Background
Caring for a family member with mental illness like schizophrenia is a laborious and time-consuming task that usually leads to adverse physical, psychological, emotional and economic impacts on family members, known as family burden [
1‐
3].Decades of international research have established the positive correlation between family burden and a range of negative caregiver outcomes such as depression, anxiety, physical disease and even mortality [
4‐
6]. While it is widely acknowledged that these conditions can be greatly ameliorated with effective psycho-social interventions [
7‐
9], how to measure family burden and define a valid cutoff to identify family caregivers in need of such interventions remains a key question.
The importance of measuring family burden has long been recognized in the literature, with a large quantity of instruments developed for assessing family burden in both physical diseases such as hemanioma, atopic dermatitis, ichthyosis and mental disease such as dementia, bipolar disorder, etc. [
10‐
13]. However, a review of past literature only detected four instruments that are specific to measuring family burden for persons with schizophrenia [
14]. Among the four instruments, the Family Burden Interview Schedule (FBIS) [
15] is proposed as the most promising one for its specificity, clinical application, and evidence [
16].
The FBIS offered a relatively short, yet comprehensive and multidimensional assessment of family burden. Originally developed by Pai and Kapur [
15] in 1981, the FBIS measures two aspects of burden (objective and subjective) encompassing six categories: financial burden, disruption of routine family activities, family leisure, family interactions, and effect on physical and mental health of others. Each category is composed of 2 to 6 items, adding up to 24 items for the whole FBIS. Each item is rated on a 3-point Likert scale from 0 (no burden) to 2 (serious burden), with a total score ranging from 0 to 48. For FBIS, the score is mostly used as a continuous variable yet no valid cutoff score has ever been proposed. However, use of continuous value for total burden score is inconvenient when it comes to deciding whether to include or not a caregiver in a burden prevention or treatment program, in this occasion, a dichotomous classification is more needed [
17].
Therefore, the present study was performed to define a valid cutoff score for the FBIS to screen caregivers in need of further assessment and intervention. Considering the well-established positive association between family burden and caregiver depression and anxiety [
4‐
6], we decided to explore FBIS cutoff score with reference to depression and anxiety score.
Discussion
Schizophrenia is a debilitating, persistent psychiatric disorder that not only affects the patients who suffer from it, but also extols significant burden on family and causes psychological distress such as depression and anxiety. Family burden has been reported to be one of the major reasons for family members to give up caregiving tasks and institutionalize the patients [
31,
32]. These conditions can be ameliorated with current psycho-educational interventions which focus on increasing caregivers’ knowledge and skills of patient management, alleviating caregivers’ feelings of stress, helplessness and burden, and improving caregivers’ sense of self-efficacy and self-value [
33‐
36]. A valid FBIS cutoff score would enable health care professionals to identify family caregivers in need of such interventions to alleviate family burden and improve caregiver’s quality of life.
To our knowledge, this is the first study to determine a statistically derived cutoff score using three methods for the most commonly used FBIS scale among schizophrenia caregivers to predict both depression and anxiety. Our findings suggest a FBIS cutoff score of 23 to identify caregivers at risk of both depression and anxiety and thus in need of further assessment and intervention. It has a positive predictive value of 76% for PHQ-9 and 74% for GAD-7, which indicates that 76% or 74% of caregivers above the FBIS cutoff are also above the depression cutoff or the anxiety cutoff. The negative predictive value is 68% for PHQ-9 and 67% for GAD-7, implying that 68% or 67% of caregivers below the FBIS cutoff are also below the depression cutoff or the anxiety cutoff.
The findings also imply some added benefits for the use of the FBIS scale by indicating that it not only measures family burden, but also assess the extent to which family burden constitutes psychological distress such as depression and anxiety for caregivers. In other words, caregivers at risk for depression and anxiety may be identified by administering the FBIS alone. In addition, the FBIS has been mostly used as a continuous variable with no cutoff point proposed in the past, the finding of the current study may fill in the research gap of lacking a FBIS cutoff value to distinguish families who are in need of further intervention from those who are not simply by FBIS score. Future family intervention program targeted at alleviating family burden and improving caregiver well-being may benefit from this cutoff as selection criterion.
In this study, we used three different analytical methods to determine a cutoff value for the FBIS score. Although they are different in definition, scope of application, terminologies, and analytic codes, they produce basically similar cutoff values for the FBIS score, implying the wide applicability and robustness of the three methods. However, cautions also need to be paid during the choice of each method. For tree-based modeling, although it has the advantage of being easy to understand, being useful in data exploration, requiring less data cleaning, with no constraint on data type, and being non-parametric method, it still confronts with the challenge of over fitting, which is one of the most practical difficulties for decision tree models and can only be solved by setting constraints on model parameters and pruning [
23‐
25]. For k-means clustering, its ease of implementation, computational efficiency and low memory consumption has kept it very popular, yet its sensitivity to the initial centroids chosen, the potential bias to create clusters of equal size, and lack of robustness to outliers require further adjustment while using this method [
29,
37]. Linear regression is the first type of regression analysis to be studied rigorously and used extensively in practical applications. However, it makes a number of assumptions about the predictor variables, the response variables and their relationship. These assumptions include weak exogeneity, linearity, constant variance, independence and lack of perfect multicollinearity. Violations of these assumptions may need various extensions based on this model to allow relaxation [
38,
39].
The study falls short in the following aspects. First of all, we used multiple statistical methods to run greedy cutoff searching, which may lead to inflated type I error. However, we re-run our analyses by splitting our sample into training set and validation set first by 1:1, then by 7:3, and found little difference of results. Considering the much smaller sample size of the split sample and related lower statistical power, we only displayed the results for the total sample testing. Future research may consider using a much larger sample size and randomly splitting it into training set and validation set with more power. Another limitation is the use of brief screening scales such as PHQ-9 and GAD-7 to assess depression and anxiety, instead of standard psychometric scales such as the Beck Depression Inventory (BDI), the Beck Anxiety Inventory (BAI), the Hamilton Rating Scale for Depression (HRSD), or the Hamilton Rating Scale for Anxiety (HRSA), which may compromise the accuracy of our measurement and thus leading to bias. However, the aim of the present study was to determine a cutoff score for the FBIS using depression and anxiety as a reference rather than to accurately measure these concepts, the results may not be affected by the choice of measurement tools. Future study may consider using standard psychometric scales and test whether brief screening scales are comparable to them. Thirdly, the use of one single cutoff value for the FBIS may introduce some kind of bias by treating persons with an FBIS score of 1 and a score of 22 as “equal” since they are both under the cutoff threshold, which is a major limitation for dichotomizing continuous variables for all scales. However, the aim of the current study was not to distinguish between various level of family burden, but to screen for those with higher burden and thus at risk for depression and anxiety for further intervention, which can be satisfied by having a cutoff value for the FBIS. Future studies focused on differentiating various levels of family burden may consider classifying the FBIS score into several levels instead of two. Also, the results of the current study are intended to serve only as a guideline for practitioners to assess their family caregivers and encourage them for further assessment and future intervention. In addition, the cutoff scores in this study warrant further test and validation in caregivers of other mental disorders.
Acknowledgements
The authors would like to thank all the families of the schizophrenia individuals we interviewed during the study for openly sharing their feelings and experiences. We’d also like to thank the health and family planning bureau of Ningxiang County and the government of the Liushahe town, Shungfupu town, Chengjiao xiang and Yutan town for their administrative support, as well as all village/community doctors for guiding us to visit each household of the schizophrenia individuals in the rural areas of Ningxiang county, Hunan province.