Lifestyle interference and emotional distress in family caregivers of advanced cancer patients
Abstract
BACKGROUND
Providing end-of-life care at home to a family member with advanced cancer can have a negative impact on the emotional well-being of the family caregiver. The current study examined the impact of providing care on lifestyle and emotional well-being in a sample of caregivers to patients with advanced cancer. The mediation of lifestyle interference between the amount of care provided and emotional distress was specifically examined.
METHODS
Forty-four family caregivers participated in a structured quantitative interview. Lifestyle interference was assessed by the Caregiving Impact Scale, amount of care provided was assessed by the Caregiver Assistance Scale, and emotional distress was assessed by the Profile of Mood States—Short Form. Pearson and partial correlations tested whether lifestyle interference mediated the relationship between caregiving assistance and emotional distress. Regression analyses determined overall correlates of emotional distress.
RESULTS
Three criteria, required to substantiate mediation, were met for total mood disturbance and the depression and tension subscales. An overall regression model identified education level and lifestyle interference to be significant and unique correlates of emotional distress.
CONCLUSIONS
The current results suggest that caregivers experience increased emotional distress, regardless of the amount of care provided, when limited in their ability to participate in valued activities and interests. In addition, caregivers with less than a high school education experience more emotional distress. Therefore, helping caregivers maintain valued aspects of their lifestyle should be an important element of home care. Cancer 2002;94:521–7. © 2002 American Cancer Society.
End-of-life care for many patients with advanced disease is provided at home by a family member who assumes and/or coordinates the majority of care. In general, family caregivers represent a significant portion of the population. In the United States, 1997 national surveys estimated that there were between 24 to 27.6 million adults providing care to a family member or friend with a chronic, disabling, or terminal illness.1 A 1996 Statistics Canada survey reported that approximately 2.7 million adult Canadians, or 1 in 8 adults, were providing such care; in women between the ages of 45 and 64, this number increased to 1 in 5.2
Unfortunately, the stress of providing care often results in family caregivers experiencing elevated levels of emotional distress.3 The current study examined lifestyle interference as a possible factor contributing to emotional distress in caregivers. Providing care can interfere with the caregiver's ability to participate in valued activities (e.g., work, recreation, social outings, etc), and this disruption to lifestyle, as shown in patient studies, can result in emotional distress.4 The purpose of the current study was to examine whether lifestyle interference mediates the relationship between the act of providing care and the negative emotional consequences.
Theoretical Framework
Family caregivers report higher levels of emotional distress than the general population.3 Available literature suggests that the amount of care provided and care recipient disability are not directly associated with caregivers' emotional well-being, but are mediated by many variables, including the perception of burden.5-7 One element of subjective burden that is not confounded with aspects of emotional well-being is lifestyle interference, i.e., the extent to which caregiving responsibilities interfere with or disrupt family caregivers' ability to participate in valued activities and interests.
Two theoretical frameworks, one each from patient literature and from caregiving literature, examine the contribution of lifestyle interference to caregivers' emotional distress. Devins4 and Nijboer et al.8 propose that the objective aspects of either an illness/treatment or caregiving situation do not directly impact on caregivers' mental health but are mediated by a subjective evaluation of their situation. Devins's Illness Intrusiveness Model proposes that this mediating variable is the extent to which an individual's illness or treatment interferes with the caregiver's ability to participate in a variety of valued activities, i.e., illness intrusiveness. Because participation in these activities contributes to good mental health, disruption results in emotional distress, or, in other words, lifestyle interference mediates the relationship between illness/treatment and emotional distress. This model has been validated in several different illness populations, including end-stage renal disease,9 multiple sclerosis,10 rheumatoid arthritis,10, 11 and, more recently, cancer.12
According to Nijboer et al.'s Family Caregiver Model, caregivers' reactions to their situations are multidimensional, including schedule disruptions, financial concerns, lack of family support, loss of physical strength, and self-esteem.8 Disrupted schedules are a sub-set of Devins's lifestyle interference construct,4 which is more comprehensive than Nijboer's, as it incorporates a wider range of important contributors to quality of life as identified in a large American study.13 Therefore, the Devins lifestyle interference approach to caregiving facilitates a more comprehensive assessment of the impact of lifestyle disruption on caregiver emotional well-being.
Studies assessing the relationship between activity restriction and negative caregiving outcomes have obtained consistent results. In family caregivers of patients suffering from the effects of stroke, dementia, hip fracture, myocardial infarction, congestive heart failure, or cancer, higher levels of activity restriction were associated with higher levels of burden14 and emotional distress.14-16 Activity restriction also predicted later institutionalization of the care recipient.17
The current study applied the central and overlapping aspects of the Illness Intrusiveness Model and Nijboer's Family Caregiver Model to examine why caregivers providing similar levels of patient care experience differing levels of emotional distress. As shown in Figure 1, family caregivers' primary stressor is the amount of care they provide. The central caregiver experience is lifestyle interference, i.e., the extent to which providing care interferes with caregivers' ability to maintain participation in valued activities and interests. The main outcome is emotional distress.
Modified Family Caregiving Model.
The current study examined lifestyle interference, associated with providing care, as a potential mediator between the work of caregiving and emotional outcomes. Mediator variables help to explain the relationship between moderately related independent and dependent variables.18 The following mediation hypotheses were tested: 1) higher levels of caregiving assistance are associated with increased lifestyle interference; 2) increased lifestyle interference is associated with increased emotional distress; and 3) after controlling for lifestyle interference, the relationship between caregiving assistance and emotional distress is negligible. In addition, sociodemographic characteristics were considered as potential contributors to emotional distress and examined in an overall model. These hypotheses were tested in a sample of family caregivers to individuals diagnosed with advanced, incurable cancer.
METHODS
Subject Selection
Family caregivers to advanced cancer patients were recruited from the University Health Network—Princess Margaret Hospital oncology programs in Toronto, Canada. Advanced cancer was defined as incurable, with a less than six-month prognosis as estimated by the patient's health care team. Family caregivers were defined as the person who conducted or coordinated the majority of the patient's home care needs without receiving financial reimbursement for the care they provided. They were identified at palliative medicine and clinical oncology rounds. To be included, caregivers had to be providing assistance with at least two activities from the Caregiver Assistance Scale (see below). They also had to be fluent in English and willing to participate in a structured interview after giving informed consent.
Measurement Instruments
Caregivers participated in a structured quantitative interview. The first section of the interview obtained demographic information for the caregiver, including age, gender, education, income, primary daily activity, number of children, relationship to the care recipient, length of time providing care, and history of previous caregiving. Limited patient information was obtained, including age, primary diagnosis, current treatment, and duration of illness. The main study variables were the amount of caregiving assistance provided, lifestyle interference, and emotional distress.
The amount of care provided was assessed by the Caregiver Assistance Scale (CAS), developed for the purpose of the current study. The CAS determined the amount of assistance provided by the caregiver with activities of daily living (e.g., bathing, eating), instrumental activities of daily living (e.g., finances), and treatment related activities (e.g., administering medications), as described by previous authors.5, 19, 20 A combination of measures was used to capture important constructs that are not covered by any known individual measure. Caregivers were asked to rate on a seven point Likert scale how much assistance they had provided for 17 items over the previous four weeks ranging from zero (none) to six (a lot). The items were added to provide a total assistance score, with higher scores indicating more assistance. Internal consistency was good in a sample of 48 stroke family caregivers (α = 0.92, unpublished data) and in the current sample of caregivers to advanced cancer patients (α = 0.87).
The caregivers' experience of lifestyle interference was assessed by the Caregiving Impact Scale (CIS), a modification of the Illness Intrusiveness Rating Scale used by Devins et al., itself based on the Illness Intrusiveness Model.21 The modified scale assessed the extent to which providing care interfered with participation in 14 domains of the caregivers' lifestyle (e.g., health, diet, employment, household responsibilities, active recreation, passive recreation) over the past four weeks. One additional domain, “household responsibilities,” was added, and the original “work” domain was relabeled “employment” because caregiving may differentially impact on employment and housework. Seven point Likert scales were used to determine the degree of interference ranging from one (not very much) to seven (very much). The items were added to provide a total score, with higher scores indicating a higher degree of lifestyle interference. Good reliability for this scale has been observed in 48 stroke caregivers (α = 0.80, unpublished data) and in the current study of caregivers to advanced cancer patients (α = 0.87). These levels are consistent with the reliability of the Illness Intrusiveness Rating Scale (α = 0.79 to 0.90) in a variety of patient populations.10, 21, 22
The most comprehensive assessment of the related construct, activity restriction, is the Caregiver Activity Restriction Scale,16 which determines the extent to which caregiving restricts caregivers' participation in nine life activities (e.g., housework, recreation). All of these aspects of life are included in the CIS. The domains examined in the CIS stem from research where respondents identified the most important contributors to their quality of life.13 The consistently top-rated items were then included in the Illness Intrusiveness Rating Scale.21 Eight items are unique to the CIS (i.e., not included in any measure of activity restriction), including health, diet, employment, and religious expression.
Emotional distress was evaluated using the short form of the Profile of Mood States.23 The short form correlates highly with the complete version (r = 0.99), as do the subscales (r's from 0.95 to 0.98). The subscales chosen to represent emotional distress have good internal consistency for depression (α = 0.91), and tension (α = 0.80).23 Good internal consistency for the overall scale has been observed in a sample of 48 stroke caregivers (α = 0.92, unpublished data) and in the current sample (α = 0.94).
Statistical Analyses
The statistical analyses evaluated whether lifestyle interference mediated the relationship between the amount of caregiving assistance provided and emotional distress. To test this mediating relationship, established methods were applied.18 These methods have also been used in testing the Illness Intrusiveness Model.22, 24 To establish mediation, the following conditions must hold: 1) the correlation between the independent variable (the amount of assistance provided) and mediator (lifestyle interference) must be significant; 2) the correlation between the mediator and dependent variable (emotional distress) must be significant; and 3) the partial correlation between independent and dependent variables, after statistically controlling for the mediator, must be significantly lower than the simple correlation between the independent and dependent variables. “Perfect” mediation exists when the correlation between the independent and dependent variables is no longer significant after statistically controlling for the mediator.22
Sociodemographic characteristics were evaluated for their univariate relationship with emotional distress using correlations for continuous variables (e.g., age) and t tests for collapsed categorical variables (e.g., education, income). Significant characteristics were included in an overall regression model. The data were examined to identify outliers and to ensure variables included in correlation and regression analyses were normally distributed.
RESULTS
Forty-four family caregivers participated (46% response rate). Demographic statistics are shown in Table 1. Fifty-seven percent of the sample population were women; 86% were spouses. Average age of respondents was 55.8 (standard deviation [SD] = 15.17) years, and 48% had at least some post-secondary education. Forty-three percent had household incomes above $50,000, 32% were employed elsewhere for pay, 7% indicated that their primary daily activity was providing care, and 23% had previous caregiving experience.
| Characteristic | N = 44 |
|---|---|
| Femalea | 25 (56.8%) |
| Relationship to patienta | |
| Spouse | 38 (86.4%) |
| Daughter | 3 (6.8%) |
| Son | 3 (6.8%) |
| Ageb | 55.8 (15.17) |
| Number of childrenb | 2.1 (1.78) |
| Educationa | |
| Primary/secondary | 22 (50.0%) |
| Postsecondary | 21 (47.7%) |
| Incomeac | |
| Less than $50,000 | 16 (36.4%) |
| Greater than $50,000 | 19 (43.2%) |
| Employmenta | |
| Working for pay | 14 (31.8%) |
| Homemaker | 11 (25.0%) |
| Retired | 13 (29.5%) |
| Caregiver | 3 (6.8%) |
| Unemployed | 3 (6.8%) |
| Duration of caregiving (months)b | 7.5 (9.07) |
| Provided care previouslya | 10 (22.7%) |
- a Number (percentage).
- b Mean (standard deviation).
- c The sum of percentages does not add to 100% due to missing data.
The care recipients had a mean age of 60.8 years (SD = 12.03 years) and had cancer for approximately 12.2 months (SD = 14.5 months). Care recipients had a variety of cancers, including colorectal carcinoma (20.5%), pancreatic carcinoma (13.6%), leukemia (6.8%), stomach carcinoma (6.8%), lung carcinoma (4.5%), and others (27.6%). The majority of the patients were currently receiving chemotherapy (38.6%), radiation therapy (27.3%), or both (18.2%). No outliers were detected in this data set, and the variables were normally distributed.
Lifestyle Interference Mediates Caregiving Assistance and Emotional Distress
The first two criteria for establishing mediation are significant simple correlations between the independent variable (assistance provided) and the mediator (lifestyle interference), and between the mediator and the dependent variable (total mood disturbance and the emotional distress subscales). The independent variable (caregiving assistance) was significantly correlated with the mediator (r = 0.45, P < 0.001, lifestyle interference). The mediator was significantly correlated with the dependent variables (depression: r = 0.56, P < 0.001; tension: r = 0.58, P < 0.001; total mood disturbance: r = 0.61, P < 0.001).
The second and third criteria for mediation are presented in Table 2, the simple and partial correlations. For each independent variable (caregiving assistance and lifestyle interference [mediator]), the first row of correlations (Pearson r) represents the univariate correlations with each of the dependent variables (depression, tension, total mood disturbance). The second row presents the partial correlations (Partial r) after adjusting for the other variable (independent or mediator). The third criterion specifies that the partial correlation between the independent and dependent variables controlling for the mediator should be smaller than the simple correlations between these two variables. After controlling for lifestyle interference, the partial correlations between caregiving assistance and depression, tension, and total mood disturbance were no longer significant or approaching significance. These results support lifestyle interference as a mediator between the amount of care provided and emotional distress.
| Variable/correlation | Depression subscale | Tension subscale | Total mood disturbance |
|---|---|---|---|
| Caregiving assistance | |||
| Pearson r | 0.35 (P = .02) | 0.29 (P = 0.06) | .24 (P = 0.11) |
| Partial r | 0.13 (P = .40) | 0.04 (P = 0.81) | −.04 (P = 0.78) |
| Lifestyle interference | |||
| Person r | 0.56 (P = .001) | 0.58 (P = 0.001) | .61 (P = 0.001) |
| Partial r | 0.49 (P = .001) | 0.52 (P = 0.001) | .58 (P = 0.001) |
Overall Model of Emotional Distress
All sociodemographic characteristics reported in Table 1 were examined for inclusion in an overall regression model. Education level was the only variable significantly associated with total mood disturbance, depression, or tension. Family caregivers with less than a high school education reported significantly more depression (t = −3.62, P < 0.01), more tension (t = −3.18, P < 0.01), and total mood disturbance (t = −3.58, P < 0.02) than those caregivers with at least a post-secondary education.
Three regression models examined the unique contribution of education level, coded as a dummy variable (0 = more than high school, 1 = high school or less), caregiving assistance, and lifestyle interference to total mood disturbance, depression, and tension. All three models were significant (Table 3). Large proportions of variance in emotional distress were explained by the variables in the model, 51% in tension, 53% in depression, and 55% in total mood disturbance.
| Variable | Statistica |
|---|---|
| Depression | adjusted R2 = 0.53 |
| Education level | 0.54b |
| Caregiving assistance | 0.26c |
| Lifestyle interference | 0.41b |
| Tension | adjusted R2 = 0.51 |
| Education level | 0.48b |
| Caregiving assistance | 0.21 |
| Lifestyle interference | 0.47b |
| Total mood disturbance | adjusted R2 = 0.55 |
| Education level | 0.50b |
| Caregiving assistance | 0.12 |
| Lifestyle interference | 0.53b |
- a Statistics presented include the adjusted R2, beta weights, and significance levels.
- b P < 0.001.
- c P < 0.05.
DISCUSSION
The results of the current study are consistent with Devins Illness Intrusiveness Model and Nijboer Caregiver Model. Lifestyle interference mediated the relationship between caregiving assistance and overall mood disturbance, including depression and tension. Specifically, more lifestyle interference was associated with more emotional distress, regardless of the level of care provided.
The best fit for the current model was with the depression subscale of the Profile of Mood States. The model fits to a somewhat lesser extent with total mood disturbance and the tension subscale. For each dependent variable, after controlling for the mediator (lifestyle interference), the partial correlations with the independent variable were no longer significant or approaching significance, suggesting “perfect” mediation.22 It must be noted that conventional levels of a statistical significance were not obtained for the correlations between the independent variable and two of the indicators of emotional distress, total mood disturbance and tension. This suggests that lifestyle interference may only partially mediate aspects of emotional distress other than depression. Overall, we can conclude that lifestyle interference is a strong predictor of emotional distress, since the adjusted R2 for the overall regression equations ranged from .51 to .55.
The results using the depression subscale as the dependent variable are consistent with the work of Devins et al. in patients with chronic disabling medical conditions.9, 11 This supports the model's robustness at evaluating health-related stressors and their impact on emotional well-being in two separate groups, patients and their family caregivers. The current findings are also consistent with previous caregiver research. Williamson applied and tested a model, developed originally for patients and the elderly, in cancer family caregivers where activity restriction mediated patient symptom severity and caregiver emotional distress.16 Similarly, in family caregivers to a variety of patient groups (myocardial infarction, congestive heart failure, stroke, and hip fracture patients), Nieboer et al. found that activity restriction mediated the relationship between the number of care tasks performed and emotional distress.15
Education level was the only sociodemographic characteristic associated with emotional distress in the current sample. When education level, caregiving assistance, and lifestyle interference were examined simultaneously in a regression model, it revealed that caregivers with less than a high school education experienced greater emotional distress. This finding is consistent with a study of informal caregivers to community-residing demented elderly, where having less education was a significant correlate of emotional distress.25 In another study, those with less education were able to derive more self-esteem from providing care, although neither education nor self-esteem were significantly associated with emotional distress.8 We must acknowledge that since we did not hypothesize that education level would be associated with caregivers' experiences of emotional distress, this relationship may represent a Type I error.
A unique aspect of the current study was the approximately equivalent number of male and female caregivers (56% female), unlike many studies, which include a larger proportion of women (66-75%).15, 16, 26, 27 This cannot be explained by the recruitment strategy, since caregivers were not targeted by gender. However, this difference may be because the patients had a variety of cancers, where the incidence rates do not differ by gender. In addition, patients in the current study were in an advanced stage of illness, requiring a relatively shorter duration of care; therefore, men were equally willing to disrupt their work and family lives to assume the caregiving role. No gender differences in the outcome variables were found, although previous research has observed differences.28, 29
Our use of a comprehensive assessment of lifestyle interference explained a larger proportion of the variance in caregiver emotional distress (51% to 55%) than previously reported studies assessing the related construct, activity restriction. Thirty-eight percent of the variance in emotional distress was explained by patient symptom severity, activity restriction, loss of intimacy, and mutually communal behaviors.16 Another study explained 49% of the variance in time two depression by time one depression scores, length of time since first interview, and the change in activity restriction between interviews.15 When examining maintenance of home care as the outcome variable, the relationship of the caregiver (spouse) and their satisfaction with participation in leisure activities explained 38% of the variance.17
The main strength of the current study was our application of a theoretical model to study the experience of emotional distress. Unfortunately, the current study also had limitations. Although the sample had adequate power to detect significant relationships between the variables of interest, a larger sample size could increase the robustness of the current findings. In addition, a 46% participation rate was obtained. The current study highlights the challenges inherent in conducting research in the area of palliative medicine.30, 31 For example, a cancer caregiver intervention study reported by Toseland et al.32 obtained a response rate of 27%. Even though we made concessions to conduct interviews over the telephone or in the hospital, it was felt, although not assessed, that the refusers were caregivers who were under more stress and, therefore, less willing to take time away from caregiving.
Another potential limitation of the current study was the exclusion of anticipatory grief as a possible correlate of emotional distress in family caregivers to individuals with advanced cancer. Although there is debate about anticipatory grief as a construct, it is thought to represent one or all of the following: an emotional reaction to a family member being diagnosed with a terminal illness, an early start to the grieving process, and/or an adaptive coping function.33 In a convenience sample of family caregivers to dementia patients, anticipatory grief explained 63% of the variance in emotional distress as assessed by the Beck Depression Inventory.34 The authors suggest that anticipatory grief is a strong correlate of emotional distress, but further research is required to confirm that it is a construct distinct from emotional distress. Research should also determine if experiencing anticipatory grief is as great a clinical concern as depression. Some researchers suggest that it is not.35, 36
The current study also exhibits limitations common to the caregiving literature. We did not employ longitudinal methods, so cannot conclude a causal relationship between lifestyle interference and emotional distress, i.e., that increased lifestyle interference results in (causes) increased emotional distress. In addition, we do not know pre-caregiving levels of emotional well-being and, therefore, cannot directly attribute observed emotional distress to providing care to a family member with advanced cancer. Future studies can address some of these challenges by recruiting family members at the time of diagnosis and following them through the illness trajectory.
Research is needed to continue to evaluate lifestyle interference and its ability to mediate the relationship between the work of providing care and emotional outcomes in family caregivers. In addition, the causal nature of this relationship should be examined. Other factors associated with emotional distress, such as mastery, social support, and coping skills, can be incorporated to obtain a more comprehensive model of caregiver emotional well-being. Future research is required to clarify the relationship between level of education and emotional distress in caregivers to patients with advanced cancer.
Family caregivers' inability to maintain participation in valued activities and interests is associated with increased emotional distress. Development and evaluation of interventions and programs aimed at assisting caregivers maintain participation in these activities should have a positive impact on their emotional well-being, which is vital to maintaining long-term home care.37 For example, modifying home-care policies to provide more services in the home or respite care may help family caregivers spend more time pursuing their own interests. In addition, providing caregivers with greater preparation for their role as care providers may make home care more manageable, which should allow caregivers more time for themselves. Both of these aspects should be pursued further.
Acknowledgements
The authors thank Keri English, Shanuka Ratnasingham, Sujitha Ratnasingham, and Jennifer Shin for assistance with data collection and LaToya Austin for comments on the article. The following University Health Network clinics assisted with subject recruitment: palliative medicine, bone marrow transplant, solid tumor, and gastro-intestinal cancer. The following individuals assisted with the identification and recruitment of family caregivers: Mary Anne Huggins, M.D., Laura Brooks, Malcolm Moore, M.D., Andrea Bezjak, M.D., M.Sc., Diane Williams, and Gary Coe.
