Nursing & Healthy Aging Essay, Research Paper Help

Family caregivers sleep disturbance and its associations with multilevel stressors when caring for patients with dementia Yi-Chen Chiua *, Yi-Nung Leeb , Peng-Chih Wangc , Ting-Huan Changd , Chia-Lin Lie , Wen-Chun Hsuf and Shwu-Hua Leeg a Graduate Institute of Nursing & Healthy Aging Research Center, College of Medicine, Chang Gung University, Taoyuan, Taiwan; b Intensive Care Unit of Neurological Surgery, Tri-Service General Hospital, Taipei, Taiwan; c Department of Clinical Psychology, College of Medicine, Fu Jen Catholic University, Xinzhuang District, New Taipei City, Taiwan; d School of Health Policy and Management, College of Health Care and Management, Chung Shan Medical University, Taichung, Taiwan; e Graduate Institute of Health Care Management, Chang Gung University, Taoyuan, Taiwan; f Department of Neurology, Chang Gung Memorial Hospital, College of Medicine, Chang Gung University, Taoyuan, Taiwan; g Department of Psychiatry, Chang Gung Memorial Hospital, College of Medicine, Chang Gung University, Taoyuan, Taiwan (Received 13 October 2012; accepted 15 August 2013) Objective: We tested a stress process model of multilevel stressors on sleep disturbance for family caregivers (FCG) of persons with dementia (PWD). Methods: For this cross-sectional study, trained research assistants collected data from a purposive sample of 180 PWD FCG dyads at two teaching hospitals, two local hospitals, and two community long-term care service programs in northern Taiwan. PWDs neuropsychiatric symptoms were assessed using the Chinese Neuropsychiatric Inventory (CNPI), FCGs distress by CNPI Caregiver Distress Scale, physical fatigue by Visual Analogue for Fatigue Scale, mental fatigue by Attentional Function Index, depressive symptoms by the Center for Epidemiological Studies Depression Scale Short Form, and sleep disturbance by the General Sleep Disturbance Scale. Results: FCGs most prevalent sleep disturbance problems were sleep quality problems (99.4%). Hierarchical regression models revealed that FCGs sleep disturbance was predicted by their physical fatigue, their depressive symptoms, and the synergistic effect of physical fatigue and depressive symptoms in the final model, explaining 57.8% of the variance. Conclusions: This study supports the model that development of caregivers sleep problems may depend on their depression, fatigue, and the synergistic effects of these two variables. These findings suggest that clinicians should educate FCGs about self-care and offer strategies for dealing with a cluster of symptoms when maintaining sleep hygiene. Key words: family caregivers; depressive symptoms; sleep disturbance; dementia Introduction Persons with dementia (PWDs) are cared for by approximately 10 million adult caregivers, two-thirds of whom will suffer from sleep disturbance while caregiving (McCurry, Logsdon, Teri, & Vitiello, 2007). In Taiwan, PWDs were estimated to account for 4.2% of Taiwans total elderly population in 2060 (Taiwan Alzheimers Disease Association, 2012). Most PWDs (85%) live in the community and are cared for by family caregivers (FCGs) (Directorate-General of Budget, Accounting and Statistic, Executive Yuan, Taiwan, 2006). Among the health problems reported by FCGs of PWDs, the most prevalent was sleep disturbance (two-thirds of FCGs had this complaint) (Chiu et al., 2010). Sleep disturbance was also identified as one of the top four health problems among FCGs in Taiwan (Tseng, 2007). Poor caregiver sleep has been linked to lowered immune function, elevated stress hormones, increased risk for cardiovascular disease, and premature mortality (von K?ªanel et al., 2006; von K?ªanel et al., 2010). A review of sleep disturbance in FCGs of PWDs found this problem to be complex and challenging, with insufficient research on the topic (McCurry, Gibbons, Logsdon, Vitiello, & Teri, 2009). In addition to sleep disturbance, FCGs often report feeling depressed. The prevalence of depression in FCGs was estimated to be 24.6% in patients with mild cognitive impairment (Lu et al., 2007) to more than 50% in PWDs (Garcia-Alberca, Lara, & Berthier, 2011). However, little is known about the presence of coexisting symptoms in FCGs of PWDs other than depression. Co-existing symptoms such as depression, fatigue, and sleep disturbance may interfere with FCGs ability to assume and fulfill the caregiving role. In addition, FCGs existing symptoms may worsen during the course of their caregiving activities. Finally, unrelieved symptoms in FCGs may affect their health and quality of life (see review by Fletcher, Dodd, Schumacher, & Miaskowski, 2008). Sleep disturbance among FCGs of PWDs has been correlated with FCGs depressive symptoms (Chiu et al., 2010; McCurry et al., 2009; Tseng, 2007), but understanding this relationship might be enhanced by considering symptoms coexisting with depression, such as fatigue. For example, higher levels of fatigue were correlated with higher levels of depression and sleep disturbance in 103 FCGs of patients with cancer (Cho, Dodd, Lee, Padilla, & Slaughter, 2006). In another study, higher levels of fatigue *Corresponding author. Email: yulandac@mail.cgu.edu.tw 2013 Taylor & Francis Aging & Mental Health, 2014 Vol. 18, No. 1, 92101, https://dx.doi.org/10.1080/13607863.2013.837141 in 67 FCGs of PWDs were correlated with higher levels of depression (Clark, 2002). Finally, FCGs of PWDs reported higher levels of perceived physical fatigue than noncaregivers (Sato, Kanda, Anan, & Watanuki, 2002). Caregiving for PWDs can be stressful; therefore, assessments of FCGs should be based on a theoretical understanding of the stress process of caregiving (Family Caregiver Alliance, 2006). One comprehensive theoretical framework for the stress process is the caregiving and stress-process model (Pearlin, Mullan, Semple, & Skaff, 1990), which includes background factors (demographic, cultural, and life-history influences), primary stressors (objective indicators such as disease severity and PWDs behavioral problems, and subjective indicators such as fatigue caused by overload and distress toward PWDs behavioral problems), secondary stressors (such as role strain and intrapsychic strain), mediators, and stress outcomes. In this model, mediating factors include resources for coping with social, economic, and internal stresses, whereas outcomes include depression, FCGs physical health, and giving up the caregiver role. This model has been successfully used to study FCGs of PWDs (Hilgeman et al., 2009). Therefore, we chose it to guide this study. Based on the caregiving and stress-process model (Pearlin et al., 1990), various outcomes should not necessarily be treated as interchangeable ways to assess the impact of caregiving stress. A more reasonable approach may be to consider the different types and levels of outcomes as interrelated. For example, elements of emotional distress such as depression are likely to surface first, and if they persist, they may eventually be inimical to physical well-being. Therefore, we hypothesized that fatigue resulting from the antecedent stressful caregiving process predicted primary outcomes such as depression. Then fatigue and depressive symptoms would interact with each other and develop synergistic effects to predict sleep disturbance. By exploring the relationships of FCGs sleep disturbance with multilevel stressors, we may discover that a cluster of symptoms and health problems in FCGs of PWDs better reflects the daily experiences of this population, allowing the development of effective interventions. Methods Study design and participants This cross-sectional study was conducted at two teaching hospitals, two local hospitals, and two community longterm care service programs in northern Taiwan. Outpatients of memory disorder clinics underwent a standard comprehensive evaluation at their respective clinics to determine their eligibility for the study. A neurologist of the affiliated hospital evaluated potential participants for dementia diagnosis and severity based on criteria of the Diagnostic and Statistical Manual (DSM)-IV (American Psychiatric Association, 1999) and guidelines of the National Institute of Neurological and Communicative Disorders and the Stroke and Alzheimer Disease and Related Disorders Association (McKhann et al., 1984). Other inclusion criteria for PWDs included (1) speak Mandarin Chinese, Taiwanese, or Hakka dialect, and (2) have a primary or secondary FCG. Exclusion criteria for PWDs included (1) acute illnesses, (2) impaired hearing loss and severe visual problems, and (3) chronic alcohol abuse or use of drugs which could affect functions of the central nervous system. Inclusion criteria for FCGs (1) provide all or most of the assistance to PWD for the past 3 months, or (2) provide secondary care to their relative by supervising a hired care assistant. Co-residency was not required for FCGs. Exclusion criteria for FCGs included (1) documented cognitive or mental disorder, such as severe memory problems or major affective disorders, (2) hearing or visual impairments that were not properly corrected, (3) use of prescription drugs known to impair or enhance attention, e.g., antidepressants, barbiturates, or other depressants, amphetamines, and (4) insufficient command of Chinese, Taiwanese, or Hakka. After all eligible participating FCGPWD dyads read or heard the consent form, they signed it. Context of home care in Taiwan Although most PWDs are cared for by their FCGs, PWDs can receive limited reimbursement from the government for housekeeping and personal care, depending on the severity of their disability (Ministry of the Interior, 2007). Most FCGs would still need to pay out-of-pocket for home services, but these services are not sufficient to support the caregiving tasks of FCGs of PWDs (Shyu, Huang, Huang, & Chen, 2008). Thus, many caregivers hire foreign care aides (Chen & Wu, 2008). FCGs not only have to provide direct care to PWDs but also have to supervise the care activities of foreign aides. Since FCGs commonly use home services and hire care aides in Taiwan, we recruited both primary and secondary FCGs (those supervising care aides) to reflect the reality of Taiwanese society. Measures Data were collected on PWDs characteristics (age, gender, education [years]), dementia diagnosis, degree of dementia severity, cognitive function, depression, and neuropsychiatric symptoms, as well as FCGs demographic factors (age, gender, marital status, relationship to the PWD), caring-related variables (caregiving distress, caring duration, weekly caring time), depressive symptoms, fatigue, and sleep disturbance. PWDs dementia severity PWDs dementia severity was determined by the Clinical Dementia Rating Scale (CDR) (Hughes, Berg, Danziger, Coben, & Martin, 1982), which rates impairment in six domains: memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care. Items are rated on a 5-point scale: 0 (intact), Aging & Mental Health 93 0.5 (questionable dementia), 1 (mild dementia), 2 (moderate dementia), and 3 (severe dementia). The Chinese CDR had appropriate psychometric properties (Lin & Liu, 2003). Higher CDR scores indicate more severe dementia. CDR data on outpatient PWDs were collected from chart review, but research assistants administered the CDR to community-dwelling PWDs. Therefore, we did not calculate Cronbachs a for this scale. PWDs cognitive function PWDs cognitive function was assessed by the Chinese Mini-Mental State Examination (MMSE) (Folstein, Folstein, & McHugh, 1975; Guo et al., 1988), a widely used instrument for screening and monitoring global cognitive impairment. The Chinese MMSE had good validity and reliability (Guo et al., 1988) and measures orientation, registration, recall, language, and spatial capacity with a total possible score of 30. Higher scores suggest higher levels of cognitive function. MMSE total scores were collected from chart review for outpatient PWDs, while scores on community-dwelling PWDs were administered by trained research assistants. We did not calculate MMSE Cronbachs a in this study. PWDs depression symptoms Depressive symptoms of PWDs were assessed using the Geriatric Depression Scale Short Form (GDS-S) (Sheikh & Yesavage, 1986), which eliminates the influence of somatic complaints on depressive symptoms in older adults (Yesavage, Brink, & Lum, 1983). The GDS-S consists of 15 yes/no questions; higher scores indicate more depressive symptoms (Sheikh & Yesavage, 1986). The internal consistency and construct validity of the Chinese GDS-S were good (Cronbachs a ¬ 0.81, r ¬ 0.91, p < 0.01) (Lu, Liu, & Yu, 1998). In this study, Cronbachs a of GDS-S was 0.80. PWDs neuropsychiatric symptoms PWDs neuropsychiatric symptoms were assessed by FCGs with the Chinese neuropsychiatric inventory (CNPI) (Leung, Lam, Chin, Cummings, & Chen, 2001), which was adapted from the original scale (Cummings, 1997) and measures 12 behavioral manifestations of PWDs. The CNPI requires FCGs to evaluate the frequency and severity of PWDs psychiatric symptoms, and multiplies both scores into a final score. A higher final score suggests a higher frequency and severity of psychiatric symptoms. The concurrent validity of the original scale correlated (p < 0.001) with the corresponding domains in the Behavioral Pathology in Alzheimers Disease Rating Scale (Reisberg et al., 1987). The overall internal consistency of the CNPI was 0.84, while the interrater reliability (kappa coefficients) ranged from 0.7 to 1.00, demonstrating an acceptable level of agreement between raters (Leung et al., 2001). In this study, the CNPIs Cronbachs a was 0.85. FCGs caregiving distress FCGs caregiving distress was assessed by the CNPI-Caregiver Distress Scale (Leung et al., 2001), which requires FCGs to rate their own distress from 0 (not at all distressing) to 5 (very severely or extremely distressing) regarding the 12 psychiatric symptoms of PWDs assessed by the CNPI; higher scores indicate greater caregiving distress. Cronbachs a of the CNPI-Caregiver Distress Scale in this study was 0.88. FCGs depressive symptoms FCGs depressive symptoms were assessed by the 10-item Chinese Center for Epidemiological Studies Depression Scale (CESD-10) (Boey, 1999), which emphasizes emotional symptoms. The Chinese CESD-10 demonstrated comparable accuracy to the original CESD (Radloff, 1977) in classifying cases with depressive symptoms (k ¬ 0.84, p < 0.01) (Boey, 1999). Respondents evaluate their mood changes in the past week on a 4-point scale (03), with scores ranging from 0 to 30; higher scores indicate more depressive symptoms. CESD-10 scores 10 indicate depressive tendency and warrant a clinicians diagnosis. The CESD-10 in this research had a Cronbachs a of 0.84. FCGs fatigue FCGs fatigue was conceptualized as mental fatigue and physical fatigue. Mental fatigue was assessed by the 16- item Chinese Attentional Function Index (AFI) (Chiu, 2002), which was designed by Cimprich (1990) to measure purposeful activities crucial for effective functioning in daily life. Items are rated on an 11-point scale (0 ¬ not at all, 10 ¬ all the time). Scores range from 0 to 160; higher scores indicate a higher level of functioning. The Chinese AFI had a three-factor solution based on principal component analysis with varimax rotation, explaining 50.1% of scale variance. The AFI had acceptable construct validities with the Cognitive Abilities Screening Instrument (Teng et al., 1994) and the Digit Span Forward and Backward (Lezak, 1995) of r ¬ 0.36 and 0.33 (p < 0.001), respectively (Chiu, 2002). In this study, the AFI had a Cronbachs a of 0.83. Physical fatigue was measured by the 18-item Visual Analogue for Fatigue Scale (VAS-F) Chinese version (Lee, Hicks, & Nino-Murcia, 1991; Lee, Lee, Rankin, Weiss, & Alkon, 2007). The self-rated VAS-F has two parts: vitality (13 items) and energy (5 items), with responses rated on an 11-point scale from 0 (not fatigued) to 10 (extremely fatigued). The Chinese VAS-F in this study had a two-factor solution (fatigue factor and energy factor), explaining 68.63% of the variance and Cronbachs a of 0.94. Since the original scale treated fatigue and energy as two different concepts, this study only used the fatigue subscale. FCGs sleep disturbance FCGs sleep disturbance was measured by the 28-item Chinese General Sleep Disturbance Scale (GSDS), with 94 Y.-C. Chiu et al. satisfactory internal consistency reliability (Cronbachs a ¬ 0.81) (Lee, 1992, 2007). The GSDS was developed to measure sleep disturbance in FCGs of newborn infants after discharge from the intensive care unit care. We chose the GSDS to measure sleep disturbance in FCGs of PWDs because they face chronic stress, similar to FCGs of newborn infants even after hospital discharge (Lee, 2007) and because the GSDS has conceptual congruency with physical fatigue. The first 20 items of the GSDS assess sleep disturbance in the past week within seven domains (GSDS subscales): difficulty falling asleep (1 item), waking up during sleep (1 item), waking up before the sleep cycle ends (1 item), sleep quality (3 items), sleep quantity or sleep time (2 items), daytime sleepiness (7 items), and consumption of substances to aid sleep (5 items). The original GSDS contains 6 items in the last subscale, but Lee (2007) deleted 1 item (frequency of smoking marijuana) because having marijuana is illegal in Taiwan. To determine our sampling adequacy, we calculated the KaiserMeyerOlkin measure of sampling adequacy, which was 0.841. Thus, we ran principal component factor analysis for the Chinese GSDS-Sleep Disturbance subscale with varimax rotation and found a five-factor solution that explained 59.38% of total variance. Another GSDS item asks whether any unusual sleep problems were experienced in the past week. If yes, respondents complete the Interference with Daily Life subscale (7 items) to assess the frequency and degree to which these seven problems interfered with their daily activities. The frequency for each item (problem) is rated from 0 (never) to 10 (every day), and the degree of interference is rated from 0 (no interference) to 10 (severe interference). Higher scores indicate greater sleep disturbance and higher levels of interference. Cronbachs a-values for the first seven subscales and interference subscales in this study were 0.82 and 0.95, respectively. The interrater reliabilities (intraclass correlation coefficients) of the first seven subscales and interference subscales, determined using data from eight FCGs, were 0.97 and 1.00 (both p < 0.01), respectively. Since these eight FCGs did not differ signifi- cantly in age, education, duration of care (months), and caring time (hours/week) from the other 172 FCGs in the study, we combined these two data sets (N ¬ 180). For hierarchical regression analyses, we used only scores on the first seven subscales because only 48 FCGs reported that their sleep disturbance interfered with their daily life, an insufficient sample for regression analysis. Procedures This study was approved by the institutional review boards of the hospital affiliated with the authors university (Case No. 95-0049B). Consecutive PWDFCG dyads were recruited by purposive sampling from the participating sites by two well-trained research assistants using a standardized interview. The research assistants, one registered nurse with a bachelors degree in nursing and another with a masters degree in gerontological nursing, were trained by a clinical psychiatrist to administer the test battery and by the principal investigator on background knowledge of dementia, dementia care, and community resources for the FCGPWD dyads. Data on outpatient PDWs diagnosis and cognitive status were collected by research assistants from PDWs charts, and this information on community-dwelling PWDs was obtained from their respective physicians. Statistical analysis All analyses were performed using SPSS, version 13.0 (SPSS Inc., Chicago, IL). Data were cleaned using frequency and descriptive statistics to check for outliers. To reduce deviations from normality, all variables were checked for skewness and kurtosis to identify those that could benefit from data transformations. Patients and FCGs demographic and main variables were analyzed by descriptive statistics. Relationships between PWDs neuropsychiatric symptoms, FCGs stressors (caregiving distress, depression, fatigue), and FCGs sleep disturbance scores were explored by Pearson correlation coefficient and hierarchical regression models. The final sample size was 180 dyads. The sample size was estimated for a medium effect size, power of 0.8, a level of 0.05, and analysis of 14 variables: background factors (FCGs gender, age, education [in years], marital status, relationship with PWD, living with PWD [yes/no], having a foreign helper [yes/no]), stressor variables (PWDs neuropsychiatric symptoms and disease severity, FCG caregiving distress and fatigue), FCG depressive symptoms, synergistic effects of depressive symptoms and fatigue, and the main outcome variable (FCG sleep disturbance). This analysis determined that a sample size from 100 to 250 dyads would be sufficient to detect R2 between 21 and 8, with 20 independent variables (Hair, 1988). Of 198 dyads contacted, 189 agreed to participate, for a response rate of 95.24%. Of these 189 participating dyads, nine failed to complete the test battery due to time limitations and schedule conflicts. Results Participants characteristics The 180 PWDs had a mean age of 77.61 years (SD 8.2), a male/female ratio of 89/91, and mean educational level of 8.16 years (SD 5.2). The majority of PWDs was diagnosed with Alzheimers disease (n ¬ 132, 73.3%). Almost half the elders (n ¬ 88, 48.9%) had a CDR score ¬ 1, with 27.8% (n ¬ 50) having a CDR score ¬ 0.05, and 22.4% (n ¬ 42) having a CDR score 2 (CDR ¬ 2, n ¬ 39; CDR ¬ 3, n ¬ 3). Their mean MMSE score was 16.7 (SD ¬ 6.1), meaning moderate global cognitive impairment, while the mean CNPI score was 16.6 (SD ¬ 19.8) (Table 1). The FCGs had a mean age of 56.0 years (SD ¬ 13.8), with more than half 41 to 60 years old (53.3%). The majority of FCGs were female (65%), most were married (90.6%), and they had a mean caring duration of 30.0 months (SD ¬ 40.6), and a mean caring time of 66.2 hours per week (SD ¬ 50.6). The majority of FCGs were PWDs adult children, including sons and daughtersin-law (55.6%), followed by spouses (40.6%) (Table 1). Aging & Mental Health 95 These background characteristics of PWDs and FCGs are similar to those of other Taiwanese FCGPWD dyads (Huang, Shyu, Chen, & Hsu, 2009). Distributions of FCG sleep disturbance FCGs mean GSDS score for the first seven subscales was 46.2 (SD ¬ 28.3). Since the original instrument has not established a cutoff score, we report here the percentage of FCGs with any scores 1 on the first seven subscales. Among 180 FCGs, 109 reported difficulty falling asleep (60.6%), 122 reported waking up during sleep (67.8%), and 110 reported waking up before the end of a sleep cycle (61.1%). With regard to self-perceived sleep quality, 179 FCGs reported worse sleep quality (99.4%); as to sleep quantity, 116 FCGs considered themselves sleeping too much or too little (64.4%). Almost everyone indicated experiences of dozing in the daytime (97.8%), and only 55 claimed they used substances to help them sleep (30.6%). Finally, 51 FCGs reported experiencing unusual sleep disturbance in the past week, and 48 reported that this problem interfered with their lives (26.7%) (Table 2). Relationships between FCGs sleep disturbance and PWDs characteristics FCGs sleep disturbance was not significantly related to PWDs demographic characteristics in Pearson correlation analysis, one-way ANOVA with Scheffes test for post hoc analysis, and independent sample t-tests. However, FCGs sleep disturbance was moderately, positively correlated with PWDs neuropsychiatric symptoms (r ¬ 0.29, p < 0.01) and highly, positively correlated with PWDs depression/bad mood (r ¬ 0.32, p < 0.01) (Table 3). FCGs mean score for caregiving distress toward PWDs neuropsychiatric symptoms was 8.61 (SD 10.56), indicating mild distress (Matsumoto et al., 2007). The results of correlation analysis indicate that FCGs sleep disturbance and interference with daily life were signifi- cantly correlated with distress regarding all but PWD sleep/nighttime activities. Specifically, FCGs sleep disturbance was strongly correlated with their distress towards patients delusions (r ¬ 0.25, p < 0.01), hallucinations (r ¬ 0.22, p < 0.01) and emotion-related behavioral symptoms (including irritation/aggression, Table 1. Demographic and clinical characteristics of PWDs and their FCGs (N ¬ 180). PWDs FCGs Mean (SD) n (%) Mean (SD) n (%) Female 91 (50.6) 117 (65) Age (years) 77.61 (8.2) 56.0(13.8) Education (years) 8.16 (5.2) 11.9(4.1) Caring duration (months) 30.0(40.6) Caring time (hours/week) 66.2(50.6) Relationship Spouse 73 (40.6) Adult children (including in-laws) 100 (55.6) Other 6 (3.3) Missing 1 (0.1) Living with PWD Yes 139 (77.2) No 41 (22.8) Hired foreign helpers Yes 40 (22.2) No 140 (77.8) Diagnosis Alzheimers disease 132 (73.3) Vascular dementia 27 (15.0) Other 21 (11.7) Clinical Dementia Rating 0.5 50 (27.8) 1 88 (48.9) 2 42 (22.4) Mini-Mental State Examination 16.7(6.1) Barthel Index 88.2 (18.8) Chinese Neuropsychiatric Inventory (CNPI) 16.6 (19.8) Geriatric Depression Scale-S (patient version) 52.0 (1.6) CNPI-Caregiver distress 8.6(10.6) CESD-10 6.6 (5.9) Attentional Function Index 129.0 (9.9) Lees Fatigue Scale 28.3(29.1) General Sleep Disturbance Scale First seven GSDS subscales 48.5(25.6) Interference 6.9(13.8) CESD-10, 10-item Center for Epidemiological Studies Depression Scale; GSDS-S, General Sleep Disturbance Scale Short Form. 96 Y.-C. Chiu et al. depression, and anxiety) (Table 3). Overall, FCGs sleep disturbance was strongly, positively correlated with their distress towards patients neuropsychiatric symptoms (r ¬ 0.32, p < 0.01) (Table 3). The relationships between FCGs caregiving distress and different aspects of sleep disturbance were then explored using Pearson correlations. Results indicate moderate and significant positive correlations between FCGs distress and sleep disturbance domains (r ¬ 0.20 0.29, p < 0.01), except for self-perceived sleep quality (reversed question, r ¬ 0.13, p ¬ 0.07). FCGs distress was most significantly correlated with dozing during the daytime (r ¬ 0.25, p ¬ 0.001) and interference with daily life (r ¬ 0.29, p ¬ 0.000). To explore possible significant predictors of FCGs sleep disturbance based on the stress-process model for family caregiving (Pearlin et al., 1990), we used hierarchical multiple regression models. Independent variables included FCGs background factors (age, gender, education, living with PWDs, care duration, care time per week, having a foreign helper, marital status, and relationship with patient), primary stressors (PWDs neuropsychiatric symptoms and disease severity), secondary stressors (FCGs caregiving distress and fatigue), and FCGs depressive symptoms. These independent variables were examined by tolerance tests for collinearity; none of the predictors had a tolerance indicator >0.1, suggesting no collinearity (Shi, 2003). To avoid multicollinearity when calculating the synergistic effects of depressive symptoms and fatigue indicators, we adopted a mean-centering approach (Aiken, West, & Reno, 1991). The results of five-level hierarchical model analyses showed that only 11.1% of the variance in sleep disturbance was explained by FCGs background factors in the level I model, but the contribution to variance increased to 18.4% after PWDs neuropsychiatric symptoms and disease severity were entered in the level II model. Moreover, the overall explained variance in sleep disturbance increased to 47.5% after FCGs caregiving distress and fatigue indicators were entered, adding 29.0% to the Table 3. Correlations among FCGs sleep disturbance, FCGs distress and PWDs psychiatric symptoms (N ¬ 180). PWDs neuropsychiatric symptoms FCGs distress regarding PWDs neuropsychiatric symptoms FCGs sleep disturbance subscale score FCGs interference subscale score GSDS total score FCGs sleep disturbance subscale score FCGs interference subscale score GSDS total score Delusion .19 .11 .18 .25 .23 .27 Hallucination .15 .11 .15 .22 .19 .22 Agitation/aggression .24 .22 .25 .20 .23 .23 Dysphoria/depression .31 .27 .32 .27 .25 .29 Anxiety .27 .18 .26 .28 .20 .28 Euphoria .16 .05 .13 .07 .13 .10 Apathy .05 .12 .08 .16 .14 .17 Disinhibition .16 .16 .17 .20 .16 .20 Irritability .22 .21 .24 .19 .21 .21 Aberrant motor behavior .09 .22 .15 .16 .26 .21 Sleep & nighttime behavioral change .06 .07 .07 .11 .08 .11 Appetite and eating behavioral change .11 .18 .15 .13 .18 .16 PWDs neuropsychiatric symptoms .34 .32 .30 .37 .26 .32 p < 0.05; p < 0.01. Table 2. Types of sleep disturbance among FCGs (N ¬ 180). Number of FCGs with scores 1 (%) Mean SD Range Sleep disturbance subscale (items) 46.2 28.3 0147 Difficulty falling asleep (1) 109 (60.6) 2.9 3.1 010 Waking up during sleep (1) 122 (67.8) 3.7 3.5 010 Waking up before the end of a sleep cycle (1) 110 (61.1) 3.0 3.2 010 Sleep quality (3) 179 (99.4) 12.9 4.1 027 Sleep quantity (2) 116 (64.4) 4.3 4.2 014 Dozing in daytime (7) 176 (97.8) 19.3 12.0 057 Consumption of sleeping pills (5) 55 (30.6) 2.5 5.1 027 Interference subscale Sleep disturbance interfering with life (7) 48 (26.7) 6.9 13.8 064 Note: One GSDS item with a dichotomized response to measure unusual sleep disturbance in the past week is not listed. Aging & Mental Health 97 variance, but the predictive effect of PWDs neuropsychiatric symptoms disappeared in the level III model. An additional 9.1% of the variance in sleep disturbance was explained by FCGs depressive symptoms in the level IV model. When the synergistic effect of FCGs depressive symptoms and physical fatigue was entered into the final level V model, this model explained 57.8% of the FCGs sleep disturbance variance. The hierarchical model analyses suggested that the most important predictors were FCGs physical fatigue and depressive symptoms. Finally, FCGs sleep disturbance was predicted by physical fatigue, rather than mental fatigue, depressive symptoms, and the synergistic effect of depressive symptoms and physical fatigue, explaining 57.8% of the total variance, even though the increased variance of the synergistic effect was not significant (Table 4). Discussion This study shows that, in general, about two-thirds of FCGs of PWDs suffered from various types of sleep disturbance, similar to a review of primarily Western FCGs of PWDs (McCurry et al., 2009) and a study of Taiwanese FCGs of PWDs (Tseng, 2007). The most prevalent sleep disturbance problems reported by our FCGs included sleep quality problems (99.4%), dozing in daytime (97.8%) and waking up before the sleep cycle ends (67.8%). Despite the sleep disturbances, FCGs of PWDs in our study reported low consumption of sleeping aids. This result might be due to Taiwanese FCGs worrying about addiction and difficulty waking up for night care (Tseng, 2007). Future investigations are warranted to explore the conditions and reasons for taking sleeping pills in this population, as well as subjective and objective measurements of sleep disturbance to comprehensively understand sleep disturbance problems in Taiwanese FCGs of PWDs. Our results indicate that FCGs sleep disturbance was not significantly correlated with PWDs age, gender, dementia diagnosis, severity of dementia, and overall cogn