Caregiver Well-Being: Intersections of Relationship and Gender.

We know much about caregiving women compared with caregiving men and caregiving spouses compared with caregiving adult children. We know less about the intersections of relationship and gender. This article explores this intersection through the well-being (burden and self-esteem) of caregivers to family members with dementia. Throughout British Columbia, Canada, 873 caregivers were interviewed in person for on average, over 1½ hours. The results reveal that daughters experience the highest burden but also the highest self-esteem, suggesting the role is less salient for their self-identities. Wives emerge as the most vulnerable of the four groups when both burden and self-esteem are considered. The data confirm the usefulness of the intersectionality framework for understanding co-occupancy of more than one status and indicate that positive cognitive well-being and negative affective well-being can be differentially related. Multivariate analyses confirm the importance of caregiver, not patient, characteristics for burden and self-esteem.

Introduction

There is much gerontological research on the importance of gender and the importance of spousal and adult–child relation for various caregiver outcomes. Wives’ and daughters’ greater involvement in providing more care, more personal care, and more tasks of caregiving is well documented (Chappell & Hollander, 2013). While not conclusive, research increasingly suggests women caregivers experience more burden than men. Research comparing spouses and adult–child caregivers is inconsistent and that examining both gender and relation sometimes suggests intersectionality (Conde-Sala, Garre-Olmo, Turró-Garriga, Vilalta-Franch, & López-Pousa, 2010) but sometimes not (Raschick & Ingersoll-Dayton, 2004). There is, therefore, a need to further explore the intersection between gender and relation, particularly given the now widespread recognition that occupying two such statuses interacts in the lives of caregivers, albeit in largely unknown ways.

In this article, we examine the intersectionality of gender and relation among caregivers to relatives with dementia in British Columbia, Canada. Wives, husbands, daughters, and sons are compared in terms of two well-being measures (burden and self-esteem), then the importance of gender and relation for these outcomes is examined in multivariate analyses for men and women, spouses and adult children.

Review of the Literature

The research on gender differences in caregiving for older adults is quite abundant, establishing that women tend to provide more hours of care, more hands-on care, and are involved in more tasks than are men (Calasanti & Bowen, 2006). This feminization of caring has been facilitated by women’s socialization into the nurturing roles (Aronson, 1992). Male caregivers have been characterized as more instrumental, focused on specific tasks in contrast with female caregivers who also tend to the emotional work, maintaining identities and relationships (Carroll & Campbell, 2008). Recent research suggests the emergence of less gendered distributions of caregiving responsibilities (Calasanti & Bowen, 2006). In a European context, Da Roit (2007) notes that daughters’ involvement in direct care has decreased with a greater willingness to arrange, supervise, and finance paid care at home. She suggests this reduced share of direct care is not the consequence of men’s increased involvement but rather because of structural factors or shifting filial obligation norms and is typically restricted to the middle and upper classes.

Women also tend to report more burden than men caregivers (Garlo, O’Leary, Van Ness, & Fried, 2010; Hooker, Manoogian-O’Dell, Monahan, Frazier, & Shifren, 2000; Kim, Chang, Rose, & Kim, 2012) and lower self-esteem (Kim, Baker, & Spillers, 2007), although some studies report no gender difference in burden (Rosdinom, Zarina, Marhani, & Suzaily, 2013) or depression or distress (Mohamed, Rosenbeck, Lyketsos, & Schneider, 2010).

We also know much about the differences between spouse and adult–child caregivers. Spouses tend to be older, living with the care recipient, have a different emotional relationship with the older adult care recipient, have fewer competing responsibilities, have worse physical health, and view caregiving as a normative part of their marriage (Pinquart & Sörensen, 2011). Adult–child caregivers, on the other hand, can experience caregiving as extra to their already busy lives and as role reversal (Bastawrous, 2013; Lee & Smith, 2012).

Research on how caregiving impacts spouse and adult–child caregivers is inconclusive; Pinquart and Sörensen’s (2011) review finds no differences in overall burden but spouses experience higher physical, financial, and relationship burden with no more emotional burden or social or job strain than adult children. Spouses also report lower levels of positive psychological well-being and higher levels of depression than adult children. Research since that review reports that spouses more satisfied with their relationships with the care recipient (Williams, 2011) are more burdened than adult–child caregivers (Kim et al., 2012; Mohamed et al., 2010) and for some, no difference (Rosdinom et al., 2013).

Some research comparing both spouse and adult–child caregivers suggests there is gender/relation intersectionality in terms of their subjective well-being. Conde-Sala, Garre-Olmo, Turró-Garriga, Vilalta-Franch, and López-Pousa (2010) find the order of least burden and better mental health to most burden and worst mental health to be husbands, wives, daughters, and then sons. Furthermore, the correlates of burden differ by group. For example, living with the care recipient is more burdensome only for adult children. However, comparing spouse and adult–child caregivers, Chappell, Dujela, and Smith (2014) find child caregivers significantly more burdened than spouses at T1 with no significant gender differences in multivariate analyses. Raschick and Ingersoll-Dayton (2004) do not find a relation–gender interaction when examining the costs of caregiving (exhausted at the end of the day, more to do than they could handle, no time for self, and never made any progress despite working hard as a caregiver) and rewards (caregiving made them feel good about themselves and made them appreciate life more). They do find that women experience greater costs than men and that adult children experience greater rewards than spouses.

In addition to gender and relation, several other correlates of well-being are of interest. Health of the care recipient can refer to many aspects of their condition, that is, functional decline in terms of activities of daily living (ADL) and instrumental ADL (IADL) (Gallagher et al., 2011; Kim et al., 2012); cognitive decline (Ryan et al., 2010); more advanced disease stage (Vetter et al., 1999); and behavioral disturbances such as agitation, aggression, irritability, delusions, hallucinations, apathy, anxiety, disinhibition, and motor overactivity (Cheng, Lam, & Kwok, 2013; Leroi et al., 2012). Research to date is inconsistent in terms of what is most predictive of lower well-being for caregivers, whether it is functional decline for those caring for persons with mild dementia and behavioral symptoms for moderate to severe dementia (Gallagher et al., 2011) or cognitive deficit in the early stages and behavioral problems and poor IADL as the disease progresses (Ornstein et al., 2013; Yeager, Hyer, Hobbs, & Coyne, 2010; Zucchella, Bartolo, Pasotti, Chiapella, & Sinforiani, 2012). Still others (Allegri et al., 2006; Zawadzki et al., 2011) find behavioral symptoms compared with cognitive or functional decline are more important. In terms of the care provided, more hours of care is related to more distress (Kim et al., 2012), confirming an earlier meta-analysis (Pinquart & Sörenson, 2003).

Caregivers’ external resources can be relevant in explaining the impact of caregiving on their well-being. A lack of informal supports (Galvin et al., 2010), an inability to take breaks when needed (Goldsworthy & Knowles, 2008), a poor relationship quality with the care recipient (Goldsworthy & Knowles, 2008), and a lack of receipt of formal services (Zarit et al., 2011) have all been related to more burden. Specific to taking cholinesterase inhibitors (ChEIs), a medication for mild to moderate dementia and being taken by the care recipients in this article, Schoenmakers, Buntinx, and Delepeleire (2009) conclude from a systematic review that ChEIs seem to lower caregiver burden irrespective of the actual effect of these drugs on the person with dementia, perhaps explained by the hope the medication brings, thereby serving as a focal point for enhanced interactions between caregivers and their relatives under treatment (Smith, Kobayashi, & Chappell, 2011).

Caregivers also have their own internal resources. Those related to distress include loneliness/social isolation (Mausbach, Coon, Patterson, & Grant, 2008), a lack of secondary intrapsychic strengths such as competence and inner strength (Pearlin, Mullan, Semple, & Skaff, 1990), feeling underappreciated, and their own health (Chappell et al., 2014). Caregiver sociodemographic factors are also important. Fewer economic resources whether measured as income or social class (Robinson, Fortinsky, Kleppinger, Shugrue, & Porter, 2009; Sun, Hilgeman, Durkin, Allen, & Burgio, 2009), less education (Navaie-Waliser et al., 2002) coresidence with the care recipient (Kim et al., 2012), and younger age (Andren & Elmstahl, 2007), often attributed to older caregivers more likely to be spouses, have all been related to lower well-being.

The differential relevance of any of these factors (care recipient disease characteristics, caregiver external or internal resources, and caregiver or care recipient sociodemographic factors) for the intersectional effects of gender and relation on the burden and self-esteem of caregivers is largely unknown. In this article, we compare men and women, spouse and adult–child caregivers to older adults, to gain a better understanding of the intersection of gender and relation in the experience of caregiving examining one positive (self-esteem) and one negative (burden) indicator of well-being.

Theoretical Framework

Intersectionality refers to simultaneous sources of disadvantage and the impact of their co-occurrence on the individuals involved. As a theoretical framework, intersectionality has become an influential paradigm in health studies (Hankivsky, 2012; Hankivsky & Christoffersen, 2008) since it was first popularized by Crenshaw (1994). In gerontology, it offers advantages over previous largely a theoretical analyses that examined one experience (e.g., being old) in isolation or as having an additive effect to one or more subordinate positions (e.g., being old plus being female and/or a lower social class, etc.) as reflected in the concepts of double, triple, and multiple jeopardy (Acker, 2006; Denis, 2008). Intersectionality seeks to understand the interactions between various statuses, recognizing their fluidity and variability. As Krekula (2007) notes, their intersections can strengthen or weaken, supplement or compete, further marginalize or neutralize each other. That is, their combined effects are complex and may or may not lead to negative consequences for the individual.

In the context of this study, the intersectionality framework serves as a useful theoretical tool to integrate social differences in caregiving, as we seek to avoid essentializing these aspects into discrete dimensions or along some predetermined hierarchical pattern. Instead, we use the framework to articulate the multidimensional and relational nature of caregiving and the social conditions under which well-being is experienced by individuals cohabiting their roles as both family member and care provider.

Our interest in intersectionality is its relevance for the subjective well-being of caregivers. Well-being is sometimes used synonymously with quality of life, sometimes not, with a lack of consensus of the meaning of either term (Sarvimaki & Stenbock-Hult, 2000). Camfield and Skevington (2008) note in their review integrating both fields, that when subjective well-being is referred to as the valuations we make about our lives and subjective quality of life is referred to as our perceptions of our position in life, the centrality of value judgments in both concepts leaves them virtually indistinguishable. Without entering the debate about the different uses of these terms found in the literature, we adopt the concept of well-being here as used by Hansen, Sladsvold, and Ingebretsen (2013), Deaton (2008), and Knight and Rose (2011) which distinguishes between cognitive well-being (examples include life satisfaction and self-esteem) and affective well-being, sometimes referred to as subjective well-being (Pinquart & Sörensen, 2004) (examples include happiness, burden, and depression). We view well-being, not as unidimensional, but rather cognitive and affective dimensions as well as positive and negative aspects as potentially related but distinct, coexisting phenomenon. They are not opposite ends of one continuum. How they are related to one another, however, is unresolved. In this article, how the intersection of gender and relationship is associated, if at all, with one negative affective well-being measure, burden, and one positive cognitive well-being measure, self-esteem, is compared with single occupancy of either status.

Method

Data come from the Caregiver Appraisal Study (CAS), part of a larger program of research (the Alzheimer’s Drug Therapy Initiative). The CAS recruited caregivers province-wide in British Columbia, Canada. All were caring for those with dementia as diagnosed by a physician, taking a ChEI (Aricept®, Reminyl®, or Exelon®) and covered by the B.C. Ministry of Health’s PharmaCare program. The study was established to assess caregivers’ perceptions of the effectiveness of these medications. The sample is not representative; it was recruited through referral (self, staff, and physician) and through calls made by PharmaCare to notify individuals that the medications were approved for coverage of the care recipient, informing them of the study and asking whether they wished to be linked by phone to the study office, sent a letter regarding the study, or not interested.

Study personnel informed potential respondents about the study and asked whether they were willing to be included in a triage phone call after the care recipient had been taking a ChEI for 6 months. Inclusion criteria for the triage were as follows: caregivers spoke English, were a family member, providing at least 3 hr of care per week, and the care recipient was currently taking a ChEI and living in the community (not in a nursing home). Of the 1,300 triages completed, 1,243 caregivers or 95.6% agreed to participate in a face-to-face interview; 29 were ineligible (patient deceased, not diagnosed with dementia, infrequent contact, etc.); 28 declined the full interview (too busy, patient too sick, caregiver too sick, etc.).

Of the 1,243 caregivers, 24.2% (301) were not interviewed (the patient died, the caregiver interviewed at triage was no longer the caregiver, etc.); that is, 942 caregivers were interviewed in person, approximately 6 months after the older adult had been taking the medication. Six months is when physicians assessed cognitive performance and recommended whether the older adult should continue, stop, or switch the medication. The average length of the interview was 97.26 min. In these analyses, only spouse and adult–child caregivers are included (873 or 92.7% of the sample) of whom most, 69.8%, are spouses and 30.2% adult children. Interviews began in 2008 and were completed in 2010.

Measures

There are two dependent variables (DVs): burden and self-esteem. Burden refers to the negative consequences of caregiving; it can be physical, financial, and/or psychosocial, but there is no theoretical (or empirical) demarcation of when high burden has been reached (Garlo et al., 2010). Burden is measured using the short Zarit Burden Interview (Bédard et al., 2001), a measure of subjective burden consisting of 12 questions such as “Do you feel that because of the time you spend with ______, you don’t have enough time to yourself?” and “Do you feel strained when you are around ______?”. Scoring is from 0 to 4, never to daily. The higher the score, the more burden (α = .89).

Self-esteem is a measure of one’s overall worthiness (Mruk, 2006; Pearlin et al., 1990) or positive well-being. It is measured using the Rosenberg Scale of Self-Esteem (The Morris Rosenberg Foundation, 2003). The scale consists of 10 questions; 5 of which are positive, such as “I feel that I have a number of good qualities,” and five are reversed, such as “I certainly feel useless at times.” Questions are recoded and scored from 0 (strongly agree) to 3 (strongly disagree), then summed with a resultant scoring where a higher number is equal to lower self-esteem. The α is .88. That is, for both DVs, the higher the score the worse the individual is feeling.

The independent variables (IVs) are gender (male/female) and relation (spouse/adult child).

The control variables referring to care recipient condition are disease stage (mild or moderate/severe dementia), type of dementia (Alzheimer’s, vascular, and other), care recipient disability (basic ADLs summed—bathing, dressing, toileting, transfers, continence, feeding, and phoning—α = .80 and IADLs summed—shopping, food preparation, housekeeping, laundry, transportation, medication management, and banking—α = .83), older adult chronic conditions (high blood pressure/hypertension, arthritis/rheumatism, ear trouble including hearing loss, stomach/digestive problems, feet/ankle problems, depression, leg problems, allergies/sinus problems, back problems, chronic pain, thyroid problems, fatigue/sleep problems, and weight loss/gain—summed), incontinence problems (no, yes), sleep/fatigue problems (no, yes), and verbally agitated behavior (no, yes).

Amount of caregiving is measured in terms of hours/week of care provided and length of time providing care in years.

External resources of the caregiver are as follows: whether he or she receives help with the care from family or friends (no, yes), whether he or she needs family or friends to be more involved (no, yes), whether he or she can take a break when needed (no, sometimes, yes), whether he or she receives emotional support (no, yes), expressive support measured using Pearlin’s Expressive Support Scale (of the original eight questions, the question—“There is really no one who understands what you are going through.”—was deleted because doing so raised the α from .79 to .91. Other questions include for example, “You have someone you feel you can trust.” and “There are people in your life who help you keep your spirits up.” (1= strongly agree, 4 = strongly disagree, items are summed, α = .91). The caregiver was also asked whether the care recipient receives emotional support (no, yes), and the caregiver’s assessment of their prior relationship with the care recipient (on a scale of 1–5, the higher the number, the closer the relationship). Caregivers were asked the number of formal services received by the care recipient (0, 1, 2, 3 or more), the type of ChEI (Aricept®, Reminyl®, Exelon®) being taken, and drug history (continuing same ChEI or switched type).

Internal caregiver resources are as follows: his or her perceptions of their own health (scale of 1–3), the number of chronic conditions he or she has (high blood pressure/hypertension, arthritis/rheumatism, ear trouble including hearing loss, stomach/digestive problems, feet/ankle problems, depression, leg problems, allergies/sinus problems, back problems, chronic pain, thyroid problems, fatigue/sleep problems, and weight loss/gain—summed), anxiety measured using the anxiety subscale of the Hospital Anxiety and Depression Scale (Mykletun, Stordal, & Dahl, 2001) consisting of 6 items such as “I feel tense or wound up,” “Worrying thoughts go through my mind,” α = .83), whether he or she feels alone or isolated as a result of providing care (no, yes), whether he or she feels appreciated on a scale of 0–10 (“On a scale of 1 to 10, with one being not appreciated at all and 10 being very appreciated, how much do you feel appreciated by the care recipient for what you do to support them?”). In addition, caregivers were asked about 10 pleasures or satisfactions they might potentially derive from their caregiving (such as “Become more aware of inner strengths,” “Gained personal satisfaction,” and “Made new social connections”). Items are summed, α = .79.

Sociodemographic variables include caregiver age, education, how well income satisfies their needs (coded 1–3), marital status (married/common-law, other), geographic distance in miles from care recipient, and religiosity (0–3, sum of 3 items, “Are you affiliated with any organized religion?”, “Do you engage in spiritual or religious activities on a regular basis?”, and “Do you think religious or spiritual beliefs have an impact on your caregiving?”, α = .81). Care recipient’s sociodemographic variables include gender, age, education, and monthly income.

Analyses initially examine frequencies and bivariate correlations comparing women, men, spouses, and adult children, then multiple regression analyses assess the significance of relation among men and among women and the significance of gender among spouses and among adult children when other factors are controlled. Analyses are also conducted for the four groups separately. The data were checked for multicollinearity, linearity, and homoscedasticity. When collinearity occurred, the variables were entered into separate regressions. When skewness occurred, variables were either truncated or categories collapsed (log transformations did not remove sufficient skewness). Listwise deletion was used in all cases (in no case did missing values exceed 5%). All scales were created by summing then dividing by the number of items answered to take into account any missing data. Only significant variables are shown in the tables.

Results

Table 1 shows selected sample characteristics for the total sample, for spouse and adult–child caregivers separately, and for male and female caregivers separately. Almost all are Caucasian (95.7%) and primarily spouses (70.2%) and female (68.7%). However, care recipients are half men and half women (48.8% women). Most have dementia of the Alzheimer’s type (59.2%) and 12.9% have vascular dementia; 21.9% of caregivers do not know the type and stage of dementia. Whether caregivers were not informed by the physician as to the stage of dementia, the physician could not determine the stage, or the caregiver has forgotten is unknown. In three fourths of the cases, the caregiver and care recipient live together (75.3%) and almost all (89.2%) are married, reflecting the percentage of spouses in the sample. Incomes vary, ranging from low to high. Caregivers are, on average, 10 years younger than the care recipients and they provide an average of over 24 hr of care/week.

Table 1.

Sample Characteristics (%).

Characteristics	All caregivers N = 873	Spousesa N = 613	Adult child N = 260	Maleb caregivers N = 273	Female caregivers N = 596
CG is adult child	29.8			26.4	31.3
CR is husband/wife	70.2			73.6	68.7
Sex of caregiver (female)	68.7	67.2	72.3
Sex of care recipient (female)	48.8	33.6	84.6	91.0	31.4
Alzheimer’s disease diagnosis	59.2	63.4	49.5	47.5	45.2
Vascular dementia diagnosis	12.9	10.2	19.0	7.3	10.8
Caregiver and care recipient live together	75.3	92.0	35.9	77.3	72.6
Caregiver marital status (married/common-law/widowed)	89.2	99.0	66.2	88.1	87.4
Caregiver monthly income
Less than $2249 per month	20.9	20.3	22.2	14.7	19.1
$2250 to $3499 per month	27.5	31.0	18.9	19.8	25.3
$3500 to $4749 per month	24.7	26.1	21.2	20.9	20.7
$4750 or more per month	26.9	22.6	37.7	31.3	19.1
	Range, mean
Hours per week caregiving	0–165, 24.58	0–165, 25.12	1–160, 23.53	0–165, 26.42	0–164, 23.90
Caregiver age at time of interview	28–93, 69.03	42–93, 74.27	28–87, 56.59	38–93, 71.66	20–89, 67.56
Care recipient age at time of interview	48–100, 79.61	48–93, 77.44	58–100, 84.73	53–99, 78.89	48–100, 80.00

Note. CG = caregiver; CR = care recipient. aSpouses and adult children are significantly different on each of the characteristics (p < .000) except for sex of the caregiver, which is ns. bMale and female caregivers are significantly different in terms of the gender of the care recipient and caregiver income.

Spouse and adult–child caregivers are significantly (p < .000) different from one another on all of the variables shown in Table 1 except in terms of the sex of the caregivers where, for both groups, about 70% are women. Not surprisingly, spouses are more likely to be caring for a male, living with the care recipient, married, of moderate income, providing more hours of care, caring for someone with a diagnosis of Alzheimer’s disease, and older but caring for someone younger than are adult children. This suggests adult–child caregivers are caring for a widowed parent.

Male and female caregivers differ from one another in far fewer ways. Male caregivers are much more likely to be caring for a female care recipient (typically their wives) to have a higher monthly income (>US$4,750) and to be older than are female caregivers (p < .000 in all instances); males are also more likely to be caring for someone who is younger than the person women are caring for (p < .05).

When intersecting gender and relation, different sample sizes are evident, with wives predominating and sons totaling less than 100 in the overall sample: wives: N = 409; husbands: N = 204; daughters: N = 187; sons: N = 73.

Table 2 reveals the relationship between each of the DVs (burden and self-esteem) and relation, gender, gender among spouses only, gender among adult children only, relation among men only, and relation among women only. Spouses and adult children are significantly different from one another in terms of both DVs, more so in terms of burden than self-esteem. Adult children experience significantly more burden but better self-esteem than spouses. Women, however, experience more burden than men, but there is no significant difference in the self-esteem, suggesting a closer examination of relation and gender is warranted.

Table 2.

Burden and Self-Esteem: t-Tests.

	Burden		Low self-esteem
	xˉ	T	xˉ	t
Spouses	10.44	5.14****	7.29	3.28***
Adult children	14.47		6.16
Men	10.07	3.15**	6.93	ns
Women	12.36		6.95
Spouses only
Husbands	8.83	3.10**	6.96	ns
Wives	11.22		7.45
Adult children only
Sons	13.50	ns	6.86	ns
Daughters	14.75		5.90
Men only
Husbands	8.83	3.02**	6.96	ns
Sons	13.50		6.86
Women only
Wives	11.22	3.97****	7.45	3.89****
Daughters	14.87		5.90

*p < .05; **p < .01; ***p < .001; ****p < .0000.

The highest burden is experienced by daughters, followed by sons, then wives, then husbands. The lowest self-esteem is experienced by wives followed by husbands, sons, and daughters without large differences other than between wives and daughters. The lower self-esteem experienced by spouses compared with adult children appears to be due to the more negative experiences of wives compared with daughters; sons and husbands do not differ in these respects. These bivariate correlations suggest the importance of studying both relation and gender within multivariate analyses, and, men and women, spouses and adult children separately. In addition, the differences, together with the differential sample sizes for each group, warrant further examination of each separately.

Table 3 shows the multiple regressions with burden and self-esteem as the DVs for each of: women with relation plus the control variables, men with relation plus the control variables, spouses with gender plus the control variables, and adult children with gender plus the control variables. There are common factors across all four groups. The more hours providing care, the more distant the relationship was with the care recipient and the younger the age of the caregiver, the greater the burden among women, men, spouses, and adult children. Burden is a significant correlate of self-esteem among all groups; the higher the burden the lower the self-esteem. Hours of caregiving, prior relationship, and age therefore are related to self-esteem indirectly, through burden, among all groups (see Table 3, Panels A and B). In addition, across all groups, those with higher levels of education have better self-esteem. It is also to be noted that for both DVs and across all groups, caregiver characteristics emerge as significant predictors but no care recipient characteristics are significant.

Table 3.

Multiple Regressions (βs).

IVs	Womena	Menb	Spousesc	Adult childrend
A. Burden
Hours/week caregiving	.27****	.28****	.29****	.22****
CG chronic conditions	.11*	n.s.	n.s.	.18**
Prior relationship	−.20****	−.25****	−.25****	−.15**
CG age	−.23****	−.29***	−.16****	−.22****
CG education	.10**	ns	.10**	ns
IVs	Womene	Menf	Spousesg	Adult childrenh
B. Lower self-esteem with burden
CG chronic conditions	.16****	.18**	.19****	ns
Prior relationship	ns	−.15*	−.11**	ns
CG gender/relation	.15*** wife worse	ns	ns	−.16** son worse
CG education	−.15****	−.13*	−.14**	−.18**
CG income	−.15***	ns	−.13**	−.15*
Burden	.21****	.20**	.18****	.24***

a R 2 = .24; df = 7 and 561; p < .000. b R 2 = .27; df = 7 and 249. p < .000. c R 2 = .24; df = 7 and 575; p < .000. d R 2 = .25; df = 7 and 235; p < .000. e R 2 = .17; df = 6 and 487; p < .000. f R 2 = .17; df = 6 and 219; p < .000. g R 2 = .16; df = 6 and 513; p < .000. h R 2 = .19; df = 6 and 193; p < .000.

*p < .05; **p < .01; ***p < .001; ****p < .000.

Other factors are significant selectively across the groups. Looking first at women, those with more chronic conditions have greater burden and lower self-esteem. Self-rated health, however, is related to neither. Those with more income also have better self-esteem. Even when controlling for numerous factors, relation is still significantly related to self-esteem. Among women caregivers, wives have lower self-esteem than do daughters. However, more education is related to higher burden among women. The most variance is explained in burden (24%) less in self-esteem (17%).

Men are very different. No additional factors to those noted previously are related to burden but those with more chronic conditions and husbands and sons who had a more distant prior relationship with the care recipient tend to have lower self-esteem. The most variance is explained in burden (27%) less in self-esteem (17%), similar to women but with fewer correlates.

Among spouses, those with more education tend to be more burdened; those with more income tend to have better self-esteem. Spouses with more chronic conditions and those who had a more distant prior relationship with the care recipient tend to have worse self-esteem. Once again, the most variance is explained in burden (24%) and less in self-esteem (16%).

Among adult children, the more chronic conditions they have, the more burden they experience. The higher their income, the higher their self-esteem. Even when controlling for numerous factors, among adult children, sons still have significantly worse self-esteem than do daughters. Self-esteem is unrelated to whether they are caring for their mother or their father. The most variance is explained in burden (25%) less in self-esteem (19%).

That is, when multivariate analyses are examined, the intersection of relation and gender emerge only in some circumstances. Relation is not significant for burden but this does not obviate the bivariate relationships; rather, the relation differences are explained by factors such as hours of caregiving, prior relationship with the care recipient, and age of the caregiver. The findings also reveal that, in addition to the several significant factors such as their burden, their own chronic conditions, education, and income, wives still reveal lower self-esteem than daughters and sons than daughters—that is, daughters emerge has having particularly high self-esteem. There are in other words, factors not taken into account here that explain this gender/relation intersection. Taken together, these findings point to wives as especially vulnerable and daughters as doing especially well.

Discussion and Conclusions

Drawing on an intersectionality framework, this article queried the intersection of gender and relation for caregivers' burden and self-esteem. The results presented here, especially the differences between wives and daughters for both burden and self-esteem suggest that measurement matters and, at least for some subgroups of caregivers to older adults with dementia, lower burden is not necessarily related to lower self-esteem. It is among wives in the sample studied here but not among daughters. Such findings suggest it is important to include more than one aspect of broader concepts such as quality of life and well-being and at minimum, not to generalize from measuring one aspect to another. These findings point to the fact that positive dimensions are not necessarily related to negative ones, that cognitive well-being is not necessarily related to affective well-being.

The multivariate analyses nuance the bivariate findings, revealing how examining gender alone or relation alone masks differences that emerge when the intersection is the focus of attention. They indicate wives’ worse burden than husbands’ can be explained in terms of other factors such as more hours of caregiving and feeling there was a distant prior relationship with the care recipient. Wives also experience lower self-esteem than daughters and sons experience lower self-esteem than daughters even after accounting for many other factors, suggesting there is something within these relationships that are not being tapped by the control variables measured here. Wives, a particular intersection of gender and relation emerge as most vulnerable on role-specific burden and a more generalized concept of self-esteem. They differ from daughters, who experience higher burden but also have significantly higher self-esteem. Wives are an example of where a positive cognitive dimension of well-being is correlated with a negative affective dimension, suggesting that the demands of caregiving infect their perceptions of self, their identities. Immersed in a couple’s identity, burden within that relationship appears associated with how they value themselves. As the most vulnerable group of the four examined here, wives should be viewed by the care system, when they come in contact, as potentially susceptible and in need of greater and/or different kinds of support than husbands, sons, or daughters. When they are feeling especially burdened, this is related to their self-esteem. Two longitudinal studies have concluded that low self-esteem can lead to depression (Orth et al., 2009), confirmed in Sowislo and Orth’s (2013) review of the literature. Daughters, on the other hand, find caregiving burdensome but this does not generalize to how they feel about themselves, suggesting it is not as salient for their perceptions of self and identity. This would be worth pursuing, perhaps using in-depth qualitative methodology to learn how daughters protect their self-esteem from the demands of caregiving. It could well be related to their identities being attached much less to the caregiving than is true of wives.

Some of the correlates confirm past research (hours of caregiving as a strong correlate of burden among all groups and burden as a predictor of self-esteem). Of note, here are the differential factors that emerge among the groups, indicating for example, that men caregivers are distinctive. In terms of self-esteem, only among men is income irrelevant. For all other groups, the more the income, the better the self-esteem. There was a good range of values on income in this sample; it could be that income is so salient for men that irrespective of their level, it has the same impact on their sense of worthiness. The receipt of paid services is unrelated.

Of note are the two control variables that are related in opposite directions for burden and for self-esteem. Higher education and more income are related to more burden and better self-esteem. Whether this is due to those with higher levels of education preferring to be spending their time elsewhere, which they can afford, or from higher expectations of themselves in the provision of care, resulting in more burden, is unknown. Furthermore, both income and education emerge independently while controlling for the other; that is, one is not a proxy for the other in this instance. Rather, education seems to protect caregivers from the burden of caregiving generalizing to broader concepts of their identities, over and above income. The higher levels of education allow the individual to maintain a higher sense of their own worthiness despite the burden they are experiencing.

These findings support Garlo, O’Leary, Van Ness, and Fried's (2010) research that caregiver characteristics are more strongly related to high burden than patient characteristics. Our findings add that this is also true of a positive cognitive dimension of well-being, namely self-esteem. It is notable that not one care recipient characteristic, including any of the disease characteristics (such as stage of disease or behavioral problems) was related to either burden or self-esteem. Such findings suggest support for caregivers should be targeted to their contexts rather than targeted to the person with dementia per se.

The findings reported here also point to important avenues for future research, including an understanding of the apparent protective factor of education (in addition to income) in preventing the burden of caregiving from dampening caregivers’ self-esteem; what the differential factors are for men’s burden that have not been uncovered here and why income, supposedly so important in men’s lives is not a significant correlate yet it is for women, spouses, and adult children.

This study has its limitations. The sample was not random, so generalizations cannot be drawn to all caregivers or even all caregivers to those with dementia. However, to the extent that the findings confirm much past research, they suggest this sample may not be so different in terms of their burden and self-esteem. This sample included only caregivers to those with dementia and prescribed ChEI, covered by the provincial PharmaCare system; how this biased the findings is unknown, although the research on the clinical effects of these medications suggests any effectiveness is small and no subsample of patients has been identified as benefitting more than others. That is, the drugs tend be prescribed without targeting subpopulations of persons with dementia (Chappell et al., 2014), so it is not clear that any bias was introduced from this source. The sample was virtually all Caucasian and limited to spouse and adult–child caregivers; it is not known whether similar findings would accrue among different populations. Also, some of the measures lack precision, such as “hours of caregiving” that refers only to the caregivers’ subjective interpretation of what caregiving entails for them or prior relationship with the care recipient, captured here in only a single question.

Despite its shortcomings, the analyses presented point to the fruitfulness of the intersectionality framework, not only for studying traditional disadvantaged statuses such as gender, social class, and race but also of its potential for the study of the co-occupancy of virtually any status, not only those considered disadvantageous. Indeed, it can be argued that relation between caregiver and care recipient is not necessarily a disadvantaged status. Yet the analyses presented here reveal the importance of the intersection of relation and gender while also pointing to the complexity of both intersectionality and the multifaceted concept of well-being.