Relationship Quality Change Among Partnered Older Adults During the COVID-19 Pandemic

Abstract

Objectives

The coronavirus disease 2019 (COVID-19) pandemic dramatically changed social life, but we know less about whether partnered relationships changed during this time. Because high-quality intimate relationships are key to many older Americans’ well-being, we explore whether, how, and for whom the pandemic changed relationship quality among partnered older Americans.

Methods

Nationally representative data from 1,642 partnered adults aged 50 and older come from the 2020 National Social Life, Health, and Aging Project COVID-19 Study. Multinomial logistic regression analyses estimate the likelihood of reporting improved, unchanged, or worsened relationship quality since the pandemic began across sociodemographic groups.

Results

Two thirds of respondents reported that relationship quality stayed the same during the pandemic, 22.8% reported their relationship quality got better, and 10.5% reported their relationship quality got worse. Although women and the oldest old were less likely to report improved relationship quality, Black respondents were more likely to do so, even after controlling for explanatory variables.

Discussion

Resilience describes many partnered older adults’ experiences during the first year of the pandemic, suggesting that they were able to protect their partnerships during this public health crisis.

The coronavirus disease 2019 (COVID-19) pandemic changed social life as people minimized in-person contact and shifted activities online. Such physical distancing measures may have been especially important for older adults, who are at higher risk of dying from COVID-19 (Centers for Disease Control and Prevention, 2021). Although some have documented how the pandemic affected older adults’ loneliness and mental well-being (NORC at the University of Chicago, 2021), fewer have examined how COVID-19 changed older adults’ partnered relationship quality and whether these changes are unequally distributed across sociodemographic groups. As high-quality intimate relationships are key resources for older adults’ physical and emotional health (Wong & Waite, 2015), it is important to understand whether, how, and for whom romantic relationship quality changed since the pandemic began.

The vulnerability–stress–adaptation model of marital quality (Karney & Bradbury, 1995) proposes that relationship quality depends on (a) how much stress a couple encounters; (b) the enduring vulnerabilities partners bring to the romantic unit; and (c) how well partners adapt to stressors and vulnerabilities. The COVID-19 pandemic represents a major stressor that, combined with vulnerability risk factors for lower relationship quality such as cohabiting (vs being married; Brown et al., 2022), living with children under 18 (Wang & Mutchler, 2020), and having poor physical and mental health (Yorgason & Choi, 2016), could worsen relationship quality. On the other hand, this external stressor may have created opportunities for partners, especially those with greater access to resources that boost relationship well-being, such as socioeconomic resources (Liu & Upenieks, 2021) and social support from friends and family (Zhaoyang & Martire, 2021), to adapt together, thereby improving relationship quality.

Relationship quality changes since the pandemic began may differ across sociodemographic groups due to an unequal distribution of vulnerability and access to resources for adaptation. We expect socially disadvantaged older adults (minoritized racial/ethnic groups, lower socioeconomic groups, women, and the oldest old) to be more vulnerable to the negative effects, and less likely to experience the potential positive effects, of the pandemic on relationship quality because they face more stressors and may have fewer financial and/or psychosocial resources for coping.

For example, relative to White Americans, Black and Hispanic Americans may report worsened relationship quality since the pandemic started because COVID-19 illnesses and deaths (Garcia et al., 2021), and the associated psychological distress (Bui et al., 2021), were higher in these groups. Although racial/ethnic minorities in the general adult population may draw on psychosocial resources like extended network ties to protect relationship quality (Bermudez & Mancini, 2013; Bryant et al., 2010), the pandemic may have reduced access to these adaptive supports. Similarly, lower socioeconomic status individuals may face greater income instability and disproportionately suffer the loss of public resources to facilitate daily activities (Morrow-Howell et al., 2020).

Moreover, COVID-19 may have created additional burdens for women compared to men, and the oldest old compared to younger groups, which may worsen relationship quality. For example, women perform more emotionally taxing care work than men (Umberson et al., 2018) and may have taken on more caregiving responsibilities, including increasing grandchild care as schools and daycares shuttered, during the pandemic (Truskinovsky et al., 2022). Women’s relationship quality also suffers when they have less access to friend and family support (Zhaoyang & Martire, 2021), and social distancing may have limited these psychosocial resources. The oldest Americans have the greatest physical and psychological health burdens (Jaul & Barron, 2017) and may have increased difficulty accessing health care during the pandemic (Morrow-Howell et al., 2020). These additional burdens along with reduced resources may create greater challenges for maintaining relationship quality among socially disadvantaged groups during the pandemic. We test the following:

Competing Hypothesis 1a: A greater proportion of older adults will report that their relationship quality worsened (vs. stayed the same or improved) since the pandemic started

Competing Hypothesis 1b: A greater proportion of older adults will report that their relationship quality improved (vs. stayed the same or worsened) since the pandemic started

Hypothesis 2: Socially disadvantaged groups are more likely than advantaged groups to report worsened relationship quality and/or less likely to report improved relationship quality since the pandemic started

Hypothesis 3: Sociodemographic differences in relationship quality change may be partially explained by vulnerability risk factors and financial and psychosocial resources

Method

Data come from the National Social Life, Health, and Aging Project (NSHAP), a nationally representative longitudinal survey of community-dwelling older Americans that began in 2005–2006 (Round 1, N = 3,005; Waite et al., 2014). NSHAP added a sample of the original respondents’ spouses and cohabiting partners in 2010–2011 (Round 2, N = 3,377; Waite et al., 2019b) and recruited a cohort of Baby Boomers and their coresidential partners to participate in 2015–2016 (Round 3, N = 4,777; Waite et al., 2019a). Details about recruitment and sample characteristics are available elsewhere (O’Muircheartaigh et al., 2009, 2014, 2021). Data for this report come from the NSHAP COVID-19 Study, which was fielded between September 2020 and January 2021 using web, phone, and paper surveys to understand older adults’ experiences during the first 6 to 9 months of the pandemic. A total of 2,672 Round 3 respondents who were eligible for the COVID-19 Study completed it, reflecting a response rate of 58%. We restrict our analysis to 1,642 COVID-19 Study respondents aged 50 and older who were married or living with a partner and had complete data on key variables.

To measure change in relationship quality since the pandemic began, respondents were asked, “Compared to before the start of the COVID-19 pandemic, would you say your relationship with your partner has gotten a lot better, gotten a little better, stayed the same, gotten a little worse, or gotten a lot worse?” We combined a lot better with a little better and a lot worse with a little worse into a three-category outcome variable with stayed the same as the reference.

Demographic covariates include: race/ethnicity (non-Hispanic White [reference], Hispanic/non-Hispanic Black, non-Black Hispanic, and Other); educational attainment (less than high school [reference], high school, vocational/some college, and bachelor’s or more); gender (man [reference] or woman); and age (50–59 [reference], 60–69, 70–79, and 80+). All covariates are measured in Round 3 except for age, which is measured in the COVID-19 Study.

We examine four measures of vulnerability from the COVID-19 Study: (a) marital status (married [reference] or unmarried, cohabiting); (b) whether any children under 18 live in the household; (c) self-rated physical, and (d) mental health (poor/fair [reference] to excellent). We analyze three measures of access to financial and psychosocial resources: (a) whether respondents experienced changes in income and ability to meet basic expenses and pay bills (I’m better off, I’m about the same as before [reference], and I’m worse off); and whether respondents received (b) tangible help, and (c) social support since the pandemic began (less often, about the same as before [reference], and more often).

We first report the proportion of older adults for whom relationship quality stayed the same, got better, or got worse during the pandemic compared to before it. We then estimate multinomial logistic regression models to predict changes in relationship quality. Model 1 includes all demographic variables. Model 2 adds vulnerability and resource measures. We formally test for mediation using the Karlson, Holm, and Breen (KHB) method (Kohler et al., 2011). Robustness checks are discussed in the online Supplementary Analyses and are available upon request. All analyses are performed in Stata 15 using survey weights to produce nationally representative estimates.

Results

Table 1 shows that 66.8% of partnered older adults report that relationship quality stayed the same since the pandemic started. Another 22.8% report that relationship quality improved, and only 10.5% say relationship quality worsened. These results partially support Hypothesis 1b: a greater proportion of older adults report that their relationship quality improved since the pandemic started.

Table 1.

Weighted Descriptive Statistics

	%		%
Change in relationship quality		Any children under 18 at home
Got worse	10.5	Yes	12.4
Stayed the same	66.8	No	87.6
Got better	22.8	Self-rated physical health
Race/ethnicity		Poor/fair	16.1
White	81.4	Good	35.7
Black	7.9	Very good	37.8
Hispanic, non-Black	6.1	Excellent	10.4
Other	4.6	Self-rated mental health
Education		Poor/fair	11.1
<High school	5.6	Good	32.9
High school	19.0	Very good	38.5
Vocational/some college	35.6	Excellent	17.4
Bachelor’s or more	39.9	Change in income/ability to meet expenses
Gender		Worse off	18.8
Men	52.0	Same as before	76.4
Women	48.1	Better off	4.9
Age group		Change in receiving tangible help
Mean: 66.0; Range: 50–97		Less often	8.8
<60	26.4	Same as before	84.7
60–69	41.5	More often	6.5
70–79	23.8	Changes in receiving support
80+	8.3	Less often	8.5
Marital status		Same as before	80.8
Married	94.3	More often	10.7
Cohabiting, unmarried	5.7

Note: N = 1,642.

Multinomial logistic regression results (relative risk ratios [RRR]) are reported in Table 2. Model 1 with sociodemographic predictors shows that, consistent with Hypothesis 2, women are 20% less likely than men (RRR = 0.8, 95% confidence interval [CI] = [0.6, 0.9]), and respondents aged 80 and older are 60% less likely than respondents aged 50–59 (RRR = 0.4, 95% CI = [0.2, 0.6]), to report improved rather than stable relationship quality since the pandemic started. Furthermore, those with a bachelor’s degree or more are twice as likely as those with less than high school (RRR = 2.2, 95% CI = [1.0–7.2]) to report improved relationship quality since the pandemic began. Contrary to Hypothesis 2, Black respondents are more likely than White respondents to report improved relationship quality (vs stayed the same) since the pandemic started (RRR = 2.2, 95% CI = [1.2, 4.2]).

Table 2.

Relative Risk Ratios From Multinomial Logistic Regression Predicting Change in Relationship Quality

Relative to stayed the same…	Model 1		Model 2
	Got better	Got worse	Got better	Got worse
Race/ethnicity (ref. White)
Black	2.2*	1.3	2.1*	1.0
Hispanic, non-Black	1.5	1.1	1.4	0.6
Other	0.7	0.3	0.6	0.1*
Education (ref. <high school)
High school	0.9	1.1	0.9	1.5
Vocational/some college	1.6	1.6	1.5	2.3
Bachelor’s or more	2.2*	1.7	2.0	2.7*
Gender (ref. men)
Women	0.8*	0.8	0.7*	0.7
Age group (ref. 50–59)
60–69	0.9	0.9	1.0	1.0
70–79	0.7	0.7	0.8	0.9
80+	0.4***	0.5	0.4**	0.5
Marital status (ref. cohabiting, unmarried)
Married			1.3	2.6
Children in household (ref. no children)
Yes children			0.9	1.1
Self-rated physical health (ref. poor/fair)
Good			1.1	1.5
Very good			1.0	1.6
Excellent			0.8	1.4
Self-rated mental health (ref. poor/fair)
Good			1.3	0.3***
Very good			1.6	0.2***
Excellent			1.4	0.1***
Change in income (ref. about the same)
I’m worse off			1.3	2.9***
I’m better off			3.7***	1.0
Change in receiving help (ref. about the same)
Less often			1.5	3.4**
More often			1.0	2.0
Change in receiving support (ref. about the same)
Less often			1.3	1.1
More often			2.7***	3.1**

Notes: N = 1,642.

*p < .5, **p < .01, ***p < .001.

Model 2 shows that women (RRR = 0.7, 95% CI = [0.5, 0.9]) and those 80 and older (RRR = 0.4, 95% CI = [0.2, 0.7]) remain less likely, whereas Black respondents remain more likely (RRR = 2.1, 95% CI = [1.1, 4.1]), to report improved relationship quality after controlling for enduring vulnerabilities and financial and psychosocial resources. There is no statistically significant association between educational attainment and reporting improved relationship quality in this model, however. KHB mediation analyses (results available upon request) confirm, in contrast to Hypothesis 3, that these vulnerability and resource variables are not significant mediators. Controlling for these factors reveals additional statistically significant associations between relationship quality change and sociodemographic characteristics that contradict Hypotheses 2 and 3. Compared to White respondents, “Other” racial/ethnic respondents are 90% less likely to report worsened relationship quality (RRR = 0.1, 95% CI = [0.1, 0.7]). Those with a college degree or more have nearly three times the risk of reporting worsened relationship quality during the pandemic compared to those with less than a high school degree (RRR = 2.7, 95% CI = [1.0, 7.2]).

Although marital status, having children at home, and physical health status are unrelated to relationship quality change, reporting better mental health is associated with a lower relative risk of reporting worsened relationship quality. As expected, those who became financially better off are more likely to report improved relationship quality, whereas those who became worse off are more likely to report worsened relationship quality. Those receiving less tangible help since the pandemic began are more likely to report worsened relationship quality. Interestingly, those receiving more social support since the pandemic started are more likely to report improved and worsened relationship quality (vs. stayed the same).

Discussion

The COVID-19 pandemic is a major stressor that could change relationship quality for older couples and affect individuals’ mental and physical health down the line. Our analysis of nationally representative data suggests that relationship quality stayed the same for 66.8% of older partnered Americans since the pandemic began. In contrast to Hypothesis 1a and in support of Hypothesis 1b, however, more reported improved (22.8%) than worsened (10.5%) relationship quality. It is heartening to note that most partnered older adults experienced stability, and only a minority experienced a decline, in relationship quality during this time. For some, the pandemic improved relationship quality, perhaps due to increased time together or greater opportunities to show mutual support. Because the NSHAP COVID-19 Study did not ask whether partners increased their time in shared activities or increased their provision of social support to one another since the pandemic began, we call on future research to test these hypotheses. Future research could also examine relationship quality change using a couple-level analysis to understand whether one partner’s report of change affects the other partner’s report, and whether this association varies by gender in different-gender couples.

Our results provide partial support for Hypothesis 2. Women and the oldest old were less likely to report improving relationship quality than men and those aged 50–59, respectively. Yet, inconsistent with Hypothesis 3, these gender and age disparities did not shrink when controlling for vulnerability risk factors and financial and psychosocial resources. Research on younger women suggests they increased their risk management behaviors to support their loved ones’ health during the pandemic (Umamaheswar & Tan, 2020), and this worry work may prevent women from experiencing relationship quality improvement. Future studies could examine whether older women also engaged in more COVID-19 protective behaviors than men, and whether this can explain differences in relationship quality change during the pandemic. The oldest age group may have been less likely to report improved relationship quality because relationship quality is more likely to decline with age (Umberson et al., 2005).

Counter to predictions, Black respondents were more likely than White respondents to report improved relationship quality. Black Americans report lower relationship quality than White Americans in the general adult population (Bulanda & Brown, 2007), so there could be more room for relationship quality to improve among Black respondents during this time. Also, it may be more necessary for Black partners living in a racist society to develop resistance strategies to protect their relationships (Bryant et al., 2010), boosting their ability to improve relationship quality despite pandemic stressors. Because the NSHAP COVID-19 data do not permit a study of these mechanisms, we call on others to explore these questions in future work.

We found two surprising results after controlling for vulnerabilities and resources: (a) “other” racial/ethnic respondents were less likely than White respondents to report worsened relationship quality since the pandemic began and (b) those with a bachelor’s degree or more were more likely to report worsened relationship quality relative to the least educated. Because our measures of vulnerability and resources did not explain such demographic differences, future research should seek to understand why “other” racial/ethnic respondents had more stable relationship quality and why the college educated reported worsened relationship quality.

Several other findings warrant continued research. Relationship quality change was not statistically related to marital status. This is consistent with other research suggesting that cohabitation is similar to marriage among older adults (King & Scott, 2005). However, because our sample of cohabitors is small (n = 85), future studies with larger samples of cohabitors could conduct stratified analyses to confirm our results. Having children under 18 at home was also unrelated to relationship quality change. The respondents’ adult children (the parents of the younger children) may be living in the household, so respondents may not experience the stress of caregiving that could harm relationship quality. Because the NSHAP COVID-19 Study does not allow us to identify whether adult children live in the household, additional research is needed to confirm this hypothesis. Receiving more social support since the pandemic began was associated with improved and worsened relationship quality. Perhaps some received more social support when the pandemic began, which bolstered relationship quality against stressors, while others received more social support because the pandemic harmed partners’ relationship quality.

Resilience best describes many partnered older adults’ experiences during the first year of the pandemic. Even as older Americans were perhaps the most vulnerable to pandemic-related health risks and changes to social life, partnered older adults may have strong emotional resources to protect their relationships (Birditt & Fingerman, 2005; Carstensen et al., 1995). Although this finding bodes well for partnered older Americans’ well-being beyond the pandemic, future research should seek to understand more about those who reported worsened relationship quality during this time. Additional research can point to how to maintain and improve these relationships so all older adults can enjoy their partnerships even under challenging circumstances.

Funding

The National Social Life, Health, and Aging Project are supported by the National Institute on Aging and the National Institutes of Health (AG043538-08S1, R01AG043538, and R01AG048511). 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 declared.