirowsky, J., & Ross, C. E. (2001). Age and the effect of economic hardship on depression. Journal of Health and Social Behavior, 42, 132–150. doi: 10.2307/3090174 [DOI] [PubMed] [Google Scholar]
  34. Morley, J. E. (2017). Cognition and chronic disease. Journal of the American Medical Directors Association, 18(5), 369–371. doi: 10.1016/j.jamda.2017.02.010 [DOI] [PubMed] [Google Scholar]
  35. Ofstedal, M. B., Fisher, G. G., & Herzog, A. R. (2005). Documentation of cognitive functioning measures in the Health and Retirement Study. University of Michigan. [Google Scholar]
  36. Peters, R., Poulter, R., Warner, J., Beckett, N., Burch, L., & Bulpitt, C. (2008). Smoking, dementia and cognitive decline in the elderly, a systematic review. BMC Geriatrics, 8(1), 1–7. doi: 10.1186/1471-2318-8-36 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Pudrovska, T., & Carr, D. (2008). Psychological adjustment to divorce and widowhood in mid-and later life: Do coping strategies and personality protect against psychological distress? Advances in Life Course Research, 13, 283–317. doi: 10.1016/s1040-2608(08)00011-7 [DOI] [Google Scholar]
  38. Raley, R. K., & Sweeney, M. M. (2020). Divorce, repartnering, and stepfamilies: A decade in review. Journal of Marriage and Family, 82(1), 81–99. doi: 10.1111/jomf.12651 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Ramírez-Moreno, J. M., Vega, P. M., Espada, S., Alberca, S. B., Aguirre, J., & Peral, D. (2020). Association between self-perceived psychological stress and transitory ischaemic attack and minor stroke: A case–control study. Neurología (English Edition), 35(8), 556–562. doi: 10.1016/j.nrleng.2017.09.004 [DOI] [PubMed] [Google Scholar]
  40. Sasson, I., & Umberson, D. J. (2014). Widowhood and depression: New light on gender differences, selection, and psychological adjustment. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69(1), 135–145. doi: 10.1093/geronb/gbt058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sbarra, D. A. (2015). Divorce and health: Current trends and future directions. Psychosomatic Medicine, 77(3), 227. doi: 10.1097/PSY.0000000000000168 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Scazufca, M., Almeida, O. P., & Menezes, P. R. (2010). The role of literacy, occupation and income in dementia prevention: The São Paulo Ageing & Health Study (SPAH). International Psychogeriatrics, 22(8), 1209–1215. doi: 10.1017/S1041610210001213 [DOI] [PubMed] [Google Scholar]
  43. Sommerlad, A., Ruegger, J., Singh-Manoux, A., Lewis, G., & Livingston, G. (2018). Marriage and risk of dementia: Systematic review and meta-analysis of observational studies. Journal of Neurology, Neurosurgery and Psychiatry, 89(3), 231–238. doi: 10.1136/jnnp-2017-316274 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. StataCorp. (2017). Stata Statistical software: Release 15. StataCorp LLC. [Google Scholar]
  45. Sundström, A., Westerlund, O., & Kotyrlo, E. (2016). Marital status and risk of dementia: A nationwide population-based prospective study from Sweden. BMJ Open, 6(1), e008565. doi: 10.1136/bmjopen-2015-008565 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Umberson, D. (1992). Gender, marital status and the social control of health behavior. Social Science and Medicine, 34(8), 907–917. doi: 10.1016/0277-9536(92)90259-S [DOI] [PubMed] [Google Scholar]
  47. Vidarsdottir, H., Fang, F., Chang, M., Aspelund, T., Fall, K., Jonsdottir, M. K., & Valdimarsdottir, U. (2014). Spousal loss and cognitive function in later life: A 25-year follow-up in the AGES-Reykjavik study. American Journal of Epidemiology, 179(6), 674–683. doi: 10.1093/aje/kwt321 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Williams, K., & Umberson, D. (2004). Marital status, marital transitions, and health: A gendered life course perspective. Journal of Health and Social Behavior, 45(1), 81–98. doi: 10.1177/002214650404500106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wörn, J., Comijs, H., & Aartsen, M. (2020). Spousal loss and change in cognitive functioning: An examination of temporal patterns and gender differences. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(1), 195–206. doi: 10.1093/geronb/gby104 [DOI] [PubMed] [Google Scholar]
  50. Yaffe, K., Falvey, C., Harris, T. B., Newman, A., Satterfield, S., Koster, A., & Simonsick, E. (2013). Effect of socioeconomic disparities on incidence of dementia among biracial older adults: Prospective study. BMJ, 347, 7051. doi: 10.1136/bmj.f7051 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yan, Y., Dominguez, S., Fisher, D. W., & Dong, H. (2018). Sex differences in chronic stress responses and Alzheimer’s disease. Neurobiology of Stress, 8, 120–126. doi: 10.1016/j.ynstr.2018.03.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Zaheed, A. B., Sharifian, N., Morris, E. P., Kraal, A. Z., & Zahodne, L. B. (2021). Associations between life course marital biography and late-life memory decline. Psychology and Aging, 36(5), 557–571. doi: 10.1037/pag0000617 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Zhang, Z., & Hayward, M. D. (2006). Gender, the marital life course, and cardiovascular disease in late midlife. Journal of Marriage and Family, 68(3), 639–657. doi: 10.1111/j.1741-3737.2006.00280.x [DOI] [Google Scholar]
  54. Zhang, Z., Liu, H., & Choi, S. E. (2021). Marital loss and risk of dementia: Do race and gender matter? Social Science and Medicine, 275, 113808. doi: 10.1016/j.socscimed.2021.113808 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Zhang, Z., Liu, H., & Yu, Y.-L. (2016). Marital biography and health in middle and late life. In Bookwala J. (Ed.), Couple relationships in the middle and later years: Their nature, complexity, and role in health and illness (pp. 199–218). American Psychological Association. [Google Scholar]
  56. Zhou, S., Zhou, R., Zhong, T., Li, R., Tan, J., & Zhou, H. (2014). Association of smoking and alcohol drinking with dementia risk among elderly men in China. Current Alzheimer Research, 11(9), 899–907. doi: 10.2174/1567205011666141001123356 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES