Grandchildren as Caregivers: Adding a New Layer to the Sandwich Generation

Abstract

Although 10% of family caregivers are grandchildren, only a few studies have examined the experience of grandchildren who provide care to grandparents. The current study examined the caregiving processes and outcomes of grandchild caregivers to grandparents. Participants were (N = 5,778) adults identified as a caregiver, including 311 adult grandchildren. Analyses showed that although caregivers to grandparents did not differ significantly from other family caregivers in terms of depression, grandchildren did differ on a variety of demographic and caregiving context variables. A hierarchical binary logistic regression showed that providing personal care and helping with household tasks contribute to the equation, however, grandchild status did not uniquely contribute to the equation after other elements of the caregiving and personal contexts were entered. Post-hoc analyses identified additional predictors within the group of grandchild caregivers. The current study is an important starting point in understanding the experiences of grandchildren caregivers.

Building on models of the Sandwich Generation (Manor, 2021), that group of adults who are simultaneously helping their aging parents and their own children, family caregiving research has expanded in the past several decades. As a newer group of family caregivers enter the public awareness, there is much to be learned about grandchildren who are caregivers to their aging grandparents. As noted elsewhere (Patrick & Goedereis, 2009), today’s grandchildren have more grandparents, will know more of their grandparents and will do so for longer lengths of time than at any other period in history. Such extended contact increases the potential to become an active caregiver.

Whereas the early research in grandparent–grandchild relationships focused on grandparents as reluctant caregivers (e.g., Troll, 1985) or examined structural facilitators and barriers to such interactions (e.g., Uhlenberg & Hammill, 1998), recent studies have focused on grandparents as caregivers to their young grandchildren (e.g., Hayslip & Patrick, 2006; Hayslip et al., 2019). Estimates suggest that about 10% of family caregivers are grandchildren, often grouped under “other family” (e.g., Pristavec, 2019). Scholars have begun to examine the experiences of grandchildren as caregivers to their aging grandparents (e.g., Dellmann-Jenkins & Blankemeyer, 2009; Fruhauf & Orel, 2008), but only a handful of qualitative and quantitative studies exist. As a new focus in the family caregiving literature, most studies have been exploratory, but have been guided by more standard models of the caregiving process and outcome (e.g., Pruchno et al., 1996). That is, most of these exploratory studies use a stress and coping approach in which various factors related to the caregiving context (e.g., help with personal care, presence of dementia) and caregiver personal resources (e.g., age, employment, finances, other obligations) are linked to caregiver health and well-being outcomes. The current analyses focus on grandchild caregivers to aging grandparents using a standard caregiving model.

Grandchildren as Caregivers

Family histories and dynamics among siblings often influence who becomes a primary caregiver for aging parents (Kokorelias et al., 2022; Tatangelo et al., 2018). This is true for caregivers to grandparents, as well. For example, grandmothers’ expectations for care from a grandchild are higher when the grandmother has been involved in raising grandchildren (Stelle et al., 2012). For these grandchildren, such expectations and adopting the role of caregiver can feel “off-time,” interrupting the completion of other developmental milestones (Fruhauf & Orel, 2008). Among these milestones for emerging adults (ages 18–25 years) and younger adults (ages 25–40 years) are the need to establish their own identity separate from their family of origin, to develop and nurture intimate personal relationships, and to pursue professional and work-related goals (Dellmann-Jenkins & Blankemeyer, 2009; Dellmann-Jenkins et al., 2001; Fruhauf & Orel, 2008).

Like other caregivers, grandchildren caregivers to grandparents do experience negative emotional well-being. Celdrán et al. (2011) noted that among grandchildren caregivers ages 14–21 years, the most frequently reported emotions included sadness associated with the grandparent’s dementia-related deterioration. However, these grandchildren also reported feelings of satisfaction with respect to better understanding their grandparent and being able to help with their caregiving. D’Amen et al. (2022) noted similar effects among 87 adolescent caregivers, ages 15–17 years, who provided informal support to their grandparents. Adolescents, like other family caregivers, reported positive aspects of caregiving, including feeling emotionally closer to family. Challenges associated with caregiving included material and financial strains, physical difficulties related to assisting with activities of daily living involving maneuvering or handling their grandparent, communication difficulties, and other emotional and psychological difficulties.

The present study seeks to build upon the literature by examining how grandchildren caregivers differ from other family caregivers in terms of their experience with a depression. Additionally, whether contextual variables influence these potential differences between caregiver groups are explored. However, we approached these data without specific a priori hypotheses about the direction of differences.

Method

Sample

We used the 2020 data from the Centers for Disease Control and Prevention (CDC) Behavioral Risk Factor Surveillance System (BRFSS) for these analyses. These were the most current data available. Data for 2021 will be ready for public use in September 2022. The BRFSS is the largest telephone health survey in the United States. Annually, more than 400,000 residents ages 18 + respond to the telephone survey (CDC, 2022). In 2020, five states completed a caregiving module, including Florida (n = 11,799), Georgia (n = 9,090), Oregon (n = 5,426), Pennsylvania (n = 5,541), and West Virginia (n = 5,880). Of these adults, 5,778 identified as a caregiver for another person with physical needs. Among these five states, 15.7% of those from Florida, 15.0% from Georgia, 13.1% from Oregon, 13.1% from Pennsylvania, and 19.2% from West Virginia identified as a caregiver. A total of 5.4% (n = 311) reported providing care to a grandparent (75 to a grandfather, 236 to a grandmother). An additional 25.1% were caring for a mother or mother-in-law, 8.4% for a father/father-in-law, 21.8% for a spouse or partner, 9.1% for their child or grandchild, 7.5% for a sibling, 7.0% for another relative, and 15.9% for a nonrelative.

Measures

We used single-item self-reports from the BRFSS to index age (Mean = 56.2, SD = 16.1), sex (61.5% female), race (81.1% White, not Hispanic; 9.9% Black, not Hispanic; 0.5% Asian, not Hispanic; 0.9% American Indian, not Hispanic; 4.3% Hispanic; and 3.28% other race, not Hispanic). Most (64.3%) had some college or a college degree. Fewer than half, 43.7%, were employed for pay and 57.9% were married or partnered. We also used a single-item variable to index income. The sample reported a mean income in the range of $35,000–$50,000. On average, 25.6% of the caregivers shared their home with a child. Finally, caregivers were asked to indicate whether they had ever been diagnosed with depression or anxiety, with 26.8% responding in the affirmative.

The caregiving context was characterized by several single-item questions. The mean length of caregiving was around 31 months (SD = 24). Most caregivers 54.2%, provided more than 8 h of care per week, with a mean of around 17.3 h. Slightly more than half of the caregivers, 54.2%, provided help with Basic Activities of Daily Living, including personal care activities. The majority, 80.0%, reported helping with household chores and tasks. A minority of caregivers (17.3%) were providing support to a person with dementia.

Analytic Plan

We examined mean differences between grandchild caregivers and other family caregivers using t-tests and chi-squared analyses, when appropriate. To examine whether grandchild caregiver status added to the correct prediction of depression, we conducted a hierarchical analysis. Because our depression outcome was dichotomous, we conducted a binary logistic regression, in which we could examine both the statistical significance of each regressor and its overall effect on classification via the Odds Ratio. In order to examine whether grandchild caregiver status represented an additional challenge to caregivers’ well-being, beyond the caregiving context and beyond their other demands, we conducted a hierarchical regression. Our plan was to first enter the variables indexing the caregiving context in Step 1. We followed those with indices of personal resources/demands in Step 2. Finally, we entered the dichotomous variable indicating grandchild caregiver status in Step 3.

Results

Mean Group Differences

As shown in Table 1, although caregivers to grandparents did not differ significantly from other family caregivers in terms of the percentage experiencing depression or income, grandchildren did differ on a variety of demographic variables. Grandchild caregivers were younger, less likely to be female, more likely to be employed for pay, less likely to be married, but were living with more children. In addition to these personal resources, caregivers to grandparents also differed from other family caregivers in terms of the caregiving context. As shown in Table 1, caregivers to grandparents were newer to the role, provided fewer hours of care each week relative to other caregivers, were more likely to be providing help with household tasks, and were more likely to be assisting a person with dementia.

Table 1.

Mean Differences.

	Grandchild caregiver	Other caregivers	T	df	p
Depression (%)	30.10 (45.94)	26.62 (44.20)	1.34	5,747	.090
Care recipient has dementia (%)	25.00 (43.60)	16.00 (37.10)	3.81	5,243	<.001
Length of care in months	25.90 (23.38)	31.23 (24.01)	3.78	5,677	<.001
Care hours per week	12.68 (12.91)	17.61 (15.42)	5.37	5,333	<.001
Help with personal care tasks (%)	54.00 (49.90)	51.00 (50.00)	1.21	5,683	.225
Help with household tasks (%)	85.00 (36.00)	80.00 (40.30)	2.14	5,557	.033
Own sex (% female)	54.34 (49.89)	61.92 (48.56)	2.67	5,776	.008
Own age	31.68 (10.12)	57.55 (15.26)	29.54	5,776	<.001
Number of own children	1.00 (1.32)	0.46 (1.00)	9.00	5,754	<.001
Employed for pay (%)	62.87 (48.40)	42.58 (49.45)	7.00	5,746	<.001
Income category	5.75 (2.16)	5.70 (2.10)	0.39	3,980	.69
Education in years	13.64 (1.89)	13.83 (1.88)	1.70	5,761	.089
Married/partnered (%)	36.36 (48.18)	59.52 (49.09)	8.06	5,743	<.001
Race		75.2		5	< .01
% White, NH	75.2	81.5
% Black, NH	9.3	9.9
% Asian, NH	1.3	0.5
% American Indian, NH	1.3	0.9
% Hispanic	8.0	4.1
% Other, NH	4.8	3.1

Model Testing

We examined whether grandchild caregiver status would contribute additionally to the classification of depression by conducting a hierarchical binary logistic regression. Here, known predictors of depression from both the caregiving context (Step 1) and personal context (Step 2) were entered into the regression equation. Grandchild caregiver status was entered last in Step 3. This is considered to be a stringent test of the contributions of the relationship to the care recipient, because any shared variance among grandchild status and the other variables is already accounted for by the first two steps of the equation. Results are shown in Table 2.

Table 2.

Hierarchical Binary Logistic Regression.

	Step 1				Step 2				Step 3
	_b_	SE(b)	p	OR	_b_	SE(b)	p	OR	_b_	SE(b)	p	OR
Caregiving context
Recipient dementia	.062	.089	.487	1.064	.078	.093	.402	1.081	.107	.093	.249	l.l 13
Length of care months	.066	.073	.363	1.069	−.030	.077	.692	.970	−.031	.077	.685	.969
Care hours per week	.180	.093	.053	1.197	.140	.097	.149	1.150	.140	.097	.147	1.151
Personal care tasks	.001	.001	.535	1.001	.003	.001	.075	1.003	.002	.001	.098	1.002
Household care tasks	.008	.002	.001	1.008	.008	.002	.002	1.008	.007	.003	.004	1.007
Personal context
Own sex (female = 1)					.615	.075	<.001	1.849	.609	.075	<.001	1.838
Own age					−.036	.036	.312	.964	−.039	.036	.276	.962
Number of own children					−.032	.003	<.001	.968	−.035	.003	<.001	.965
Employed for pay					−.777	.078	<.001	.460	−.800	.078	<.001	.449
Income category					−.280	.029	<.001	.750	−.279	.029	<.001	.756
Education in years					−.471	.071	<.001	.624	−.482	.071	<.001	.617
Married/partnered					.078	.093	.402	1.081	−.571	.162	<.001	.565
Relationship
Caregiver to GP									.107	.093	.249	l.l 13
	X2 (DF = 5) = 28.984, p < .001, Nagelkerke R2 = .01				Step: X2 (DF = 7) = 420.08, p < .001 Model : X2(DF = 12) = 443.20, p < 001; Nagelkerke pseudo R2 = . 152				Step: X2 (DF = 1) = 5.90, p<.001 Model: X2 (DF = 113 = 440.10, p < .001, Nagelkerke R2 = . 154

As shown, in Step 1, providing personal care and household support significantly contributed to the equation. Although significant, the pseudo R² was very small and the equation only correctly classified 72.8% of the sample, with no persons with depression being correctly classified.

Step 2 added elements of the caregiver’s personal context. The equation was significant and correctly classified 74.1% of the sample, with true positives at 20.7%. Providing personal care and helping with household tasks continued to contribute to the equation. Being younger, living with fewer children, being employed for pay, having lower income, and having fewer years of education significantly contributed to the classification. Of note, although not reaching statistical significance, an inspection of the Odds Ratio (OR) shows that female caregivers were 1.8 times more likely to report depression.

Finally, we examined the contributions of grandchild caregiver status. The equation continued to reach significance. The pseudo R² was small, but nontrivial. About 73.3% were correctly classified, with 13.9% of those reporting depression being correctly classified. Of note, although grandchild status did not uniquely contribute to the equation after other elements of the caregiving and personal contexts were entered, grandchild caregivers were 111% more likely to report depression.

Post-hoc Analyses

To better understand grandchild caregivers’ experiences, we conducted a post-hoc binary logistic regression examining the first two steps of our model among the 311 grandchild caregivers. Results are shown in Table 3. Focusing on the entire two-step hierarchical model, the equation correctly classified 73.1% of the grandchild caregivers, with a true positive rate of 25.8%. Although no regressor uniquely contributed to the prediction of depression among grandchild caregivers, the odds ratios are informative. Grandchildren caring for a grandparent with dementia were more likely to report depression. Those providing care for more hours per week and longer periods of time were about 1.2 times more likely to report depression. Granddaughters were 1.5 times more likely to report depression. Those grandchildren who were employed for pay were also more likely to report depression.

Table 3.

Post-hoc Binary Logistic Regression, Grandchildren-Only.

	Step 1				Step 2
	_b_	SE(b)	p	OR	_b_	SE(b)	p	OR
Caregiving context
Recipient dementia	−.363	.372	.330	.696	−.385	.392	.326	.680
Length of care months	.136	.109	.211	1.146	.155	.116	.184	1.167
Care hours per week	.282	.147	.055	1.325	.171	.160	.286	1.187
Personal care tasks	−.143	.325	.660	.867	−.091	.351	.795	.913
Household care tasks	−.643	.496	.195	.526	−.735	.540	.173	.480
Personal context
Own sex (female = 1)					.431	.358	.229	1.538
Own age					−.001	.020	.946	.999
Number of own children					.040	.133	.765	1.041
Employed for pay					.115	.079	.143	1.122
Income category					−.210	.127	.098	.811
Education in years					−.016	.184	.931	.984
Married/partnered					.078	.096	.413	1.082
	X2 (DF = 5) = 10.09, p = .07, Nagelkerke R2 = .07 71.2% overall; 12.1% true positive/depressed				Step: X2 (DF = 7) = 13.15, p = .07 Model : X2 (DF = 12) = 23.24, p <05; Nagelkerke pseudo R2 = .146 73.1% overall; 25.8% true positive/depressed

Discussion

With grandchildren having more grandparents and having more contact with them over extended periods of time, it is not unrealistic to expect that some of these grandchildren will become caregivers to aging grandparents. Grandchild caregivers have not received significant empirical attention, despite large caregiving surveillance studies showing that many family caregivers are providing care to someone other than spouses, parents, or siblings. Because the needs and challenges facing grandchild caregivers may differ from these other family members, we examined the caregiving process and outcomes among more than 5000 caregivers who participated in the 2020 CDC BRFSS.

At the level of mean comparisons, grandchild caregivers differed from other family caregivers on a variety of both personal factors and the caregiving context indicators. For example, grandchildren caregivers were about 26 years younger than other family caregivers and more racially and ethnically diverse. Although grandchildren caregivers were less likely than others to be married, they were co-residing with more children. In terms of the caregiving context, grandchild caregivers were newer to the role and were providing fewer hours of care per week. Grandchild caregivers were as likely to be providing help with personal care tasks, but were more likely to be helping with household tasks than other family caregivers.

In order to examine predictors of caregiver depression, we drew from models in the literature (e.g., Pruchno et al., 1996). We conducted a three-step hierarchical binary logistic regression to predict depression. We first entered indicators of the caregiving context, then personal context, and finally whether a person was a grandchild caregiver. This is a “hard” test, asking the question, “when we allow known predictors of depression to enter the model first, does grandchild caregiver status tell us any more about the predictors of depression?.” The short answer is: “no.” Group differences in caregiving context and personal resources account for depression, even among the grandchild caregivers. Specifically, the four regressors that significantly predict caregiver depression are: helping with personal care tasks, helping with IADLs, being younger in age, and having more children at home. However, the odds ratios also tell a story. Looking at the OR in the Step 3 model of Table 2 shows that care recipient dementia (OR = 1.11), care hours per week (OR = 1.15), and caregiver sex (OR = 1.8 female) are associated with higher reports of depression.

Examining these predictors within the grandchild caregivers only, those caring for a grandparent with dementia were more likely to report depression. Helping with personal care tasks was also associated with depression. Other responsibilities in one’s life, such as employment were associated with higher reports of depression, too.

Although these results suggest that more empirical, policy, and programmatic attention should be directed toward this newer group of caregivers, we do note limitations. The BRFSS is not designed for in-depth examinations of such caregiving processes (CDC, 2022). Other large, national data sets include more detailed and nuanced measures of the caregiving process (e.g., Pristavec, 2019). Notably, we encourage replication and extension of these findings within the National Health and Aging Trends Study (NHATS) and National Study of Caregiving (NSOC) data. Longitudinal data would enable an examination of whether these grandchildren are being mentored into taking on a larger caregiver role within the family. Another limitation to our data is that we do not know whether the grandchild is the primary caregiver or whether they are providing back-up to someone else who serves as the primary caregiver. Additionally, beyond a few demographics, we do not know what other developmental or daily challenges these grandchildren are facing. Finally, the BRFSS module was only completed by residents in five states. Including this module in the broader BRFSS survey would provide important information about these families. Moreover, doing so would allow more sophisticated questions to be posed, such as whether caring for someone younger, the same age, or older brings additional challenges.

However, despite the limits of the data set, it is an important starting point. We are especially intrigued by the notion of being “off-time.” Future studies might examine whether grandchild caregivers are delaying their own adult dreams. Finally, it may be critical to examine how much of the depression is compounded with interactions with the COVID-19 pandemic. We are hopeful that other research teams will join us in investigating this new layer in the sandwich generation of family caregiving.

Biographies

Julie Hicks Patrick is the Editor of IJAHD and the P.I. of the Appalachian Gerontology Experiences: Advancing Diversity in Aging Research (AGE-ADAR Program).

Laura E. Bernstein is a graduate student mentor in the AGE-ADAR Program and a doctoral student in Psychology.

Arianna Spaulding is undergraduate AGE-ADAR Scholar, whose interests lie in clinical psychology and family processes.

Bianca E. Dominguez is undergraduate AGE-ADAR Scholar, whose interests lie in clinical psychology and family processes.

Carly E. Pullen is a graduate student mentor in the AGE-ADAR Program and a doctoral student in Psychology.