The Impact of Living Arrangements on Dementia Caregiver’s Sleep Quality

Abstract

In the United States half of the 15 million informal caregivers of persons with Alzheimer's disease or a related dementia (PWDs) do not live with the PWD. This paper compares the sleep quality and health of 59 community-dwelling caregivers living with the PWD and 21 living apart from the PWD. Variables of interest were caregiving experience (hours caregiving, problematic behaviors of the PWD, caregivers' perception of dementia severity), sleep quality, and health (perceived health, stress, and depressive symptoms). Parametric unpaired t tests were used to calculate the differences between key variables. Multiple regression models were constructed, controlling for age, gender, behavior index, and dementia severity to examine the variance explained by living arrangements on sleep quality and health. Caregivers living apart from the PWD experienced the same level of poor sleep quality as did caregivers living with the PWD. The living arrangements of the caregiver did not make a unique contribution to sleep quality or health variables except for reports of unhealthy days. Given the importance of good quality sleep for health, the findings highlight the importance of evaluating caregivers living apart from the PWD for sleep problems with the same level of concern as one would have for those living with the PWD.

Eighty percent of all care for persons with Alzheimer’s or related dementias (PWDs) is provided by unpaid caregivers, most of whom are family members caring for the PWD in the home, and 30% of all caregivers are caring for 2 generations, adult children caring for their parents while their own children still live with them. ^1,2 Living arrangements of the family caregiver–PWD dyad vary, with about half of the caregivers living in the home with the PWD while half do not. The latter consists of long-distance caregivers (more than an hour away ¹ ) who do not have easy access to the PWDs as well as caregivers living in close proximity accessible for supportive and direct care on a daily basis. For this study, the noncohabitating, yet self-identified primary caregivers are called distal caregivers. A few studies of caregivers providing long-distance caregiving ^3,4 have described challenges of communication, but there has been less focus on the distal caregiver. These distal caregivers represent a subpopulation of caregivers that may experience the stress of caregiving differently given living arrangements, yet research describing the experience of distal caregiving is limited.

Although living with the PWD is a predictor of caregiver burden, ⁵ some research supports differences in the caregiving experience based on residential status. Cox and Albis, ⁶ for example, have provided evidence that emotional burdens differ depending on residential status. Noncohabitating (distal) caregivers were more likely to report constantly being worried or stressed; caregivers living with the PWD reported being overwhelmed. Other caregiving characteristics that differed between distal caregivers and cohabitating caregivers included shorter duration of caregiving role and fewer hours of care per week for distal caregivers; in addition, distal caregivers were more likely to be employed while providing care. Caregivers living with the PWDs were more likely to report worse health, loss of independence, and loss of social life in comparison with distal caregivers. ⁶ Evidence supporting a difference in caregiving experience based on living arrangements and race has been presented by Siegler et al ⁷ who found that white distal caregivers experienced more depression than white caregivers living with the PWD, whereas black caregivers reported more depression when living with the PWD. These studies provide some evidence that the caregiving experience is different for distal caregivers as opposed to caregivers living with the PWD; these 2 groups experience different emotional stressors. The effect on sleep, an important health habit, was not examined in these studies however.

The literature provides abundant evidence of the impact of chronic poor sleep on physical and emotional heath as well as quality of life. ^{8 -12} Caregivers experience poorer subjective sleep than noncaregivers do, with poorer daytime function as a result. ¹³ However, much of the research on sleep quality in informal caregivers has been limited to spouses or has included only those living with the PWD. In a recent review of 18 studies, 13 were spousal or cohabitating caregivers. ¹⁴ The remaining 5 studies had no inclusion or exclusion criteria addressing living arrangements but instead required certain levels of caregiving. None of the 5 studies controlled for living arrangements. Living arrangements were not considered a confounding variable or risk factor for sleep disturbance. However, a secondary analysis of the data from the Study of Osteoporotic Fractures did identify living with the care recipient as a high-stress caregiving situation. ¹⁵ This was a study of informal women caregivers older than 65 years who were providing some assistance in activities of daily living or instrumental activities of daily living for care recipients who physically, cognitively, or psychologically could not perform certain tasks (28% had dementia). Caregivers living with the care recipients had a higher risk of sleep disturbance than did distal caregivers, supporting the idea that differences in sleep quality are associated with the living arrangements of the caregiver–PWD dyad. However, another study of 80 caregivers ¹⁶ reported that living arrangements did not separate good sleepers from poor sleepers. The goal of the present study is to examine similarities and differences between distal caregivers and caregivers living with the PWD further, in order to inform a better understanding of the distal caregiving experience. Our theoretical framework is provided by Pearlin’s stress process model, ¹⁷ in which contextual factors such as living arrangements influence how the caregiver experiences stressors such as the severity of dementia or the PWD’s problematic behaviors. Primary stressors lead to secondary stressors—in the present study, to poor sleep quality and ultimately to poor health outcomes.

Method

This is a secondary analysis of a nonexperimental, cross-sectional study that described the relationship between the behavioral and psychiatric symptoms of PWDs and caregivers’ sleep. ¹⁶ In the present exploration, the relationships between living arrangements and caregiver sleep quality and health outcomes (perceived health, stress, and depressive symptoms) were compared while considering confounding variables of caregiving experience (hours per week caregiving, problematic behaviors of the PWD, and caregivers’ perception of dementia severity).

Sample

Caregivers of community-dwelling PWDs in the metropolitan area of Austin, Texas, were recruited as a convenience sample, using flyers and word of mouth in clinic practices and community outreach programs. Participants were at least 21 years of age, self-identified as the primary caregiver of a PWD, and able to communicate in English; all consented freely to participate. Caregivers were excluded if they had untreated obstructive sleep apnea, a medical condition that could affect nighttime sleep and daytime function. Caregivers were also excluded if they had been diagnosed with a major depressive disorder prior to caregiving. For our primary study, ¹⁶ a power analysis was conducted using Nquery Advisor software (Statistical Solutions, Ltd.) to determine an appropriate sample size. ¹⁸ The relationship between sleep quality and mastery as reported by Carter and Acton ¹⁹ was used to calculate the need for a sample size of 75 to provide power at 80% with an α of .05. A total of 80 caregivers were enrolled; 59 were living with the PWD and 21 were living apart but within the same city.

Data Collection

Approval for all study procedures was granted by a University institutional review board. Data were collected via interview by a single interviewer who was an advanced practice nurse, at a time and place convenient for the caregiver who also allowed for supervision of the PWD. Interviews most often occurred in caregivers’ homes but would also include respite program locations or coffee shops. Instruments used for data collection included the following:

• Caregiver demographics: Standard demographics (eg, age, race/ethnic background, education level, socioeconomic status, and work status) and caregiver information (eg, relationship with PWD, amount of hours spent caregiving per day, years of caregiving, and perceived health status) were collected from the caregivers. Caregivers also reported information about the PWD (eg, age, diagnosis, time since diagnosis, education level, and comorbidities).

• Problematic behavior of persons with Alzheimer’s or related dementia: The PWDs’ behavioral and psychological symptoms common in dementia (BPSD) were assessed using a Likert-type scale for frequency of occurrence of behaviors over the last 30 days, modeled from the Neuropsychiatric Inventory (NPI). ²⁰ The 9 items were worded like the NPI stems and included agitation/aggression, apathy, dysphoria or depression, anxiety, inappropriate behaviors, psychoses such as hallucinations or paranoia, changes in nighttime sleep, daytime napping, and wandering. Wandering is not specifically included in the NPI but is a symptom that can be distressing to caregivers. ²¹ Frequency of occurrence was scored as not at all = 0, rarely = 1, frequently = 2, and daily = 3. The frequency scores were summed to provide an index called “behaviors.” Higher scores indicated more frequent occurrence of BPSD. The α for behavior index was .63.

• Dementia severity: Cognitive and functional status of PWD as perceived by the caregiver was measured using the Dementia Severity Rating Scale. ²² This is a measure of the caregiver’s assessment of the PWD’s functioning, with 12 items covering 11 domains of cognition (memory, orientation, judgment, social interactions, home activities, personal care, speech/language, recognition, feeding, incontinence, and mobility). Each item was scored on a Likert-type scale. For each domain, scores range from 0 = normal function to a high score of 3 to 7 depending on the domain. Total scores range from 0 to 51, with higher scores indicating less functionality. Convergent validity tests ²¹ showed correlations with the Mini-Mental State Examination, r = −.77, and with the Consortium to Establish a Registry for Alzheimer’s Disease Neuropsychological Assessment Battery, r = −.73. The αs for the present study were .94 consistently.

• Caregiver’s sleep quality: The original Pittsburgh Sleep Quality Index (PSQI) ²³ consists of 19 items used to calculate scores on 7 subscales (sleep quality, sleep latency or time needed to fall asleep, sleep duration, sleep medication use, sleep efficiency calculated as percentage of time asleep while in bed, sleep disturbances, and daytime dysfunction). The subscale of sleep disturbance includes a question in the original PSQI of “other reasons” for sleep disturbances with the opportunity to write in other reasons not listed. For this study, “other reasons” indicated whether the caregiver’s sleep had been disturbed by the PWD. Possible reasons (eg, help with toileting) were recorded. Subscales were calculated following PSQI guidelines, using the combined scores of the 7 subscales to produce an overall global sleep quality score for the analysis. A global score greater than 5 indicates poor sleepers and has been reported ²³ to have a diagnostic sensitivity of 89.6% and a specificity of 86.5%. Cronbach’s α has ranged from .61 to .73 in previous studies ^24,25 and was .70 in the present study.

• Caregiver’s perceived stress: Perceived stress was measured with the Perceived Stress Scale (PSS), ²⁶ a 14-item instrument designed to measure the degree to which situations in one’s life are felt to have been unpredictable, uncontrollable, and overloading over the past week. The items were rated on a 5-point Likert-type scale ranging from 1 = never to 5 = very often. Higher scores equate to higher stress. Cohen et al ²⁶ reported coefficient α reliability of .84 to .85 in 3 validation studies, and in the present study the αs ranged from .87 to .89.

• Caregiver depressive symptoms: Caregiver depressive symptoms were measured using the Center for Epidemiological Studies Depression Scale (CES-D). The CES-D is a 20-item self-report instrument that assesses the frequency of depressive symptoms experienced over the past week. Scores of 16 or higher indicate risk of clinical depression. Criterion validity has been studied in a population of community-dwelling adults aged 55 to 85 years ²⁷ and found to have a sensitivity of 100% when the cutoff was 16, with a specificity of 88% and a positive predictive value of 13.2%. In the present study, the αs ranged from .87 to .89.

• Caregiver self-report of health status: The Centers for Disease Control and Prevention’s (CDC’s) health-related quality-of-life measure ²⁸ was used to assess the caregivers’ perception of their health. A single item asking how the caregiver rates his or her overall health, measured on a Likert-type scale from 0 = excellent to 4 = poor, is a representation of past, present, and anticipated health. An index of unhealthy days that explains recent health was calculated by combining the number of days on which the caregiver reported that his or her physical or mental health was not good. For instance, if the caregiver reported 2 days on which his or her physical health was not good and 6 days on which his or her mental health was not good, the unhealthy days index score would be 8 of a possible 30. The final question assesses the number of days on which the caregiver was unable to perform normal activities because of experiencing unhealthy days. These core questions of the Behavioral Risk Factor Surveillance System conducted by the CDC have been tested and found valid in both the general community and older adult samples. ²⁹

Analysis

The data were analyzed using SPSS 19 software. Items within the instruments that needed to be reverse coded were transformed into a different variable. There were no missing data. Descriptive statistics were performed, and frequency distributions were graphed. For caregiver characteristics and caregiving factors, Welch’s unpaired t test (to accommodate the uneven sample size) was used to calculate the continuous variables, and chi-square was used for the categorical variable. Bonferroni adjustments were made for the multiple univariate analysis of caregiver characteristics (age, gender, race, ethnicity, work status, income, and roles) with a significance level of P = .007 (.05/7). For the caregiving experience (PWD age, length of months caregiving, hours per week caregiving, behavior index, and dementia severity), the Bonferroni adjustment was .05/5 = .01. The main variables of sleep quality (PSQI, total hours of sleep, and percentage of sleep) as well as overall health, health index, and CES-D and PSS scores were normally distributed, and Welch’s unpaired t tests were therefore used to calculate the differences in these variables between caregivers living with the PWD and those living distally. The caregivers’ reports of sleep latency, physical health, and mental health days were skewed, so the nonparametric Mann-Whitney U test was used. To further examine the importance of living arrangements on the outcome variables of sleep quality and health (depression, perceived stress, overall health, and unhealthy days), a multiple regression model was constructed that controlled for other contextual factors (caregiver age, gender, and relationship with PWD), roles (caregiving hours and the status of caring for children and total roles), and primary stressors (behavior index and dementia severity).

Results

A comparison of demographic and caregiving characteristics for those living with and those living distal to the PWD is presented in Table 1. The caregivers were similar in gender, race, ethnicity, and income level, with no statistical differences between the 2 groups. The percentage of spousal caregivers was higher for caregivers living with the PWD (64.4%) than for distal caregivers (9.5%), and the distal caregivers were predominantly adult children providing care (85.7%; χ² = 22.32, P < .001). This difference was reflected in the mean age of caregivers, which was lower yet not statistically significant after the Bonferroni adjustment for distal caregivers (t = 2.32, P = .03). Thirty-three percent of the distal caregivers were also providing care for children in the household, but this was not statistically different in relation to caregivers living with the PWD (χ² = 1.92, P = .16).

Table 1.

Comparison of Demographic and Caregiving Factors of Cohabitating Caregivers and Distal Caregivers of the PWD.^a

	Cohabitating Caregivers (n = 59)	Distal Caregivers (n = 21)
Age, years	65.08 (12.87)	58.14 (11.36)
Gender, %
Female	89.80	85.70
Race, %
White	89.80	90.50
Black	10.20	9.50
Ethnicity, %
Hispanic	11.90	19
Not Hispanic	88.10	81
Work status, %
Full time	22.00	23.80
Part time	16.90	23.80
Not working	10.20	28.60
Retired	50.80	23.80
Income, %
More than enough	52.50	66.70
Just enough	42.40	33.35
Not enough	5.10
Roles, %
Working	45.80	47.60
Caring for children	18.60	33.30
Volunteer	49.20	38.10
Total rolesb	1.14	1.29
Relationship
Spouse or partner	38	2
Adult child	15	18
Other	6	1
Care recipient’s age	77.97	83.05
Caregiving length, months	51.56 (36.42)	60.05 (40.02)
Caregiving hours per week	118.44 (60.0)	43.76 (46.29)c
Behavior index	9.29 (4.5)	11.29 (4.9)
Dementia Severity Scale Score	24.31 (11.23)	28.43 (10.86)

Abbreviation: PWD, person with dementia.

^aMeans (and standard deviation) or percentages.

^bTotal roles refer to the total number of roles the caregiver participates in besides caregiving and included caring for children, working, and volunteering.

^cSignificant difference at <.001 level.

The caregiving experience was similar for the 2 groups with respect to the length of time they had been providing care, but caregivers living with the PWD provided a significantly higher number of hours of care per week (t = 5.85, P < .001). The PWDs living with their caregivers were younger than those living apart (t = −2.51, P = .02), but there was no reported difference between the 2 groups in dementia severity of the PWD (t = −1.48, P = .15) or in the number of problematic behaviors exhibited over the last 30 days (t = −1.63, P = .11), see Table 1 for means and standard deviations comparing the 2 groups.

Comparison of Sleep Experience

Caregiver’s sleep quality measured by the global sleep quality score did not differ between distal caregivers and caregivers living with the PWD (t = −.38, P = .71; Table 2). Both groups reported poor sleep, with PSQI scores greater than 5. Caregivers reported approximately the same number of minutes to fall asleep, the same sleep efficiency (percentage of time asleep while in bed), and less than 7 hours of total sleep time on average over a 2-week period. As expected, caregivers living with the PWD reported more frequent disruptions at night by the PWD (M 1.56, SD 1.33 for caregivers living with PWD vs M .33, SD .79 for distal caregivers; t = 5.0, P < .001). Of the cohabitating caregivers, 35 (59%) reported being disrupted by the PWD at least once a week, with 21 (36%) reporting disruptions 3 or more times a week. Yet distal caregivers also experienced disruptions at night. Two caregivers either stayed with the PWD or had the PWD stay with them for some period of time during the 2 weeks prior to the survey because of an acute illness or injury. These 2 caregivers reported having to get up at night to help the PWD. Two other distal caregivers reported being disturbed “less than once a week” to help the PWD at night. In both cases, the caregivers received phone calls. One was directly from the PWD who was disoriented and confused about the time of day/night. The other was from a sibling who needed help calming the PWD who was agitated and resistant to care.

Table 2.

Comparison of the Sleep Experience and Health Outcomes of Cohabitating Caregivers and Distal Caregivers.^a

	Cohabitating Caregivers (n = 59)	Distal Caregivers (n = 21)
PSQI global scores	7.41 (3.35)	7.81 (4.46)
Percentage of sleep	80.19 (13.22)	81.25 (15.23)
Sleep latency	21.89 (23.88)	24.09 (28.04)
Total sleep hours	6.36 (1.30)	6.4 (1.67)
Perceives stress Scale	37.80 (9.59)	39.10 (9.02)
Depression (CESD score)	13.41 (9.57)	11.62 (8.65)
Overall health rating	1.41 (1.04)	1.43 (1.08)
Days physical health was not good	3.52 (7.02)	8.09 (10.89)
Days mental health was not good	9.15 (10.79)	16.76 (12.96)b
Unhealthy days index	11.85 (11.2)	20 (11.7)c
Days poor health (mental or physical) stopped activities	2.27 (5.05)	5.09 (9.04)

Abbreviations: CESD, Center for Epidemiologic Studies Depression; PSQI, Pittsburgh Sleep Quality Index.

^aMeans (and standard deviations).

^bSignificant difference P < .05 level.

^cSignificant difference P = .009.

Comparison of Health Outcomes

The comparison of health outcomes showed no statistical difference between caregivers living with the PWD and distal caregivers except in poor mental health days. Both groups of caregivers experienced the same level of perceived stress and the same level of depression (which was below the cut score of 16), and both reported similar levels of overall health that fell between the “very good” and “good” scores. However, the report of number of days on which caregivers’ physical health was not good was higher in distal caregivers, but not statistically significant, whereas the report of days on which distal caregivers’ mental health was not good was significantly higher (t = −2.41, P = .02). This resulted in an unhealthy days index that was significantly higher for distal caregivers (t = −2.77, P = .009). However, the number of days on which poor physical or mental health kept the caregiver from performing normal daily activities was not statistically different between the 2 groups.

Amount of Variance in Outcomes Explained by Living Arrangements

Multivariate models examining the impact of living arrangements on sleep quality, depression, perceived stress, and overall health proved to be a good fit for the outcomes of sleep, depression, and unhealthy days but not overall health or perceived stress (see Table 3). Variables included age and gender (contextual factors), dementia severity and frequency of behavioral problems (primary stressors), and factors of interest that were different (if not statistically) between the 2 groups (relationship status with PWD, hours caregiving, caring for children in the home, total roles performed by caregiver besides caregiving, and the PWD’s age). Although living arrangement did not make a significant contribution in the final model for sleep quality, the total number of roles was significant and in a negative relationship such that caregivers fulfilling higher numbers of roles had better sleep (β = −0.32, P = .04). Total roles explained some of the variance in depression (β = −0.31, P = .04) as did the frequency of behaviors by the PWD (β = 0.30, P = .02). The model for unhealthy days index was a good fit and represented 30% of the variance. In this model, living arrangement (β = −0.27, P = .05), caring for children in the home (β = 0.35, P = .01), and age of the PWD (β = 0.32, P = .04) made significant contributions to the variance in the unhealthy index for caregivers.

Table 3.

Results for Sleep Quality and Health Outcomes Regressed on Living Arrangements Controlling for Age, Gender, Dementia Severity, and Frequency of Behaviors.

	Sleep Quality β	Depression β	Perceived Stress β	Overall Health β	Unhealthy Days Index β
Caregiver’s age	−.21	−.25	−.20	.08	−.18
Gender	.16	−.11	−.11	−.06	−.09
Relationship with PWD	.05	−.20	−.15	.04	−.02
Hours a day caregiving	.16	.14	.02	.11	.24
Caring for childrena	.10	.06	.18	.33a,b	.35a,b
Total number of rolesc	−.32b,c	−.31b,c	−.21	−.11	−.20
PWD’s aged	.13	.08	.03	.24	.32b,d
Dementia severity	−.02	−.02	.02	−.06	−.08
Frequency of behaviors	.08	.30b	.32e	.15	.01
Living arrangementse	−.03	.07	.00	.06	−.27b,e
F 10,69	2.31, P = .02	2.41, P = .02	1.41, P = .20	1.05, P = .41	2.99, P = .003
R 2	.25	.26	.17	.13	.30
R 2 change	.00	.00	.00	.00	.04

Abbreviation: PWD, person with dementia.

^aCaring for children was scored no = 0 and yes = 1, therefore the positive significant relationship indicates caring for children was associated with increased report of worse overall health and more unhealthy days by caregivers.

^bSignificant difference at P < .05.

^cTotal roles refer to the total number of roles the caregiver participates in besides caregiving and included caring for children, working, and volunteering; and the significant negative relationship with sleep quality and depression indicate the more roles the caregiver participated in, the better their sleep and level of depression.

^dThe relationship between older PWD and more unhealthy days reported by caregivers was significant.

^eLiving arrangements were scored as living with the PWD = 1, distal caregivers = 0, thus there was a significant negative association between distal caregivers and more unhealthy days reported.

Discussion

The purpose of this study was to examine the experience of distal caregivers to those living with the PWD with a focus on sleep quality. Previous research has shown that the main cause of poor sleep in caregivers is disruption by the PWD at night, ³⁰ but our prior study ¹⁶ showed that distal caregivers have poor sleep as well. In this secondary analysis, we considered factors that might contribute to poor sleep and other health outcomes such as self-reported health, depression, and perceived stress. The results indicated that factors that measure sleep (eg, sleep efficiency) did not differ between the 2 groups. However, being involved with more total roles (including caregiving for children and volunteering) was related to better sleep quality and less depression. In this sample, caregivers living distally to the PWD were involved in more total roles, and although this may be protective for sleep quality, it was not associated with better mental health; caring for children and living distal to the PWD, along with the PWD’s age, explained some of the variance in self-reported unhealthy days. These findings provide new information about the experiences of distal caregivers of PWDs.

The findings of this study are consistent with previous research. For example, the association between living distally to the PWD and unhealthy days is similar to Kim et al ⁵ who found that living arrangements explained a small percentage of variance in caregiving burden. This burden included a factor of self-reported emotional stress. In the current study, the distal caregivers’ self-reported bad mental health days increased the unhealthy days index. Our finding of no statistical differences in elements of sleep quality (eg, total sleep time and sleep efficiency) despite living arrangements is consistent with the findings of Kochar et al, ¹⁵ where caregivers with low levels of depression experienced similar risk in sleep quality measures of waking too early regardless of living arrangements in comparison with noncaregivers. However, Kochar et al found that the highest risk for decreased sleep efficiency because of not maintaining sleep was associated with living with the care recipient and with high depressive symptoms. This would not be consistent with our finding of similar sleep efficiency in both groups, but caregivers living with the PWD did have more disruptions by the PWD as expected. The unexpected finding was the report of distal caregivers still experiencing nighttime disruptions by the PWD. Differences in the findings of the present study and those of Kochar et al could be contributed to the lower levels of depressive symptoms found in the present study. Furthermore, the sample from Kochar et al’s study of osteoporotic fractures included women caring for a wide variety of ill and disabled family members. This makes comparisons difficult because research indicates that the experience of caregivers of PWDs is different from that of other caregivers with higher levels of burden. ³¹ Furthermore, comparison of the 2 studies is limited due to differences in sample size and strategies for measuring subjective sleep.

In the present study, the models used to explain the variance in sleep quality that included key contextual, background, and primary stressors represented 25% of the variance in sleep quality. This small variance emphasizes the complexity of the caregiving experience. Given that sleep quality was poor for distal caregivers despite fewer hours caregiving and less nighttime disruptions by the PWD, future research should explore, and clinicians should probe to identify, what other precipitating and perpetuating factors lead to poor sleep quality in distal caregivers. Our previous research shows that daytime behaviors of the PWDs such as agitation and apathy are related to caregivers’ perception of their own sleep quality. ¹⁶ Perhaps the impact of these behaviors goes beyond the physical proximity of the behavior and contributes to the experience of distal caregivers, who constantly feel worried or stressed, ⁶ adding to rumination that can be disruptive to nighttime sleep or contributing to a sense of always being “on” even when they are away from the PWD. Furthermore, caregivers living distally to the PWDs were involved in more roles, and although this was related to better sleep quality, it did not provide a significant difference between distal caregivers and those living with the PWDs. Future research should explore the benefits of volunteering on sleep quality because volunteering has been proven to reduce stress for some individuals. ³² Another confounding variable not accounted for in this secondary analysis was the quality of the caregivers’ relationship with their bed partner, which can influence sleep quality. ³³ Similarly, the quality of the relationship between the caregiver and PWD may have contributed to the mental health and poor sleep quality of caregivers and may be a contributing factor to why the caregiver did not live with the PWD. Issues of quality in the relationship of bed partners or quality of the relationship with the PWD could be additional stressors for caregiving that contribute to poor sleep and should be considered in future research.

Certain other limitations to the present study must be acknowledged. Because this was an exploration of a secondary outcome, possible confounding variables impacting sleep quality and living arrangements such as the quality of the relationship in the caregiver–PWD dyad may have been omitted by the primary research design. This limitation should be addressed in future research that focuses on distal caregivers, those not living with the PWD, given that they represent a large proportion of the caregivers of PWDs and that the present study indicates that they are equally at risk of poor sleep quality. The internal reliability of the behavior index was weak at .63 and may be a result of the limited number of questions in comparison with the full NPI, but the questions were limited to frequency in order to decrease the burden on participants. Validity of the behavior index was not confirmed and therefore future research will consider a more robust and established measure of behavioral and psychotic symptoms of dementia. The use of unequal sample sizes is a limitation but was handled in the statistical analysis by assuming that variances were not equal. This sample is consistent with most caregiver research (high percentage of white women), but a larger, more culturally diverse sample would allow for increased confidence in generalizability.

In conclusion, caregivers living distally to PWDs were found to experience the same level of poor sleep quality (eg, total hours of sleep and sleep efficiency) as did caregivers living with PWDs. The caregivers’ living arrangements were significant for the caregivers’ reports of unhealthy days; the distal caregivers reported more unhealthy days, with mental health being significantly different. Given the importance of good-quality sleep for health, the present findings highlight the importance of evaluating distal caregivers of PWDs for sleep problems with the same level of concern that one would have for caregivers living with PWDs.