Mastery and Longevity in Spousal Caregivers of Persons with Dementia

Abstract

Objective

Researchers have consistently shown that providing care in a gradually deteriorating situation, such as dementia, can be stressful and detrimental to the caregiver’s (CG) health. Although stressor appraisal is important in understanding variability in CG outcomes, the role of personal mastery, a coping resource, in shaping CG’s health outcomes has not been considered. The primary goal of this paper was to determine whether personal mastery is associated with a survival advantage for spousal CGs of persons with dementia.

Methods

This study assessed the association of CG burden and personal mastery with longevity over a 10-year period in 71 spousal CGs of persons initially diagnosed with mild cognitive impairment.

Results

Over the 10 years, 16 of 71 CGs (23%) died. Cox regression models with right censoring of CGs’ time to death showed that after adjusting for the health of family CG, spousal CGs who reported high levels of burden had an 83% reduced risk of death when they also reported high personal mastery (hazard ratio [HR] = 0.17, 95% confidence interval [CI] 0.04, 0.65).

Conclusions

Findings have implications for support programs that help build personal mastery.

Numerous studies have documented the profound effects of long-term exposure to physical and psychosocial caregiving demands on the health and well-being of family caregivers (CG) (e.g., Pinquart & Sörensen, 2007; Vitaliano, Zhang, & Scanlan, 2003; Wolff, Spillman, Freedman, & Kasper, 2016), including higher risk of mortality among CGs who experienced greater caregiving strain (Perkins et al., 2013; Schulz & Beach, 1999). More recently, however, investigators have failed to replicate the findings of higher mortality among family CGs (Roth, Brown, Rhodes, & Haley, 2018; Roth, Fredman, & Haley, 2015) and in contrast have found that prosocial helping behaviors may be beneficial in buffering CGs from health consequences of the caregiving strains (Brown & Brown, 2015). One explanation for these contradictory findings comes from the Transactional Model of Stress (Lazarus & Folkman, 1984), which posits a dynamic interplay of different factors that interact to produce considerable individual variability in how stressors may affect CGs. Such factors include the CG–care recipient relationship (e.g., spouse vs adult child), length of time in the caregiving role, stressors associated with care recipient’s disabilities (e.g., responding to cognitive and behavior problems vs providing help with physically demanding tasks such as bathing), CG’s subjective appraisal of the situation, and coping resources utilized to respond to the stressors.

In recent years, several studies have focused on identifying the mechanism by which caregiving stressors upset physiological processes and proliferate into chronic health conditions. Various cross-sectional and prospective studies have found that spousal CGs of persons with dementia (PwD) who report higher personal mastery, or the belief that one has control over ones’ life and environment, have lower physiological arousal to stressors as evidenced by lower norepinephrine reactivity to an acute psychological stressor and higher allostatic load indices (see Roepke & Grant, 2011 for meta-analysis). Prospective studies in middle-aged and older adults (but not family CGs) also have shown a strong relationship between mastery and better cardiovascular health and reduced risk for all-cause mortality (Roepke & Grant, 2011). A key question from the positive health psychology perspective is, does high mastery, in the context of dementia caregiving, confer protection to spousal CGs of PwD?

Given evidence on the protective effects of mastery, probing the buffering effect of mastery on longevity in spousal CGs of persons with dementia is a first step in this inquiry. Therefore, the primary goal of this paper was to contribute preliminary evidence to support the hypothesis that psychological resources such as personal mastery may be protective for CGs who are exposed to long-term demands of caregiving and may explain variability in long-term health outcomes such as mortality among dementia CGs. The present study extends previous research in two ways. First, we examined all-cause mortality over a 10-year period in spousal CGs who were caring for their partner initially diagnosed with mild cognitive impairment. Second, we contribute to research outlining the positive effects of personal mastery by examining whether mastery offers protective effects for longevity among spousal CGs.

Method

Sample

This study originated with 136 adults, aged 60+, recruited through six U.S. memory clinics based on a clinical assessment of Mild Cognitive Impairment (MCI; Roberto, Blieszner, McCann, & McPherson, 2011). Each older adult diagnosed with MCI (PwD) identified one family CG; the PwD-CG dyads were interviewed up to four times over a 10-year period to understand the experiences of the dyad partners as the PwDs’ memory declined. In the present analysis, only the spouse/partner CGs with data on all key predictor variables collected in the first wave (N = 71) were included. The Institutional Review Boards of Virginia Tech and each of the memory clinics approved the study.

All deaths among the CGs were recorded using the following procedures. The research team confirmed existing contact information via telephone. When a participant could not be contacted via telephone, searches were conducted using internet search engines, obituaries posted electronically on newspaper and genealogy websites, and the Social Security Death Index. Personal identifying information previously collected from participants including age, residence, and names of relatives or friends were used to confirm whether a located individual was a study participant. Unless evidence confirmed otherwise, all spousal CGs were assumed to be alive.

Measures

Separate face-to-face interviews were conducted with the PwD and CG at Wave 1. The following items and measures from these interviews were used for the present analysis: subjective appraisal of the CG Burden was measured using the 12-item Zarit Burden Interview-Short version (Bédard et al., 2001). CGs rated on a five-point scale (0 = never, 4 = nearly always) the amount of burden they experienced in the process of taking care of their spouse with cognitive impairment. The burden scale score was constructed by calculating the mean across each set of questions (α = 0.86). CG’s Personal Mastery was measured using the mastery subscale of the Ryff’s Psychological Well-Being Scales (Ryff & Singer, 1996). This subscale contains nine items measuring individuals’ ability to control their environments, manage the demands of everyday life, and take advantage of the opportunities in a way that is suitable to their physical and psychological health. CGs were asked to indicate on a four-point Likert-type scale how strongly they agreed to each statement (1 = strongly disagree, 4 = strongly agree). The mastery scale score was constructed by calculating the mean across each set of questions (α = 0.82). Because mortality is highly associated with age and health, we considered CG’s age and CG’s health (i.e., “How would you rate your overall health at the present time?” rated on a four-point Likert-type scale from 1 = poor to 4 = excellent) as control variables in the analysis. However, due to the small sample size, a parsimonious model controlling for CG’s health is presented in this paper. Expanded models with CG’s age and health are available in the Supplementary Appendix.

Analytic Procedures

A Cox proportional hazard model with robust standard errors was used to examine whether burden and personal mastery were associated with all-cause mortality, controlling for CG’s health. Survival time was defined as the number of days between initial interview at Wave 1 and event time (i.e., decease date due to any cause). If the CGs were living throughout the course of the study, they were right-censored using the number of days between initial interview date and end of study date as their survival time. All predictors were centered before entering the Cox regression model. To test the associations among burden, personal mastery, and mortality, a burden × personal mastery interaction was included in the model. The observed statistical power was reported for the key predictors and interaction effects. Analyses were conducted in Stata (Version 14) and plotted in R (V. 3.2.2). Sensitivity analysis with and without additional control variables were conducted and are provided in the Supplementary Appendix.

Results

Descriptive statistics for the main predictors are shown in Table 1. The Cox regression model specified in the analytic strategy section was fitted. Results from this model are depicted in Table 2. Overall, 16 out of the 71 CGs (23%) died before the end of the study. Schoenfeld residuals on the Cox model suggested that the proportional hazard assumption held, χ²(df) = 3.69 (3), p < .30. The main effect of burden was not significant (Observed Power = 0.38). However, mastery was significantly associated with mortality. Specifically, each unit increase in mastery was associated with a 78% reduced likelihood of death (hazard ratio [HR] = 0.22, 95% confidence interval [CI] 0.06, 0.83); Observed Power = 0.84). More importantly, the interaction between burden and mastery was also significant, suggesting that high levels of mastery in the presence of high levels of burden was significantly associated with lower mortality by 83% (HR = 0.17, 95% CI 0.04, 0.65; Observed Power = 0.64). To further facilitate the interpretation of the interaction effect, the estimated survival probabilities adjusted for covariates were graphed for four combinations of burden and personal mastery: 1 SD above the mean for burden and 1 SD below the mean for personal mastery, 1 SD below the mean for burden and 1 SD above the mean for personal mastery, 1 SD above the mean for both, and 1 SD below the mean for both. As shown in Figure 1, among CGs with high burden, high personal mastery predicted the highest survival probability (Survival_Day=3478 = 0.88; 2 out of 13 CGs died), whereas low personal mastery predicted the lowest survival probability for CGs (Survival_Day=3478 = 0.59; 8 out of 16 CGs died). Among CGs with low burden, personal mastery did not differentiate survival probabilities.

Table 1.

Descriptive Statistics of Main Study Variables and Correlations

	CG’s age	CG’s health	Burden	Mastery
Mean	71.81	2.72	0.79	3.05
SD	7.80	0.70	0.60	0.47
Range	52–89	1–4	0–2.42	2–4
Possible Range	--	1–4	0–4	0–4
		Correlations
CG’s health	−0.11	1
Burden	0.10	−0.16	1
Mastery	−0.15	0.37*	−0.39*	1

Note: CG = Caregiver.

*p < .001. All predictors are from Wave 1 interview.

Table 2.

Cox Proportional Hazards for All-Cause Mortality Among Family Caregivers of Persons with Cognitive Impairment

	HR	95% CI for HR	Interpretation
Predictors a
CG’s health	0.42*	(0.20, 0.87)	Better health is associated with 58% lower mortality risk
Burden	1.85	(0.70, 4.88)	-
Mastery	0.22*	(0.06, 0.83)	Each unit increase in mastery reduces the mortality risk by 78%
Interaction Effects
Burden × Mastery	0.17**	(0.04, 0.65)	See Figure 1 for interpretation

Note: Total N = 71; Deceased n = 16.

Wald Test_{(df =4)} = 27.23, p < .001; Log Pseudolikelihood = −54.34.

CG = Caregiver; CI = Confidence interval; HR = Hazard ratio.

^aAll predictors are mean-centered. All predictors are from the Wave 1 (Time = 0) interview.

*p < .05; **p < .01.

Figure 1.

Adjusted estimated survival probability at four different levels of burden (Burden) and personal mastery (Mastery): high (i.e., 1 SD above the mean) burden and high mastery, high burden and low (i.e., 1 SD below the mean) mastery, low burden and high mastery, and low burden and low mastery. All four curves reflect average survival probabilities adjusted for caregiver’s health.

Discussion

The current study findings offer a first look at linkages between mastery and longevity among spousal CGs of PwD. Recognizing the limitations of the study (i.e., small sample size, no control group of non-CGs), and the modest overall pattern of effects, the data nonetheless reveal consistencies that call for further investigation.

Persons with dementia survive an average of 4–10 years after diagnosis, yet some live as many as 20 years with dementia. During this time, spouse CGs take on increasingly more household chores and responsibilities, as well as providing physical care and handling various behavioral and mood conditions such as increasing cognitive and functional deterioration, depression, anger, agitation, and paranoia in their partner. Many spousal CGs stop participating in valued activities and disengage from their existing social networks because providing care consumes most of their day (Alzheimer’s Association, 2017). This chronic exposure to physical and psychological stress leads to CG distress, burden, overload, feelings of entrapment, and social isolation. It is not surprising then that spousal CGs who feel more burden and are biologically vulnerable due to their own age are at higher risk for serious illnesses and mortality (Schulz & Beach, 1999). Unlike recent studies (Roth et al., 2015, 2018), the current investigation found that in and of itself, CG burden does not relay the whole story about CG well-being. Instead, the CGs’ appraisal of their ability to cope with obstacles, that is, their sense of mastery, protected their longevity perhaps by reducing their risk for downstream diseases (Roepke & Grant, 2011). Specifically, CGs who reported high CG burden and high personal mastery had lower risks of mortality; however, those who reported higher CG burden, but lower personal mastery, showed a higher mortality rate. These findings validate the Transactional Stress Model, adding preliminary and valuable new information about sources of heterogeneity among CGs that may be associated with a survival advantage for spousal CGs. This survival advantage merits further investigation and should be validated in population-based studies.

Several limitations of this preliminary study should be noted. The sample is small and the PwD were diagnosed with MCI within an average of 6 months before Wave 1 data collection. Therefore, we did not necessarily observe a full range of CG burden during this early stage of the caregiving career. Furthermore, as symptoms of dementia worsened over time, CGs may have relied on their care network (informal and paid helpers) to aid them in providing care, thus lowering their burden. We did not have specific measures of these variables. Finally, although previous studies compared the mortality risk of CGs to non-CGs, we did not have such a control group. Nevertheless, the current paper provides evidence of the potency of personal mastery that scientists and health care professionals can integrate as a component of psychoeducational and skill-building programs supported by both public (e.g., National Family Caregiving Support Program) and private funds (e.g., Alzheimer’s Association). Designing more comprehensive support programs, like COPE (Gitlin, Winter, Dennis, Hodgson, & Hauck, 2010) and REACH (Wisniewski et al., 2003), which attend to the evolving and changing needs of attendees over time, will also help mitigate the harmful effects of stress and burden on spousal CGs.

Funding

This research was supported by the Commonwealth of Virginia (Alzheimer’s and Related Diseases Research Award Fund #17-2), the Alzheimer’s Association (IIRG-03-5926 and IIRG-07-59078).

Conflict of Interest

None reported.