Validation of ‘personal strain’ and ‘role strain’: Subscales of the short version of the Japanese version of the Zarit Burden Interview (J-ZBI_8)
Dr Yumiko Arai, Research Unit for Nursing Caring Sciences and Psychology, National Institute for Longevity Sciences, 36-3 Gengo Morioka-cho, Obu-shi, Aichi 474-8522, Japan. Email: address: firstname.lastname@example.org
Abstract The eight-item short version of the Japanese version of the Zarit Burden Interview (J-ZBI_8) has been confirmed for reliability and validity, and its two subscales (personal strain and role strain) are based on the factor structure of the ZBI. It has been demonstrated that these subscales have good reliability. The aim of the present study was to confirm their construct validity. A total of 51 impaired elderly who had been receiving regular nurses’ visits in Kyoto Prefecture, Japan and their family primary caregivers, participated in the present study. Each caregiver was asked to complete a questionnaire which included the J-ZBI_8, the hours spent in caregiving, and the physical and cognitive disability of the impaired elderly. A principal component analysis identified the following two principal components of these variables: ‘Activities of Daily Living deficits’ and ‘behavioral disturbances’. Consequently, Barthel Index (BI) and Troublesome Behavior Scale (TBS) were selected as the representative variable, respectively, for each component. We subsequently calculated the Spearman's rank correlations among the subscales of J-ZBI_8, BI and TBS. Personal strain was found to be correlated with TBS (rho = 0.48, P < 0.01), while role strain was correlated with the BI (rho = −0.29, P = 0.04). The correlation between personal strain and BI was not significant, nor was the correlation between role strain and TBS. These findings indicate that the personal strain and role strain measured by J-ZBI_8 correspond to the personal strain and role strain of the ZBI originally described by Whitlatch et al. Therefore, the construct validity of the J-ZBI_8 subscales is confirmed.
The Zarit Burden Interview (ZBI) is the instrument most widely used in North America and Europe for assessing the burden experienced by family caregivers who look after the community-residing impaired elderly.1,2 The Japanese version of the ZBI (J-ZBI), developed by Arai and colleagues, has been widely used in Japan for the same purpose.3,4 Recently, the eight-item short version of the J-ZBI, the J-ZBI_8, was developed by Arai and colleagues for easier administration of the instrument in clinical settings. The reliability and validity of the J-ZBI_8 have been confirmed,5 and cross-validation of the J-ZBI_8 has also been conducted in a different sample.6
The J-ZBI_8 consists of the following two subscales: personal strain (five items) and role strain (three items), which are based on the factor structure of the ZBI described by Whitlatch.7 Specifically, the ZBI is comprised of two factors which have been described as follows: personal strain means ‘how personally stressful the experience is’; and role strain is ‘the stress due to role conflict or overload’.7–9
So far, it has been demonstrated that the two subscales of the J-ZBI_8 (i.e. personal strain and role strain) have good reliability.5,6 However, the validity of these subscales has yet to be fully confirmed. Indeed, it has been stated that there is no one single study which can unequivocally prove construct validity and that construct validation is an ongoing process.10
Therefore, the aim of the present study was to elucidate the psychometric properties of the subscales of the J-ZBI_8, that is, personal strain and role strain, in order to confirm their construct validity.
As the first step, 53 impaired elderly were identified who had been registered with a practice nurse clinic attached to U general hospital in the vicinity of Kyoto City. In the second step, these 53 impaired elderly and their family principal caregivers were contacted by letter to explain the objectives of the present study. Informed consent was obtained from all of these pairs in March 2001. This study was endorsed by the ethical committee of the National Institute of Longevity Sciences. All of the primary caregivers were co-residing with the patients at the time. Among these 53 pairs, 51 pairs participated in the present study.
Each caregiver was asked to complete a questionnaire in relation to the caregiving situation, their feelings of burden, and the disabilities of the impaired elderly under their care. Specifically, the questionnaire included: (i) questions regarding demographic variables of the caregivers and the patients; (ii) questions regarding the duration of caregiving and the hours spent in caregiving; (iii) the Japanese version of the Zarit Burden Interview (J-ZBI); and (iv) the physical and cognitive disability of the impaired elderly.
Caregivers were asked to indicate how many months’ duration they had cared for the impaired elderly. They were then asked to estimate how many hours per day that they provided assistance for the Activities of Daily Living (ADL) of the impaired elderly, as well as how many hours per day they spent in supervising the elderly other than providing the ADL assistance. In addition, they were to estimate the number of hours per day they were able to be temporarily relieved of their duties and/or to leave the side of the patient and go out.
Caregivers’ burden was assessed by the short version of the J-ZBI, that is, J-ZBI_8. The score of the J-ZBI_8 was calculated from the J-ZBI score obtained from the questionnaire distributed to the caregivers in the present study.
The ADL of the elderly was assessed using the Barthel Index (BI), the widely used 10-item ADL scale.11,12 Cognitive impairment of the elderly was assessed with the Japanese version of the Short-Memory Questionnaire (SMQ),13,14 a 14-item screening test for the assessment of memory difficulties found in dementia, with scores ranging from 0 to 46. The cut-off point for cognitive impairment was less than 40.14,15 The frequency of behavioral disturbances associated with dementia observed by primary caregivers of the elderly was assessed using the Troublesome Behavior Scale (TBS).16,17
A principal component analysis was used to examine the structure of variables concerned with the caregiving of the impaired elderly, that is, hours of ADL assistance/day, hours of supervision/day, hours caregivers can go out/day, BI, SMQ, and TBS. The principal components were retained when their eigenvalues were greater than 1.0, yielding two principal components. We then selected variables which showed the highest loading for each principal component. We subsequently calculated the Spearman's rank correlation coefficients among the above-selected variables. Personal strain and role strain, the J-ZBI_8 subscales, were employed in order to identify differences between them.
Statistical significance was set at P < 0.05. SPSS (version 11.5.1 J; SPSS Inc. Chicago, Illinois, USA) was used for the above statistical analyses.
Table 1 shows the characteristics of the subjects in our study. The mean age of the impaired elderly was 79.5 (SD 9.5) years old, and 65% were female. The mean age of the caregivers was 60.4 (SD 13.4) years old, and 75% were female. The mean score of the J-ZBI_8 was 13.4 (SD 7.8); the mean score of personal strain was 7.6 (SD 5.0); and that of role strain was 5.8 (SD 3.7). A total of 46 out of the 51 impaired elderly who scored less than 39 on the SMQ were regarded as having some memory difficulties. The average hours spent for ADL assistance/day was 3.1 (SD 2.8), while the hours spent for supervision was 1.4 (SD 2.8), which was half the time spent for ADL assistance.
Table 1. . Characteristics of the impaired elderly and their caregivers
| Age||79.5|| 9.5|
| Barthel Index|| 8.0|| 6.3|
| TBS|| 5.6|| 7.9|
| No. family members|| 3.1|| 1.8|
| Duration of caregiving (months)||55.0||52.0|
| Hours of ADL assistance per day|| 3.1|| 2.3|
| Hours of supervision per day|| 1.4|| 2.8|
| Hours caregivers can go out per day|| 2.9|| 2.8|
| J-ZBI 8||13.4|| 7.8|
| Personal strain|| 7.6|| 5.0|
| Role strain|| 5.8|| 3.7|
| Husband|| 5|| 9.8|
| Son|| 6||11.8|
| Other|| 3|| 5.9|
As shown in Table 2, the principal component analysis yielded two principal components which explained 65.5% of the total variance. On the basis of the item loadings, the first component was regarded as related to the caring for the ADL deficits of the impaired elderly impaired elderly, and thus was labelled ‘ADL deficits’. The second component was regard as related to behavioral disturbances of the impaired elderly, which we labeled ‘behavioral disturbances’. The following two variables, which showed the highest loading on each principal component, were selected: BI from the first component, and TBS from the second component.
Table 2. . Factor loading of variables concerning caregiving of impaired elderly
|Barthel Index||−0.85|| 0.37|
|Hours of ADL assistance/day|| 0.63||−0.25|
|Hours caregivers can go out/day||−0.62||−0.22|
|TBS|| 0.01|| 0.91|
|Hours of supervision/day|| 0.22|| 0.72|
|Eigenvalue|| 3.44|| 1.80|
|Contribution rate (%)|| 43.04|| 22.50|
Table 3 presents Spearman's rank correlations among the subscales of J-ZBI_8, BI, TBS, and age of caregivers and impaired elderly. First, personal strain was correlated with TBS (rho = 0.48, P < 0.01). Second, role strain was correlated with the Barthel Index (rho = −0.29, P = 0.04). The correlation between personal strain and BI was not significant, nor was the correlation between role strain and TBS. Neither caregivers’ age nor the age of the impaired elderly was correlated with the other variables.
Table 3. . Correlation coefficients among subscales of J-ZBI_8, Barthel Index, TBS, Caregivers’ age and age of impaired elderly
|Personal strain|| 1.00|| || || || || |
|Role strain|| 0.57**|| 1.00|| || || || |
|Barthel Index||−0.02||−0.29*|| 1.00|| || || |
|Troublesome Behavior Scale|| 0.48**|| 0.20|| 0.32*|| 1.00|| || |
|Caregivers’ age||−0.05||−0.01||−0.19||−0.12||1.00|| |
|Age of impaired elderly||−0.16||−0.19||−0.03||−0.25||0.01||1.00|
The principal component analysis in the present study revealed that there were two principal components, that is, ‘ADL deficits’ and ‘behavioral disturbances’, among the variables concerned with the caregiving of the impaired elderly (Table 2).
It is assumed that ‘ADL deficits’, which includes the time constraints imposed by providing ADL assistance, may have prevented caregivers from fulfilling any role but that of a caregiver. On the other hand, ‘behavioral disturbances’ are known to be related to caregivers’ stress.18,19
The analyses of the subscales of the J-ZBI_8 in the present study demonstrated a significant relationship between personal strain and TBS of the impaired elderly, while role strain was significantly related to their BI. Neither the correlation between personal strain and BI nor between role strain and TBS was significant (Table 3). These findings indicated that personal strain was related to ‘behavioral disturbances’, and role strain was related to ‘ADL deficits’. Neither the correlation between personal strain and ‘ADL deficits’ nor between role strain and ‘behavioral disturbances’ was significant.
As mentioned above, ‘behavioral disturbances’ of the impaired elderly was associated with caregivers’ stress. Therefore, it is postulated that personal strain, measured by the J-ZBI-8, reflects ‘how personally stressful the experience is’ as earlier described by Whitlatch et al.7
As stated above, ‘ADL deficits’ of the impaired elderly deterred a caregiver from fulfillling anything but his/her role as a caregiver. Thus, it is concluded that role strain, measured by the J-ZBI-8, is associated with the ‘role conflicts’ as described by Whitlatch et al.7
In conclusion, the present study revealed that personal strain and role strain, measured by J-ZBI-8, correspond to the description of these two factors of the ZBI originally described by Whitlatch et al. Therefore, the construct validity of the J-ZBI_8 subscales is confirmed.
Our special thanks to the participants in the present study, which was in part supported by research grants provided by the Ministry of Health, Labor and Welfare (Grant Nos. H15-C-025 and H14-S-031), the Ministry of Education, Culture, Sports, Science and Technology, Japan (Grant No. 14570375), and the Uehara Memorial Foundation.