Abstract
We used data on 718 dementia caregivers and multivariable linear regression to test associations between residential locale and psychosocial outcomes (grief, wellbeing, burden, quality of life [QOL], self-efficacy/mastery, and social networks). Rural residence (versus urban or suburban) was not associated with the psychosocial outcomes. However, for rural caregivers, greater self-efficacy/mastery was associated with lower grief (versus urban/suburban) and burden (versus suburban), and greater social network quality was associated with lower burden (versus suburban) and higher QOL (versus urban). Interventions targeting self-efficacy/mastery and social networks may be particularly effective at improving rural caregivers’ mental health and QOL.
INTRODUCTION
While Alzheimer’s disease (AD) is the most common cause of dementia among older adults, other causes include dementia with Lewy bodies (DLB), Parkinson’s disease (PD), frontotemporal degeneration, and cerebrovascular disease. Each of these dementia etiologies have distinct clinical syndromes and neuropathological features. Yet, each share common challenges to the persons living with AD and related disorders and their caregivers.
Individuals with dementia often rely on formal (paid) and/or informal caregivers (e.g., family) to manage medical, financial, and personal care. Caring for and emphasizing the health and wellbeing of dementia caregivers is essential to support individuals with dementia and caregivers themselves. Caregivers experience significant physical, financial, and emotional burdens from their caregiving responsibilities [1–6]. For example, 24%of dementia caregivers report depression and 13%report frequent poor physical health [7]. Concerns for dementia caregivers include inadequate access to caregiving resources including social support, paid time off, respite care, and educational opportunities. Improving access to these resources may bolster self-efficacy (i.e., belief in ability to successfully perform an action/behavior) in caring for persons with dementia and resiliency against caregiver burdens [7–9].
Few studies have examined differences between rural and non-rural caregivers of persons with dementia [10, 11]. In this study, we assess if: 1) caregiver characteristics differ by residential locale (rural, suburban, or urban), 2) residential locale is associated with caregiver psychosocial characteristics (e.g., grief, burden), and 3) depending on residential locale, associations differ between self-efficacy/mastery and social network quality and caregiver psychosocial outcomes. Access to resources and services may be limited in rural areas compared to non-rural areas because of physical distance and lower socioeconomic status of rural caregivers and patients [12, 13]. Beliefs, preferences, and lifestyles may also differ between rural and non-rural caregivers [10], and these differences may translate into geographic differences in caregiver burden and wellbeing. Thus, this study examined differences in caregiver characteristics and psychosocial outcomes by residential locale (proxy for resource, access, and cultural differences).
MATERIALS AND METHODS
Sample
Data came from an Institutional Review Board-approved online survey of US dementia caregivers who were recruited via email from partner organizations (i.e., Lewy Body Dementia Association, Alzheimer Foundation of America, Family Caregiver Alliance, Alzheimer’s Daily News, Parkinson’s Disease Foundation, and National Institute on Aging Alzheimer Disease Center Program). Survey questions and instruments assessed caregiver and patient demographics and caregiver psychosocial characteristics (e.g., self-efficacy/mastery, social networks, grief, burden, quality of life [QOL], and psychological wellbeing). Additional details provided elsewhere [14, 15].
Caregiver and patient characteristics
Caregiver characteristics included age, sex, race, Hispanic ethnicity, relationship to patient (e.g., spouse, child), education level, marital status, living with patient, primary caregiver status (yes/no), and self-reported residential locale (“Is the area you live in urban, suburban, or rural?”). Caregivers reported the patient’s dementia-related diagnosis (e.g., DLB, AD, etc.) and disease stage (mild, moderate, or severe). Disease stage was also assessed using the Quick Dementia Rating System (QDRS) [16], which assesses ten domains based on caregiver interview (orientation, memory and recall, activities outside home, decision-making and problem-solving, functioning in home and hobbies, toileting and personal hygiene, language and communication, behavior and personality changes, mood, and attention and concentration). A global Clinical Dementia Rating [17] equivalent score was derived from the first six domains (0 = no impairment, 0.5 = possible, 1 = mild, 2 = moderate, 3 = severe).
Caregiver psychosocial characteristics
The psychosocial instrument details are published elsewhere [14]. Briefly, self-efficacy/mastery was measured using an adaptation of the Revised Scale for Caregiving Self-Efficacy [18]. The total score was a sum of scores on 15 questions (e.g., “I am confident I could find information on how to best care for my loved one”) measured on a 4-point scale (1 = strongly disagree to 4 = strongly agree; range: 1–60).
Caregiver grief was measured using the Marwit-Meuser (MM) Caregiver Grief Inventory Short Form [19]. Total score was derived by summing 18 questions (e.g., “I’ve had to give up a great deal as a caregiver”) measured on a 5-point scale (1 = strongly disagree to 5 = strongly agree; range: 1–90).
Caregiver burden was measured using the Zarit Burden Inventory [20], which asks questions such as “overall, how burdened do you feel in caring for the patient” and “are you afraid of what the future holds for the patient”. Total score came from summing the 12 questions measured on a 5-point scale (0 = never to 4 = nearly always; range: 0–60).
Caregiver QOL was assessed using an adaption of the Quality of Life in Alzheimer’s Disease scale [21], which includes questions such as “how do you feel about your energy level” and “how would you describe your life as a whole”. Total score was calculated by summing 14 questions measured on a 4-point scale (1 = poor to 4 = excellent; range: 1–56).
The caregiver’s social network quality was assessed using six questions generated by investigators (e.g., friends see/hear from at least once/month, someone to talk to about important decisions). Total score was calculated by summing responses to 5 questions and 1 question collected on 6-point and 4-point scales, respectively (e.g., 1 point if no friends seen; 6 points for ≥9 friends seen; range: 1–34).
Caregiver wellbeing was measured using the Ryff psychological wellbeing scale [22]. Total score came from summing responses to 24 questions (e.g., “the demands of everyday life often get me down”) measured on a 5-point scale (1 = strongly disagree to 5 = strongly agree; range: 1–120). Some items were reverse coded so that a higher total score equates to better wellbeing.
Analyses
We describe caregiver and patient characteristics overall and by residential locale, and tested differences in characteristics by residential locale using unadjusted linear or logistic regression. In adjusted linear regression, we tested associations between residential locale (rural versus urban, rural versus suburban) and the six psychosocial outcomes (grief, wellbeing, burden, QOL, self-efficacy/mastery, and social network quality), with each outcome considered in six separate regression models. The models controlled for hypothesized confounders: age, sex, race (non-white versus white), education (high school or less, some college, or college degree or higher), marital/partner status, living with patient, primary caregiver status, disease stage, and spouse of patient.
Lastly, we ran eight separate regression models to test whether associations between self-efficacy/mastery and social network quality and four psychosocial outcomes (grief, burden, wellbeing, QOL) differed by residential locale, controlling for the same covariates as the adjusted models above. Individuals missing data were listwise deleted from models. Interaction terms were included in the models to test if associations varied by residential locale. Our reporting of results focused on associations with p < 0.05 based on the false discovery rate method (FDR) [23] to account for multiple comparisons.
RESULTS
Participants (n = 718) were on average 60 years old (standard deviation [SD] = 10.7), 89%were women, 97%were White, and 66%had a college degree (Table 1). Nineteen percent (n = 135) lived in rural areas, 55%(n = 398) in suburban areas, and 26%(n = 185) in urban areas. The mean Zarit Burden score was 26.4 (SD = 8.3) and the mean MM Caregiver Grief score was 60.7 (SD = 13.2). The majority of patients had DLB (68%), PD dementia (11%), or AD (11%) [24]. Caregiver characteristics including the psychosocial measures did not differ substantially by residential locale.
Caregiver characteristics by residential locale
SD, standard deviation; HS, high school; amissing data: race, n = 6; Hispanic, n = 13; living with patient, n = 4; married, n = 1; relationship to patient, n = 1; primary caregiver, n = 12; CDR, n = 221; patient diagnosis, n = 19; income, n = 101; PHQ2, n = 70; Social support scale, n = 60; social network scale, n = 58; Caregiver wellbeing, n = 66; Caregiver grief inventory, n = 89; self-efficacy and mastery, n = 123; Quality of life, n = 14; Zarit Burden, n = 101. bUnadjusted linear or logistic regression. cPossible range of scores: social network scale (1 to 24), Ryff Psychological Wellbeing (1 to 120), MM Caregiver Grief Inventory (1 to 90), Self-efficacy and mastery (1 to 60), quality of life (1 to 56), Zarit Burden Inventory (0 to 60). dp-value when calculated using false discovery rate to account for multiple testing was 0.56 (i.e., associations became insignificant after accounting for multiple testing).
In adjusted analyses, living in a rural versus urban/suburban areas was not associated with caregiver psychosocial characteristics in any model (Table 2). Irrespective of residential locale, greater self-efficacy/mastery and social network quality were associated with lower grief and better wellbeing and QOL, and greater self-efficacy/mastery was associated with lower caregiver burden (Table 3).
Adjusted associations between residential locale and caregiver outcomes
CI, confidence interval. a Linear regression model controlling for age, sex, race (non-white, white), education (high school or less, some college, or college degree or higher), marital status, living with patient, primary caregiver, disease stage, spouse of patient.
Associations between caregiver self-efficacy/mastery, social network score, and caregiver outcomes by residential locale
CI, confidence interval; a Linear regression model controlling for age, sex, race (non-white versus white), education (high school or less, some college, or college degree or higher), marital status, living with patient, primary caregiver, disease stage, spouse of patient (versus not); *p-value when calculated using false discovery rate to account for multiple testing was < 0.05.
Although there were no associations between residential locale and the psychosocial constructs, associations between self-efficacy/mastery and the psychosocial outcomes and between social network quality and the psychosocial outcomes differed by residential locale (Table 3, Supplementary Figure 1). With greater self-efficacy/mastery, rural caregivers demonstrated lower self-rated grief (versus urban/suburban) and lower caregiver burden (versus suburban). With greater social network quality, rural caregivers showed lower caregiver burden (versus suburban) and greater self-rated QOL (versus urban).
In post-hoc sensitivity analyses, we removed primary caregiver and spouse from the models stratified by residential locale to ensure estimates were not unstable due to correlation between those factors and the covariate for living with the patient. The main estimates of interest (e.g., residential locale) changed minimally (data not shown).
DISCUSSION
We found few differences in dementia caregiver and patient characteristics by residential locale and no associations between residential locale and caregiver psychosocial characteristics (i.e., self-efficacy/mastery, social network quality, grief, burden, wellbeing, and QOL). However, compared to non-rural participants, rural participants demonstrated stronger associations between self-efficacy/mastery and better outcomes (lower grief, burden) and between greater social network quality and better outcomes (lower burden, greater QOL).
Two known studies explored differences in dementia caregiver outcomes by caregiver’s residential locale [10, 11]. In the first, twelve rural and 11 urban dwelling caregivers of persons with dementia were interviewed in a qualitative study [10]. Compared to urban caregivers, rural caregivers placed more importance on accepting the disease and preservation of previous ways of life. Social networks did not play a significant role for rural caregivers, and there was more emphasis on the family unit compared to urban caregivers. In contrast, we found that better social network quality was more strongly associated with better psychosocial outcomes (lower burden, greater QOL) among rural than non-rural caregivers. In the second study, access to formal supports was reduced among rural versus urban caregivers, and unlike urban caregivers, greater perceived caregiver burden was associated with a decrease in healthy behaviors (e.g., exercise, seeking counseling) [11].
Other studies that contrasted rural and urban caregivers more generally (not dementia specifically) found similarities in health barriers [25] and perceived burden [26], but differences in financial barriers [25] and services used [26] (e.g., health care versus respite). Compared to urban caregivers, rural caregivers may rely more on family assistance and less on formal services [11, 27]. Altogether, the full impact of residential locale on dementia caregivers remains unknown but may be significant and distinct given the unique and complex symptoms, burden, and time course of dementia disorders compared to other health conditions.
Study strengths included the larger sample of rural dementia caregivers and assessments of multiple psychosocial characteristics. Limitations included generalizability given the convenience-based sample of mostly White caregivers of high socioeconomic status and caregivers of DLB/PD patients. Although our sample was comprised primarily of caregivers of patients with DLB, a previous study found no differences in QOL, wellbeing, caregiver burden, or grief between caregivers of patients with DLB, PD dementia, and AD [24]. We lacked data on the caregivers’ specific geographic location (relied on self-reported locale) and available community resources. Therefore, we could not examine specific differences in access to care and available resources by residential locale. The number of participants missing data on key characteristics (e.g., self-efficacy/mastery) may have biased our findings. Seventeen percent of participants did not complete the self-efficacy/mastery questionnaire, with differential missingness by residential locale (13%among rural, 17%suburban, 20%urban) that may have biased estimates. While the direction of the potential bias is unknown, the observed stronger associations between self-efficacy/mastery and psychosocial outcomes among rural participants may also have been observed for non-rural participants if they had less missingness. While multiple testing may have resulted in spurious associations, FDR methods helped account for this. Finally, we used cross-sectional data which prohibits the determination of a causal association between self-efficacy/mastery and social network quality and the psychosocial outcomes.
Self-efficacy/mastery and social network quality were more strongly associated with better psychosocial outcomes (grief and burden, and burden and QOL, respectively) among rural caregivers than suburban and urban caregivers. Future studies would benefit from longitudinal follow-up of caregivers and a more racially/ethnically and socioeconomically diverse sample. While our findings will need to be replicated, the results may suggest that interventions targeting self-efficacy/mastery [28, 29] and social networks [30] may improve mental health and QOL for rural dementia caregivers to a greater degree than suburban/urban caregivers. The increase in online and telehealth services may increase accessibility to these types of interventions for dementia caregivers in rural areas.
