Copyright 7997 by The Cerontological Society of America

This paper critiques and discusses 10 measures of burden that have been used with caregivers of individuals with dementia. Current measures target diverse caregiver and care recipient populations, and vary in conceptualization and measurement of burden. Although many measures contain reliability data, relatively few report content validity, convergent/divergent validity, criterion validity, or sensitivity to change. Implications of these limitations for research and clinical practice are discussed. Key Words: Alzheimer's disease, Specificity, Psychometric properties, Distress

Burden: A Review of Measures Used Among Caregivers of Individuals with Dementia1

Historically, Grad and Sainsbury (1963) were the first to acknowledge the impact or burden felt by family members who care for mentally ill relatives. Since that time, burden has become a commonly recognized phenomenon. Caregiver burden has been defined as "the physical, psychological or emotional, social, and financial problems that can be experienced by family members caring for impaired older adults" (George & Gwyther, 1986, p. 253). Alzheimer's disease (AD), in particular, is an illness that places unrelenting and progressive demands for care on caregivers because behavioral and cognitive deficits restrict the ability of the impaired elder to perform activities of daily living. Zarit, Reever, and BachPeterson (1980) were among the first researchers to study burden in caregivers of demented relatives. Since this pioneering work, the burden associated with the care of individuals with AD has been the focus of much research interest. Burden has been examined in various samples and has been conceptualized in several ways (Poulshock & Deimling, 1984; Stephens & Kinney, 1989). Thompson and Doll (1982) explicated the distinction between objective burden (e.g., disruption of family life) and subjective burden (e.g., caregiver response to the situation). These authors suggested that objective burden reflects disruptions in finances, role, family life, supervision, and neighbor relations, whereas subjective burden refers to feeling embarrassed, overloaded, trapped, resentful, and excluded. In this view, objective and subjective burden are distinct and are hypothesized to relate differently to care recipient and caregiver variables. Yet the con-

"This research was supported by the National Institute of Aging, RO1 AGO-6770-03, and National Institute of Mental Health, RO1 MH-43267-02. We thank Roslyn Siegel and loanna Plattner for their assistance. Address correspondence to: Dr. Peter P. Vitaliano, Department of Psychiatry and Behavioral Sciences, University of Washington, RP-10, Seattle, WA 98195. •department of Psychiatry and Behavioral Sciences, University of Washington.

Vol. 31, No. 1,1991

founding of "objective" and "subjective" sources of burden remains problematic in many measurement approaches. The importance of conceptual clarity was discussed cogently by Stephens and Kinney (1989), who urged distinction between sources of stress and appraisal of stressors. Establishing construct validity for burden measures has been hampered by two issues: uncertainty about expected relationships between burden and functioning of the impaired elder, and limited elucidation of potential relationships of burden and more general psychological criteria. Conflicting relationships have been reported between self-reported caregiver burden and care recipient functioning (Deimling & Bass, 1986; Pearson, Verma, & Nellett, 1988; Pruchno & Resch, 1989; Zarit et al., 1980). A number of authors have reported increased caregiver burden or stress with higher impairment (Miller, 1987; Greene et al., 1982; Nygaard, 1988; Pearson, Verma, & Nellett, 1988; Poulshock & Deimling, 1984). In contrast, Zarit et al. (1980) found that burden was not associated with severity of behavioral problems in their sample. Documented relationships between situational stress and caregiver responses over time might vary according to changes in functioning of the impaired individual, caregiver adaptation/decompensation, or methods variance. For example, Pruchno and Resch (1989) have suggested that burden does not follow a linear trajectory with functional/behavioral impairment, but peaks in the middle phases of AD when behavioral disturbances are most problematic. Haley and Pardo (1989) provide evidence of the multidimensional problems associated with dementia, suggesting that different aspects of the situation are stressors at different times on the trajectory. Interactions between age, sex, and level of impairment also influence reported burden (Fitting et al., 1986), confounding the relationship between burden and functioning. Using the stress, appraisal, and coping framework of Lazarus and Folkman (1984), Townsend and col67

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

Peter P. Vitaliano, PhD,2 Heather M. Young, RN, PhC,2 and Joan Russo, PhD2

tionships with general psychological criteria, and other psychometric properties. A methodological review of burden measures is in order because their use and misuse can have major implications for both the interpretation and utilization of research findings. In a recent conceptual review of caregiver stress instruments, Stephens and Kinney (1989) assessed the fit of eight measures to the cognitive-phenomenologic model of stress (Lazarus & Folkman, 1984). The authors reviewed the psychometric properties and content of the instruments and found considerable variation in the focus of the measures and the conceptualizations of caregiver stress and burden. This paper complements their review by offering additional methodological information about these and other burden measures. Tables 1 and 2 provide outlines of our review of 10 burden measures. These include content validity and conceptualization (Table 1) and psychometric properties, such as reliability and construct validity (Table 2). Measures were included only if they had been used in populations with dementia. (The Cost of Care Index [Kosberg & Cairl, 1986] was not included in our review because it has been used primarily as a clinical as opposed to research tool. Moreover, psychometric properties were not available for this measure.) The measures were identified in a computer-

ized search using Medline and Psychological Abstracts since 1980; subsequent work was identified using the Scientific Citations Index. Each measure is

Table 1 . Content Validity of Burden Measures: Specificity and Conceptualization of Burden

Specificity

Study

Conceptualization Relationship to care recipient (N)

Care recipient population

Objective burden (OB)

Subjective burden (SB)

Zarit etal. (1980)

Senile dementia

S (18)

Yes-

Yes-

Greene etal. (1982)

Senile dementia

S (15) C(16),O(7)

Yes"

Yes"

Rabins etal. (1982)

Dementia

S, C (55)

Yes

Yes

Robinson (1983)

Postdischarge hip surgery/ heart disease

S(32) C(27) O(25)

Yes

No

Poulshock & Deimling (1984)

Frail elderly

S (307) C (307)

Yes

Yes

Montgomery etal. (1985)

Frail elderly

Yes"

Yes"

C(62),O(17) Lawtonetal. (1989)

Demented elderly respite/ institution

S (284)/(24) C (272)/(158) O (51)/(34)

Yes-

Yes"

Kinney & Stephens (1989)

Alzheimer's disease

S(28) C(25) 0(7)

Yes

Yes

Novak & Guest (1989)

Alzheimer's disease

S(46) C(50) O(9)

Yes

Yes

Vitalianoetal. (1991)

Alzheimer's disease

S(79)

Yes

Yes

Note: S = spouse; C = child; P = parent; O = other. -Distinct scores do not exist; however, items reflect both OB and SB. "OB and SB scales tap different content domains.

68

The Gerontologist

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

leagues (1989) demonstrated that the mental health of caregivers did not deteriorate over time, but rather actually improved, indicating successful adaptation to the situation. Finally, a number of methodological factors may have contributed to the lack of consistent relationships between burden and functioning among studies: diverse definitions and conceptual models of burden and related variables; the use of varied measurement approaches with, in some cases, limited psychometric justification; and the application of nonspecific burden measures to different populations of caregivers and care recipients. A more clinical issue relevant to the construct validity of burden measures concerns potential relationships between caregiver burden and general psychological criteria (e.g., depression, anxiety, wellbeing). Psychological symptoms are potentially treatable, yet relationships between these constructs and burden have not been well explicated. As suggested by Anthony-Bergstone, Zarit, and Catz (1988), identification of sources of psychological distress would facilitate the development of treatment programs for caregivers. Furthermore, exclusion of general psychological criteria from caregiver research has limited the comparability of results to noncaregiver groups and to normative data (George & Gwyther, 1986). Given these issues, the purpose of this paper is to review burden measures used with dementia samples. The measures are examined for specificity, rela-

Table 2. Psychometric Properties of Burden Measures Study

Number of items

Construct validity

Reliability

Zaritetal. (1980)

22

.91' .71"

Brief Symptom Inventory; morale

Greene etal. (1982)

15

.85"

Patient behavior, functional limitations, & mood disturbance; caregiver distress

Rabins et al. (1982)

52

NA

NA

Robinson (1983)

13

.86"

Care recipient mental status and ADLs; caregiver perceptions, anxiety, and emotions

6'

NA

Care recipient cognitive and ADL ability; impact on family relationships; depression

Poulshock & Deimling (1984)

13d Montgomery etal. (1985)

9

.88W.80'11

Depression, disruptive care recipient behavior, social functioning

.851 .86'

Assisting with physical transfer

47

.85V. 87'

Emotional burden; negative and positive quality of relationship

Kinney & Stephens (1989)

42

.91V.83"

Care recipient impairment; caregiver hostility, anxiety, depression

Novak & Guest (1989)

24

.85, .85, .86, .73, .77

NA

Vitalianoetal. (1991)

25

.85-.891 .70"

Care recipient cognitive/ADL functioning; caregiver anger, morale, anxiety, depression

•'Internal consistency. "Test-retest. 'Burden. ''Impact. 'Negative changes in family relationships subscale, eight items. 'Activity restrictions subscale, five items. ^Criterion: relationship (p < .05) with burden.

presented chronologically and discussed relative to the outline. We employed three criteria in evaluating each measure: specificity with respect to both care recipient population and relationship of caregiver to care recipient (content validity); conceptualization in the measurement of burden (objective and subjective burden); and psychometric properties, including reliabilities and construct validity. The measures reviewed ranged broadly along these criteria. Only three scales (Kinney & Stephens, 1989; Novak & Guest, 1989; Vitaliano et al., 1991) were designed specifically for AD, whereas five others targeted individuals with dementia of mixed etiology. The remaining measures were developed for frail elderly (i.e., Greene et al., 1982) and postdischarge patients (i.e., Robinson, 1983). They had been used in subsequent studies of caregivers of individuals with dementia. Ten Measures of Burden

The Burden Interview The Burden Interview (Bl) (Zarit et al., 1980; Zarit & Zarit, 1987) is a 22-item self-report inventory that examines burden associated with functional/behavioral impairments and the home care situation. The items possess content validity as they were derived from clinical and research experience with caregivers of individuals with dementia and reflect common areas of concern, namely, health, finances, social Vol. 31, No. 1,1991

life, and interpersonal relations. The items are worded subjectively, focusing on the affective response of the caregiver. A 5-point scale (ranging from never to nearly always present) is used. Psychometrically, such an approach may have inherent problems in that the occurrence of an experience or problem does not necessarily imply distress in response to that experience (Vitaliano et al., 1984). Many of the items on the Bl, however, are worded so that distress can be inferred (e.g., "Do you feel angry when you are around your relative?"). The Bl has high internal consistency (alpha = .91) and test-retest reliability (alpha = .71) (Gallagher et al., 1985). Construct validity was examined in relation to both a single global rating of burden and the Brief Symptom Inventory (Zazit & Zarit, 1987). Both studies by Zarit et al. (1980) and Fitting et al. (1986) report no relationship between severity of dementia and Bl scores; however, the latter authors found that burden increased with dysfunction when controlling for caregiver age and sex. Pratt, Schmall, and Wright (1986) provide evidence of construct validity in that burden was negatively related to morale and positively related to hours spent giving care. Whereas the Bl focuses on consequences of caregiving, problems of the impaired elder and perceptions of these demands as stressful are measured by using the Memory and Behavior Problems Checklist (Zarit & Zarit, 1987). This 30-item checklist is scored on two grids, one to measure frequency of behavior, and the other to measure caregiver reactions (e.g.,

69

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

Lawton etal. (1989)

Behavior and Mood Disturbance Scale, Relatives' Distress Scale Greene and colleagues (1982) developed two scales, the Behavior and Mood Disturbance Scale (BMDS) and the Relatives' Stress Scale (RSS), in a sample of community-dwelling individuals with dementia and their primary caregivers. The BMDS provides a frequency count of 34 common behavioral problems, scored on a 5-point scale (never to always; test-retest reliability = .84). The 15-item RSS is also scored on a 5-point scale (not at all to considerably) assessing severity of affective responses and disruption of family and social life (test-retest reliability = .85). Construct validity of the RSS was supported in relation to the BMDS (Greene et al., 1982), disruptive behaviors and depression of the impaired elder, and caregiver distress (Pearson, Verma, & Nellett, 1988). Factor analysis established three factors corresponding to subscales of each scale; however, these results are questionable given the very small sample size (n = 38) relative to number of items (K = 49). A major strength of this approach is the inclusion of both the frequency of and distress associated with specific caregiving experiences. The items, however, tap distinct content domains, with no correspondence between measures of actual problems (objective burden) and appraised distress associated with specific caregiving experiences (subjective burden). Rabins, Mace, and Lucas' Structured Interview Rabins, Mace, and Lucas (1982) developed a structured interview (52 items) that is similar to the Bl (Zarit et al., 1980), but that also includes an assess70

ment of the prevalence of problems associated with the care of an individual with dementia. Caregiving situations are rated on a4-point scale (not occurring; occurring occasionally, but not a problem; occurring sometimes and causing problems; causing serious problems). An advantage of this approach is borne out in the observed discrepancies between the rank order of a problem's occurrence and its seriousness — for example, violent behaviors occurred infrequently, but when they did occur, they were rated as serious. Such information is lost when one simply asks whether one is upset by the presence of violent behavior. Unfortunately, no psychometric data have been reported for this measure. Caregiver Strain Index In the tradition of Pearlin and Schooler (1978), some authors have viewed burden (or "strain") as tantamount to distress, so that the presence of a problem is considered evidence of strain. The 13item Caregiver Strain Index (CSI; Robinson, 1983) is an example of such an approach. The CSI was developed inductively. Scoring is dichotomous, and as with the Bl (Zarit et al., 1980), subjective burden can be inferred only through the endorsement of certain items (e.g., feeling completely overwhelmed) . The scale has high internal consistency (alpha = .86). Construct validity was examined and was supported in relation to care recipient characteristics, caregivers' subjective views of the caregiving situation, and the emotional health of the caregivers. The CSI reflects common stressors associated with caring for a patient postdischarge, following acute hospitalization for hip surgery or heart disease. This may limit its usefulness in studies with caregivers of different types of care recipients (e.g., individuals with AD). In fact, the content validity of burden measures designed for one type of care recipient or caregiver population is questionable in another type of population. The practice of adapting scales, such as the CSI, for use in populations with AD (Pruchno, 1990) warrants careful consideration. The particular caregiving demands related to features of specific illnesses has been discussed by Silliman and Sternberg (1988). Poulshock and Deimling's Model Problems in comparing studies of burden have been documented by Poulshock and Deimling (1984). They proposed a model of burden that treats functional/behavioral impairment as the independent variable, impact on life as the dependent variable, and emotional responses of the caregiver ("burden") as the mediator in this relationship. Impairment is operationalized as dependency (number of ADL's requiring assistance) and three mental impairment indicators (factors derived from an analytic model using 23 items). The two burden measures correspond to dependency and mental impairment, respectively. For dependency, caregivers rate their assistance with each ADL as tiring, difficult, or upsetting (scoring one point for each, range = 0-3). For The Gerontologist

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

how upset one is in response to a behavior, scored 04). Cuttman split-half reliabilities for the two scales were .65 and .66, respectively. Validity was established in relation to both care recipient mental status and caregiver symptoms (Zarit & Zarit, 1987). Major strengths of the approach taken by Zarit et al. (1980) include the broad scope of situational stimuli considered as sources of burden (i.e., affective responses and life changes that might contribute to burden beyond the burden associated with impairment alone) and the subjective nature of the Bl (i.e., items are worded as the affective response of the caregiver). The scoring of the Bl, however, precludes separate evaluation of prevalence of experiences and caregiver distress (i.e., both objective and subjective burden). Although the Bl is complemented by the Memory and Behavior Checklist, which provides an excellent assessment of care recipient-centered problems, demands such as the degree of disruption in family and social life related to caregiving are not examined thoroughly. When used together, the Bl and Memory and Behavior Checklist offer a complete view of problems of the impaired elder and caregiver responses to such problems. However, there is no measure of caregiver-centeredproblems to complement the Memory and Behavior Checklist. This omission limits comparison of care recipientcentered and caregiver-centered sources of burden.

Montgomery, Gonyea, and Hooyman's Inventories Montgomery, Conyea, and Hooyman (1985) extended the conceptualization and measurement of objective and subjective burden. Objective burden, defined as changes and disruptions, was measured by using a 9-item inventory with each item rated on a 5-point scale (Cronbach's alpha = .85). Items are worded generally, and cover finances, privacy, social activity, health, and interpersonal relationships. SubVol. 31, No. 1,1991

jective burden, defined as attitudes and emotional responses of the caregivers, was measured by using a 13-item inventory, each item rated on a 5-point scale (Cronbach's alpha = .86). Construct validity was supported only in relation to one ADL item, assisting with physical transfer. These items were adapted from the Burden Interview (Bl; Zarit et al., 1980); see critique above of the Bl. Also, there is no conceptual correspondence between the items on each scale, and it is difficult to assess which specific demands are distressing to the caregiver. Finally, the validity of using an inventory developed on demented elderly (i.e., the Bl), for the study of community-dwelling frail elderly has not been addressed. Caregiver Appraisal Measure A recent conceptual contribution suggests that caregiver burden be expanded to a more general construct, namely, caregiver appraisal (Lawton et al., 1989). Such an approach is in keeping with the cognitive-phenomenologic stress model (Lazarus & Folkman, 1984), in which appraisal is not inherently negative or positive. Using this model, Lawton etal. (1989) developed a Caregiver Appraisal Measure by supplementing current measures of burden to include potentially positive dimensions of the appraisal of caregiving experiences. The items (e.g., "Embarrassed over impaired person's behavior?"; "Helping the impaired person has made me feel close to him/her") are scored by endorsement using one of two 5-point scales (never to nearly always true or strongly agree to strongly disagree). The 5-point scale provides an indication of frequency (as with the Bl; Zarit et al., 1980), and distress/satisfaction must be inferred. Five dimensions (subjective caregiver burden, caregiver satisfaction, caregiver impact, caregiver mastery, and traditional caregiver ideology) were hypothesized a priori and received some support in a factor analysis. The instrument was tested on two groups (n = 632 and n = 234) of caregivers of demented elderly. Internal consistency reliability coefficients of the subscales ranged from .65 to .87. Construct validity was established in relation to affective states, quality of relationship with care recipient, and emotional burden. Broadening the scope of relevant caregiver experiences is a major strength of this approach. By using the appraisal model, the complexities of the interrelationships between positive and negative aspects of caregiving may be explored. Although the dimensions of appraisal were not conclusively confirmed, this study has provided an exciting new foundation for future caregiver research. As Lawton et al. (1989) suggest, improvement of the psychometric quality of measures of burden, satisfaction, and impact should enhance our understanding of caregiver appraisal. Caregiver Hassles Scale The Caregiver Hassles Scale (CHS) was developed to measure the day-to-day demands of caregiving (Kinney & Stephens, 1989). This 42-item measure was rationally derived, and subsequently refined, to rep71

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

mental impairment, caregivers rate how upset they are (on a 3-point scale) in relation to the three mental impairment indicators (sociability, disruptive behavior, and cognitive incapacity). The dependent variable, impact, is operationalized as disruption in family life (e.g., job conflict, finances, interpersonal relationships) and is measured by using items akin to the Bl (Zarit et al., 1980). The items were derived by using factor analysis, resulting in two subscales: Negative changes in family relationships (eight items, coefficient alpha = .88) and restrictions in activity (five items, coefficient alpha = .80) (Deimling & Bass, 1986). There is evidence of construct validity through observed relationships between the burden measures and care recipient cognitive and ADL ability, impact on family relationships and depression, and decision to institutionalize (Poulshock & Deimling, 1984; Deimling & Poulshock, 1985); and between disruptive care recipient behaviors and both negative changes in relationships and activity restrictions (Deimling & Bass, 1986). Poulshock and Deimling (1984) made a major contribution to the conceptualization of burden by measuring this construct in relation to actual caregiving problems and care recipient impairments. Moreover, there is a conceptual correspondence between functional/behavioral impairments and the burden items. The construct of impact is a more distal indicator of caregiver stress that is similar to the conceptualizations of other researchers (e.g., Zarit et al., 1980; Robinson, 1983). A limitation of the model is that distress associated with the disruption of family life (impact) is not considered. Psychological distress, operationalized as depression, is considered primarily as an antecedent to burden. Such a conceptualization does not account for the impact itself as a further source of burden. The feedback loops among these variables warrant further exploration. To reduce this circularity, alternate caregiver outcomes, such as symptoms of psychological distress or life satisfaction, might be considered. Another problem with these measures is that the scaling of the burden items (tiring, difficult, upsetting) is not continuous, limiting their usefulness as a scale and precluding exploration of the measures' psychometric properties. Finally, the study sample on which these measures were developed consisted of caregivers of frail elderly. The Negative Impact on Family Relationships subscale has been used in a sample with dementia (Haley et al., 1987); the content validity of these measures among caregivers remains to be established.

Caregiver Burden Inventory The Caregiver Burden Inventory (CBI) (Novak & Guest, 1989) is a multidimensional measure of caregiver burden. Items were derived from the experiences of caregivers of individuals with AD and review of the burden literature. The 24 items load on five factors: time-dependence burden, developmental burden, physical burden, social burden, and emotional burden. Responses to each item range from 0 (not at all descriptive) to 4 (very descriptive), yielding maximum subscale scores of 20 for Factors 1, 2, 4, and 5, and 16 for Factor 3. As with the Bl (Zarit et al., 1980), the items include both affective responses and task-related sources of burden, mixing subjective and objective burden. Moreover, evidence of construct validity needs to be provided. George and 72

Gwyther (1986) have noted that summary burden scores mask underlying sources of burden. The CBI overcomes this problem in that subscale scores can be used to generate a caregiver burden profile, an important step in designing therapeutic approaches for caregivers. The CBI offers a relatively short, yet comprehensive measure of burden that holds both research and clinical promise. Screen for Caregiver Burden The Screen for Caregiver Burden (SCB) is a 25-item measure designed to assess objective and subjective burden among caregivers of spouses with AD (Vitaliano et al., 1989b, 1991). The SCB is used to target potentially distressing experiences. To specify the content domain, caregivers were asked to identify the most upsetting caregiver experience in their lives. The most frequently cited stressors were translated into structured items, which tap several domains, including care recipient behaviors, disruptions in family and social life, and caregiver affective responses. Scoring of the SCB yields two scores: objective burden (OB) and subjective burden (SB). The former consists of a prevalence count of caregiving experiences, whereas the latter reflects ratings (from 1 to 4) of distress in relation to each experience. The psychometric properties of the SCB have been demonstrated in two independent samples (Vitaliano et al., 1989a, 1991). Internal consistency coefficients were .85 and .88 for OB and SB, respectively. Construct validity (convergent/divergent) was supported by relationships of care recipient behavioral and cognitive functioning with OB and caregiver distress and personality variables with SB. Criterion validity (differences in burden between AD caregivers versus controls) was demonstrated by using age- and sex-matched controls. Finally, sensitivity to change was evident by using a sample of individuals with early- to mid-stage AD: Over 15 to 18 months, a majority of the caregivers changed by a value that was greater than what would be expected by measurement error alone. Moreover, relationships occurred between increases in SB/OB and deterioration of functional/behavioral functioning, as well as between SB/OB and increases in caregiver distress. Limitations of the SCB include the specificity of content to spouses, the lack of subscales for various burden dimensions, and the inclusion of subjective statements in the OB score. For example, although the SCB provides separate measures of OB and SB, several of the actual items involve appraisals (lack of control, frustration) so that the OB scale may actually be affected by these "subjective" statements. In spite of this, OB and SB appear to reflect objective and subjective burden by virtue of their respective convergent and divergent associations with care recipient behaviors and caregiver distress. The SCB also offers some other advantages. These include parallel measures of OB and SB and data demonstrating criterion validity and sensitivity to change. The Gerontologist

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

resent five categories of hassles: those associated with assistance with basic and instrumental activities of daily living (9 and 7 items, respectively), care recipient cognitive status (9 items) and behavior (12 items), and caregiver support network (5 items). Items are scored for occurrence over the past week (yes vs. no), and for those that occurred, rated on a 4point scale ranging from "not at all a hassle" to "a great deal of a hassle." The instrument was refined and tested in a group of primary caregivers (n = 60) of individuals with probable AD. Alpha and testretest reliabilities were high (.91 and .83, respectively); however, these values, estimated with the same sample that was used to develop the scale, warrant replication with an independent sample. Construct validity was supported in relation to caregiver reports of care recipients' physical limitations, behavior and cognitive impairment, and also in relation to measures of caregiver well-being (e.g., anxiety, hostility, depression, and somatization). The CHS assesses caregiver concerns specific to the daily care of an individual with AD. Although this measure yields information about the occurrence and degree of hassle associated with each item, it does not assess subjective burden in relation to that item. The presence of a hassle does not necessarily imply burden or distress associated with that experience. For example, preparing meals for the care recipient might involve effort and planning, yet not be perceived as a burden by the caregiver. Another conceptual concern is the decision to limit responses to occurrence over the past week. Some items, such as giving medications, are time limited and task oriented, whereas other items, such as incurring extra expenses, may be pervasive worries that are felt even during weeks in which no actual extra expenses are incurred. Although designating a response frame enhances precision (Stephens & Kinney, 1989), limiting responses to such a short time period may increase measurement error. Major strengths of this study include the establishment of construct validity in relation to indicators of psychological symptoms and the specificity of the items to the population of individuals with AD and their caregivers. The CHS complements more global measures of burden.

Summary

Vol. 31, No. 1,1991

73

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

In this review of burden measures we provide evidence of the diversity of approaches, measures, and populations under study. The content validity of burden measures used for different care recipient populations has not been addressed in the burden literature nor in a recent review of available measures (Stephens & Kinney, 1989). Silliman and Sternberg (1988) have provided evidence concerning differences in demands posed to caregivers of individuals with hip fracture, stroke, and dementia. It is clinically evident that both caregiving demands and care recipient abilities vary markedly according to particular illnesses. As such, measures of burden designed for caregivers of individuals with congestive heart failure may not be appropriate for caregivers of individuals with dementia, and vice versa. For example, caregivers of the former group might be asked to provide assistance with activities of daily living, transportation, procurement of medical supplies, and advocacy within the health care system. Alternatively, caregivers of individuals with dementia may face problems of communication, supervision, and inappropriate or embarrassing care recipient behaviors. The content validity of burden measures must be considered when investigators make use of these instruments in new populations. To enhance measurement precision, we suggest that items should reflect common problems that are characteristic features of the particular illness in the target population. A second component of specificity concerns the relationship of the care recipient to the caregiver. Heterogeneous populations, including spouses, children, and others, were used in almost every study we reviewed. Yet Cantor (1983) found that burden (as caregiver strain) varies by the nature of the relationship between the caregiver and the recipient (e.g., spouse vs. child vs. friend). In light of these findings, researchers should scrutinize the content validity of measures used with diverse caregiving groups. From the perspective of role theory (Hardy & Conway, 1978), one might predict different expectations for giving and receiving care between marital dyads, parents and children, and friends. Caring for an ailing spouse might be viewed as part of the marriage commitment, and often occurs when other external demands, such as career and growing children, are diminished. Alternatively, caring for an ill parent might pose conflicts of equity when a middle-aged son or daughter is also involved in nurturing a young family and developing a career (Brody, 1981; George, 1982). Relationships with others in the social network might constitute less proximal and binding commitment and reciprocity over time, contributing to yet another kind of care expectation within friendship dyads. Furthermore, living arrangements between caregiver and care recipient influence whether the care is continuous or episodic in nature. The ecological validity of burden measures is related to this issue of specificity. With the exception of Vitalianoetal. (1991), all the burden measures

targeted diverse caregiver groups; one might expect that patterns of item endorsement might differ among spouses, adult children, and other caregivers. The contributions of various sources of burden may be masked when a global score is used. The alternate approach, using items derived in a particular caregiving group (Vitaliano et al., 1991) limits comparisons among groups and entails a more cumbersome measurement approach when examining several types of caregivers. Empirical consideration of the issue of caregiver specificity is an important next step in burden research. The second criterion examined, conceptualization of burden, yielded diverse representations of the construct. The distinction between objective caregiver experiences (i.e., prevalence) and subjective response to these experiences (i.e., distress) is crucial in understanding sources of distress and in identifying treatment priorities. Only five measures (Rabins et al., 1982; Poulshock & Deimling, 1984; Kinney & Stephens, 1989; Vitaliano et al., 1991; Novak & Guest, 1989) examined both prevalence of caregiver demands (OB) and caregiver response to these specific demands (SB). Two other measures (Green et al., 1982; Montgomery et al., 1985) include both OB and SB scales; however, there is no conceptual correspondence between the items in these scales. The scoring method of the Burden Interview (Zarit et al., 1980) and the Caregiver Appraisal Measure (Lawton et al., 1989) precludes making a clear distinction between OB and SB. The remaining measure (i.e., Robinson, 1983) explicitly assesses only OB. The importance of separating the various components of burden has been discussed by Stephens and Kinney (1989), who suggest distinguishing stressors from appraisals from impact. This aim is facilitated by scoring methods that assess both prevalence or frequency of experiences and also level of distress in relation to each specific experience. Finally, the psychometric properties of some of the reviewed measures have not been systematically and rigorously tested. Both our review of Psychological Abstracts and Medline and our follow-up review of the Scientific Citations Index revealed that with the exceptions of the CSI (Poulshock & Deimling, 1984), the CHS (Kinney & Stephens, 1989), the Bl (AnthonyBergstone et al., 1988), and the SCB (Vitaliano et al., 1991), caregiver burden has not been linked to symptoms of psychological distress (e.g., depression and anxiety). Such relationships would anchor burden to established, recognizable, and potentially treatable criteria. Criterion validity and sensitivity to change were examined in only one of the studies we reviewed (Vitaliano et al., 1991). This limitation has major implications for both clinical work and research on caregiving. Burden measures that are sensitive to changes in care recipient functioning and caregiver characteristics should be especially useful for interventions with caregivers. However, to judge intervention programs we will need measures of burden with strong psychometric properties that are also specific to particular samples of individuals with AD and their caregivers. Clearly, more work needs to

74

dence, the development and refinement of parallel measures of OB and SB is indicated. Second, the applicability of specific burden measures to diverse populations must be examined, as well as the demands made by specific care recipient populations on caregivers. The challenge for researchers of caregiver burden is to address these basic issues while maintaining relevance to broader theoretical stances and general psychological criteria. References Anthony-Bergstone, C. R., Zarit, S. H., & Catz, M. (1988). Symptoms of psychological distress among caregivers of dementia patients. Psychology and Aging, 3, 245-248. Brody, E. M. (1981). Women in the middle and family help to older people. The Cerontologist, 21, 471^80. Cantor, M. H. (1983). Strain among caregivers: A study of experience in the United States. The Cerontologist, 23, 597-604. Deimling, C. T., & Bass, D. M. (1986). Symptoms of mental impairment among elderly adults and their effects on family caregivers. Journal of Gerontology, 41, 778-784. Deimling, C. T., & Poulshock, S. W. (1985). The transition from family-inhome care to institutional care. Research on Aging, 7, 563-576. Fitting, M., Rabins, P., Lucas, M. )., & Eastham, J. (1986). Caregivers for dementia patients: A comparison of husbands and wives. The Cerontologist, 26,248-252. Gallagher, D., Rappaport, M., Benedict, A., Lovett, S., & Silver, D. (1985, November). Reliability of selected interview and self-report measures with family caregivers. Paper presented at The Annual Scientific Meeting of the Gerontological Society of America, New Orleans. George, L. K. (1982). Models of transitions in middle and later life. Annals of the Academy of Political and Social Sciences, 464, 22-37. George, L. K., & Gwyther, L. P. (1986). Caregiver well-being: A multidimensional examination of family caregivers of demented adults. The Cerontologist, 26,253-259. Grad, J., & Sainsbury, P. (1963). Mental illness and the family. The Lancet, 1, 544-547. Greene,). G., Smith, R., Gardiner, M., & Timbury, G. C. (1982). Measuring behavioral disturbance of elderly demented patients in the community and its effects on relatives: A factor analytic study. Age and Ageing, 11, 121-126. Haley, W. E., & Pardo, K. M. (1989). Relationship of severity of dementia to caregiving stressors. Psychology and Aging, 4, 389-392. Haley, W. E., Levine, E. G., Brown, S. L, Berry, ). W., & Hughes, G. H. (1987). Psychological, social, and health consequences of caring for a relative with senile dementia, journal of the American Geriatrics Society, 35, 405-411. Hardy, M. E., & Conway, M. E. (1978). Role theory: Perspectives for health professionals. Norwalk: Appleton-Century-Crofts. Kinney, J., & Stephens, M.A.P. (1989). Caregiving Hassles Scale: Assessing the daily hassles of caring for a family member with dementia. The Gerontologist, 29, 328-332. Kosberg, J. I., & Cairl, R. (1986). The cost of care index: A case management tool for screening informal caregivers. The Cerontologist, 26, 273-278. Lawton, M. P., Kleban, M. H., Moss, M., Rovine, M., & Glicksman, A. (1989). Measuring caregiver appraisal. Journal of Gerontology, 44, 61-67. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer. Miller, B. (1987). Gender and control among spouses of the cognitively impaired: A research note. The Gerontologist, 27, 447-453. Montgomery, R. ). V., Gonyea,). G., & Hooyman, N. R. (1985). Caregiving and the experience of subjective and objective burden. Family Relations, 34,19-26. Novak, M., & Guest, C. I. (1989). Application of a multidimensional caregiver burden inventory. The Gerontologist, 29, 798-803. Nygaard, H. A. (1988). Strain on caregivers of demented elderly people living at home. Scandinavian Journal of Primary Health Care, 6, 33-37. Pearlin, L. I., & Schooler, C. (1978). The structure of coping. Journal of Health and Social Behavior, 19, 2-21. Pearson, T., Verma, S., & Nellett, C. (1988). Elderly psychiatric patient status and caregiver perceptions as predictors of caregiver burden. The Cerontologist, 28, 79-83. Poulshock, S. W., & Deimling, G. T. (1984). Families caring for elders in residence: Issues in the measurement of burden. Journal of Gerontology, 39, 230-239. Pratt, C , Schmall, V., & Wright, S. (1986). Family caregivers and dementia. Social Casework: The Journal of Contemporary Social Work, February, 119-124. Pruchno, R. A. (1990). The effects of help patterns on the mental health of spouse caregivers. Research on Aging, 12, 57-71. Pruchno, R. A., & Resch, N. L. (1989). Aberrant behaviors and Alzheimer's

The Gerontologist

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

be done in the arena of assessment in order to facilitate caregiving research. In summary, studies of caregiver burden have progressed from relative isolation to become a major body of research. Currently, such research occupies an entire section of the presentations at the annual meetings of the Cerontological Society of America. Burden research now includes greater use of theoretical models than 5 years ago, and it has been integrated into general research on stress. In spite of major advances, burden research would still benefit from more refined measures that are specific to the type of caregiver/care recipient dyads being studied. Moreover, although more and more data are becoming available on reliability, and to some extent, validity, burden measures generally have not demonstrated sensitivity to change. This is important because examining change is essential to understanding relationships between variables both crosssectionally and longitudinally. Based on these comments, several issues emerge that researchers and clinicians should consider when selecting a burden measure. The first involves one's goals. For example, does a researcher wish to generalize to caregivers of individuals with AD, individuals with dementia of mixed etiology, or other care recipient groups such as those terminally ill with cancer or heart disease? If only individuals with AD are examined, then one would want to consider a measure as specific to this population as possible. To make statements about stress in general or to allow comparison with other studies of AD caregivers, a researcher should include measures of general distress (e.g., depression, anxiety) as more distal indicators of the impact of caregiving. If, however, several care recipient groups are involved, a researcher would have to supplement measures that are specific to the problems generated by such groups, with measures that involve more general caregiver demands. For example, scales should include items specific to AD (e.g., asking the same questions over and over; getting lost) as well as items reflecting generic problems faced by caregivers (e.g., physical, emotional, economic burden). To minimize total testing time, a brief measure of AD problems is desirable if it is necessary to add more general burden scales. One's particular study design also will affect one's choice of a measure. If there are many repeated measures, the researcher may have to consider a relatively brief measure with items sensitive to the illness trajectory of the care recipient. On a more practical level, both participant factors (e.g., fatigue) and the cost of administering and scoring a lengthy measure might influence one's choice. Stephens and Kinney (1989) made a cogent argument for conceptual precision and exploration of discriminant validity among the plethora of available caregiver stress measures. Our review complements their work and suggests two important areas of future research. First, understanding sources of caregiver burden can be enhanced by careful measurement of objective demands and responses to those demands. To achieve this conceptual correspon-

disease: Mental health effects on spouse caregivers. Journal of Gerontology, 44, S177-S182. Rabins, D., Mace, N., & Lucas, M. (1982). The impact of dementia on the family. Journal of the American Medical Association, 248, 333-335. Robinson, B. C. (1983). Validation of a caregiver strain index. Journal of Gerontology, 38, 344-348. Silliman, R. A., & Sternberg,). (1988). Family caregiving: Impact of patient functioning and underlying causes of dependency. The Gerontologist, 28, 377-382. Stephens, M. A. P., & Kinney, ). M. (1989). Caregiver stress instruments: Assessment of content and measurement quality. Gerontology Review, 2(1), 40-54. Thompson, E. H., & Doll, W. (1982). The burden of families coping with the mentally ill: An invisible crisis. Family Relations, 31, 379-388. Townsend, A., Noelker, L, Deimling, C , & Bass, D. (1989). Longitudinal impact of interhousehold caregiving on adult children's mental health. Psychology and Aging, 4, 393-401. Vitaliano, P. P., Becker, )., Russo, )., Magana-Amato, A., & Maiuro, R. D.

(1989a). Expressed emotion in spouse caregivers of patients with Alzheimer's disease. Journal of Applied Social Sciences, 13, 215-250. Vitaliano, P. P., Maiuro, R. D., Ochs, H., & Russo, ). (1989b). A model of burden in caregivers of DAT patients. In E. Light, & B. Lebowitz (Eds.), Alzheimer's disease treatment and family stress: Future directions for research (pp. 267-291). Washington, DC: U.S. Government Printing Office. Vitaliano, P. P., Russo,)., Carr, ]. E., & Heerwagen,). (1984). Medical school pressures and their relationship to anxiety. Journal of Nervous and Mental Disease, 172, 730-736. Vitaliano, P. P., Russo,)., Young, H. M., Becker, J., & Maiuro, R. D. (1991). The Screen for Caregiver Burden. The Gerontologist, 31, 76-83. Zarit, S. H., Reever, K. E., & Bach-Peterson, J. (1980). Relatives of the impaired elderly: Correlates of feelings of burden. The Gerontologist, 20,649-655. Zarit, S. H., & Zarit, ). M. (1987). The Memory and Behavior Problems Checklist — 1987R and the Burden Interview (Technical report). University Park, PA: Pennsylvania State University.

EMOTIONAL PROBLEMS IN LATER LIFE Intervention Strategies for Professional Caregivers By Dan Blazer, MD 1990 272pp hd $29.95 (outside US $34.50) THE MOSAIC OF CARE Frail Elderly and Their Families in the Real World By Jaber F. Gubrium, PhD 1991 192pphd $31.95 (outside US $35.80) GENDER. HEALTH. AND LONGEVITY Multidisciplinary Perspectives Marcia G. Ory, PhD and Huber R. Warner, PhD, Editors 1990 256pp hd $34.95 (outsideUS $38.50) THE VULNERABLE AGED People, Services, and Policies By Zev Harel, PhD 1990 336pp hd $40.95 (outside US $45.50)

ANXIETY IN THE ELDERLY Treatment and Research Carl Salzman, MD and Barry D. Lebowitz, PhD, Editors 1990 320pp hd $38.95 (outside US $43.80) LIFE SPAN EXTENSION Consequences and Open Questions Frederic C. Ludwig, MD, DSc, Editor 1991 240pp hd $34.95 (outside US $38.80)

ANNUAL REVIEW OF GERONTOLOGY AND GERIATRICS Volume 10: Special Focus on the Biology of Aging Editor-in-Chief: M. Powell Lawton, PhD Guest Editor: Vincent J. Cristofalo, PhD 1990 270pp hd $36.95 (outside US $41.50) PRACTICING REHABILITATION WITH GERIATRIC CLIENTS J. D. Frengley, MB, ChB, MRCP, P. Murray, MD, and M.L. Wykle, PhD, RN, Editors 1990 256pp hd $32.95 (outside US $36.80)

Springer Publishing Company 536 Broadway, New York, NY 10012* 212-431-4370 • FAX 212-941-7842

Vol. 31, No. 1,1991

75

Downloaded from http://gerontologist.oxfordjournals.org/ at University of Windsor on September 13, 2015

New In Gerontology From Springer

Burden: a review of measures used among caregivers of individuals with dementia.

This paper critiques and discusses 10 measures of burden that have been used with caregivers of individuals with dementia. Current measures target div...
1MB Sizes 0 Downloads 0 Views