Factor analysis of the Resilience Scale for Brazilian caregivers of people with Alzheimer’s disease

Abstract Introduction Resilience is a dynamic process that acts to modify the effects of an adverse life event. In this study, we aimed to test the construct validity of the Resilience Scale by employing exploratory and confirmatory procedures, and to investigate the relationship between caregiver’s resilience and clinical status of people with Alzheimer’s disease. Methods A sample of 143 dyads of people with Alzheimer’s disease and their primary caregivers were included. Results The total Resilience Scale mean score was 140.3 (standard deviation [SD] = 16.289), ranging from 25 to 175, indicating a high level of resilience. Cronbach’s alpha was high (α = 0.77), indicating excellent internal consistency. The mean of corrected item-total correlation coefﬁcients was moderate. The Resilience Scale presented a four-factor solution with a well-deﬁned structure: sense of life and self-sufficiency, perseverance, self-confidence and equanimity, and meaningfulness. Conclusion The findings indicate excellent internal consistency of the Resilience Scale when used to evaluate psychological and emotional difficulties of caregivers, even though the correlations observed between the Resilience Scale and clinical variables were not significant for functionality, mood, awareness, neuropsychiatric symptoms, or burden.


Introduction
Resilience may be defined as a dynamic process involving the interaction between both risk and protective factors, both internal and external to the individual, that act to modify the effects of an adverse life event. 1,2 Risk factors are individual or environmental obstacles that would increase an individual's vulnerability to negative outcomes, such as the development of physical and mental illness or non-effective coping. 3 Protective factors are characteristics that reduce or prevent the occurrence of problems and result in positive and adaptive results. 4 Consequently, resilience does not involve invulnerability to stress, but, rather, the ability to recover from negative events. 5 Resilience is a construct that is cognitive in nature, present in all people, differing only in its level. 3 Since it is presented differently from subject to subject, there is an interest in developing methods of evaluating resilience, seeking to prevent diseases and promote mental health.
In general, to identify whether an individual is resilient or not, two evaluations are required. 1,2 First, the individual should be threatened by a high-risk state, or exposed to severe adversity or trauma. 6 Second, the quality of the individual's adaptation to the adverse life event or development should be good. 6 Good adaptation may be operationally defined through indicators associated with functional competency in specific developmental domains, which imply behavioral achievements expected in specific areas. 7 In dementia research, the stress and burden associated with caregiving, which impacts the caregivers' physical health and increases their mortality risk, is well known. 8 Caregivers of people with Alzheimer's disease (AD) have been shown to present more stress, burden, and depression compared to caregivers of people with other diseases. 4 However, despite the negative aspects involving the care of people with AD, many caregivers report a variety of positive experiences related to caregiving. 2 The theoretical resilience framework developed by Windle & Bennett 3 recognizes that caregivers will draw on individual resources, but also interact with their environment by employing community and societal resources which may facilitate or hinder resilience. 3 The absence of resources may lead to poor outcomes or further caring challenges. Considering this framework, resilience can be described as "the process of negotiating, managing, and adapting to significant sources of stress or trauma." 3 Assets and resources within the individual, their life, and environment facilitate this capacity for adapting and "bouncing back" in the face of adversity. 3 A gap in the literature about the resilience of caregivers of people with AD still deserves attention. It is well known that neuropsychiatric symptoms, impaired awareness of disease, and deficits in functionality are related to increased burden and decreased quality of life (QoL) of the caregiver. 9 Nevertheless, few studies consider the influence of these clinical symptoms on the level of caregiver's resilience. 10 Therefore, in dementia research, it is important to assess people with ADcaregiver dyads to evaluate whether resilience is an individual characteristic or whether it is related to the level of severity of dementia symptoms. Rosa et al. 10 found no significant difference in resilience between caregivers of mild and moderate people with AD, but upon analyzing the factors related to resilience in both groups, the results suggested that caregivers' resilience is driven by different factors according to disease severity.
Understanding the resilience of caregivers of people with AD is crucial to the development of intervention strategies that can contribute to the improvement of their emotional disorders, such as anxiety, stress, and depression. 5 Studies indicate that the general condition of the caregiver can interfere with the quality of care provided to people with AD, and may even lead to neglect or abuse of the elderly, as well as to their early institutionalization. 5 11 and search for challenges. 11 There were also associations with decreased burden, 11,12 stress, 7 neuroticism, 13 perceived control, 14 and distress. 14,15 The second area was biological, meaning that higher levels of resilience were associated with less depression 13 and better physical health. 13 Finally, there was the social area, 16

Study design
This was a cross-sectional study. Measures of cognition, 22 awareness of disease, 23 and depression 24,25 were administered to people with AD. Caregivers were assessed with measures of resilience 18 and burden. 26 Also, they were asked to provide information about the people with AD on the following topics: demographics, awareness of disease, 23 mood, 24,25 neuropsychiatric symptoms, 27 and dementia severity. 21 The dyad was interviewed separately, simultaneously, with higher values indicating greater resilience. Scores below 125 indicate low resilience, between 125-145 medium resilience, and above 145 high resilience. 18 Cognition. suggested accepting a value > 0.5 and recommended a minimum of 5 observations per variable. 31 We used maximum likelihood extraction instead of principal component analysis because the latter procedure does not discriminate between shared and unique variance, inflating variance estimates. 32 Maximum likelihood has been recommended as an extraction method even in cases when data is not normally distributed. 33,34 Factor rotation was performed through an oblique method (promax, δ = 0), because of potential conceptual correlation among the factors.

Participants
Examination of scree-plot and parallel analysis 35 40 The clinical data of people with AD and their caregivers are shown in Table 1.

Exploratory factor analysis
The KMO analysis revealed a value of 0.63, indicating that the correlation matrix was appropriate for factor analysis. The examination of scree plot and parallel analysis led to a four-factor solution, which accounted for 47% of the variance. Results from the pattern and structure matrix were similar; we report the pattern matrix because results are typically more conservative and not inflated by the overlap between factors. 41,42

Hierarchical factor analysis
The second-order factor analysis led to a onefactor solution with an eigenvalue equal to 2.6, which accounted for 54.1% of the variance, and three other factors with eigenvalues smaller than one. This result suggests, indeed, that the Resilience Scale has a hierarchical factor structure, in which the four lowerorder factors are loaded on a single higher-order factor. Table 3 shows that the higher-order structure was good and simple. Because the Schmid-Leiman procedure allows the higher-order factor to account for as much of the correlation among the items as possible, while the lower-order factors are reduced to residual factors uncorrelated with each other and with the higher-order factor, factor loadings are generally lower than those observed in the original exploratory factor analysis presented in Table 2. Therefore, factor loadings equal to or greater than 0.25 are generally considered satisfactory. 40 The higher-order factor accounted for and III (self-confidence). Item loadings across these factors were similar to the pattern observed in Table 2, suggesting the same factor labels, the exception being items from factor I (sense of life and self-sufficiency), which loaded heavily on the higher-order factor.
Hyperplane items from the first-order analysis loaded better on the higher-order factor. Table 3 shows the hierarchical Resilience Scale structure with loadings for one higher-and four lower-order factors.

Correlations between factors of the Resilience Scale
As shown in Table 4

Correlations between Resilience Scale factors and clinical variables
Pearson's correlations were calculated to explore the relationship between total Resilience Scale scores and clinical variables of the people with AD, such as cognition, mood, functional disability, neuropsychiatric symptoms, and awareness of disease, in addition to caregiver burden. To avoid inflation of the family error rate, the p-values were adjusted with the Bonferroni-Hochberg corrections. The results can be seen in  The coefficients of item-total correlations were within acceptable levels, suggesting adequacy of the items of the Resilience Scale. Our results showed that resilience is a construct with multiple dimensions. The first exploratory factor analysis detected four dimensions related to resilience: sense of life and self-sufficiency (factor I), perseverance (factor II), self-confidence (factor III), and equanimity and meaningfulness (factor IV). Second-order exploratory factor analysis indicated that the four lower-order factors loaded onto a single higher-order factor. Most items yielded expressive loadings in the higher-order factor. In addition, the high-order factor accounted for more than half of the variance.
It is important to highlight that our findings in the exploratory factor analysis differ from the findings obtained with the original scale 1 and also from those related to the Brazilian validation study whose sample comprised students from public schools. 18  Pesce et al., 18 in turn, in the exploratory factor analysis, found three factors that did not distinguish between sense of competence and self-sufficiency. Our findings showed that the four-factor solution of the Resilience Scale presented a well-defined structure. With the exception of some items (#2, 4, 15, 17, 23, and 24) that loaded more than one factor, the main charging items are fully highlighted. Conversely, we may suppose that the five hyperplane items identified (#1, 7, 11, 20, and 22) do not seem to be suitable to our target population due to specific disease characteristics, i.e., people with AD have cognitive deficits and behavioral disturbances that are difficult to be managed by the caregivers. Furthermore, there are some possible explanations for the differences observed between the exploratory factor analysis performed by Pesce et al. 18 and our findings. First, we may assume that resilience may vary according to the age of the target population. In the study by Pesce et al., 18 the sample comprised students with an age range between 12 and 19 years. were not statistically significant. Therefore, we assume that resilience may not be related to the clinical characteristics of people with AD. 43 The subjective assessment of caregivers' individual characteristics may be a determinant of the situation, i.e., the perception, interpretation, and sense attributed to the stressor event may be or not classified as a stressful condition. 18 However, the lack of relationship between resilience and the clinical aspects of AD is derived from a correlation analysis that does not allow to infer any causality.

Limitations
The present study has some limitations that should be considered. Firstly, we recruited caregivers from only one center of treatment for people with dementia.
Secondly, this was a cross-sectional study focused on caregivers of people with AD, and the findings cannot be generalized to caregivers of people with other types of dementia. In addition, we did not study the concurrent validity of the Resilience Scale as compared to any other resilience assessment. Finally, we have not assessed the history of neurological and psychiatry disorders in caregivers, which would allow to investigate their impact on caregivers' resilience.

Conclusion
To the best of our knowledge, there is no standardized resilience measure in Brazil other than the Resilience Scale, used to investigate caregivers of people with AD. The factor structure of the scale obtained through exploratory factor analysis provides strong evidence for construct validity, indicating that the Resilience Scale is a reliable measure of resilience in AD. The multidimensional nature of the resilience concept was confirmed by the factor analysis, which identified four dimensions of resilience related to sense of life and self-sufficiency, perseverance, self-confidence, and equanimity and meaningfulness. Investigating the underlying mechanisms associated with the differences and degrees of resilience will allow to enhance the understanding of risk factors associated with resilience and to improve coping strategies and self-efficacy among caregivers of people with AD.