Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2022 Nov 21;78(Suppl 1):S15–S26. doi: 10.1093/geronb/gbac145

Validation of a Measure of Role Overload and Gains for End-of-Life Dementia Caregivers

Shelbie G Turner 1,2,✉, Fayron Epps 2, Minghui Li 3, Amanda N Leggett 4,3, Mengyao Hu 5
Editor: Toni C Antonucci
PMCID: PMC10010474  PMID: 36409299

Abstract

Objectives

Caregiving stress process models suggest that heterogeneous contexts differentially contribute to caregivers’ experiences of role overload and gains. End-of-life (EOL) caregivers, especially EOL dementia caregivers, facing unique challenges and care tasks, may experience role overload and gains in different ways than other caregivers. This study evaluates measurement invariance of role overload and gains between EOL caregivers and non-EOL caregivers and between EOL dementia and EOL non-dementia caregivers.

Methods

We utilized role gains and overload data from 1,859 family caregivers who participated in Round 7 of the National Study of Caregiving. We ran confirmatory factor analyses to investigate the factorial structure across all caregivers and then examined the structure’s configural, metric, and scalar invariance between (a) EOL caregivers and non-EOL caregivers and (b) EOL dementia and EOL non-dementia caregivers.

Results

Across the entire sample, the two-factor overload and gains model had good fit (χ 2(19) = 121.37, p < .0001; RMSEA = .053, 90% CI = [.044, .062]; CFI = .954; TLI = .932). Tests of invariance comparing EOL caregivers to non-EOL caregivers and EOL dementia caregivers to EOL non-dementia caregivers maintained configural, metric, and partial scalar invariance. Latent mean comparisons revealed that EOL caregivers had higher role overload (p = .0002), but no different role gains (p = .45), than non-EOL caregivers. Likewise, EOL dementia caregivers had higher role overload (p = .05), but no different role gains (p = .42), than EOL non-dementia caregivers.

Discussion

Results offer both a deeper theoretical understanding of end-of-life dementia caregivers’ experiences of role overload and gains, and a practical tool to measure those experiences.

Keywords: Dementia, End-of-life caregiving, Measurement invariance, National Study of Caregiving, Survey validation

For decades, researchers have understood that in addition to reporting role overload from caregiving, family caregivers also report positive appraisals of their caregiving experiences, or role gains (Cohen et al., 2002; Epps, 2014; Kramer, 1997; Lawton et al., 1991; Zarit, 2012). Alongside causing stress (Son et al., 2007; Tsai, 2003) and burden (Zarit, 2008), caregiving offers opportunities for personal growth (Skaff & Pearlin, 1992) by heightened self-esteem, self-confidence, and self-efficacy. Both role overload and role gains impact caregivers’ well-being. Greater role overload is associated with caregiver psychological distress (Nah et al., 2022). Role overload negatively impacts sleep quality (Liang et al., 2020), work (Gordon et al., 2012; Wang et al., 2011), and leisure activities that would otherwise give caregivers a break from caregiving (Bastawrous et al., 2015). On the contrary, positive caregiving appraisals are associated with caregivers feeling as though caring is an opportunity to give back, connect with others, and develop a closer relationship and commitment to the care recipient (Peacock et al., 2010). Perceived gains from caregiving can buffer the impact of negative appraisals by increasing caregivers’ intrinsic motivation to offer care (Rapp & Chao, 2010; Yu et al., 2018), and caregivers who report positive appraisals of caregiving have personal benefits even after their caregiving has ended (Bangerter et al. 2019). Notably, some research suggests that positive appraisals are actually associated with a variety of adverse health consequences, such as more depressive symptoms and more physical complaints (Pendergrass et al, 2019). Such a connection between positive appraisals of caregiving and adverse health consequences is likely because more intense caregiving is actually associated with greater subjective caregiving gains (Quinn et al., 2012).

In addition to role overload and gains being constructs that predict caregiver well-being, they are outcomes that are dependent on the caregiving context. Pearlin et al.’s (1990) caregiver stress process model outlines contextual factors that impact caregivers’ experiences of role overload and gains. Such theoretical propositions have borne out in the empirical analysis. Caregivers who provide more medication management (Polenick et al., 2020) and have more financial strain due to caregiving (Liu et al., 2019) report more role overload. Additionally, behavioral and psychological symptoms experienced by persons living with dementia often lead to increased caregiver burden (Chiao et al., 2015). Regarding role gains, Quinn et al. (2012) found that caregivers who had a better relationship with the care recipient, who were more religious, and who provided more hours of care reported more personal meaning from caregiving. Carbonneau et al.’s (2010) conceptual model of positive aspects of caregiving outlines caregivers’ self-efficacy and day-to-day experiences of enrichment in the role (i.e., small events or occurrences such as the care recipient smiling) as pivotal to predicting how positively a caregiver appraises their role as a caregiver.

Indeed, theoretical and empirical studies support the notion that caregiving role overload and gains are dramatically affected by the caregiving context. The estimated 2.3 million end-of-life (EOL) caregivers in the United States (Ornstein et al., 2017) have unique experiences of the factors presented in Pearlin et al.’s (1990) model. EOL caregivers often provide care that is more intense, which is associated with increased caregiver burden (Ornstein et al., 2017). EOL caregivers report providing care for more hours a day than non-EOL caregivers, and they report more care-related challenges (Ornstein et al., 2017). EOL caregivers also report managing additional stress and guilt when bringing in supplemental care support—or even moving the care recipient to a nursing home or skilled care facility—as EOL care gets more challenging (Martz & Morse, 2017). Moreover, EOL caregivers have to emotionally prepare for death and grief (i.e., Durepos et al., 2019) in ways that affect—and are impacted by—their experience of caregiver role loss (Supiano et al., 2022).

In Pearlin et al.’s (1990) stress process model, care recipients’ cognitive status—along with behavioral and psychological symptoms and activities of daily living and instrumental activities of daily living dependencies—affect their caregivers’ experiences of role overload and gain. Forty-seven percent of people receiving EOL family care have probable dementia (Ornstein et al., 2017), meaning nearly half of EOL caregivers are EOL dementia caregivers. EOL caregivers who are caring for a person with dementia have two times greater odds of experiencing caregiving strain than EOL non-dementia caregivers (Vick et al., 2019). Many caregivers report a lack of knowledge on late-stage dementia and, thereby, confusion over how to make informed decisions about long-term care plans and medical treatment (Broady et al., 2018). This lack of knowledge also creates confusion and uncertainty that leads to greater caregiver stress (Broady et al., 2018). Moreover, EOL dementia caregivers have more difficulty navigating healthcare support services than EOL non-dementia caregivers (Davies et al., 2014). For example, people with dementia are less likely to receive palliative care before their death than people without dementia (Shah et al., 2016). EOL dementia caregivers report additional challenges from having communication barriers as the person with dementia loses their ability to communicate (Lawrence et al., 2011), and care recipient communication problems are associated with increased caregiver burden (Savundranayagam et al., 2005). EOL dementia caregivers’ grief and coping process are also different from EOL non-dementia caregivers as they have to manage ambiguous loss, whereby the psychological, emotional, and personality characteristics of the care recipient are perceived to be lost faster or more severely than the physical characteristics (Blieszner et al., 2007; Noyes et al., 2010).

Ultimately, EOL caregiving may be fundamentally different from non-EOL caregiving. Moreover, among EOL caregivers, those who care for someone with dementia at the EOL may have fundamentally different caregiving experiences. As such, EOL and, more specifically, EOL dementia caregivers may not necessarily report more negative appraisals and less positive appraisals; rather, the factors they use to inform their appraisals may simply be different from the factors that non-EOL and EOL non-dementia caregivers use. Some researchers have even argued that caregiver burden, for example, be “redefined” for EOL caregivers because of how different EOL caregiving can be from non-EOL caregiving (Choi & Seo, 2019). Precisely and accurately understanding EOL dementia caregivers’ experiences rests on ensuring data collection instruments are appropriate for the unique nature of EOL dementia caregiving.

Measurement Invariance

Measurement invariance analyses test whether a latent factor structure varies between groups or across time and offers psychometric validation to quantitative surveys—that is, measurement invariance tests whether the same survey questions measure the same construct in the same way for different groups. Tests of measurement invariance are fundamental to psychosocial research methods (Putnik & Bornstein, 2016). Should measures not be invariant across certain groups, they should not be used to compare groups. Tests of measurement invariance can reveal potential differences in experiences of gains and overload between EOL and non-EOL caregivers. Therefore, they offer a deeper theoretical understanding of the EOL caregiving experience and methodological insight so that researchers can adjust their measures in response to the uniqueness of EOL caregiving.

With the increasingly diverse older adult and caregiver population, there is a crucial need to evaluate measurement invariance across groups for existing survey questions and develop items that work equally well for various population groups. Measurement invariance studies help researchers tailor quantitative measures to empirically established heterogeneity, and they help capture unrealized heterogeneity amongst the older adult and caregiver population. As such, researchers are calling for more gerontological measurement invariance studies to equip quantitative measures for the increasingly diverse older adult population (Cabrera & Gallego-Alberto, 2020). Examples of tests of measurement invariance in gerontological literature include well-being between young-old and old-old adults (Maitland et al., 2001), subjective well-being between White, Black, and Hispanic older adults (Kim et al., 2020), self-perceptions of aging between older men and older women (Turner et al., 2021), and cognitive abilities across ethnicity, gender, and time (Blankson & McArdel, 2015). In caregiving literature specifically, measurement invariance tests have supported scale validation across demographic factors. For example, Ayalon (2015) found that a seven-item scale of elder neglect had configural, metric, and scalar invariance across age, gender, and education.

The Present Study

In this study, we used the National Study of Caregiving (NSOC), which is a supplemental study to the nationally representative National Health and Aging Trends Study (NHATS) in the United States, to bring analyses of measurement invariance to scholarship on EOL caregiving. Together, NHATS and NSOC provides a nationally representative survey of older adults and their unpaid caregivers residing in the United States. Caregivers to NHATS older adults who participate in the NSOC answer four questions about role gains and four questions about role overload. Though the items are commonly used as composite scales in studies (e.g., Liang et al., 2020; Polenick et al., 2020; Pristavec, 2019), to our knowledge the questions have yet to be validated as a full eight-item two-factor measure, nor has there been any analysis about how the constructs may be similar or different between EOL and non-EOL caregivers, or between EOL dementia and EOL non-dementia caregivers. To offer such validation, we ran confirmatory factor analyses to establish the eight-item two-factor solution across all caregivers. We then tested configural, metric, and scalar invariance to analyze how the solution may differ between (a) EOL and non-EOL caregivers and (b) EOL dementia and EOL non-dementia caregivers. Doing so allowed us to gain a deeper theoretical understanding of EOL and EOL dementia caregivers’ experiences of role overload and gains, and a practical tool to measure those experiences.

Method

Study Procedure and Participants

Data came from participants of Round 7 (2017) of the NSOC, which is a companion study with the 2017 NHATS. NHATS includes a nationally representative sample of Medicare beneficiaries who are 65 years of age and older. During the NHATS 2017 in-person interview participants were asked if a family member or nonfamily helper provided them with assistance on household chores, mobility, or self-care tasks. If so, NHATS participants provided the names of individuals who helped them with each of these tasks. If more than five names were listed, five were randomly selected for inclusion in NSOC. If NHATS participants had probable dementia (see “Dementia Versus Non-Dementia” section below for information about how participants were identified as having dementia), proxy respondents who were familiar with the participant completed the interview.

Measures

Caregiving role gains

Caregiving role gains are the positive feelings derived from the care role. The four caregiving gains questions are as follows: (a) Helping the care recipient has made you feel more confident in your abilities, (b) Helping the care recipient has taught you to deal with difficult situations, (c) Helping the care recipient has brought you closer to him/her, and (d) Helping the care recipient has given you satisfaction that he/she is/was well cared for. Caregivers answered each question on a 3-point Likert scale from 1 (not at all) to 3 (very much so), with higher scores indicating higher caregiving role gains. Across the entire sample, the role gains scale had a Cronbach’s alpha of .72.

Caregiving role overload

Caregiving role overload is the negative feeling derived from the care role. Four caregiving role overload questions are as follows: (a) You are exhausted when you go to bed at night, (b) You have more things to do than you can handle, (c) You don’t have time for yourself, and (d) As soon as you get a routine going, the care recipient’s needs change. Caregivers answered each question on a 3-point Likert scale from 1 (not at all) to 3 (very much so), with higher scores indicating higher caregiving role overload. Across the entire sample, the role overload scale had a Cronbach’s alpha of.76.

End-of-life versus non-end-of-life

NHATS collects yearly data on participants’ death, if applicable. To identify end-of-life caregivers in the NSOC dataset, we looked prospectively at Round 8 of NHATS data to identify whether the care recipient died at any point in the year after the caregiver completed NSOC Round 7 of data collection, as has been done in a prior study of EOL caregiving (Ornstein et al., 2017). If the care recipient died in the following year, we coded the NSOC caregiver as an end-of-life caregiver (1); if not, we coded them as a non-end-of-life caregiver (0). This prospective identification of EOL caregivers was required because researchers did not ask role gains questions once the caregiver reported the care recipients’ death (though they did continue to ask role overload questions). We ran t-tests to evaluate the extent to which EOL caregivers were different than non-EOL caregivers and to help ensure that the way we operationalized EOL caregivers captured some of the known differences between EOL caregivers and non-EOL caregivers. We found that EOL caregivers in our sample cared for significantly more days per week that non-EOL caregivers (EOL: M = 6.13, SD = 1.61; non-EOL: M = 5.50, SD = 2.03; p = .0002). They also cared for significantly more hours per day than non-EOL caregivers (EOL: M = 5.39, SD = 5.34; Non-EOL: M = 4.28, SD = 4.25; p = .01). Both statistics reinforce EOL caregivers as operationalized in our study as different than non-EOL caregivers.

Dementia versus non-dementia

To identify dementia caregivers in the NSOC data set, we utilized NHATS Round 7 data. NHATS labels participants as having “probable dementia,” “possible dementia,” or no dementia,” via self or proxy report of a doctor’s diagnosis of dementia, a score of two or more on the eight-item AD8 Dementia Screening Interview, or cognitive tests administered as a part of NHATS (see Kasper et al., 2013 for more information). If a care recipient had probable dementia, we coded the caregiver as a dementia caregiver (1); if a care recipient did not have dementia, we coded the caregiver as a non-dementia caregiver (0). We excluded participants who had “possible dementia.”

Analytic Strategy

We first ran a confirmatory factor analysis across the entire sample (n = 1,859) to establish the two-factor gains and overload structure. We then tested configural invariance in the factor structure between (a) EOL caregivers and non-EOL caregivers and (b) EOL dementia caregivers and EOL non-dementia caregivers. That is, we analyzed whether the general factor structure of a construct is the same across groups and allow thresholds and factor loadings to be freely estimated in each group. We assessed global model fit using a chi-square test of overall fit, the root mean square error of approximation (RMSEA; > .08 = poor fit, .05–.08 = acceptable fit, .00–.05 = close fit), the comparative fit index (CFI; < .90 = poor fit, .90–.95 = acceptable fit, >.95 = good fit), and the Tucker–Lewis index (TLI; <.90 = poor fit, .90–.95 = acceptable fit, >.95 = good fit). We treated all variables as ordinal, thereby comparing thresholds rather than means. After establishing invariance, we tested convergent validity (Supplementary Table 1).

Finally, we tested for metric/weak invariance (whether the factor loadings were equivalent between groups) and scalar/strong invariance (whether the item thresholds were equivalent between groups). After moving from both configural to metric and from metric to scalar, we compared models by analyzing the change in model fit statistics using DIFFTEST command in MPlus. If the change in chi-square statistics was not significant, we concluded that the two models fit the data similarly well and thus considered the model to be metric-invariant (when compared with the configural model) or scalar-invariant (when compared to the scalar model) between the groups. We also tested for partial scalar invariance for both the EOL and non-EOL comparison and the EOL dementia and EOL non-dementia comparison; we freely estimated thresholds for individual items based on the thresholds of the metric model.

We ran demographic and descriptive statistics in SAS Version 9.4. We conducted all confirmatory factor analyses in MPlus Version 8.5 using weighted least square mean and variance adjusted estimation to treat the indicators as ordinal. All analyses were weighted and account for the complex survey design features of NHATS and NSOC. NSOC includes caregivers of care recipients who are both community dwelling and those who live in residential care facilities (i.e., assisted living, nursing homes). For the full sample confirmatory factor analysis and the invariance testing comparing EOL caregivers to non-EOL caregivers, our sample only included caregivers of community-dwelling care recipients. For the invariance testing comparing EOL dementia caregivers and EOL non-dementia caregivers, our sample included all caregivers regardless of the residence of the care recipient in order to have a reasonable sample size.

Results

Sample Description

The sample included 1,634 non-EOL caregivers and 225 EOL caregivers to community-dwelling older adults. There were 280 EOL caregivers to older adults in all residential statuses, of whom 166 were providing care for a person living with dementia. Full sample description can be found in Table 1.

Table 1.

Sample Description

End-of-life compared with non-end-of-life n Mean age % Female Race/ethnicitya % <4-year degree Mean number of caregiving days/week Mean number of caregiving hours/day % dementia caregivers
End-of-life caregivers 225b 60.56 65.50 76.25% White, 12.63% Black, 6.09% Hispanic, 0.07% more than one race, 3.07% Otherc 71.45 5.81 4.89 44.46
Non-end-of-life caregivers 1,634 59.18 64.09 68.85% White, 15.02% Black, 9.46% Hispanic, 0.17% more than one race, 3.08% Otherc 74.23 5.22 3.92 22.86
Among end-of-life caregivers, dementia compared to non-dementiad
End-of-life dementia caregivers 166 62.13 61.06 80.05% White, 8.4% Black, 7.29% Hispanic, 0.10 more than one race, 1.55% Otherc 66.82 5.64 4.65 100
End-of-life non-dementia caregivers 114 60.35 68.07 77.89% White, 7.94% Black, 2.47% Hispanic, 0% more than one race, 3.87% Otherc 70.21 5.41 4.44 0
Open in a new tab

Notes: Demographic statistics computed via SAS PROC SURVEYFREQ AND SAS PROC SURVEYMEANS in order weigh data to account for the complex survey design.

aPercentages may not add up to 100% because some participants did not offer their race or ethnicity.

bSample only included caregivers caring for a community-dwelling care recipient; thus, n = 225.

cOther included American Indian/Asian/Native Hawaiian/Pacific Islander/other specific caregivers.

dEnd-of-life dementia and end-of-life non-dementia caregiver samples include caregivers caring for a care recipient from all residential statuses (not just community-dwelling care recipients). Thus, n = 280.

Confirmatory Factor Analyses

Across the entire community-dwelling sample (n = 1,859), the two-factor model with caregiver gains and caregiver overload factors had good fit (χ 2(19) = 121.37, p < .0001; RMSEA = .053, 90% CI = [.044, .062]; CFI = .954; TLI = .932).

Invariance testing: EOL versus non-EOL caregivers (community dwelling)

Configural.

—The two-factor configural model comparing EOL caregivers to non-EOL caregivers had good fit (χ 2(38) = 129.79, p < .0001; RMSEA = .051, 90% CI = [.042, .061]; CFI = .963; TLI = .946; Table 4). Item mean scores are available in Table 2, and factor loadings are available in Table 3. The role gains scale had a Cronbach’s alpha of .69 for EOL caregivers and .73 for non-EOL caregivers. The role overload scale had a Cronbach’s alpha of .79 for EOL caregivers and .73 for non-EOL caregivers.

Table 4.

Model Fit Statistics

Model χ 2 (df), p value Δχ 2 (df)2, p value RMSEA (90% CI) CFI TLI
EOL versus non-EOL
 Configural invariance (factor structure) 129.786 (38), <.0001 — .051 (.042, .061) .963 .946
 Metric invariance (factor loadings) 129.296 (44), <.0001 9.373 (6), .154 .046 (.037, .055) .966 .956
 Scalar invariance (item thresholds) 152.521 (58), <.0001 35.147 (14), .0014 .042 (.034, .050) .962 .963
 Partial scalar invariance 141.230 (55), <.0001 19.352 (11), .06 .041 (.033, .050) .965 .965
EOL dementia versus EOL non-dementia
 Configural invariance (factor structure) 41.67 (38), .31 — .03 (.000, .068) 1.0 1.0
 Metric invariance (factor loadings) 48.061 (44), .31 7.178 (6), .305 .026 (.000, .065) 1.0 1.0
 Scalar invariance (item thresholds) 69.29 (58), .15 25.33 (14), .032 .038 (.000, .068) 1.0 1.0
 Partial scalar invariance 64.64 (57), .23 18.86 (13), .13 .031 (.000, .064) .99 .99
Open in a new tab

Notes: Δχ 2 = change in chi-square; ΔCFI = change in CFI value; CFI = comparative fit index; CI = confidence interval; EOL = end-of-life; RMSEA = root mean square error of approximation; TLI = Tucker–Lewis index.

Table 2.

Item Mean Scores

Mean Score
Factor Item EOL Non-EOL EOL dementia
EOL non-dementia
Role gains Helping the care recipient has made me more confident in my abilities 2.34 2.36 2.25 2.22
Helping the care recipient has taught you to deal with difficult situations 2.57 2.46 2.63 2.38
Helping the care recipient has brought you closer to him/her 2.65 2.66 2.56 2.65
Helping the care recipient has given you satisfaction that he/she is/was well cared for 2.86 2.87 2.83 2.81
Role overload You are exhausted when you go to bed at night 1.72 1.59 1.81 1.49
You have more things to do than you can handle 1.64 1.49 1.68 1.58
You don’t have time for yourself 1.73 1.58 1.73 1.72
As soon as you get a routine going, the care recipient’s needs change 1.49 1.29 1.55 1.43
Open in a new tab

Notes: EOL, end-of-life. Response items ranged from 1 (not so much) to 3 (very much).

Table 3.

Factor Loadings

Factor loadings
EOL Non-EOL EOL dementia EOL non-dementia
Factor 1: role gains
 Helping the care recipient has made me more confident in my abilities .86 .83 .86 .79
 Helping the care recipient has taught you to deal with difficult situations .74 .72 .80 .81
 Helping the care recipient has brought you closer to him/her .77 .77 .75 .79
 Helping the care recipient has given you satisfaction that he/she is/was well cared for .73 .72 .78 .56
Factor 2: role overload
 You are exhausted when you go to bed at night .70 .73 .67 .75
 You have more things to do than you can handle .90 .93 1.00 .90
 You don’t have time for yourself .89 .68 .74 .96
 As soon as you get a routine going, the care recipient’s needs change .70 .55 .60 .79
Open in a new tab

Notes: EOL = end-of-life. EOL caregivers compared with non-EOL caregivers model fit statistics: χ 2(38) = 129.786, p< .0001; RMSEA = .051, 90% CI = [.042, .061]; CFI = .963; TLI = .946. EOL dementia caregivers compared with EOL non-dementia caregivers model fit statistics: χ 2(38) = 41.67, p = .31; RMSEA = .03, 90% CI = [.000, .068]; CFI = 1.0; TLI = 1.0. CFI = comparative fit index; CI = confidence interval; EOL = end-of-life; RMSEA = root mean square error of approximation; TLI = Tucker–Lewis index.

Metric

.—The two-factor metric model comparing EOL caregivers to non-EOL caregivers had good fit (χ 2(44) = 129.30, p < .0001; Δχ 2(6) = 9.373, p = .15; RMSEA = .046, 90% CI = [.037, .055]; CFI = .966; TLI = .956; Table 4). The DIFFTEST change in chi-square was not significant, suggesting that the metric model was no different than the configural model and thus, the model maintained metric invariance.

Scalar

.—The two-factor scalar model comparing EOL caregivers to non-EOL caregivers had good fit (χ 2(58) = 152.521, p < .0001; Δχ 2(14) = 35.147, p = .0014; RMSEA = .042, 90% CI = [.034, .050]; CFI = .962; TLI = .963; Table 4). The DIFFTEST change in chi-square was significant, suggesting that the scalar model was significantly different than the metric model and thus, the model did not maintain scalar invariance.

Partial scalar

.—Because the model comparing EOL caregivers to non-EOL caregivers was not scalar invariant, we tested partial scalar invariance. Partial scalar invariance testing offers insight into item-by-item invariance, and without at least partial scalar invariance, means should not be compared. To test partial scalar invariance, we went back to the metric invariance results, noting the differences in item thresholds between EOL and non-EOL caregivers (thresholds are available in Supplementary Table 2). Then, in the scalar invariance code, we allowed the threshold with the highest difference between groups to be freely estimated. We freely estimated the thresholds with the next highest differences between groups, one-by-one, until the model had good fit. We reached partial scalar invariance after holding the thresholds with the three highest differences between groups. The associated items were as follows: Helping the care recipient has taught you to deal with difficult situations (gains factor), You have more things to do than you can handle (overload factor), and As soon as you get a routine going, the care recipient’s needs change (overload factor). Once we held these three thresholds constant, the two-factor model assessing partial scalar invariance between EOL caregivers and non-EOL caregivers had good fit (χ 2(55) = 141.230, p < .0001; ∆χ 2(11) = 19.352, p = .055; RMSEA = .041, 90% CI = [.033, .050]; CFI = .965; TLI = .965; Table 4). The DIFFTEST change in chi-square was not significant (p = .06), suggesting that the model maintained partial scalar invariance.

After establishing partial scalar invariance, we tested group mean differences and found that EOL caregivers had significantly higher role overload (EOL: M = 6.56, SD = 2.36; non-EOL: M = 5.92, SD = 2.06; p = .0002), but no different role gains (EOL: M = 10.38, SD = 1.73; Non-EOL: M = 10.29, SD = 1.90; p = .45), than non-EOL caregivers.

Invariance testing: EOL dementia caregivers versus EOL non-dementia caregivers (all residential statuses)

Configural

.—The two-factor configural model comparing EOL dementia caregivers to EOL non-dementia caregivers had good fit (χ 2(38) = 41.67, p = .31; RMSEA = .03, 90% CI = [.000, .068]; CFI = 1.0; TLI = 1.0; Table 4). Item mean scores are available in Table 2, and factor loadings are available in Table 3. The role gains scale had a Cronbach’s alpha of .73 for EOL dementia caregivers and .67 for EOL non-dementia caregivers. The role overload scale had a Cronbach’s alpha of .84 for EOL dementia caregivers and .71 for EOL non-dementia caregivers.

Metric

.—The two-factor configural model comparing EOL dementia caregivers to EOL non-dementia caregivers had good fit (χ 2(44) = 48.061, p = 0.31; ∆χ 2(6) = 7.178, p =.305; RMSEA = .026, 90% CI = [.000, .065]; CFI = 1.0; TLI = 1.0; Table 4). The DIFFTEST change in chi-square was not significant, suggesting that the model maintained metric invariance.

Scalar

.—The two-factor scalar model comparing EOL dementia caregivers to EOL non-dementia caregivers had good fit (χ 2(58) = 69.29, p = 0.15; ∆χ 2(14) = 25.33, p = .032; RMSEA = .038, 90% CI = [.000, .068]; CFI = .99; TLI = .99; Table 4). The DIFFTEST change in chi-square was significant, suggesting that the model was not scalar invariant.

Partial scalar.—

Because the model comparing EOL dementia caregivers to EOL non-dementia caregivers was not scalar-invariant, we tested partial scalar invariance. Just as with our partial invariance testing for the model comparing EOL caregivers to non-EOL caregivers, to test partial scalar invariance, we went back to the metric invariance results, noting the differences in item thresholds between EOL dementia and EOL non-dementia caregivers (thresholds are available in Supplementary Table 3). Then, in the scalar invariance code, we allowed the threshold with the highest difference between groups to be freely estimated. We reached partial scalar invariance after holding the threshold with the highest difference between groups. The associated item was as follows: Helping the care recipient has brought you closer to him/her (gains factor). The two-factor model assessing partial scalar invariance between EOL dementia caregivers and EOL non-dementia caregivers had good fit (χ 2(57) = 64.34, p = .23; ∆χ 2(13) = 18.86, p = .13; RMSEA = .031, 90% CI = [.000, .064]; CFI = .99; TLI = .99; Table 4). The DIFFTEST change in chi-square was not significant (p = .13), suggesting that the model maintained partial scalar invariance.

After establishing partial scalar invariance, we tested group mean differences and found that EOL dementia caregivers had significantly higher role overload (dementia: M = 6.75, SD = 2.50; non-dementia: M = 6.17, SD = 2.14; p = .05), but no different role gains (dementia: M = 10.24, SD = 1.79; non-dementia: M = 10.05, SD = 1.86; p = .42), than EOL non-dementia caregivers.

Discussion

Our study offers validation of a measure of caregiving role-related gains and overload for EOL and EOL dementia caregivers. Results from confirmatory factor analyses supported an eight-item, two-factor solution, with four items constructing the gains factor and four items constructing the overload factor. Given how salient positive and negative appraisals of caregiving are to well-being outcomes and how challenging EOL caregiving can be, especially EOL dementia caregiving, validation of such a measure was overdue and necessary.

Researchers can confidently use the measure to evaluate caregiving role-related gains and overload among EOL and EOL dementia caregivers either in the nationally representative NSOC dataset or with their own samples. Additionally, practitioners and clinicians working directly with dementia caregivers can use the measure to support caregivers’ well-being. For example, researchers are working to design and evaluate interventions aimed at improving caregivers subjective appraisals of caregiving (e.g., Wharton et al., 2019), and the measure we validated here could be used as a short, convenient tool to assess the effectiveness of these programs. Moreover, clinicians working directly with people with dementia could use the measure in their efforts to support a family-centered approach to dementia care, especially EOL dementia care. Indeed, research suggests that caregivers’ health impacts care recipients’ health (Amjad et al., 2021; Schulz et al., 2021), further underscoring the import of considering caregivers’ well-being when designing care plans for people with dementia.

Beyond the practical implications of offering a tool to measure caregiving role gains and overload, our study also offers a deeper theoretical understanding of caregiving role gains and overload, and how EOL and non-EOL caregivers may experience role gains and overload differently. Namely, our study also adds to the longstanding theoretical proposition that caregivers have both negative and positive appraisals of their experiences (e.g., Pearlin et al., 1990), and the growing understanding that negative and positive experiences of caregiving are two separate constructs rather than two ends of the same spectrum (e.g., Rapp & Chao, 2010). Indeed, in addition to the confirmed two-factor structure, the correlation between role gains and overload was not significant across the entire sample, nor for any of the groups tested in invariance testing (EOL caregivers, non-EOL caregivers, EOL dementia caregivers, and EOL non-dementia caregivers). That is, caregivers’ experiences of role gains were unrelated to their experiences of role overload. Moreover, EOL caregivers experienced higher role overload, but no different role gains, than non-EOL caregivers, as did EOL dementia caregivers when compared to EOL non-dementia caregivers. Such a finding is surprising given the research suggesting that more intense caregiving is associated with greater role gains (Quinn et al., 2012). It is important to note that the role gains scale had a Cronbach’s alpha of .69 for EOL caregivers, which may explain in some the lack of difference between role gains in EOL and non-EOL caregivers. Even still, the finding offers an additional example of the ways in which role gains and role overload operate independently.

It is possible that the independence of positive and neagative appraisals is a reflection of—and a contributor to—feelings of ambivalence or indifference (Ross et al., 2019). That is, caregivers can have high positive appraisals of caregiving alongside high negative appraisals of caregiving (ambivalence), and they can have low positive appraisals of caregiving alongside low negative appraisals of caregiving (indifference); the two do not necessarily work in diametric opposition. Much of the work on ambivalence as it pertains to caregiving is focused on intergenerational ambivalence (Lüscher & Pillemer, 1998), and the ambivalence that women feel towards social relationships (Connidis & McMullin, 2002). This existing work is helpful given nearly one-third of dementia caregivers are adult-daughters (Friedman et al., 2015; Kasper et al., 2015). However, our study supports the idea that other types of caregivers, and caregivers navigating other contextual dynamics such as EOL care, may also experience ambivalence. In so doing, our results here might inform future work that aims to reach a deeper theoretical understanding of both positive and negative subjective caregiving experiences as well as caregiving ambivalence.

Additionally, in the partial scalar invariance testing wherein we found certain items to be noninvariant offers theoretical insight into differences between EOL and non-EOL caregivers and, among EOL caregivers, between dementia and non-dementia caregievers. In comparing EOL caregivers to non-EOL caregivers, we found the following items to be noninvariant: You have more things to do than you can handle (overload factor) and As soon as you get a routine going, the care recipient’s needs change (overload factor). EOL caregivers treated those three items differently, an example of differential item functioning (McDonald, 1999). EOL caregivers report having to continue learning new skills in order to keep up with dying care recipients’ changing needs (Morgan et al., 2020; Stajduhar et al., 2013), which is likely related to their feelings of the care recipients’ needs changing and feeling as though they are learning to deal with difficult situations, and could contribute to their feelings of having more to do than they can handle. As such, there is empirical support for EOL caregivers being more likely to endorse those items.

Similarly, in the partial scalar invariance testing comparing EOL dementia caregivers to EOL non-dementia caregivers, we found the following item to be noninvariant: Helping the care recipient has brought you closer to him/her (gains factor). It is possible that the ambiguous loss EOL dementia caregivers report (feeling as though the care recipient is no longer the same person, despite their physical presence) alters the ways in which they feel close with care recipients. Similarly, care recipients with dementia may be unable to communicate their care needs, affecting their caregivers ability to feel close with the care recipient.

Though the differential item functioning seen in our analyses can be justified theoretically and empirically, further exploring differential item functioning is vital for both practical and theoretical advancements in EOL caregiving research, and is an opportunity for further quantitative analysis, such as Item Response Theory (Embretson & Reise, 2013), and qualitative analysis, such as cognitive interviewing to determine how EOL caregivers may view survey questions differently than non-EOL caregivers (Castillo-Díaz & Padilla, 2013). An avenue for future research also includes redesigning items that were noninvariant. Such efforts are especially important given the ongoing debates on the literature about the acceptable proportion of invariant to noninvariant items within a scale to declare partial scalar invariance. Such debates have led to a lack of de-facto guidelines for partial invariance and, therefore, limit our ability to declare the partial scalar invariance found in our study sufficient for the trustworthy comparison of latent means between groups. For now, as it pertains to the measure in its current form, we suggest researchers use the scale confidently with caregivers of all types, regardless of whether they are EOL or EOL dementia caregivers. But, we urge caution when comparing latent means between different types of caregivers until noninvariant items are redesigned. Notably, however, even redesigning measures may not lead to scalar invariance; indeed, some researchers consider scalar invariance an “unachieveable ideal” (Marsh et al., 2018, p. 524).

Limitations

Though our prospective identification of EOL caregivers was similar to other research (i.e., Ornstein et al., 2017), such a method for identifying EOL caregivers is a limitation of our study. Namely, it is possible that EOL caregivers did not know they were EOL caregivers while responding to the gains and overload questions in Round 7. Likewise, this method for identifying EOL caregivers may have included caregivers who were caring for a relative who died suddenly. As such, though included in statistical models as EOL caregivers, those caregivers may not have ever truly felt like an EOL caregiver or managed unique experiences related to EOL care. If a care recipient died suddenly, for example, the EOL caregiver may also not have experienced anticipatory grief. Because anticipatory grief is connected to caregivers’ subjective caregiving appraisals (Holley & Mast, 2009), there are likely real differences in EOL caregivers’ experiences depending on whether the care recipients’ death was anticipated or if they died unexpectedly and suddenly. As such, if our sample only included caregivers who were anticipating the care recipients death, the differences between EOL and non-EOL caregivers found in our study may be even more pronounced.

In this study, we use the language “role overload and gains” to construct positive and negative appraisals of caregiving. Doing so was informed by Pearlin et al.’s (1990) coinage of the four overload questions as indicating “role overload,” and of their caregiving stress process model, which suggests role-related appraisals of caregiving impact health outcomes. Yet, some questions are more related to role-related strain (“You don’t have time for yourself”) than others (“As soon as you get a routine going, the care recipient’s needs change”). The ways in which caregiving is a personal identity-related role that becomes embraced or rejected contribute to caregivers’ behavior, such as willingness to access support services, and thus caregiving outcomes (Eifert et al., 2015). Therefore, more research should unearth whether the scale we validate in this study is a scale of role overload, more specifically, or overload generally. Relatedly, there are multiple constructs used to measure negative appraisals of caregiving, including caregiver stress (Son et al., 2007; Tsai, 2003) and caregiver burden (Bédard et al., 2001; Zarit, 2008). Future research is necessary to unpack the extent to which caregiver role overload is a unique construct that is separate from these other negative appraisals of caregiving.

Lastly, our invariance testing between EOL dementia and EOL non-dementia caregivers relied on small sample sizes. The sample sizes suggested for confirmatory factor analysis are highly dependent on the nature of the analysis, such as factor loadings and the number of factors and items within them (MacCallum et al., 1999; Wolf et al., 2013). Wolf et al. (2013) suggest that samples can be as small as 200 with similar factor structures as ours presented in this manuscript, but their analysis did not account for dividing groups to conduct invariance testing. Thus, results from invariance testing between EOL dementia and EOL non-dementia, though informative, should be taken with caution.

Conclusion

Our study adds empirical support that the ever-increasing set of heterogeneous contexts of caregiving impact how caregivers appraise their experience of caregiving and, therefore, should be considered in measurement. By establishing the factor structure of an eight-item measure of caregivers’ role overload and gains, we offer a practical tool that researchers and practitioners can trust as a valid resource for measuring both EOL and EOL dementia caregivers’ caregivers experiences. The measure can be used by researchers and practitioners alike to support the growing number of EOL and EOL dementia caregivers.

Supplementary Material

gbac145_suppl_Supplementary_File
Click here for additional data file. (224.8KB, pdf)

Acknowledgments

National Health and Aging Trends Study (NHATS) is sponsored by the National Institute on Aging (grant number NIA U01AG32947) and was conducted by the Johns Hopkins University. This research was partly supported by awards from the National Institute on Aging (K01AG056557 and P30 AG053760 for Leggett, K23AG065452 for Epps, R01-AG054004 for Hu, and R36AG070451 and T32 AG049666 for Turner). The authors would like to acknowledge this work also resulted from the development plan and activities of one author’s career development award through the National Institute on Aging, a division of the National Institutes of Health (K23AG065452 [F.E.]). The contents are those of the author(s) and do not necessarily represent the official views of, nor an endorsement of the National Institutes of Health.

Contributor Information

Shelbie G Turner, School of Social and Behavioral Health, Oregon State University, Corvallis, Oregon, USA.

Fayron Epps, Nell Hodgson Woodruff School of Nursing, Emory University, Atlanta, Georgia, USA.

Minghui Li, Department of Clinical Pharmacy and Translational Science, University of Tennessee Health Science Center, Memphis, Tennessee, USA.

Amanda N Leggett, Department of Psychiatry, University of Michigan Medical School, Ann Arbor, Michigan, USA.

Mengyao Hu, Survey Research Center, Institute for Social Research, University of Michigan, Ann Arbor, Michigan, USA.

Funding

This work was supported by the Michigan Center for Contextual Factors in Alzheimer's Disease (MCCFAD) [P30 AG059300], funded by the National Institute on Aging of the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflict of Interest

None declared.

References

  1. Amjad, H., Mulcahy, J., Kasper, J. D., Burgdorf, J., Roth, D. L., Covinsky, K., & Wolff, J. L. (2021). Do caregiving factors affect hospitalization risk among disabled older adults? Journal of the American Geriatrics Society, 69(1), 129–139. doi: 10.1111/jgs.16817 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ayalon, L. (2015). Reports of elder neglect by older adults, their family caregivers, and their home care workers: A test of measurement invariance. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 70(3), 432–442. doi: 10.1093/geronb/gbu051 [DOI] [PubMed] [Google Scholar]
  3. Bangerter, L. R., Griffin, J. M., & Dunlay, S. M. (2019). Positive experiences and self-gain among family caregivers of persons with heart failure. The Gerontologist, 59(5), e433–e440. doi: 10.1093/geront/gny162 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bastawrous, M., Gignac, M. A., Kapral, M. K., & Cameron, J. I. (2015). Adult daughters providing post-stroke care to a parent: A qualitative study of the impact that role overload has on lifestyle, participation and family relationships. Clinical Rehabilitation, 29(6), 592–600. doi: 10.1177/0269215514552035 [DOI] [PubMed] [Google Scholar]
  5. Bédard, M., Molloy, D. W., Squire, L., Dubois, S., Lever, J. A., & O’Donnell, M. (2001). The Zarit Burden Interview: A new short version and screening version. The Gerontologist, 41(5), 652–627. doi: 10.1093/geront/41.5.652 [DOI] [PubMed] [Google Scholar]
  6. Blankson, A. N., & McArdle, J. J. (2015). Measurement invariance of cognitive abilities across ethnicity, gender, and time among older Americans. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 70(3), 386–397. doi: 10.1093/geronb/gbt106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Blieszner, R., Roberto, K. A., Wilcox, K. L., Barham, E. J., & Winston, B. L. (2007). Dimensions of ambiguous loss in couples coping with mild cognitive impairment. Family Relations, 56(2), 196–209. doi: 10.1111/j.1741-3729.2007.00452.x [DOI] [Google Scholar]
  8. Broady, T. R., Saich, F., & Hinton, T. (2018). Caring for a family member or friend with dementia at the end of life: A scoping review and implications for palliative care practice. Palliative Medicine, 32(3), 643–656. doi: 10.1177/0269216317748844 [DOI] [PubMed] [Google Scholar]
  9. Cabrera, I., & Gallego-Alberto, L. (2020). Analyzing the invariance of measurement instruments for dementia caregivers: An opportunity for cross-cultural research. International Psychogeriatrics, 32(9), 1023–1025. doi: 10.1017/S1041610220001039 [DOI] [PubMed] [Google Scholar]
  10. Carbonneau, H., Caron, C., & Desrosiers, J. (2010). Development of a conceptual framework of positive aspects of caregiving in dementia. Dementia, 9(3), 327–353. doi: 10.1177/1471301210375316 [DOI] [Google Scholar]
  11. Castillo-Díaz, M., & Padilla, J. L. (2013). How cognitive interviewing can provide validity evidence of the response processes to scale items. Social Indicators Research, 114(3), 963–975. doi: 10.1007/s11205-012-0184-8 [DOI] [Google Scholar]
  12. Chiao, C. Y., Wu, H. S., & Hsiao, C. Y. (2015). Caregiver burden for informal caregivers of patients with dementia: A systematic review. International Nursing Review, 62(3), 340–350. doi: 10.1111/inr.12194 [DOI] [PubMed] [Google Scholar]
  13. Choi, S., & Seo, J. (2019, April). Analysis of caregiver burden in palliative care: An integrated review. Nursing Forum, 54(2): 280–290, doi: 10.1111/nuf.12328 [DOI] [PubMed] [Google Scholar]
  14. Cohen, C. A., Colantonio, A., & Vernich, L. (2002). Positive aspects of caregiving: Rounding out the caregiver experience. International Journal of Geriatric Psychiatry, 17(2), 184–188. doi: 10.1002/gps.561 [DOI] [PubMed] [Google Scholar]
  15. Connidis, I. A., & McMullin, J. A. (2002). Sociological ambivalence and family ties: A critical perspective. Journal of Marriage and Family, 64(3), 558–567. doi: 10.1111/j.1741-3737.2002.00558.x [DOI] [Google Scholar]
  16. Davies, N., Maio, L., Rait, G., & Iliffe, S. (2014). Quality end-of-life care for dementia: What have family carers told us so far? A narrative synthesis. Palliative Medicine, 28(7), 919–930. doi: 10.1177/0269216314526766 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Durepos, P., Sussman, T., Ploeg, J., Akhtar-Danesh, N., Punia, H., & Kaasalainen, S. (2019). What does death preparedness mean for family caregivers of persons with dementia? American Journal of Hospice and Palliative Medicine, 36(5), 436–446. doi: 10.1177/1049909118814240 [DOI] [PubMed] [Google Scholar]
  18. Eifert, E. K., Adams, R., Dudley, W., & Perko, M. (2015). Family caregiver identity: A literature review. American Journal of Health Education, 46(6), 357–367. doi: 10.1080/19325037.2015.1099482 [DOI] [Google Scholar]
  19. Embretson, S. E., & Reise, S. P. (2013). Item response theory. Psychology Press. [Google Scholar]
  20. Epps, F. (2014). The relationship between family obligation and religiosity on caregiving. Geriatric Nursing, 35(2), 126–131. doi: 10.1016/j.gerinurse.2013.11.003 [DOI] [PubMed] [Google Scholar]
  21. Friedman, E. M., Shih, R. A., Langa, K. M., & Hurd, M. D. (2015). US prevalence and predictors of informal caregiving for dementia. Health Affairs, 34(10), 1637–1641. doi: 10.1377/hlthaff.2015.0510 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Gordon, J. R., Pruchno, R. A., Wilson-Genderson, M., Murphy, W. M., & Rose, M. (2012). Balancing caregiving and work: Role conflict and role strain dynamics. Journal of Family Issues, 33(5), 662–689. doi: 10.1177/0192513X11425322 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Holley, C. K., & Mast, B. T. (2009). The impact of anticipatory grief on caregiver burden in dementia caregivers. The Gerontologist, 49(3), 388–396. doi: 10.1093/geront/gnp061 [DOI] [PubMed] [Google Scholar]
  24. Kasper, J. D., Freedman, V. A., & Spillman, B. C. (2013). Classification of persons by dementia status in the National Health and Aging Trends Study (Technical Paper #5, 5, 1–4). http://nhatspubdemo.westat.com/scripts/documents/NHATS_Dementia_Technical_Paper_5_Jul2013.pdf
  25. Kasper, J. D., Freedman, V. A., Spillman, B. C., & Wolff, J. L. (2015). The disproportionate impact of dementia on family and unpaid caregiving to older adults. Health Affairs, 34(10), 1642–1649. doi: 10.1377/hlthaff.2015.0536 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kim, G., Wang, S. Y., & Sellbom, M. (2020). Measurement equivalence of the subjective well-being scale among racially/ethnically diverse older adults. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(5), 1010–1017. doi: 10.1093/geronb/gby110 [DOI] [PubMed] [Google Scholar]
  27. Kramer, B. J. (1997). Gain in the caregiving experience: Where are we? What next? The Gerontologist, 37(2), 218–32. doi: 10.1093/geront/37.2.218 [DOI] [PubMed] [Google Scholar]
  28. Lawrence, V., Samsi, K., Murray, J., Harari, D., & Banerjee, S. (2011). Dying well with dementia: Qualitative examination of end-of-life care. The British Journal of Psychiatry, 199(5), 417–422. doi: 10.1192/bjp.bp.111.093989 [DOI] [PubMed] [Google Scholar]
  29. Lawton, M. P., Moss, M., Kleban, M. H., Glicksman, A., & Rovine, M. (1991). A two-factor model of caregiving appraisal and psychological well-being. Journal of Gerontology, 46(4), P181–P189. doi: 10.1093/geronj/46.4.p181 [DOI] [PubMed] [Google Scholar]
  30. Liang, J., Aranda, M. P., & Lloyd, D. A. (2020). Association between role overload and sleep disturbance among dementia caregivers: The impact of social support and social engagement. Journal of Aging and Health, 32(10), 1345–1354. doi: 10.1177/0898264320926062 [DOI] [PubMed] [Google Scholar]
  31. Liu, Y., Dokos, M., Fauth, E. B., Lee, Y. G., & Zarit, S. H. (2019). Financial strain, employment, and role captivity and overload over time among dementia family caregivers. The Gerontologist, 59(5), e512–e520. doi: 10.1093/geront/gnz099 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lüscher, K., & Pillemer, K. (1998). Intergenerational ambivalence: A new approach to the study of parent–child relations in later life. Journal of Marriage and the Family, 60(2), 413–425. doi: 10.2307/353858 [DOI] [Google Scholar]
  33. MacCallum, R. C., Widaman, K. F., Zhang, S., & Hong, S. (1999). Sample size in factor analysis. Psychological Methods, 4(1), 84–99. doi: 10.1037/1082-989X.4.1.84 [DOI] [Google Scholar]
  34. Maitland, S. B., Dixon, R. A., Hultsch, D. F., & Hertzog, C. (2001). Well-being as a moving target: Measurement equivalence of the Bradburn Affect Balance Scale. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 56(2), P69–P 77. doi: 10.1093/geronb/56.2.p69 [DOI] [PubMed] [Google Scholar]
  35. Marsh, H. W., Guo, J., Parker, P. D., Nagengast, B., Asparouhov, T., Muthén, B., & Dicke, T. (2018). What to do when scalar invariance fails: The extended alignment method for multi-group factor analysis comparison of latent means across many groups. Psychological Methods, 23(3), 524–545. doi: 10.1037/met0000113 [DOI] [PubMed] [Google Scholar]
  36. Martz, K., & Morse, J. M. (2017). The changing nature of guilt in family caregivers: Living through care transitions of parents at the end of life. Qualitative Health Research, 27(7), 1006–1022. doi: 10.1177/1049732316649352 [DOI] [PubMed] [Google Scholar]
  37. McDonald, R. P. (1999). Test theory: A unified treatment. Erlbaum. [Google Scholar]
  38. Morgan, T., Bharmal, A., Duschinsky, R., & Barclay, S. (2020). Experiences of oldest-old caregivers whose partner is approaching end-of-life: A mixed-method systematic review and narrative synthesis. PLoS One, 15(6), e0232401. doi: 10.1371/journal.pone.0232401 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Nah, S., Martire, L. M., & Zhaoyang, R. (2022). Perceived gratitude, role overload, and mental health among spousal caregivers of older adults. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 77(2), 295–299. doi: 10.1093/geronb/gbab086 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Noyes, B. B., Hill, R. D., Hicken, B. L., Luptak, M., Rupper, R., Dailey, N. K., & Bair, B. D. (2010). The role of grief in dementia caregiving. American Journal of Alzheimer’s Disease & Other Dementias, 25(1), 9–17. doi: 10.1177/1533317509333902 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Ornstein, K. A., Kelley, A. S., Bollens-Lund, E., & Wolff, J. L. (2017). A national profile of end-of-life caregiving in the United States. Health Affairs, 36(7), 1184–1192. doi: 10.1377/hlthaff.2017.0134 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Peacock, S., Forbes, D., Markle-Reid, M., Hawranik, P., Morgan, D., Jansen, L., Leipert, B. D., & Henderson, S. R. (2010). The positive aspects of the caregiving journey with dementia: Using a strengths-based perspective to reveal opportunities. Journal of Applied Gerontology, 29(5), 640–659. doi: 10.1177/0733464809341471 [DOI] [Google Scholar]
  43. Pearlin, L. I., Mullan, J. T., Semple, S. J., & Skaff, M. M. (1990). Caregiving and the stress process: An overview of concepts and their measures. The Gerontologist, 30(5), 583–94. doi: 10.1093/geront/30.5.583 [DOI] [PubMed] [Google Scholar]
  44. Pendergrass, A., Mittelman, M., Graessel, E., Özbe, D., & Karg, N. (2019). Predictors of the personal benefits and positive aspects of informal caregiving. Aging & Mental Health, 23(11), 1533–1538. doi: 10.1080/13607863.2018.1501662 [DOI] [PubMed] [Google Scholar]
  45. Polenick, C. A., Stanz, S. D., Leggett, A. N., Maust, D. T., Hodgson, N. A., & Kales, H. C. (2020). Stressors and resources related to medication management: Associations with spousal caregivers’ role overload. The Gerontologist, 60(1), 165–173. doi: 10.1093/geront/gny130 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Pristavec, T. (2019). The burden and benefits of caregiving: A latent class analysis. The Gerontologist, 59(6), 1078–1091. doi: 10.1093/geront/gny022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Putnik, D. L., & Bornstein, M. H. (2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Developmental Review, 41, 71–90. doi: 10.1016/j.dr.2016.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Quinn, C., Clare, L., & Woods, R. T. (2012). What predicts whether caregivers of people with dementia find meaning in their role? International Journal of Geriatric Psychiatry, 27(11), 1195–1202. doi: 10.1002/gps.3773 [DOI] [PubMed] [Google Scholar]
  49. Rapp, S. R., & Chao, D. (2010). Appraisals of strain and of gain: Effects on psychological wellbeing of caregivers of dementia patients. Aging & Mental Health, 4(2), 142–147. doi: 10.1080/13607860050008664 [DOI] [Google Scholar]
  50. Ross, K. M., Rook, K., Winczewski, L., Collins, N., & Dunkel Schetter, C. (2019). Close relationships and health: The interactive effect of positive and negative aspects. Social and Personality Psychology Compass, 13(6), e12468. doi: 10.1111/spc3.12468 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Savundranayagam, M. Y., Hummert, M. L., & Montgomery, R. J. (2005). Investigating the effects of communication problems on caregiver burden. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 60(1), S48–S55. doi: 10.1093/geronb/60.1.S48 [DOI] [PubMed] [Google Scholar]
  52. Schulz, R., Beach, S. R., & Friedman, E. M. (2021). Caregiving factors as predictors of care recipient mortality. The American Journal of Geriatric Psychiatry, 29(3), 295–303. doi: 10.1016/j.jagp.2020.06.025 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Shah, S. M., Carey, I. M., Harris, T., DeWilde, S., Victor, C. R., & Cook, D. G. (2016). The mental health and mortality impact of death of a partner with dementia. International Journal of Geriatric Psychiatry, 31(8), 929–937. doi: 10.1002/gps.4411 [DOI] [PubMed] [Google Scholar]
  54. Skaff, M. M., & Pearlin, L. I. (1992). Caregiving: Role engulfment and the loss of self. The Gerontologist, 32(5), 656–64. doi: 10.1093/geront/32.5.656 [DOI] [PubMed] [Google Scholar]
  55. Son, J., Erno, A., Shea, D. G., Femia, E. E., Zarit, S. H., & Parris Stephens, M. A. (2007). The caregiver stress process and health outcomes. Journal of Aging and Health, 19(6), 871–887. doi: 10.1177/0898264307308568 [DOI] [PubMed] [Google Scholar]
  56. Stajduhar, K. I., Funk, L., & Outcalt, L. (2013). Family caregiver learning—How family caregivers learn to provide care at the end of life: A qualitative secondary analysis of four datasets. Palliative Medicine, 27(7), 657–664. doi: 10.1177/0269216313487765 [DOI] [PubMed] [Google Scholar]
  57. Supiano, K. P., Luptak, M., Andersen, T., Beynon, C., Iacob, E., & Wong, B. (2022). If we knew then what we know now: The preparedness experience of pre-loss and post-loss dementia caregivers. Death Studies, 46(2), 1–12. doi: 10.1080/07481187.2020.1731014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Tsai, P. F. (2003). A middle-range theory of caregiver stress. Nursing Science Quarterly, 16(2), 137–145. doi: 10.1177/0894318403251789 [DOI] [PubMed] [Google Scholar]
  59. Turner, S. G., Hooker, K., & Geldhof, G. J. (2021). Self-perceptions of aging: Factorial structure and invariance by gender. The Gerontologist, 61(3), 425–429. doi: 10.1093/geront/gnaa059 [DOI] [PubMed] [Google Scholar]
  60. Vick, J. B., Ornstein, K. A., Szanton, S. L., Dy, S. M., & Wolff, J. L. (2019). Does caregiving strain increase as patients with and without dementia approach the end of life? Journal of Pain and Symptom Management, 57(2), 199–208. doi: 10.1016/j.jpainsymman.2018.11.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Wang, Y. N., Shyu, Y. I. L., Chen, M. C., & Yang, P. S. (2011). Reconciling work and family caregiving among adult-child family caregivers of older people with dementia: Effects on role strain and depressive symptoms. Journal of Advanced Nursing, 67(4), 829–840. doi: 10.1111/j.1365-2648.2010.05505.x [DOI] [PubMed] [Google Scholar]
  62. Wharton, W., Epps, F., Kovaleva, M., Bridwell, L., Tate, R. C., Dorbin, C. D., & Hepburn, K. (2019). Photojournalism-based intervention reduces caregiver burden and depression in Alzheimer’s disease family caregivers. Journal of Holistic Nursing, 37(3), 214–224. doi: 10.1177/0898010118801636 [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Wolf, E. J., Harrington, K. M., Clark, S. L., & Miller, M. W. (2013). Sample size requirements for structural equation models: An evaluation of power, bias, and solution propriety. Educational and Psychological Measurement, 73(6), 913–934. doi: 10.1177/0013164413495237 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yu, D. S. F., Cheng, S.-T., & Wang, J. (2018). Unravelling positive aspects of caregiving in dementia: An integrative review of research literature. International Journal of Nursing Studies, 79, 1–26. doi: 10.1016/j.ijnurstu.2017.10.008 [DOI] [PubMed] [Google Scholar]
  65. Zarit, S. H. (2008). Diagnosis and management of caregiver burden in dementia. Handbook of Clinical Neurology, 89, 101–106. doi: 10.1016/S0072-9752(07)01209-2 [DOI] [PubMed] [Google Scholar]
  66. Zarit, S. H. (2012). Positive aspects of caregiving: More than looking on the bright side. Aging and Mental Health, 16(6), 673–674. doi: 10.1080/13607863.2012.692768 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbac145_suppl_Supplementary_File
Click here for additional data file. (224.8KB, pdf)

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

RESOURCES