Support Exchanges Among Very Old Parents and Their Children: Findings From the Boston Aging Together Study

Abstract

Background and Objectives

Very old parents and their “old” children are a growing group in industrialized countries worldwide. Care needs of very old parents can be substantial, while children may also face their own age-related issues. Continued mutual support represents an important pathway to addressing emerging care needs. This study aimed to identify patterns of support exchanges occurring in very old parent–child dyads and to ascertain associated individual and relationship factors.

Research Design and Methods

Participants were 114 very old parents (aged ≥90) and their children (aged ≥65) from the Boston Aging Together Study. Data were collected using comprehensive, semistructured in-person interviews with both dyad members, including standardized assessments of support exchanges, relationship quality, health, and perceptions of family norms. Actor–Partner Interdependence Models were used to predict upward and downward support reported by children and parents.

Results

Both dyad members not only reported substantial upward support (child to parent) in all domains but also notable amounts of downward (parent to child) in the domains of emotional support, listening, and socializing. Findings showed significant associations of parent functional impairment, parent and child relationship quality, and child perceptions of family obligation with upward support and of relationship quality with downward support.

Discussion and Implications

Continued support exchanges among very old parents and their children indicated that intergenerational theories still hold up in very late-life relationships. Health care professionals should be aware that attention to relationship quality and family norms might be vital to ensure that support needs are met.

Increases in life expectancy give rise to a new phenomenon that has received little attention to date—family members reaching very old age together (Christensen et al., 2009). The U.S. Census 2010 documents almost 2 million very old individuals aged 90 years and older (U.S. Census Bureau 2010). Two thirds of them have children of advanced age, who are likely to become the main source of support, because most very old individuals have outlived spouses and friends (Jopp, Boerner, Ribeiro et al., 2016). The prevalence of very old parents with an advanced-age child is expected to increase substantially, but virtually nothing is known about the nature and implications of this relationship constellation.

Although large population-based studies include social network indicators, these investigations lacked attention to the issues specific to this age group. Furthermore, most dyadic studies on later-life support exchanges involve spouses or middle-aged children (Carr & Utz, 2020), generating findings that may not apply to parent–child dyads involving very old parents. The fourth age represents a time of “testing the limits” in many life domains (e.g., plasticity of cognitive capacity; Baltes, 1997), and balanced support exchanges may be one of them. For the children of very old adults, continued responsibility for a parent in one’s own late life could be considered an off-time life event (Neugarten, 1976). Preliminary evidence suggests that this scenario can be perceived as challenging by both sides. Han et al. (2004) reported that very old Koreans were ashamed of being unable to die, while their children felt resentment for being denied seniority status. A Portuguese study found elevated anxiety levels among children of centenarians and that their own life plans needed to be changed due to caregiving responsibilities (Brandão et al., 2017). Similarly, narrative accounts of perceived challenges and rewards from the Boston Aging Together Study revealed that the balance of challenges and rewards was notably less favorable for children compared to parents, with references to children’s own advanced age and health problems, and the prolonged caregiving involvement due to their parents’ longevity (Boerner et al., 2021). This article builds on these insights by investigating the support exchanges that take place in the context of the challenges associated with the parents’ fourth age affecting or overshadowing the third age of their children.

Support Exchanges Between Generations in Late and Very Late Life

Research on mid- to late-life parent–child relationships shows that intergenerational support exchanges constitute essential aspects of both parents’ and children’s lives (Fingerman et al., 2020). Support between generations generally flows downstream in families, with parents providing more support to offspring than the reverse until parents develop health problems and the balance starts to shift toward equilibrium or greater support given to parents (in particular, instrumental support; Lin, 2008). As virtually all very old persons suffer from multiple, chronic health problems and many have substantial care needs (Jopp, Boerner, & Rott, 2016), this shift is likely to become more extreme in very old parent–child dyads, necessitating intensified upward support and resulting in a significant burden for children. Compared to spouse caregivers, adult children who cared for oldest-old parents tended to express higher levels of stress from caregiving (Takagi et al., 2013).

While downward support has been observed in older parents with adult children (Mandemakers & Dykstra, 2008), the extent to which it remains possible in very old parent–old child dyads is not known (Fingerman et al., 2013). What types of downward support can continue despite increasing care needs of very old parents is also unknown, even though intangible types of support are still likely to occur.

Support exchanges have also been shown to depend on relationship quality (Stuifbergen et al., 2008). Evidence suggests that when parents advance into an age associated with increasing dependency, negativity in relationships can no longer be avoided (Hogerbrugge & Silverstein, 2015). This is in line with the observation that relationship reports tend to become worse when parent health issues emerge (Kim et al., 2017). Ambivalence or negativity can also become more prevalent because of conflicting expectations; advanced-aged children may expect their later life to be a time offering regained freedom and leisure, while very old parents expect support and caregiving from the children they raised (Boerner et al., 2021).

Conceptual Guidance: Interdependence, Contingency, and Solidarity

Interdependence Theory (Kelley & Thibaut, 1978) posits that the study of interactions between two people should consider individual and dyadic characteristics. We accordingly examine how both dyad members’ perspective on the relationship and support exchanges can uniquely contribute to support exchanges in the dyad.

Contingency Theory (Eggebeen & Davey, 1998), which has primarily guided the research with younger dyads, posits that intergenerational support exchanges are a function of support recipients’ needs and support providers’ resources. Compared to younger dyads, very old parent–child dyads are likely to be characterized by intensified care needs and fewer resources. Children of very old adults are more likely to face their own health problems and reduced resource capacity. They are also often the only or one of few remaining support sources for their parent. Moreover, the main burden is often placed on one focal child, rather than being equally distributed among several children (Lin & Wolf, 2020). Thus, while support could flow both ways depending on needs and resources, upward support levels are likely to be higher compared to downward support; needs and resources of both parents and children could play a role, but parent characteristics would likely be the stronger determinant.

Intergenerational Solidarity Theory (Bengtson & Roberts, 1991) views support exchanges among parents and children during adulthood as a function of solidarity. Relationship quality and individual perceptions of societal norms or family obligation toward one another constitute important drivers of solidarity (Gans & Silverstein, 2006). Support exchanges could thus vary based on the quality of their relationship and dyad members’ sense of obligation for looking after one another. Generally, better relationship quality and a greater sense of family obligation should lead to more support exchanges. For very old parent–child dyads whose relationship rests on life-long experiences, these features may be particularly relevant for children’s levels of upward support.

Other Factors Associated With Support Exchanges

Daughters generally provide more support to their aging parents than sons (Pillemer & Suitor, 2014), and older mothers tend to report closer relationships with their children than older fathers (Pillemer et al., 2012). Indicators of socioeconomic advantage or disadvantage (e.g., education, income, and racial/ethnic minority) can be conceptually grouped as resources or needs. Parent–child coresidence, marital status, and help from people outside of the dyad are also likely to be associated with support exchanges (Fingerman et al., 2020). Coresidence facilitates access to one another, being married could mean additional support or burden, and the presence of other helpers could reduce or indicate high support needs.

Study Aims and Hypotheses

The primary aim of this study was to identify patterns of support exchanges occurring in very old parent–child dyads, as well as to ascertain individual and dyadic factors associated with these patterns. For patterns of support, we expect that, following Contingency Theory, upward support will be more prominent than downward support (i.e., with children providing more support to parents), but that some downward support (e.g., intangible support) may be present despite the parents’ fourth age.

For individual and dyadic factors associated with support exchanges, we expect that, as per Contingency Theory, support exchanges will be associated with both dyad members’ needs and resources, but that, considering the assumed greater support need of very old parents, parental health problems will be of primary importance. As per Solidarity Theory, we anticipate that support exchanges will also be associated with dyad members’ relationship quality and norms of family obligation. However, given available literature, we predict these associations to particularly apply to upward support. The notion of downward support and correlates in case of very old parents is less well understood and thus needs to be explored.

As per Interdependence Theory, we acknowledge that parent and child perspectives on their relationship could be interdependent, affecting these support patterns. However, we generally expect that own characteristics would be more likely to be associated with own support reports. For relationship quality, we expect that both parent and child reports could be associated with their own and the other dyad member’s support reports, as each dyad member’s relationship perception can reflect the overall emotional climate of the relationship.

Method

Data Collection and Procedures

The Boston Aging Together Study is a dyadic, mixed-method study on very old parents and their children. This article focuses on support exchange data from this larger study as the outcome and examines sociodemographic, health, relationship quality, and perception of family norm variables as potential correlates of support exchanges.

Participants were 114 very old parent–child dyads. We recruited study participants through announcements via various media outlets (e.g., senior magazines and town newspapers) and collaboration with local service providers (e.g., homecare agencies and senior programs). Recruitment focused on community-dwelling dyads, to ensure that both dyad members had the cognitive capacity to provide reliable information about their relationship. Furthermore, the eligibility age for parents was set at an age of at least 90 so that we could recruit from a sufficiently large pool of dyads with children aged at least 65, an age widely agreed to demarcate older adulthood. Additional eligibility criteria were Mini-Mental State Examination (shortened version) score of at least 12 (out of 21 maximum), both dyad members agree to participate, and at least three interactions between dyad members per week. To select children, we identified the child most involved in supporting the parent. Initial contact and screening took place via telephone and email. In 25 cases, we accepted participants with a lower age (lowest age = 62 for children and 85 for parents) because they had specific characteristics that we felt were important to represent (e.g., racial/ethnic minority background).

In-person interviews were conducted at participants’ residences (parent and child interviews separately), lasting about 2 h. Parent interviews were typically split into two sessions to reduce the burden of the interview. Interviewers were doctoral students in gerontology, trained extensively in applying the study protocol. Participants received U.S. $40 in acknowledgment of their time.

Measures

Support exchanges

We assessed support exchanges by asking participants how often they provide support to their parent/child, and how often their parent/child provides support to them in six domains: emotional, practical, socializing, advice, financial, and talking about daily events (total of 12 items; Fingerman et al., 2009). Answering categories were on an 8-point scale, ranging from 1 (less than once a year/never) to 8 (daily). Mean scores of upward and downward support were computed for children and parents separately, with higher scores reflecting more frequent support exchanges (α = .62 for child upward, α = .66 for child downward, α = .70 for parent upward, and α = .51 for parent downward).

Main predictors

Cognitive functioning was assessed with a shortened version of the Mini-Mental State Examination (MMSE; Folstein et al., 1975), focusing on items unlikely to be biased by poor sensory functioning prevalent in very old adults (Holtsberg et al., 1995). Subscales included orientation (0–10), registration (0–3), attention (0–5), and recall (0–3). Due to our at least 12 eligibility criterion, the possible score range for participants was 12–21.

Functional impairment was assessed with the Older Americans Resources and Services Multidimensional Functional Assessment Questionnaire (Fillenbaum, 1988), which included physical (activities of daily living [ADL]; seven items) and instrumental ADL (IADL; seven items). Answering options ranged from 1 (no difficulty) to 4 (can’t do without help). Mean scores of ADL/IADL items were computed, with higher scores reflecting more functional impairment (α = .76 for children and α = .88 for parents).

Relationship quality was measured using an abbreviated 11-item version of the Quality Relationship Index (Pierce et al., 1997), composed of three subscales: support (three items; e.g., How much can you rely on your [parent/child] for emotional support), conflict (four items; e.g., How much do you argue with [parent/child]; reverse-coded), and centrality (four items; e.g., How much do you depend on [parent/child]?). Responses were rated from 0 (not at all) to 4 (very much/a lot). Mean scores of 11 items were calculated, with higher scores indicating better relationship quality (α = .83 for children and α = .65 for parents).

Family obligation was assessed with a six-item scale, reflecting views about younger family members’ responsibilities to their older adults (e.g., looking after their older adults, assisting them financially, and making them happy; Gallois et al., 1999). Responses were rated from 1 (strongly disagree) to 5 (strongly agree). Mean scores of six items were calculated, with higher scores reflecting a greater sense of obligation (α = .66 for children and α = .57 for parents).

Person and dyadic characteristics

Person characteristics included individuals’ age, gender, marital status, education, work status, perceived income adequacy, number of children, and chronic health conditions. Education was rated from 0 (never attended/kindergarten only) to 21 (doctoral degree). Income adequacy was assessed with one item asking whether the income would suffice given needs, with response options ranging from 0 (can’t make ends meet) to 3 (money is not a problem). Chronic health conditions were assessed with an 18-item illness checklist (Lawton et al., 1982). While all person characteristics served to describe the sample, only gender, marital status, education, and perceived income adequacy were included as covariates in multivariate models.

As dyadic characteristics, we considered gender composition of the dyad (for sample description), and, as covariates, race/ethnicity (1 = racial/ethnic minority, 0 = non-Hispanic white), coresidence status (1 = coresiding or living within 1 mile, 0 = not coresiding), and other help. For other help, children were asked: “When your parent needs help with daily things, who usually helps you (by helping you and/or your parent)?” Parents were asked: “When you need help with daily things, who (other than your child) usually helps you?” We created two binary indicators (1 = yes, 0 = no): (a) having informal helpers and (b) having formal helpers. Because nearly all reported some informal help, we decided not to include this variable in multivariate analyses.

Analytic Strategy

First, we examined means and standard deviations of upward (child to parent) and downward (parent to child) support reported by parent and child. We examined these for the six support types individually and for all mean scores (i.e., child reports upward and downward, parent reports upward and downward). Second, we evaluated bivariate associations among all study variables in preparation for multivariate analyses (Supplementary Table 1).

To predict upward and downward exchanges of support reported by children and parents, we used Actor–Partner Interdependence Models (APIMs; Cook & Kenny, 2005), which handle interdependence between dyad members’ reports on support exchanges. We examined children’s and parents’ functional and cognitive health, relationship quality, and family obligation as main predictors, accounting for individual (e.g., gender, education) and dyadic covariates (i.e., minority status and coresidence). All models were estimated using Mplus 7.4 (Muthén & Muthén, 1998–2015). The analytic sample for all analyses beyond sample description was reduced to 112 dyads due to two dyads with incomplete support exchange data. All predictor and control variables were complete except for income inadequacy (no response from seven parents). These missing data were handled using full information maximum likelihood procedures.

Results

Table 1 presents a descriptive summary of participant characteristics. As is typical for older populations, the majority of parents (M_age = 93.31, SD = 3.10) and children (M_age = 67.67, SD = 3.04) were female and mother–daughter dyads. However, about one third of the sample were mother–son (12%), father–daughter (6%), and father–son (10%) dyads. Dyads with minority background included 11 dyads identifying as African American and one as Hispanic, resulting in a total minority representation of 11%. Parents indicated on average 5.40 (SD = 2.12) chronic health conditions, children on average 2.39 (SD = 1.67). Whereas intraclass correlations (ICCs) of our two main health predictors (functional and cognitive health) were nonsignificant, the other two main predictors, relationship quality and family obligation, showed significant ICCs (.28 and .19, respectively), indicating some similarity between dyad members.

Table 1.

Descriptive Summary of Sample Characteristics

Variable	Child		Parent
	M (SD)	Range	M (SD)	Range
Person characteristics
Age	67.67 (3.04)	62–76	93.31 (3.10)	85–101
Female, %	78		84
Educationa	17.63 (2.20)	9–21	14.91 (3.14)	4–21
Income adequacyb	2.35 (0.82)	0–3	2.28 (0.77)	0–3
Marital status, %
Married	54		10
Never married	17		0
Divorced	19		6
Separated	3		3
Widowed	7		84
Working, %	42		5
Number of children	1.59 (1.46)	0–10	3.40 (1.92)	1–14
Chronic conditionsc	2.39 (1.67)	0–7	5.40 (2.12)	0–11
Dyadic characteristics
Gender composition, %
Mother–daughter	72		72
Mother–son	12		12
Father–daughter	6		6
Father–son	10		10
Racial/ethnic minority, %	11		11
Coresiding or living within 1 mile, %	32		32
Having formal helpers, %	75		75
Having informal helpers, %	98		98
Functional and cognitive health
Cognitive functioningf	20.81 (0.48)	18–21	18.82 (2.35)	12–21
Functional impairmentg	1.02 (0.08)	1.00–1.64	1.52 (0.67)	1.00–3.83
Relationship quality
Total relationship qualityh	2.62 (0.70)	1.09–3.80	3.28 (0.47)	1.55–4.00
Family norms
Family obligationi	4.19 (0.59)	2.17–5.00	3.86 (0.65)	2.00–5.00

Note: Dyad N = 114.

^aRated from 0 (never attended/kindergarten only) to 21 (doctoral degree).

^bRated from 0 (can’t make ends meet) to 3 (money is not a problem).

^cSum of 18 chronic conditions.

^dHaving paid help from professional helpers.

^eHaving unpaid help from friends or family members.

^fSum of 21 items from the modified Mini-Mental State Examination.

^gMean of 14 items rated from 1 (no difficulty) to 4 (can’t do without help).

^hMean of 11 items rated from 0 (not at all) to 4 (very much/a lot).

ⁱMean of six items rated from 1 (strongly disagree) to 5 (strongly agree).

Table 2 presents descriptive findings for the six types of support and their total mean, including child and parent reports on upward and downward exchanges. As expected, levels of upward support were generally higher compared to downward support (also see Supplementary Table 2 for comparisons between upward and downward support). However, mean levels also indicate that the average difference between upward and downward support, albeit significant, was not as large as one could have expected (ranging on average from 1.21 to 1.23). Upward and downward support were also significantly different for each of the six individual support types, but as expected, intangible types of support (e.g., listening) showed relatively high levels of downward support. Financial support exchanges generally had the lowest means and did not show drastic mean level differences between upward and downward support. Practical support showed the clearest pattern of higher upward support levels, as could be anticipated.

Table 2.

Upward and Downward Support Exchanges Reported by Parents and Children

Variable	Child reports	Parent reports	Paired t
	M (SD)	M (SD)
Upward support a
Total support (mean of six items)	5.58 (0.97)	5.02 (1.33)	5.30***
Emotional	6.30 (1.67)	5.21 (2.53)	4.53***
Listening	7.29 (0.89)	6.45 (2.02)	4.67***
Advice	5.74 (1.54)	5.13 (2.26)	2.74**
Socializing	5.41 (1.59)	5.44 (1.78)	−0.16
Practical	6.13 (1.75)	5.75 (1.90)	2.38*
Financial	2.59 (2.20)	2.17 (1.96)	1.87
Downward support b
Total support (mean of six items)	4.38 (1.16)	3.79 (1.24)	5.21***
Emotional	4.43 (2.27)	3.69 (2.40)	2.87**
Listening	6.56 (1.55)	6.00 (2.12)	2.41*
Advice	5.28 (2.00)	3.52 (2.54)	6.64***
Socializing	5.41 (1.59)	5.44 (1.78)	−0.16
Practical	2.69 (2.37)	2.20 (2.15)	2.15*
Financial	1.86 (1.48)	1.85 (1.62)	0.06

Note: Dyad n = 112 (two dyads with incomplete support exchange data were excluded).

^aHow often children provide support to parents in each domain, rated from 1 (less than once a year/never) to 8 (daily).

^bHow often parents provide support to children in each domain, rated from 1 (less than once a year/never) to 8 (daily).

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

Comparing parent and child reports, mean levels of child reports were mostly higher compared to parent reports. However, this reporting discrepancy did not appear large (i.e., total child mean of 5.58 vs. total parent mean of 5.02 for upward support; total child mean of 4.38 vs. total parent mean of 3.79 for downward support). However, it is noteworthy that practical downward support at least monthly was reported by 29% of children and 18% of parents. ICCs of parent and child reports (i.e., similarity between dyad members) were significant and substantial for the mean upward and downward support variables (.56 and .51, respectively; not shown) and showed a wider range for the six support types individually, with the lowest ICC for downward listening (.12) and the highest for upward practical support (.55).

Factors Associated With Upward Support

Using the APIMs to handle the interdependence of child’s and parent’s reports, we examined factors associated with upward support given to parents. Table 3 presents comprehensive models for upward and downward support. Our preliminary stepwise models leading up to the final models can be found in Supplementary Tables 3 and 4.

Table 3.

Models for Support Exchanges Among Very Old Parents and Their Children

Variable	Upward supporta		Downward supportb
	Child reports	Parent reports	Child reports	Parent reports
	B (SE)	B (SE)	B (SE)	B (SE)
Functional/cognitive health
Child: Cognitive functioningc	−0.11 (0.20)	−0.19 (0.29)	−0.26 (0.24)	−0.50† (0.29)
Parent: Cognitive functioningc	0.04 (0.04)	−0.06 (0.05)	0.01 (0.04)	0.03 (0.05)
Child: Functional impairmentd	0.14 (1.06)	1.23 (1.54)	1.05 (1.24)	1.38 (1.51)
Parent: Functional impairmentd	0.24† (0.14)	0.37† (0.20)	0.09 (0.16)	0.15 (0.20)
Relationship quality e
Child: Relationship quality	−0.23† (0.13)	0.17 (0.19)	0.62*** (0.15)	0.37* (0.19)
Parent: Relationship quality	0.46* (0.19)	1.05*** (0.28)	0.14 (0.23)	0.32 (0.28)
Family obligation f
Child: Family obligation	0.36* (0.15)	0.05 (0.22)	0.14 (0.17)	0.05 (0.21)
Parent: Family obligation	0.04 (0.13)	0.02 (0.19)	0.03 (0.15)	0.19 (0.19)
Child characteristics
Female	0.58** (0.21)	0.30 (0.31)	0.17 (0.25)	0.39 (0.30)
Educationg	0.05 (0.04)	0.08 (0.06)	0.10* (0.05)	0.07 (0.06)
Married	−0.28 (0.18)	−0.18 (0.26)	−0.16 (0.21)	−0.19 (0.25)
Income adequacyh	−0.01 (0.11)	−0.10 (0.16)	−0.23† (0.13)	−0.27† (0.16)
Parent characteristics
Female	0.01 (0.24)	−0.05 (0.35)	0.35 (0.29)	−0.33 (0.35)
Educationg	0.01 (0.03)	−0.01 (0.04)	−0.03 (0.03)	0.03 (0.04)
Married	0.28 (0.28)	0.20 (0.41)	0.19 (0.33)	0.18 (0.40)
Income adequacyh	−0.17 (0.11)	−0.20 (0.16)	0.03 (0.14)	0.06 (0.16)
Dyadic characteristics
Racial/ethnic minority	−0.25 (0.27)	0.33 (0.40)	−0.40 (0.32)	0.41 (0.39)
Coresiding/living within 1 mile	0.75*** (0.20)	0.17 (0.29)	1.13*** (0.23)	0.38 (0.28)
Having formal helpersi	0.33† (0.19)	−0.27 (0.28)	0.22 (0.23)	−0.14 (0.28)
R 2	.38	.29	.40	.21
Covariance (child, parent)	.55***		.44***

Note: Dyad n = 112 (two dyads with incomplete support exchange data were excluded).

^aHow often children provide support to parents in each domain, rated from 1 (less than once a year/never) to 8 (daily).

^bHow often parents provide support to children in each domain, rated from 1 (less than once a year/never) to 8 (daily).

^cSum of 21 items from the modified Mini-Mental State Examination.

^dMean of 14 items rated from 1 (no difficulty) to 4 (can’t do without help).

^eMean of 11 items rated from 0 (not at all) to 4 (very much/a lot).

^fMean of six items that range from 1 (strongly disagree) to 5 (strongly agree).

^gRated from 0 (never attended/ kindergarten only) to 21 (doctoral degree).

^hRated from 0 (can’t make ends meet) to 3 (money is not a problem).

ⁱHaving paid help from professional helpers.

†p < .10. *p < .05. **p < .01. ***p < .001.

Parent functional impairment showed marginally positive associations with upward support reported by both children and parents. As expected, parents with higher levels of impairment received more upward support from their children.

Parent reports of relationship quality were positively associated with upward support reported by both dyad members, whereas child reports of relationship quality showed a marginal, negative association with upward support reported by children. These findings suggest that, as expected, when parents perceived the relationship with their child as more positive, both parents and children were more likely to report more support provided to parents. In contrast, unexpectedly, children with more positive perceptions of the relationship tended to provide less support to their parents.

For family obligation, only a child’s sense of obligation was positively associated with upward support reported by children. As expected, children who reported a greater sense of obligation were likely to report more support provided to their parents.

Significant covariate effects emerged for child gender, coresidence, and professional help. Specifically, children who are a daughter, coresiding with the parent, and having professional help to meet the parent’s needs were more likely to report more support provided to their parent.

Factors Associated With Downward Support

Among health indicators, child MMSE was negatively associated with downward support reported by parents (Table 3). The child’s perceptions of relationship quality were significantly and positively associated with downward support reported by both dyad members. The parent’s perceptions of relationship quality had no significant effects, and neither did family obligation. Findings suggest that children who perceived the relationship with the parent as more positive were also more likely to report more support from the parent. Regarding covariates, child education and coresidence were significant for downward support reported by the child; the child’s perceived income adequacy was marginally significant for downward support reported by the parent.

Discussion

To the best of our knowledge, this is the first study highlighting the relationship constellation of very old parents and their children who are also of advanced age. Our study expands available literature by examining support exchanges in very old parents and their children, as well as by determining personal and relationship factors associated with upward and downward support.

Patterns of Support Exchange

Findings were in line with Contingency Theory and prior literature, revealing higher levels of upward support compared to downward support, but also evidence of downward support, particularly for intangible support types. These findings demonstrate that previously established support exchange patterns continue to exist when the parent grows into very old age and the child has also reached an advanced age. However, given the high age and average of five health conditions among parents, we were surprised that the mean difference between upward and downward support overall was not larger, and that very old parents were still providing practical assistance to their children in some dyads. Notably, financial support, a type of practical support that even parents in poor health may still be able to provide, was generally the least reported support type.

Although parent and child accounts did reflect a considerable extent of shared views regarding support exchanges, these were not so similar that assessment of only one dyad member seemed sufficient. Comparisons of parent and child reports indicated some discrepancies in reporting. This is an important issue to address which we plan to investigate in more detail in future analyses. Differences in predictive patterns based on reporting perspective also emerged. Parent-reported relationship quality appeared to be more important for upward support reported by both dyad members, while child-reported relationship quality appeared more important for downward support reported by both dyad members. These findings suggest that it is important to capture both parent and child perspectives, underscoring the need for dyadic designs in relationship research (Carr & Utz, 2020).

Factors Associated With Support Exchanges

Findings showed consistency with predictions from both Contingency and Solidarity Theories for upward support. Significant effects in the expected direction emerged for functional impairment, relationship quality, and family obligation. As expected, parent functional impairment and child family obligation predicted upward support. Yet, the effects of relationship quality and family obligation appeared to be relatively stronger than health indicators, suggesting a potentially greater role of solidarity over contingency in these very old parent–child dyads. It is possible that given the high age of the parent, their children were poised to provide significant support to their parents to begin with, so that acute support needs triggered by health issues were less impactful. The notion of support contingent on need may thus be less applicable to very old parent–child dyads than to younger dyads. However, because we only recruited the most involved or closest child, we cannot conclude that this would also be the case for less involved children within a family.

Interestingly, parent and child reports of relationship quality on upward support showed opposing effects, suggesting that positive relationship reports of parents were associated with more upward support, whereas positive relationship reports of children were (marginally) associated with less upward support. As these are cross-sectional findings, the temporality of these associations cannot be determined. However, it makes sense that parents receiving more upward support from their child would report better relationship quality with the child. On the other hand, children may be equally, if not more, likely to report a good relationship with a parent who is healthy enough to be fairly independent, not wanting or needing much upward support. This would be in line with previous findings showing that relationship strain increases when parent health worsens (Kim et al., 2017).

Our findings on downward support did not provide direct evidence for Contingency Theory; child health issues did not emerge as a predictor of support provision parent to child. Our sample may not have included enough children with significant health problems to detect such effects. It is possible that by recruiting the child most involved in supporting the parent, we inadvertently recruited the healthiest child best able to actively engage with the parent. However, we cannot assume that the burden is necessarily placed on the most able child. Other factors such as child gender, living close by, and emotional closeness to the parent are likely to have played a decisive role. Future research specifically looking at very old parents with children who have more serious health problems could yield insight into how contingency operates when one or more older children develop health problems before or simultaneously with their parents.

Family obligation also did not show any effects for downward support, likely because the traditional family obligation expectation in late life is that children provide more toward their parent than vice versa (Stein et al., 1998). Relationship quality was important, albeit only for children’s perceptions—when child reports were positive, downward support levels (reported by children and parents) were higher. Perhaps, positive relationship views of the child signaled appreciation to the parent, making downward support more likely. However, it is also possible that children perceive more downward support when they experience the relationship with their parents as positive.

Regarding potential effects of covariates, child but not parent reports of support were higher for coresiding dyads. Perhaps, the role of children in daily tasks was less visible to parents living with their child, or parents did not realize the extent of their codependent arrangements (e.g., child helping a parent with doing laundry). Having professional help was positively associated with child upward support, probably reflecting greater care needs of the parent. This is consistent with evidence suggesting that formal support given to families to assist with older adults’ care often complements rather than substitutes family care (Davey et al., 2005).

Limitations and Future Directions

Due to our cross-sectional study design, we were unable to ascertain the directionality of effects. Ultimately, only longitudinal data can shed light on causality. However, subsequent analyses of our qualitative data could also aid us in better understanding the dynamics between support exchanges and relationship quality.

We must also acknowledge not having a representative sample. Even though we advertised the study widely in community settings, our recruitment depended on those willing to respond to us. We were not able to track response rates, and our inclusion criteria certainly added to the selectivity of the sample. However, only by following these criteria, we were able to recruit and speak with a portion of the older population that tends to be underrepresented in larger-scale, population-based studies, allowing for a window into lives often unrecognized.

Probably the most consequential sampling decision was requiring sufficient cognitive capacity to participate in a research interview and focusing on community-dwelling parents. While this was necessitated by our dyadic study design, it forced us to leave out an important segment of the very old, namely, those with cognitive impairment, who have high support needs, and their children who are likely even more challenged than the children in our sample. Recognizing this issue, we initiated a supplemental study on older children of very old parents with dementia. These data will allow us to address some of the shortcomings of the present study, albeit with reliance on child reports only. Our findings on support exchanges would have likely looked differently with parents in the sample whose poor health and/or cognitive impairment did not allow study participation. We would expect to find no or much less downward support and similar or higher levels of upward support. How the determinants of support exchanges might differ seems less obvious to us, but we plan to explore this question to the extent possible with our supplemental data.

Conclusions

Findings demonstrated that support is actively exchanged upward and downward in very old parent–old child dyads. Upward support was more prominent than downward support, and parent–child relationship quality and norms of family obligation emerged as main predictors. Downward support still had a notable presence and relationship perceptions played a role. Thus, the evidence for continued support exchanges among very old parents and their children, and the role of relationship quality and family norms for these exchanges, shows that intergenerational theories still hold up in very late-life relationships. However, because support exchanges were found to vary as a function of relationship quality and family norms, health care professionals should be aware that attention to these aspects might be vital to ensure that support needs are met when parents and children grow old together.

Funding

This work was funded by the National Institute on Aging (Aging Together: Relationship Dynamics Between the Very Old and Their Old Children; R21 AG054668; PI: K. Boerner).

Conflict of Interest

None declared.

Author Contributions

K. Boerner and D. S. Jopp planned the study together. K. Boerner and Y. K. Kim oversaw the data collection effort. K. Kim performed statistical analyses. K. Boerner drafted the manuscript. All authors revised the manuscript.