| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
|
|
|
|||||||
|
|||||||||
, PhD6Correspondence: Address correspondence to Kevin J. McKee, Sheffield Institute for Studies on Ageing, University of Sheffield, Community Sciences Centre, Northern General Hospital, Sheffield S5 7AU, United Kingdom. E-mail: k.j.mckee{at}sheffield.ac.uk
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Assessment • Burden • Caregiving • Intervention
We developed the COPE Index (McKee et al., 2003) as a brief, first-stage assessment that identifies those carers who may be in need of supportive intervention and requiring a comprehensive assessment of their needs. It was hoped that such an instrument could facilitate a currently laborious and ad hoc assessment process (Audit Commission, 2004). The development of the COPE Index was underpinned by a theoretical framework that identified the importance of positive as well as negative aspects of caregiving (Kramer, 1997; Lopez, Lopez-Arrieta, & Crespo, 2005), and that prioritized the carer's subjective assessment of his or her situation as opposed to ‘objective’ measures of burden or need (Nolan, Grant, & Keady, 1996). Before we developed the index, we conducted a literature review that confirmed that there was no available instrument that met the criteria of being short and easy to complete, tapping both positive and negative aspects of caregiving, and that reflected caregivers' subjective perceptions rather than their objective circumstances (Borgemans, Nolan, & Philp, 2001).
An initial validation study of the COPE Index among 600 carers drawn from five European countries (McKee et al., 2003) reported two subscales present in the 15 items of the instrument, one 6-item scale measuring negative impact of caregiving, and one 5-item scale measuring positive value in caregiving. A further 3 items measured aspects of quality of support, and 1 item measured financial burden. The negative impact subscale had the better internal consistency of the two subscales, and it also had stronger associations with selected criterion validation instruments. However, the positive value subscale items were on the whole more strongly endorsed by the participants than were the negative impact items, suggesting that positive aspects of caregiving represent an important domain of the caregiving situation to assess.
A recent study described the use of the COPE Index with a small sample of carers of older people with dementia (n = 45) and health care practitioners (n = 10; see Roud, Keeling, & Sainsbury, 2006). Although the negative impact subscale demonstrated the strongest associations with selected criterion validity measures, the positive value subscale alone was significantly associated with a measure of carers' "personal gain" achieved through engagement with their role. A majority of carers responded favorably to all questions exploring their perceptions of the instrument's utility, and all 10 practitioners engaged in the study reported "possibly" or "definitely" using the instrument as part of a future assessment.
This study highlights the promise of the COPE Index as a brief, first-stage assessment of carers' needs. However, to date the validation and acceptability work on the index has been with relatively small, convenience samples, and the identified weaknesses of the positive value subscale require further evaluation within robust research that explores the index as a whole. The EUROFAMCARE study (Lamura et al., 2007) of informal carers of older people collected data from approximately 1,000 carers in each of six European countries. In this article we report data on the COPE Index, exploring its psychometric properties and its association with criterion validation measures.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
We carried out this strategy by means of the following: (a) identification of at least three representative macro regions in each country; (b) selection of sample regions within each macro region, based on the rural, urban, and metropolitan distribution of the population older than 65 years of age; and (c) identification of one or more representative sample zones for each sample region on the basis of the socioeconomic context and availability of services (in order to ensure the inclusion of communities with high, medium, and low availability of services, compared with available national standards). Within this sampling frame, we had carers recruited through a saturation approach, according to which the population of carers living in the chosen sample zones was contacted through a wide range of recruitment channels (i.e., door-to-door census, contacts through existing carer or older people organizations, advertisements in newspapers and on local radios, etc.).
Participants
We drew carers (N = 5,923) from six European countries: Greece (n = 1,014), Italy (n = 990), the United Kingdom (n = 995), Sweden (n = 921), Poland (n = 1,000), and Germany (n = 1,003). The definition of a carer that we used in the study was this: someone who perceived himself or herself to be a carer and who provided at least 4 hours of unpaid support per week (including organizing support but excluding only financial support or companionship) to an older person (
65 years old) living in the community or a residential or long-term-care setting.
Materials
We developed a common assessment tool (CAT) for use in the EUROFAMCARE survey; it was composed of a series of items and scales specifically developed by the partners, or selected for use from among a range of validated published instruments. When equivalent versions were not available in all languages, we had a backtranslation performed that followed established protocols. The questionnaire was tested in the field in two pilot studies carried out in each partner country. In this article we report on the CAT items and instruments used in the analyses that are relevant to the psychometric evaluation of the COPE Index.
In addition to participants' basic sociodemographic characteristics (age, gender, marital status, and occupational status), we measured the following caregiving variables: carer–elder relationship and cohabitation status, duration of caregiving in months, hours of caring to care receiver per week, and care receiver's dependency level. The latter variable was an item drawn from the COPE questionnaire (McKee et al., 2003) that had four response categories, varying from "independent" to "severely dependent."
We gathered information on carers' psychological well-being by using the World Health Organization-5 Well-Being Index (WHO-5; World Health Organization, 1998), a 5-item scale in which each item addresses an aspect of psychological health over the previous 2 weeks (e.g., I have felt cheerful and in good spirits); it asks the participant to record how she or he have felt on a 6-point scale, anchored by "all of the time" (5) and "at no time" (0). Items had a good internal consistency (Cronbach's 
= 0.87). Higher total scores on this tool meant better well-being. We took a single item assessing carer quality of life for the preceding 2 weeks from the 36-item Short-Form Health Survey (SF-36; Brazier et al., 1992); responses for this item were recorded on a 5-point scale anchored by "very good" (1) and "very poor" (5). From the same instrument a single item assessed carer general health status, with responses recorded on a 5-point scale anchored by "excellent" (1) and "poor" (5).
We assessed care-receiver dependency in detail through the use of the 10-item Barthel Index (Mahoney & Barthel, 1965). Higher total scores on this tool meant higher independence on activities of daily living. We assessed the behavioral problems of elders by means of 3 items adapted from the behavioral component of the Behavioural and Instrumental Stressors in Dementia instrument (BISID; Keady & Nolan, 1996). The items assessed engaging in unsafe behaviors, exhibiting lack of insight or engagement, and engaging in upsetting behaviors. Response options were "most of the time" (3), "sometimes" (2), "rarely" (1), or "never" (0). The items' internal consistency was satisfactory (Cronbach's = 0.78). Higher total scores on the tool indicate a higher frequency of behavioral problems.
We also assessed caregiver role inflexibility and included it in the present analyses. Two adapted items drawn from the Social Restriction Scale (McKee et al., 2001) assessed this construct. The respondent was asked, "If you were ill is there anybody who would step in to help with the care receiver?" and "If you needed a break from your caring role is there someone who would look after the care receiver for you?" The response options for both items were "Yes, I could find someone quite easily" (1), "Yes, I could find someone but with some difficulty" (2), and "No, there is no one" (3). The internal consistency of the scale comprising these two items was good ( = 0.89). Higher total scores on this measure indicate greater social restriction.
We adapted an interviewer rating procedure for measuring carers' willingness to continue caring (Gilhooly, 1986) and developed it into an item addressing the respondent's perception of his or her future role. The item asked this question: "In the next year, are you willing to continue to provide care to the care receiver?" Response options were "Yes, and I would even consider increasing the care I give if necessary" (1); "Yes, and I would consider increasing the care I give for a limited time" (2); "Yes, I am prepared to continue to provide care if the situation remains the same" (3); "Yes, I am prepared to continue to provide care to the elder but only if I have some more support" (4); and "No, I am not prepared to continue to provide care to the elder no matter what extra support I receive" (5).
Procedure
Following contact by means of one of the recruitment channels described earlier, potential participants were provided with information sheets describing the nature and purpose of the EUROFAMCARE study. We established that the potential participant was the primary carer for an older person; in other words, in those instances in which the participant was jointly supporting an older person, then we recruited the participant if he or she provided the most support to the older person relative to other informal carers. If the potential participant confirmed an interest in the study, then we had arrangements made between the interviewer and participant to meet at a place and time that was convenient to the participant (this could include a telephone interview).
At the commencement of the interview, the interviewer again explained the purpose of the study and went through informed consent procedures with the participant. In those instances in which the participant provided support to more than one older person, the interviewer asked the participant to focus on the most relevant or primary caregiving situation. The interviewer and participant then completed the CAT together. On average, this process took 75 minutes.
EUROFAMCARE partners obtained ethical committee approval for the study from within their own country.
Data Analysis
We used SPSS version 11.5 for our analysis. We examined descriptive statistics and requirements for a multivariate analysis. We subjected all items of the COPE Index to an exploratory principal component analysis (PCA) with orthogonal rotation (varimax) with Kaiser normalization employed following extraction. We assessed the adequacy of component extraction through (a) use of the scree test, (b) inspection of the residual correlation matrix for residuals with r >.05, and (c) examination of extracted components for reliability. As in our earlier study (McKee et al., 2003), the component loading cutoff was set at 0.40. We assessed the internal consistency of the components of the COPE Index through Cronbach's alpha reliability. We examined factor components as summative scales for confirmatory correlations with the WHO-5, the SF-36 quality of life and health status items, the Barthel Index, the BISID, hours of care per week, the Social Restriction Scale, and willingness to continue care. We set the significance level at p <.05. We deleted cases with missing data on any of the COPE Index variables from the analysis.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
|
Reliability Analyses and Scale Construction
We used the component structure and item loadings obtained in the analysis of the overall data set as the framework for further reliability analyses and scale construction. We found satisfactory internal consistency for those items that loaded on the negative impact component (Cronbach's = 0.83), with the internal consistency of the positive value and quality of support components more modest ( = 0.64 and 0.66, respectively). For the negative impact and positive value components, internal consistency could not be improved through trial item removal. For the quality of support component, the alpha value was improved to = 0.67 through removal of the "Do you feel well supported by health and social services?" item, but we decided to retain the item during scale construction because of the marginality of improvement.
We then separately summed the seven items drawn from the negative impact component, the four items drawn from the positive value component, and the four items drawn from the quality of support component, with high scores indicating high levels of the measured construct; we considered the properties of these summative scales overall and for each country. The results of these analyses are presented in Tables 5a and 5b. Internal consistency of the negative impact scale was satisfactory across countries, with alpha values ranging from = 0.79 for Sweden to = 0.84 for the United Kingdom and Greece. For the positive value scale, alpha values ranged from = 0.54 in Italy to = 0.66 in Poland. We could not improve the internal consistency of these two scales by means of item removal. For the quality of support scale, alpha values ranged from = 0.56 in Germany to = 0.70 in Poland. Internal consistency of the scale within some countries varied by item removal: removing the "Do you feel well supported by health and social services?" item raised the value in Germany from = 0.56 to = 0.60, in Greece from = 0.69 to = 0.74, in Poland from = 0.70 to = 0.71, and in the United Kingdom from = 0.65 to = 0.69. However, removal of the same item decreased alpha values in Italy from = 0.62 to = 0.60 and in Sweden from = 0.60 to = 0.58. With no common pattern apparent, we decided to retain the item.
|
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Our analysis of the combined data set indicated that the negative impact component of the COPE Index had satisfactory internal consistency, whereas the internal consistency of the positive value and quality of support components was marginally lower than the value most commonly regarded as acceptable ( = 0.70; Nunnally, 1967). When considered as a summative scale, the seven-item negative impact scale had satisfactory internal consistency across all countries. The four-item positive value scale had stable internal consistency across most countries ( = 0.64–0.66), except in Italy, where the alpha value was only = 0.54. For the four-item quality of support subscale, internal consistency varied from a low of = 0.56 for Germany to a high of = 0.70 for Poland.
The criterion validity analyses demonstrated strong associations in the expected direction between the negative impact scale and measures of psychological well-being, quality of life, care hours per week, elder dependency and behavioral disturbance, and carer social restriction. The positive value scale was also correlated with most of these measures in the expected directions, although the absolute size of the correlations was lower. Associations were highest with the well-being and quality of life measures, but of particular interest was the association with carer willingness to continue care, which was greater than the associations obtained between this measure and the other two scales. The quality of support scale had a generally similar pattern of associations with the criterion validity measures as that obtained with the positive value scale, with which it was strongly correlated. However, the quality of support scale had lower associations with the measures of elder behavioral disturbance and carer willingness to continue care, but a substantially higher association with the Social Restriction Scale.
A Three-Component Model of the Caregiving Experience
A potential criticism of the theoretical framework from which the COPE Index is derived is that caregiving is best understood as a one-dimensional psychological experience, with positive and negative ends to that continuum. However, previous research has demonstrated that positive and negative appraisals of the caregiving role are largely independent (Rapp & Chao, 2000), and that caregiving satisfaction can exist independently of stressful aspects of the caregiving situation (Lopez et al., 2005). These findings fit well with the recent burgeoning of research on the concept of attitudinal ambivalence (Conner & Armitage, 2008), which has demonstrated that holding both positive and negative attitudes toward an attitude object has implications for how someone will behave toward that attitude object relative to an individual who holds clearly positive or clearly negative attitudes (Armitage & Conner, 2004).
Differing from the earlier study of the COPE Index, in the present analyses a third component emerged that was clearly interpretable in the combined data set and in each single-country data set, had good item consistency across countries, and obtained moderate internal consistency as a scale. This quality of support scale shared only 14% of variance with the negative impact scale, and 25% variance with the positive value scale. This suggests that the support networks available to a carer are potentially an important correlate of how positively that carer experiences his or her situation but may be less related to negative perceptions of caregiving. This is important, as to date most caregiving interventions based upon providing support are primarily targeted at reducing the negative outcomes of care, such as impacts on carer health and depression.
Previous research has demonstrated that negative aspects of the daily caregiving experience are strongly associated with psychological well-being (Kinney, Stephens, Franks, & Norris, 1995). It was to be anticipated, therefore, that out of the three COPE Index subscales the negative impact scale would have the strongest associations with our criterion validity measures. However, the quality of support and positive value scales, although sharing a moderate amount of variance, demonstrated some divergence from each other—and from the negative impact scale—in their associations with the criterion measures, with positive value most apparently related to the carer's willingness to continue care, and with quality of support most substantively associated with carer social restriction.
Limitations
For most clearly defined cognitive states (e.g., intelligence, depression), an alpha value of = 0.8 would be desirable for a measurement scale, but it has been argued that, for less clearly delimited psychological constructs, measurement scales that attain an alpha value of = 0.6–0.7 can be regarded as acceptable (Kline, 1999). It should also be noted that the computation of the alpha value is related to the number of items a scale contains, with a high number of items inflating this value (Cortina, 1993). Although it would be desirable for the internal consistency of the positive value and quality of support scales to be better, for scales measuring relatively nebulous psychological states by means of only four items, the alpha value attained is arguably sufficient. Further research might improve these scales' internal consistency through item addition, but as one of the strengths of the COPE Index as a first-stage assessment is its brevity, additional items should be kept to a minimum.
The associations between the positive value and quality of support subscales and the selected criterion validity measures also give rise to a little concern, because, though significant, they are not of the magnitude that might be desired. The COPE Index was developed to assess caregivers' subjective perceptions of positive and negative aspects of their circumstances. Because such perceptions have been found to influence mental health and quality of life, in previous work on the COPE Index (McKee et al., 2003) we selected gold-standard criterion validity measures that measured mental health and quality of life and were available in the languages of each of the countries of the COPE partnership. These measures included the General Health Questionnaire (Goldberg & Williams, 1988), the Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983), and the World Health Organization Quality of Life Assessment–BREF (The WHOQOL Group, 1998). In this earlier work, the associations between the positive value scale and the criterion validity measures were, although still less than equivalent associations with the negative impact scale, generally higher than those obtained in the present study. For the current validation analyses we felt little value would be gained from simply replicating previous work, and so although a small selection of brief quality of life–psychological well-being measures was retained, a broad range of criterion validity measures was additionally selected that tapped important domains of the caregiving experience. This is justified because positive value and quality of support in caregiving are not pure psychological constructs that should map directly onto a given criterion variable, but instead are more nebulous constructs whose content should be expected to overlap with a range of other variables.
Although it could be argued that the criterion validity associations obtained in the current study offer less concrete evidence for the construct validity of the positive value scale than the previous study, a counterargument is that the analyses advance the knowledge gained in the previous work on the COPE Index by exploring how the different index subscales are related to different caregiving-related variables, and by more subtly elucidating the psychological meaning of the subscales.
Value for Practitioners and Future Development
Guberman (2005) has called for a paradigm shift in the way that caregiver assessment is both conceptualized and conducted, in order to provide a "new lens" through which to view caregivers' needs and circumstances (Fancey, Keefe, Guberman, & Barylark, 2005). Practitioners require instruments that help them understand the complexities of the caregiving role and facilitate a "meaningful dialogue" between caregivers and practitioners (Zarit & Leitsch, 2001), as service systems increasingly seek to work in a way that complements and supplements, rather than replaces, the efforts of family carers (Lyons, Zarit, & Townsend, 2000). We believe that the COPE Index can help to achieve these goals.
The COPE Index was not developed with the intention of providing a comprehensive assessment of the caregiver and his or her circumstances; it is therefore recommended as a first-stage assessment instrument to screen carers for identification of those requiring in-depth assessment (Nolan, Grant, & Keady, 1998), thus reducing the load on the assessment system. It is short and simple to complete and yet explores the caregiving situation as a multidimensional experience, characteristics that distinguish the index from other caregiver assessment instruments that focus only on negative aspects of the caregiving process or that lack brevity (Livingston, Mahoney, Regan, & Katona, 2005; Rankin, Haut, & Keefover, 1992; Schene, Tessler, & Gamache, 1994; Thornton & Travis, 2003). The COPE Index moves the focus away from burden and from the so-called objective problems of daily living of the care receiver, and onto the caregiver's subjective perceptions of his or her circumstances, the caregiver–care receiver relationship, and of the quality of formal and informal support. This meshes with recent intervention approaches based on partnership models of care in which the strengths and skills of the carer are accorded greater significance than has been the case in the past (Audit Commission, 2004). Thus, deploying the index as a first-stage assessment in concert with instruments that focus in more depth on the needs and abilities of the care receiver (e.g., Cox, Green, Seo, Inaba, & Quillen, 2006) would represent an assessment suite that provides a balanced and broad description of the caregiving situation.
The inclusion in the index of measures of positive value and quality of support in caregiving enables a more sensitive assessment that allows for more innovative forms of intervention (Nicholas, 2001). As we discussed earlier, caregivers' feelings regarding their circumstances will be complex, and health and social care practitioners need to understand that carers cannot be categorized in terms of simple feelings toward their caregiving experience. Rather, if intervention and support is to be appropriate and effective, the complexity of the caregiving situation requires unpacking (Sorensen, Pinquart, & Duberstein, 2002). One way of facilitating this process is to use appropriate assessment instruments. Practitioners using the index with caregivers will obtain profiles in terms of negative impact, positive value, and quality of support, and attention to variation in such profiles will enable more targeted interventions. For example, a caregiver who scores high on negative impact, moderate on quality of support, and high on positive value presents a different need profile than a carer who scores high on negative impact, high on quality of support, and low on positive value. The former caregiver may feel stressed, but gain satisfaction from his or her role and may strongly wish to continue if the quality of support can be improved, options for which can be discussed between caregiver and practitioner. The latter caregiver may feel fully supported in his or her role but is still experiencing stress and is unsatisfied. The practitioner can then explore through more detailed assessment and dialogue with the latter caregiver what exactly are the aspects of his or her role that are negative, and if the provision of training in care skills may help improve satisfaction and lower stress.
Trials of the acceptability of the first version of the COPE Index to carers, practitioners, and decision makers have been promising (Roud et al., 2006), and suggest that the newly revised index, which differs from the former primarily in methods of scoring and interpretation, should be equally positively received. We anticipate work that would explore the development of items for use in the positive value and quality of support subscales to enhance their reliability and construct validity. Further analyses will consider how caregiving negative impact, positive value, and quality of support vary as a function of caring for older people who differ in their care needs. The availability of longitudinal data will enable both tests of index component equivalence across countries via structural equation modeling as well as analyses of the predictive validity of the index for a variety of caregiving outcomes.
|
Footnotes |
|---|
1 Department of Cognitive Science and Education, University of Trento, Italy.
2 Department of Medical Sociology and Social Gerontology, Hamburg University Medical Centre of Hamburg-Eppendorf, Germany.
3 Sheffield Institute for Studies on Ageing, University of Sheffield, United Kingdom.
4 INRCA, Department of Gerontological Research, Ancona, Italy.
5 Department of Medical and Health Science, University of Linköping, Sweden.
6 Department of Geriatrics, Medical University of Bialystok, Poland.
7 Department of Health Services Management, SEXTANT Group, National School for Public Health, Athens, Greece.
Decision Editor: William J. McAuley, PhD
Received for publication April 10, 2007. Accepted for publication September 12, 2007.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| All GSA journals | Journals of Gerontology Series A: Biological Sciences and Medical Sciences | Journals of Gerontology Series B: Psychological Sciences and Social Sciences |
