Depression and Burden 1

Running head: DEPRESSION AND BURDEN

The Relationship Between Caregiver Depression and Burden Using the CES-D and ZBI Scales

An Undergraduate Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Honours Bachelor of Science in Psychology

Matthew P. Canning

LakeheadUniversity

Supervisor: Dr. Michel Bédard

Second Reader:Dr. Dwight Mazmanian

Abstract Caregiver Burden and depressionare frequently studied topics in psychology and gerontology. Given the currentand anticipated future rise in dementia and other geriatric conditions, the need for caregiving and caregiver burden and depression measures is of increasing concern. The Center for Epidemiologic Studies Depression (CES-D) scale is an important instrument to measure depression, and The Zarit Burden Interview (ZBI) as well as a new abridged 12-item version of the scale(SF-12)are twoimportantinstruments to measure burden. Bédard and his colleagues collected data derived from one hundred spouses for the study. Caregiver status (e.g., burden) and subsequently care recipient function (e.g., behaviour problems) were collected by research assistants, as well as caregiver psychological and health status data with the Zarit Burden Interview (short form; ZBI-S), the Short Form Health Survey (SF-36), and the Health-Promoting Lifestyle Profile (HPLP). Findings showed a relationship between depression and burden.

The Relationship Between Depression and Burden Using the CES-D and ZBI

The stress associated with caregiving is afacet of psychology that has been extensively studied in recent years (Bédard et al., 2001; Bradley et al., 2004; Jeglic et al., 2005;Yumiko et al., 2004). Caregivers can experience various types of stress throughout the length of their service, such as lack of mental well being which includesfeelings of both emotional and physical burden and depression. Studies have suggested that caregivers who are in poorer states of health, or possess fewer financial resources, are at greater risk for depression (Donaldson, Tarrier, & Burns, 1997; Schulz, O'Brien, Bookwala, & Fleissner, 1995). Previous studies have shown that both caregiver and patient characteristics are relatedto caregiver depression (Schulz, O'Brien, Bookwala, & Fleissner, 1995). In a 2003 study by Covinski and his colleagues, caregivers reported a mean of 4.4 symptoms of depression. Thirty-two percent had 6 or more symptoms of depression and received a classification of “depressed”. Patient qualitieslinkedto higher rates of caregiver depression included age (those younger had higher rates), being married, male gender, Hispanic ethnicity, higher levels of ADL dependence, lower levels of education, incontinence, lower MMSE scores, and presence of problem behaviours (Covinski et al., 2003). Researchers use the term “caregiver burden” in reference to the type of stress experienced by the caregiver as a result of providingcare. Caregivers provide support for various reasons including schizophrenia (Foldemo, Mats, Ek, & Bogren, 2005), depression (Jeglic et al., 2005), and other severe mental disorders (Ivarsson, Sidenvall, & Carlsson, 2004). Burden can result from a number of different caregiving situations including those listed above as well as others.The term "caregiver burden" refers to the emotional response of a person resulting from changes and difficulties that arise as they give assistance and support to the person they are providing care for. Reports of fair or poor condition are more likely to come from informal caregivers than from noncaregivers (Baumgarten, 1992). Theoretical research on caregiver burden has shifted to other instruments, which have been developed in order to tap burden as a multidimensional construct and are grounded in theoretical analyses of burden (Vitaliano, Russo, Young, Becker, & Maiuro, 1991). Numerous tests have been created for the express purpose of measuring this stress. Some studies have used single item indicators of stress,the Screen for Caregiver Burden (Vitalino et al., 1991), the Perceived Stress Scale (Cohen, Kamarck, & Mermelstein, 1983), the Caregiver Strain Index (Robinson, 1983), and Montgomery and Borgatta’s caregiver burden scale (1989), all of which are among psychometric tools used by researchers to assess burden. Of significance in this thesis are The Center for Epidemiologic Studies Depression (CES-D) Scale (Radloff, 1977), The Zarit Burden Interview (ZBI), designed by Professor Steven Zarit, and the 12 item brief BI (SF-12) proposed by Bédard and colleagues (2001).

The Center for Epidemiologic Studies Depression (CES-D) Scale (Radloff, 1977) is an incorporation of previously devised depressive inventories, including Zung’s depression scale (Zung, 1965), Beck’s Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961), and a scale developed by Raskin (Raskin, Schulterbrandt, & Reatig, 1966), among others. Twenty items are included on the scale which survey mood, somatic complaints, interactions with others, and motor functioning. The responses are valued on a 4-point Likert type scale, with range 0-3, and anchor points in terms of days per week ‘rarely or none of the time (less than one day)’ to ‘most or all of the time (5-7 days)’. Possible scores on the scale range between 0 and 60. High depressive symptoms are indicated by a score of 16 or higher (Frerichs, Aneshensel, & Clark, 1981; Goldberg, Van Natta, Comstock, 1985). The scale has been determined to be a valid screening tool used to detect depressive symptoms in both the general population as well as in psychiatric populations (Weissman et al., 1977). A score of 16 generally represents an individual who has reported at least six items to be habitually present over the course of the previous week, or most of all 20 items to be present for a shorter period of time. The CES-Dwas originally developed to measure levels of depressive symptomatology in adults (Motl, Dishman, Birnbaum, & Lytle, 2005). The scale contains a series of 20 questions and has been used in numerous studies (Comstock and Helsing, 1976; Frerichs et al., 1981;Murrell, Himmelfarb, & Wright, 1983;Radloff, 1977; Roberts, 1980; Roberts & Vernon, 1982; Weissman et al., 1977). The scale meets all necessary qualities for psychological tools as it is valid, reliable, and internally consistent (Roberts, 1980; Roberts & Vernon, 1982). The CES-D has been an important tool in the mental health community since its implementation and first use in the nineteen seventies. The scale has also been used in the Community Mental Health Assessment Surveys in the seventies (Comstock & Helsing 1976; Radloff & Locke, 1986), and use in the National Health and Nutrition Examination Surveys (Eaton & Kessler, 1981). It has since been transformed in use to the telephone and self administered version and has the added value of being understandable and usable to underrepresented segments of the population such as the economically depressed or elderly, ergo explaining its role in caregiver burden measures as elderly citizens characterize a segment of the population in which care is needed more frequently. The CES-D is a very popular scale as well as being a widely used instrument in psychiatry (Murphy, 2002; Naughton, & Wiklund; Nezu et al., 2002, 1993; Snaith, 1993). The scale is quite respected and widely used, as various searches for the CES-D term bringup thousands of results on the PsychINFO electronic database available online (current as of April, 2005). Internal consistency estimates range from 0.8 to 0.9; test-retest reliability, ranging from two weeks to one year, is reported to be between 0.4 to 0.7 (Devins, Orme, Costello, & Binik, 1988; Radloff, 1977). Four factors have consistently, but not always, been reported for the CES-D (Devins et al., 1988; Golding & Aneshensel 1989; Radloff, 1977; Sheehan et al., 1995): depressed affect, positive affect, somatic and retarded activity, and interpersonal relationships. The factors appear to be normally strong over time (Sheehan, Fifield, Reisine, & Tennen, 1995) and in populations varying in ethnicity (Golding & Aneshensel 1989; Golding, Aneshensel, & Hough, 1991) and health status (Devins et al., 1988). An important early validation of the CES-D was done by Weissman and colleagues (Weissman et al., 1977). The CES-D displayed concurrent and discriminate validity in a study of five samples (Weissman et al., 1977). For example, CES-D scores were highly correlated with the SCL-90 (Derogatis et al., 1973), and it was able to discriminate between psychiatric patients being treated for depression and other psychiatric patients, as well as to identify patients with a secondary diagnosis of depression among those with a primary diagnosis of alcoholism, drug addiction, and schizophrenia. High sensitivity and specificity (based upon a cut off score of 16+) were reported for people with major depression, schizophrenia, and alcohol dependence, though it was less accurate for people with a history of depression and those currently in treatment for drug dependence. Scores do not appear vulnerable to sensitivity of symptom overlap between the seriously medical ill, nor type of administration, that is, whether the respondent or a rater scores the survey (Devins et al., 1988). Recent studies involving the CES-D have been aimed at understanding depressive symptomatology among chronically ill persons (Chwastiak et al., 2002, Clark et al., 2002) as well as their caregivers (Tong et al., 2002). The average estimate of internal consistency as measured by Cronbach’s alpha was .88 (SD = .05, median = .89). This average suggests a strong internal consistency across studies (O’Rouke, 2004), (i.e., .90 ≥ α ≥ .80; Clark & Watson, 1995; DeVellis, 1991).

The Burden Interview (ZBI) by Zarit and his colleagues is the most commonly used burden measure in the intervention research literature (Knight et al., 2000). The ZBI is a commonly used test of caregiver burden due to its structure and detail, as well as its ease of implementation and evaluation. There are a number of factors which manipulate the stress levels felt by those giving care. Burden score is subject to variance depending on the setting. Examples may include living at home with the care recipient, placing him or her in a nursing home or different care setting, or living away from the individual requiring such care, while they live in their own home. The potential for burden is present in each of these instances. Coyne (1976) postulated that depression has disadvantageous effects on interpersonal relationships. In general, wives who are caregivers tend to receive higher scores of burden than husbands who are caregivers with negative comments by depressed wives more likely to be met with negative comments by their husbands (Hops et al., 1987). A 2005 study by Jeglic and her colleagues found that the effects of subjective burden were not correlated with gender, but that gender of the depressed spouse may affect communication of the burden between members. As with many other responsibilities, both emotional and physical qualitiesadd to caregiver burden. Individuals may feel anger or resentment or be otherwise more emotionally invested in the caregiving role, while others may have anindolentdeportment. As with any task, experience will help caregivers adapt to the role and allow for better control of the burden felt. With the psychological stress associated with caregiver burden recognized (Zarit et al., 1980), the need for additional psychometric instruments to measure burden is present. Zarit expressed that burden arises as a function of one’s subjective perceptions that current and future resources are inadequate to meet the demands of caregiving. Caregiving can affect a person in multiple ways, which creates a demand for burden measure scales which offer more than a single summary score. Novack and Guest identified five factors that they named Developmental Burden, Emotional Burden, Physical Burden, Time Dependence Burden, and Social Burden. Lawton and colleagues (1989) identified five factors for the construct of burden: Impact of Caregiving, Subjective Burden, Caregiving Mastery, Caregiving Satisfaction, and Traditional Caregiving Ideology (Lawton, Kleban, Moss, & Rovine, 1989). Similar to the CES-D scale, the ZBI includes factors which include aspects of the emotional costs of caregiving as well as the social impact on the individual providing care. The Burden Interview (Zarit, Orr, & Zarit, 1985) consists of 22 questions regarding subjective strain for caregivers providing aid for those with dementia. A 5-point Likert type scale is used to rate the users endorsement of the questions. The possible range for the Burden Interview (BI) scores is 0-88, with higher totals reflecting greater burden. Knight and his colleagues (2000) found that 14 of the 21 non-global items were found to tap three underlying factors of Embarrassment/Anger, Patient’s Dependency, and Self-criticism. The remaining seven factors did not load on these three factors nor coalesce to form a meaningful additional factor (Knight et al., 2000). In a study by Hassinger and Zarit (1986), three factors referred to as Caregiver Anger, Patient Dependency, and Caregiver Lack of Privacy were identified. This factor structure was replicated in a second study of 108 caregivers (Knight et al., 2000). More recently, shortened versions of scales such as the ZBI have been developed (Bédard et al., 2001). The need for brief scales is based on the premise that burden will be measured more consistently if such a scale can be identified while maintaining reliability and validity (Knight, Fox, & Chou, 2000). In 2001, Bédardand his colleagues produced a 12-item (SF12; short) and a 4-item (screening) version of the ZBI. The group cited correlations between the short and full version which ranged from 0.92 to 0.97, and from 0.83 to 0.93 for the screening version. In essence, a reduced version of the longer itemed scale can be just as effective while retaining many of the same properties. This fact is of significant interest to researchers in the field who will be able to produce the same results using the shortened version of the scale while saving time in the data collection portion of their studies. This makes execution and use of the instrument simpler with no significantdeficiencies in the effect. The shorter 12 item BI proposed by Bédard and his colleagues (2001) is said to have responses which represent two fundamental constructs (personal strain and role strain). Acceptable measures of internal consistency (i.e., α = .88 and α = .78, respectively) were found. Based on results of their 2001 study, Bédard and colleagues conclude that their scale is an effective measure of caregiver burden despite its brevity in comparison with the original 22 question ZBI. Results of a 2003 study by O’Rourke and Tuokko, show support for the psychometric properties of this brief BI. Despite reduction from the original scale, no loss of reliability or validity appears to be to present (O’Rourke & Tuokko, 2003). The conclusion can be drawn that this brief BI is appropriate for implementation despite its briefness.Understanding the factor structure of the CES-D scale, ZBI, and the SF12 have important implications in interpreting the wide range of literature which directly uses one of these scales in their respective analyses.

The use of the CES-D and ZBI are implemented individually to assess depression and caregiver burden, respectively, often for the same group of subjects. Of interest is any type of relationship between the results of either scores, i.e., the question arises of whether or not depression varies with burden in some meaningful way, or if either is mediated by a separate variable. In a study on Alzheimer’s disease conducted by Clyburn and his colleagues in 2000, results showed that effects of the caregiver’s well-being were mediated by appraisals of burden. In Model 2 of the study, a good fit of the data (X2[4] = 4.86, p = .301; GFI = .997; AGFI = .987) was found. The parameters were significant which indicated linkages to burden from high depressive behaviour, community residence of the person receiving care, and low informal help, as well as caregiver burden to caregiver depression. A moderate total coefficient of determination for the structural equations (.304) was found (Clyburn, Stones, Hadjistavropoulos, & Tuokko, 2000). Other studies have shown that caregivers of dementia patients are at a significantly greater risk to be depressed than are noncaregiver controls (Dura, Stukenberg, & Kiecolt-Glaser, 1991).

In 1987, Drinka and her colleagues did a study comparing burden and depression among caregivers. In their study, caregivers were identified as persons having daily contact with the patient and who assumed primary caregiver responsibility for the patient’s care. Mean age of the 127 caregivers was 70 years, with an age range of 25 to 92 years. A later regression analysis showed significant positive correlations between caregiver depression and burden, and patient depression scores. These findings significantly correlate depression and a caregiver’s feelings of burden, which shows that thefurther depressed a caregiver is,the more burdensome the caregiver perceives the care. A proceeding repercussion is that the caregiver may resist giving even the minimal amount of care if the burden is perceived as severe (Drinka, Smith, & Drinka, 1987).

Other studies have also shown that feelings such as depression have been experienced by those providing care. Two distinct samples of caregivers were used in a study on prevalence of depression in family caregivers (Gallagher et al., 1989). Both help seekers and non-help seekers volunteered themselves and their relative for participation in a longitudinal study focused on biological and psychosocial changes that occur during the course of Alzheimer’s disease. The overall data for the help-seeking sample showed that 46% met criteria for one of three types of depressive disorders (major, minor, or intermittent). An additional 22% had evidence of depressive features (depressed mood and one or two other symptoms), with the latter not having significant enough results to warrant diagnosis. In a 1995 study by Fuller-Jonap and Haley of caregiving for spouses with Alzheimer’s disease, results from the analysis of the BSI (Derogatis & Spencer, 1982) indicated that the mental health of caregivers was less positive than that of married men of the same age, income, and educational status who did not care for a cognitively impaired wife. Later analysis showed that this difference was due solely to differences in depression (Fuller-Jonap & Haley, 1995). Based on empirical evidence, depression levels will like exhibit a relationship with burden.

Methods

Participants

100 spouses were recruited (50 husbands / 50 wives) who were caring for a person with Alzheimer’s disease. Half of the sample was obtained in a remote urban region (Thunder Bay) and the other half in a dense urban region (Hamilton). Level of care recipient’s cognitive impairment was balanced across spouses and region by stratification. Sources of participants included the local chapters of the Alzheimer’s society, local geriatricians, and public advertising in local media (press releases, papers, cable television). The recruitment strategy was intended to recruit a varied sample of caregivers to maximize the generalizibility of the findings.

Presentations were made to support groups of the Alzheimer’s Society to obtain contact information of caregivers who expressed interest and met inclusion criteria. Society’s staff and geriatricians were also encouraged to discuss the study with potential participants. For these potential participants, the Society’s staff and geriatricians asked for permission to release names and contact information to the investigating team. The caregivers were then contacted. The study was briefly explained over the phone before an in-person interview was scheduled. The inclusion criteria for the project included a care recipient diagnosis of AD according to NINCDS-ADRDA criteria (McKhann et al., 1984) or DSM-IV criteria, and no other major concurrent illness requiring substantial management (e.g., diabetes). Caregivers lived alone with the care recipient and answered questions in English.