Health and Social Care in the Community (2016) 24(4), 428–438

doi: 10.1111/hsc.12219

Predictors of caregiver burden across the home-based palliative care trajectory in Ontario, Canada Denise Guerriere PhD1, Amna Husain MD2, Brandon Zagorski MSc1, Denise Marshall MD3, Hsien Seow PhD4, Kevin Brazil PhD5, Julia Kennedy MHSc1, Sheri Burns BA6, Heather Brooks BScH Student7 and Peter C. Coyte PhD

1

1

Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada, 2Department of Family and Community Medicine, Temmy Latner Centre for Palliative Care, Mount Sinai Hospital, Toronto, Ontario, Canada, 3Department of Family Medicine, McMaster University, Hamilton, Ontario, Canada, 4Department of Oncology, Juravinski Cancer Center, McMaster University, Hamilton, Ontario, Canada, 5School of Nursing and Midwifery, Queen’s University Belfast, Belfast, UK, 6St Joseph’s Healthcare Hamilton, Hamilton, Ontario, Canada and 7Department of Psychology, Queen’s University, 62 Arch Street, Kingston, ON, K7L 3N6 Canada

Accepted for publication 21 January 2015

Correspondence Denise Guerriere Institute of Health Policy, Management and Evaluation University of Toronto 155 College Street Toronto, Ontario, Canada M5T 3M6 E-mail: [email protected]

What is known about this topic

•

•

Family caregivers of palliative care patients often experience burden due to the enormous physical, emotional and financial responsibilities associated with care-giving. Caregiver burden is noteworthy as it may predict anxiety and depression and extend into bereavement.

What this paper adds

• •

Predictors of caregiver burden across the palliative care trajectory encompass both modifiable and non-modifiable variables. Increased monthly unpaid caregiving time costs, monthly public personal support worker costs, emergency department visits and low patient functional status were associated with higher caregiver burden in this longitudinal study.

Abstract Family caregivers of patients enrolled in home-based palliative care programmes provide unpaid care and assistance with daily activities to terminally ill family members. Caregivers often experience caregiver burden, which is an important predictor of anxiety and depression that can extend into bereavement. We conducted a longitudinal, prospective cohort study to comprehensively assess modifiable and non-modifiable patient and caregiver factors that account for caregiver burden over the palliative care trajectory. Caregivers (n = 327) of patients with malignant neoplasm were recruited from two dedicated home-based palliative care programmes in Southern Ontario, Canada from 1 July 2010 to 31 August 2012. Data were obtained from bi-weekly telephone interviews with caregivers from study admission until death, and from palliative care programme and home-care agency databases. Information collected comprised patient and caregiver demographics, utilisation of privately and publicly financed resources, patient clinical status and caregiver burden. The average age of the caregivers was 59.0 years (SD: 13.2), and almost 70% were female. Caregiver burden increased over time in a nonlinear fashion from study admission to patient death. Increased monthly unpaid care-giving time costs, monthly public personal support worker costs, emergency department visits and low patient functional status were associated with higher caregiver burden. Greater use of hospice care was associated with lower burden. Female caregivers tended to report more burden compared to men as death approached, and burden was higher when patients were male. Low patient functional status was the strongest predictor of burden. Understanding the influence of modifiable and non-modifiable factors on the experience of burden over the palliative trajectory is essential for the development and targeting of programmes and policies to support family caregivers and reduce burden. Supporting caregivers can have benefits such as improved caregiver health outcomes, and enhancing their ability to meet caregiving demands, thereby potentially allowing for longer patient care in the home setting. Keywords: cancer, family caregivers, home-based care, palliative care, predictors of caregiver burden, public and private expenditures

428

© 2015 John Wiley & Sons Ltd

Predictors of caregiver burden

Introduction The provision of palliative care in the home setting provides the opportunity for potential improved outcomes for individuals with terminal cancer. Patients may be able to actualise their preferred place of death and have a higher quality of life (Gomes et al. 2013). These potential benefits are often only possible with the time and intensive resources that are dedicated by family caregivers who provide unpaid care and assistance with daily activities to family members (i.e. relatives and spouses) who are ill (McNamara & Rosenwax 2010). As a result, family caregivers confront enormous physical, emotional and financial responsibilities while providing and co-ordinating care for their dying family member (Brazil et al. 2003, Hauser & Kramer 2004, Hudson & Payne 2011, Chai et al. 2014). Caregiver burden is often experienced by family caregivers of palliative care patients, and has been shown to be an important predictor of anxiety and depression (Grunfeld et al. 2004, Rhee et al. 2008, Papastavrou et al. 2009, Francis et al. 2010), and to be associated with poor physical health (Williams & McCorkle 2011, Shieh et al. 2012, Lee et al. 2013) and reduced quality of life (Hughes et al. 1999, Hudson & Payne 2011, Song et al. 2011). Furthermore, many of these outcomes extend into bereavement, increasing morbidity and mortality (Schulz & Beach 1999, Schulz et al. 2008, Kapari et al. 2010). Burden may be elevated in caregivers facing additional strain if they are elderly or ill, or balancing their time with other adult responsibilities such as employment and family (Grunfeld et al. 2004). It is essential to identify which caregivers are most at risk of experiencing high levels of burden in order to direct additional support to those in need. The appropriate identification of those caregivers who are most likely to experience high levels of burden requires a thorough evaluation of both modifiable and non-modifiable patient and caregiver factors including socio-demographics, clinical variables, health services utilisation and psychosocial variables. Understanding the influence of these factors on the experience of burden over the palliative trajectory is essential for the development and targeting of programmes and policies to support family caregivers and reduce burden during this incredibly onerous period. Supporting caregivers can have benefits such as improved caregiver health outcomes, and enhancing their ability to meet care-giving demands, thereby potentially allowing for longer patient care in the home setting.

© 2015 John Wiley & Sons Ltd

Existing studies of family caregiver burden are primarily retrospective and cross-sectional. These studies span a variety of contexts, including a range of patient diagnoses, stages of illness, sites of care, and both palliative and non-palliative settings (Emanuel et al. 2000, Brazil et al. 2003, Hwang et al. 2003, Goldstein et al. 2004, Rhee et al. 2008, Papastavrou et al. 2009, Francis et al. 2010, Burton et al. 2012, Park et al. 2012, Shieh et al. 2012, Teixeira & Pereira 2013, Utne et al. 2013). Five longitudinal studies have addressed family caregiver burden (Given et al. 2004, Grunfeld et al. 2004, Doorenbos et al. 2007, Garlo et al. 2010, Lee et al. 2013). Each of these studies examined the relationship between burden and independent patient and caregiver variables (e.g. socio-demographic factors, clinical features and physical functioning), and one considered the influence of family care-giving time on burden (Lee et al. 2013). None of these studies included a comprehensive assessment of potential economic determinants such as health services utilisation and use of private resources that may shift the balance of care tasks between family caregivers and the healthcare system, and subsequently, contribute to burden outcomes. The purpose of our study was to identify the socio-demographic, clinical, health services utilisation and economic factors, measured at various points along the palliative care trajectory, that account for caregiver burden among caregivers of patients enrolled in a dedicated home-based palliative care programme. Our study adds to the existing body of literature surrounding caregiver burden by using a prospective study design with a frequent interview schedule, a large sample size and an extensive list of predictors that have not collectively been considered to date.

Methods A longitudinal, prospective cohort design was employed to determine the association between independent variables (socio-demographic variables, clinical variables, health service characteristics and psychosocial variables) and caregiver burden at regular time intervals across the palliative care trajectory. The study findings reported herein represent a component of a larger study of home-based palliative care (Guerriere et al. 2016). Burden was operationalised according to Dumont et al. (2008), who define burden as a psychological and emotional experience related to the perception of the demands that are specifically associated with accompanying and providing care for a dying person

429

D. Guerriere et al.

(Dumont et al. 2008). While all individuals share the common experience of practising the tasks associated with care-giving, each individual experiences different magnitudes of burden and the associated psychological distress and emotion (Bainbridge et al. 2009). We explored, from a multidimensional perspective, variables that may predict caregivers’ perceptions of burden. This group of determinants includes time-stable covariates (caregiver and patient characteristics) and time-varying covariates (patient health, caregiver psychosocial status and resource utilisation) (Figure 1). The selection of potential determinants of caregiver burden was guided by previous empirical work (Given et al. 2004, Grunfeld et al. 2004, Doorenbos et al. 2007). Demographic variables allow for the identification of specific groups who are at risk of high burden. In addition to these previously studied variables, we contribute further to this body of research by including resource utilisation variables that may also play an important part in determining the experience of burden for caregivers. The amount and type of health services and private resources presumably influence the context of the care-giving experience as the workload and responsibility are shared between family caregivers and the healthcare system. Caregiver study participants were recruited from one of two regionally separate but dedicated homebased palliative care programmes in Southern Ontario, Canada: the Temmy Latner for Palliative Care at Mount Sinai Hospital (Toronto); and the Hamilton Niagara Haldimand Brant Palliative Care Teams. These palliative care programmes provide community and team-based multidisciplinary palliative care, including symptom and case management,

and practical/emotional support to individuals at home 24 hours/day, 7 days/week. Care components include medical care from programme-dedicated palliative care physicians, home-care agency services (e.g. nursing, personal support care), family practice physicians, outpatient clinics, palliative hospice care, and co-ordinated acute and tertiary palliative care when necessary. These programmes have operated for several years, and therefore reflect current established service delivery models of home-based palliative care in Ontario. Palliative programme staff were responsible for identifying and approaching caregivers about study participation if they met the following eligibility criteria: (i) primary caregivers (relative or friend) of patients diagnosed with a malignant neoplasm; (ii) fluent in English; and (iii) at least 18 years of age. A written consent was completed by a research officer for those consenting by telephone, and the first interview was scheduled. Our study was approved by Research Ethics Boards at the University of Toronto, Mount Sinai Hospital and McMaster University. Data were obtained from telephone interviews with caregiver participants, regional databases for the palliative care programmes and home-care agency databases. Primary caregivers were interviewed by telephone every two weeks from study admission until patient death, and were asked to report resource utilisation, burden and clinical status over the previous two weeks. A two-week data collection period was selected to minimise recall bias (Severens et al. 2000) and to avoid overburdening caregivers with frequent interviews. In each interview, participants were asked to complete three questionnaires: the Caregiver Burden Scale in End-of-Life Care

Figure 1 Study framework.

430

© 2015 John Wiley & Sons Ltd

Predictors of caregiver burden

(CBS-EOLC); the Ambulatory and Home Care Record (AHCR); and the Palliative Performance Scale (PPS). A demographic data form was used in the first interview. Perceived burden of caregiver participants was measured using the CBS-EOLC, a 16-item questionnaire using a 4-point Likert scale (Dumont et al. 2008), which provides a total score, ranging from 16 to 64, with no subscale score. This scale was developed with caregivers of palliative patients with cancer, and measures caregivers’ cognitive appraisal of the gap between potential assistance and support perceived to be accessible for dealing with the demands of providing care, as well as caregivers’ perception of the meaning of interpersonal relationships. Assessment of the psychometric properties indicated appropriate reliability of the scale (Cronbach’s alpha = 0.95) and good levels of convergent validity with fatigue (r = 0.69) and depression (r = 0.54) (Dumont et al. 2008). Private and public resource utilisation was measured using the AHCR (Guerriere & Coyte 2011). The details of this instrument are described in our previous work (Guerriere et al. 2006, 2008). Public resources measured by this instrument include all publicly financed healthcare services, such as consultations with health providers (home and ambulatory), diagnostic tests, medications, equipment/supplies, inpatient care and home-care services. Private resources comprise any costs not paid for by the public sector which are incurred by the families and private health insurance companies. Additionally, private resources include the unpaid care-giving costs of family caregivers in the form of time dedicated to providing care. Such care-giving refers to the time spent performing tasks associated with the receipt and provision of healthcare services (e.g. time travelling to appointments, personal care tasks). Patient functional status was measured using the PPS, which assigns functional performance levels based on caregiver assessments. Components of ambulation, activity and evidence of disease, selfcare, intake and conscious level were scored where values were assigned for levels of functional performance ranging from full ambulation and healthy to death. The scores for each component were summed as a total score for each interview. Higher scores can be interpreted as decreased functional performance. Reliability testing of the PPS reported intra-class correlation coefficients ranging from 0.93 to 0.96 (Downing et al. 2007, Ho et al. 2008). A multilevel modelling for change (MLMC) approach was used to assess the extent to which each of the independent variables was associated with © 2015 John Wiley & Sons Ltd

variations in caregiver burden for an episode of palliative care and at various time points over the palliative care trajectory (Singer & Willett 2003). This approach allows for the examination of the extent to which the independent variables (time-stable covariates and time-varying covariates) account for variations in burden. The advantage of this technique, which is described in detail by Raudenbush and Bryk (2002), is that it relaxes the independence of observations assumption, allowing for correlated error structures across a series of observations (Raudenbush & Bryk 2002). The MLMC addresses questions similar to those posed for other types of panel models (e.g. fixed effects models), but is flexible in terms of its ability to handle unbalanced data (i.e. different numbers of observations between people) as well as data that are collected at varying intervals, making it the appropriate strategy in this study. In addition, the trajectory of burden was studied using a mixed model regression that allowed the intercept and slope of the time to death versus caregiver burden relationship to vary across individuals, and we explored whether the variation in intercepts and slopes was related to any of the covariates (Singer 1998). Data were analysed using Stata IC 12 (StataCorp LP, College Station, Texas, USA).

Results From 1 July 2010 to 31 August 2012, 805 caregivers of patients met the study inclusion criteria. Eightyeight of these potential participants were not approached as the attending physicians noted that the caregiver was in extreme stress. Of the remaining 717 potential participants, 552 caregivers (77.0%) were contacted by the Research Officer, of whom 367 (66.5%) agreed to receive further information. Three hundred and forty-one caregivers consented to the study, but 14 of these were ineligible at the time of first interview because the patient was hospitalised (n = 4), had died (n = 9) or moved (n = 1). Thus, our results are based on a sample of 327 caregiver participants, corresponding to a total of 1940 interviews. The mean number of interviews per caregiver per month was 2.2. Patient and caregiver characteristics The characteristics of the 327 patients and their caregivers are shown in Tables 1 and 2 respectively. Mean and standard deviation values of caregiver burden for the overall cohort and for each characteristic are shown in Table 2. The average age of the patients was 71.7 years (SD: 12.9), and a nearly equal split of 431

D. Guerriere et al.

Table 1 Patient characteristics

All Age ≤60 61–70 71–80 80+ Sex Female Male Education High school or less Any vocational/college Any university Postgraduate Living arrangement Lives alone Lives with spouse Living with children Living with others Marital status Married Never married Divorced/separated Widow

Table 2 Caregiver characteristics N

%

327

100

51 93 85 98

15.6 28.4 26.0 30.0

173 154

52.9 47.1

173 55 66 33

52.9 16.8 20.2 10.1

57 191 53 26

17.4 58.4 16.2 8.0

194 10 34 89

59.3 3.1 10.4 27.2

males and females. The majority of patients lived with someone (82.6%). The average number of days from study admission to death was 110 days. The average age of the caregivers was 59.0 years (SD: 13.2), and 68.2% were female. Most of the caregivers were married and were either the spouse (48.0%) or offspring (38.2%) of the patient. Less than one-third of the caregivers reported a high school education or less (27.8%). Over half indicated that they were not working in the labour force while caring for their family member (57.2%). Multilevel model for change results: caregiver burden The results of the multilevel model are shown in Table 3. It is noteworthy that at the net of every covariate, caregiver burden increased over time, and did so in a non-linear fashion (i.e. the slope increased at each time point). Beyond time (linear and quadratic) alone, several other covariates were statistically significant in terms of their ability to predict burden. The results of each variable are discussed and grouped according to the type of variable (timevarying covariates and time-stable covariates). Two of the economic time-varying covariates – monthly family care-giving time costs and monthly public personal support worker (PSW) costs – were positively associated with higher caregiver burden, with 432

Total Buren score

All Age ≤60 61–70 71–80 80+ Sex Female Male Education High school or less Any vocational/college Any university Missing Postgraduate Marital status Married Never married Divorced/separated Missing Widow Relationship of caregiver Child Friend Other Other family Sibling Spouse Employment of caregiver Disability Full-time employed Missing Not employed On leave from full/ part-time employment Part-time/ casual employment Retired Student

N

%

Mean

327

100

25.7

8.6

SD

167 100 40 20

51.1 30.6 12.2 6.1

26.5 25.5 23.7 22.7

8.5 8.3 9.6 6.4

223 104

68.2 31.8

26.3 24.1

8.8 8.0

91 45 86 62 43

27.8 13.8 26.3 19.0 13.1

24.2 24.4 26.0 30.7 27.6

9.0 6.4 8.1 11.0 9.0

242 27 27 25 6

74.0 8.3 8.3 7.6 1.8

25.8 23.7 27.0 30.5 23.3

8.8 7.5 7.8 12.0 8.6

125 9 20 1 15 157

38.2 2.8 6.1 0.3 4.6 48.0

25.4 24.3 24.3 38.5 26.0 26.0

7.9 8.9 7.4 14.9 9.9 9.0

9 96 21 41 20

2.8 27.4 6.4 12.5 6.1

24.3 26.3 31.3 26.4 30.2

6.5 8.3 10.8 8.1 10.4

22

6.7

26.3

7.0

117 1

35.8 0.3

23.9 22.0

8.6 1.6

SD, standard deviation.

the effect of PSW costs being one of the stronger predictors in the model. Emergency department (ED) visits and PPS were also found to increase burden consistently at any given time; however, an increase in the number of hospice days was associated with lower burden scores. PPS was the strongest predictor of burden. Three time-stable covariates were associated with caregiver burden: caregiver gender, caregiver education and patient education. Caregiver gender was not predictive of burden at the start of the observation period (where t = 0), but as death approached, female © 2015 John Wiley & Sons Ltd

Predictors of caregiver burden

Table 3 Multilevel model for change results: caregiver burden Full model coefficient Burden (log) Time (linear) Time2 (quadratic) Time-varying covariates Family care-giving time cost Personal support worker costs† Nursing costs† General practitioner costs† Palliative care physician costs† Emergency department visits† Days overnight in hospital† Days in hospice† Patient functional status (Palliative Performance Scale score)‡ Time-stable covariates Male caregiver Male caregiver 9 time Married caregiver Caregiver education – any postgraduate Caregiver education – any vocational Caregiver education – any university Patient education – any postgraduate Patient education – any vocational Patient education – any university Male patient Patient lives alone Toronto region Comorbidity score Constant

SE

0.02** 0.03**

0.01 0.01

0.15*** 0.04** 0.01 0.02 0.00 0.05*** 0.02 0.04* 0.32***

0.02 0.02 0.02 0.02 0.02 0.01 0.02 0.02 0.02

0.04 0.26*** 0.02 0.19*

0.07 0.06 0.06 0.08

0.01

0.07

0.11

0.08

0.17*

0.08

0.06

0.06

0.04

0.06

0.18** 0.00 0.16* 0.02 0.26***

0.07 0.06 0.07 0.06 0.08

Reference category for education is high school or less. All covariates are standardised and grand-mean-centred (mean = 0; SD = 1). SE, standard error. † Public resources. ‡ Palliative Performance Scale assesses patient functional performance level. *P < 0.05; **P < 0.01; ***P < 0.001.

caregivers tended to report more burden compared to men (Figure 2). This finding is represented by the interaction between caregiver gender and time (linear). Burden tended to be higher when patients were male and this effect was relatively strong (P < 0.001). Both caregiver and patient education were also modelled (categorically), where the reference category was ‘high school or less’ respectively. Compared to lower educated caregivers, those with postgraduate education tended to report higher levels of burden. However, caregivers reported less burden when they © 2015 John Wiley & Sons Ltd

were treating patients with postgraduate education (compared to patients in the lowest educational category).

Discussion This is the first study to examine the association between caregiver burden and a comprehensive set of predictor variables in caregivers of patients enrolled in home-based palliative care programmes. Several modifiable and non-modifiable variables influenced caregivers’ burden including family care-giving time costs, the cost of personal support workers (PSW), ED visits, hospice use and patient functional status. Identification of non-modifiable variables helps to recognise those caregivers most at risk of caregiver burden and in need of support to reduce burden; identification of modifiable variables has implications for programme planning and allocation of resources that may serve to reduce caregiver burden. Herein, we compare our results to previous studies that have assessed predictors of burden for caregivers of palliative cancer patients; in the majority of these studies, researchers used burden scales that were divided by domains of burden. Our findings are only compared to those studies that yielded multivariate regression results, i.e. controlled for potential confounders; studies that reported univariate results are not discussed. In our study, caregiver burden increased over time in a non-linear fashion. This finding is similar to that of another study that found that burden increased as patient illness progressed (Grunfeld et al. 2004). In contrast, two other longitudinal studies found no change in burden as death approached (Garlo et al. 2010, Lee et al. 2013). However, in one study, several caregivers withdrew when the patient was near death, which may have yielded biased results if those who withdrew were experiencing greater burden (Garlo et al. 2010). We did not find that patient or caregiver age was related to caregiver burden. The findings in the literature on caregiver age are varied. Two studies found no relation to caregiver age and burden (Garlo et al. 2010, Lee et al. 2013), one study reported higher caregiver burden in older caregivers (Brazil et al. 2003), while another study found that increasing caregiver age decreased ‘scheduling’ burden (a subscale of burden), but not for the other subscales (Francis et al. 2010). Given the variability in reports, the influence of caregiver age may be context-specific, although the absence of significant results regarding burden and patient age in available studies suggests that patient age likely does not influence caregiver burden. 433

D. Guerriere et al.

Figure 2 Relationship between caregiver (CG) burden and CG gender over time. Mean CG burden score is shown in the months of observation prior to death. Female CGs tended to report more burden compared to men as death approaches.

In our study, patient gender was a significant predictor of burden. Caregivers of male patients had higher burden compared to caregivers of female patients. Two studies of caregivers of cancer patients (who were not palliative) examined patient gender and found no relation with caregiver burden (Papastavrou et al. 2009, Garlo et al. 2010). In addition to not being palliative, these studies differed from our study. One study employed a one-time cross-sectional design and a self-selected sample (Papastavrou et al. 2009). The other study included patients suffering from a variety of conditions (cancer, heart failure and chronic obstructive pulmonary disease), and the authors identified that their sample was limited in ethnical and racial diversity (Garlo et al. 2010). In our study, caregiver gender was not predictive of burden at the start of the observation period, but as time progressed, male caregivers tended to report less burden compared to female caregivers (Figure 2). Gender differences in care-giving may be explained by the findings of Brazil et al. (2009) who identified significant differences between male and female caregivers in overall strain, patterns of healthcare utilisation and patterns of support to the care recipient. Specifically, female caregivers were more likely to provide support in toilet-related tasks, and less likely to receive outside support from family and friends for personal care tasks. Male caregivers were more likely to provide support for mobility-related tasks and seek assistance from pain and symptom management consultants (Brazil et al. 2009). In another study, male caregivers of terminal cancer patients who could share care-giving activities perceived a lesser burden 434

of ‘disrupted schedule’ (Park et al. 2012). In addition, a cross-sectional study found that male caregivers of patients with advanced cancer reported less burden (Utne et al. 2013). It should be noted that several researchers have reported that caregiver gender was not associated with caregiver burden (Brazil et al. 2003, Given et al. 2004, Papastavrou et al. 2009, Garlo et al. 2010, Lee et al. 2013), although in the majority of these studies, the caregiver samples were primarily female. One-third of our sample consisted of male caregivers, which is higher than most of the previously conducted studies. Marital status of both the patient and caregiver, relationship of the caregiver to the patient and living arrangements were not predictors of burden. Marital status has not been found to predict burden in previous studies (Garlo et al. 2010, Shieh et al. 2012, Teixeira & Pereira 2013). However, one study found that married caregivers experienced reduced burden, as measured using one dimension of burden (i.e. caregivers’ perceptions of amount of family help in delivering care) (Park et al. 2012). Several studies have examined the association between caregiver relationship with the patient and caregiver burden (Brazil et al. 2003, Papastavrou et al. 2009, Garlo et al. 2010, Park et al. 2012, Lee et al. 2013), and two of the six studies found an association between caregiver relationship and burden (Brazil et al. 2003, Lee et al. 2013). Living arrangement has been found in previous work to be associated with caregiver burden (Francis et al. 2010), and not associated (Brazil et al. 2003). The contexts were very different in these studies, suggesting that there may be a more complicated © 2015 John Wiley & Sons Ltd

Predictors of caregiver burden

triangular relationship between living arrangement, type and intensity of health services, and burden outcomes. Caregivers who had higher family care-giving time costs reported higher burden. It is expected that caregivers who spend more time care-giving could experience greater subjective burden because of the work. We did not identify other studies that evaluated the relationship between care-giving costs and burden. Time spent care-giving is a proxy for time costs. Yoon et al. (2014) found that increasing care-giving hours per day was associated with increased caregiver burden, and suggest that more time spent care-giving limits time available for daily activities that may help to reduce burden (i.e. hobbies, personal relationships). A Taiwanese study measured care-giving hours as their unit of analysis (while our study measured costs) and that study found that caregivers who reported higher care-giving hours (13–24 hours per day) had lower burden than those reporting lower time (8–12 hours) (Lee et al. 2013). It was postulated that this finding was the result of a greater proportion of caregivers who reported lower care-giving hours were concurrently employed in the labour market, in comparison with those with higher care-giving hours (Lee et al. 2013). Furthermore, the study was in Taiwan which has a culture with a high sense of familial obligation, and where Chinese cultural expectations discourage women from voicing difficulty in caring for their family members (Lee et al. 2013). Caregivers of patients who had higher PSW costs (public) had higher burden than caregivers of patients who had lower PSW costs. This implies that patients receiving high amounts of PSW care correspond to a care-giving context that has an associated experience of high burden. This is an unexplored area in the literature as we were not able to identify another study that has examined this relationship. We did not find an association between nursing costs, general practitioner costs and palliative care physician costs with burden. Increased burden was found in caregivers of patients who had relatively higher ED visits. This is a new finding in the palliative care literature. Higher ED visits may be a symptom of declining health of the patient, which can lead to caregivers feeling overburdened. Future research may seek to identify the more common triggers leading to ED visits, so that caregivers can be informed about how best to manage these circumstances. Caregivers of patients who used hospice services reported less burden. It is probable that these caregivers would experience less burden as some of their responsibility is eased as the hospice assumes a por© 2015 John Wiley & Sons Ltd

tion of the care-giving tasks. A recent literature review found the relationship between hospice use and caregiver burden to be mixed – one study reviewed found positive caregiver outcomes were associated with hospice use (reduced mortality), while others found hospice use did not reduce adverse outcomes such as burden and low quality of life (Pottie et al. 2014). This inconclusive finding may be a result of variation in the range of services offered by different hospices that has been noted in both Canada and the United States (Carlson et al. 2007, Towns et al. 2012). Caregivers of patients with lower functional status (as measured by the PPS) had higher burden. Decreases in patient functional status may increase the amount and intensity of care-giving tasks required, resulting in higher caregiver burden. While the relationship between caregiver burden and scores specifically on the PPS has not been assessed elsewhere, other researchers have looked at the relationship between burden and functional status through the measurement of assistance required for activities of daily living (ADL). Brazil et al. (2003) found that increased need for assistance with ADLs was the most important predictor of high caregiver burden (Brazil et al. 2003). In contrast, two studies found no relationship between burden and ADLs (Given et al. 2004, Garlo et al. 2010). However, increased assistance with one or more ADLs was part of the inclusion criteria for one study (Garlo et al. 2010). In the other study, nearly all patients were independent with respect to ADLs, even when close to death (Given et al. 2004). In our study, caregivers who attained postgraduate education (independent of employment status) experienced greater levels of burden, while caregivers of patients who had attained postgraduate education experienced lower levels of burden. This finding may be partially explained by Nijboer et al. (1999) who postulated that caregivers with higher education may find caregiver tasks less rewarding when compared with their professional lives, as opposed to those caregivers with lower education (Nijboer et al. 1999). Caregiver education and burden have been examined in several other studies with various findings, although different measurements of education have been used (i.e. number of years of education versus level of education attained). Four studies found no relationship between caregiver education and burden (Brazil et al. 2003, Francis et al. 2010, Lee et al. 2013, Teixeira & Pereira 2013), whereas four other studies found higher levels of caregiver education were associated with decreased burden (Papastavrou et al. 2009, Park et al. 2012, Shieh et al. 2012, Utne et al. 435

D. Guerriere et al.

2013). This is an area that requires further investigation, and may also be context-specific given the differences in educational systems between countries. Limitations There are limitations associated with this study that require consideration. First, there may be an overall lower burden score than that determined because of social desirability bias; caregivers may report less burden because they do not want to portray a negative experience with care-giving. This may be an important consideration in burden research, particularly when the sample is derived from an ethnically diverse population, as there are some cultures where family values discourage individuals from articulating difficulty in caring for family members (Knight et al. 2002, Lee et al. 2013). This study measured predictors of burden in caregivers of patients receiving formal care from a publicly funded home-based palliative care programme. While we have compared our findings with that of other published work, it should be noted that true comparisons with other studies are challenging, as burden can be measured using different instruments, and the terms ‘burden’, ‘stress’ and ‘strain’ are often used interchangeably in the literature and confounds our understanding of which determinants are important in understanding caregiver burden (Thornton & Travis 2003, Bainbridge et al. 2009). Furthermore, differences in burden outcomes between studies may be more of a reflection of the availability and accessibility of local health services and private resources, which can differ from one setting to another. Future research may consider examining the relationship between burden outcomes and local health environment. For ethical reasons, our study did not examine ethnicity or race and their relationship with perceived caregiver burden. Although it could introduce bias, certain caregivers were excluded from our study because of the extreme stress to protect patients and families, as identified by attending physicians. Finally, while our study results are specific to the palliative care programmes studied, we believe that our study findings are potentially generalisable to caregivers of individuals receiving palliative care in different home-based palliative care programmes, and in different healthcare environments.

Conclusion The longitudinal design of our study allowed for the identification of both time-varying and time-stable 436

variables and their relationship with caregiver burden. Our study adds to the existing body of literature as it examines the role of health services costs (RNs, PSWs and physician) and family caregiver time costs in predicting caregiver burden. Non-modifiable variables associated with increased burden (caregiver gender, patient gender, caregiver and patient education, and patient functional status) may serve to identify those most in need of additional support. The identification of modifiable service variables related to burden outcomes suggests that interventions aimed at lessening family care-giving time, increasing PSW availability and encouraging those at risk for burden to explore hospice services may reduce burden in caregivers of home-based palliative care patients. While a general increase in palliative care servicing may lower caregiver burden, there may be opportunities to assess whether a more targeted approach is more cost-effective. Consequently, further research is required to identify the most effective strategy in the allocation of publicly financed palliative care services to reduce caregiver burden.

Acknowledgement This work was supported by the Canadian Institutes of Health Research (grant number PHE-101530).

References Bainbridge D., Krueger P., Lohfeld L. & Brazil K. (2009) Stress processes in caring for an end-of-life family member: application of a theoretical model. Aging & Mental Health 13 (4), 537–545. Brazil K., Bedard M., Willison K. & Hode M. (2003) Caregiving and its impact on families of the terminally ill. Aging & Mental Health 7 (5), 376–382. Brazil K., Thabane L., Foster G. & Bedard M. (2009) Gender differences among Canadian spousal caregivers at the end of life. Health & Social Care in the Community 17 (2), 159–166. Burton A.M., Sautter J.M., Tulsky J.A. et al. (2012) Burden and well-being among a diverse sample of cancer, congestive heart failure, and chronic obstructive pulmonary disease caregivers. Journal of Pain and Symptom Management 44 (3), 410–420. Carlson M.D., Morrison R.S., Holford T.R. & Bradley E.H. (2007) Hospice care: what services do patients and their families receive? Health Services Research 42 (4), 1672–1690. Chai H., Guerriere D.N., Zagorski B. & Coyte P.C. (2014) The magnitude, share and determinants of unpaid care costs for home-based palliative care service provision in Toronto, Canada. Health & Social Care in the Community 22 (1), 30–39. Doorenbos A.Z., Given B., Given C.W., Wyatt G., Gift A., Rahbar M. & Jeon S. (2007) The influence of end-of-life cancer care on caregivers. Research in Nursing and Health 30 (3), 270–281. © 2015 John Wiley & Sons Ltd

Predictors of caregiver burden

Downing M., Lau F., Lesperance M., Karlson N., Shaw J. & Kuziemsky C. (2007) Meta-analysis of survival prediction with Palliative Performance Scale. Journal of Palliative Care 23 (4), 245–252. Dumont S., Fillion L., Gagnon P. & Bernier N. (2008) A new tool to assess family caregivers’ burden during end-of-life care. Journal of Palliative Care 24 (3), 151–161. Emanuel E.J., Fairclough D.L., Slutsman J. & Emanuel L.L. (2000) Understanding economic and other burdens of terminal illness: the experience of patients and their caregivers. Annals of Internal Medicine 132 (6), 451–459. Francis L.E., Worthington J., Kypriotakis G. & Rose J.H. (2010) Relationship quality and burden among caregivers for late-stage cancer patients. Supportive Care in Cancer 18 (11), 1429–1436. Garlo K., O’Leary J.R., Van Ness P.H. & Fried T.R. (2010) Burden in caregivers of older adults with advanced illness. Journal of the American Geriatrics Society 58 (12), 2315–2322. Given B., Wyatt G., Given C.W. et al. (2004) Burden and depression among caregivers of patients with cancer at the end of life. Oncology Nursing Forum 31 (6), 1105– 1117. Goldstein N.E., Concato J., Fried T.R., Kasi S.V., JohnsonHurzeler R. & Bradley E.H. (2004) Factors associated with caregiver burden among caregivers of terminally ill patients with cancer. Journal of Palliative Care 20 (1), 38–43. Gomes B., Calanzani N., Curiale V., McCrone P. & Higginson I.J. (2013) Effectiveness and cost-effectiveness of home palliative care services for adults with advanced illness and their caregivers. The Cochrane Database of Systematic Reviews 6, CD007760. Grunfeld E., Coyle D., Whelan T. et al. (2004) Family caregiver burden: results of a longitudinal study of breast cancer patients and their principal caregivers. Canadian Medical Association Journal 170 (12), 1795–1801. Guerriere D.N. & Coyte P.C. (2011) The ambulatory and home care record: a methodological framework for economic analyses in end of life care. Journal of Aging Research 2011. DOI: 10.4061/2011/374237. Guerriere D.N., Tullis E., Ungar W.J. et al. (2006) Economic burden of ambulatory and home-based care for adults with Cystic Fibrosis. Treatments in Respiratory Medicine 5 (5), 351–359. Guerriere D.N., Wong A.Y., Croxford R., Leong V.W., McKeever P. & Coyte P.C. (2008) Costs and determinants of privately financed home-based health care in Ontario, Canada. Health and Social Care in the Community 16 (2), 126–136. Guerriere D., Husain A., Marshall D. et al. (2016) Predictors of place of death across the home based care trajectory. Journal of Palliative Care 24 (4), 428–438. Hauser J.M. & Kramer B.J. (2004) Family caregivers in palliative care. Clinics in Geriatric Medicine 20 (4), 671–688. Ho F., Lau F., Downing M. & Lesperance M. (2008) A reliability and validity study of the Palliative Performance Scale. BMC Palliative Care 7, 10. Hudson P. & Payne S. (2011) Family caregivers and palliative care: current status and agenda for the future. Journal of Palliative Medicine 14 (7), 864–869. Hughes S.L., Giobbie-Hurder A., Weaver F.M., Kubal J.D. & Henderson W. (1999) Relationship between caregiver burden and health-related quality of life. The Gerontologist 39 (5), 534–545. © 2015 John Wiley & Sons Ltd

Hwang S.S., Chang V.T., Alejandro Y. et al. (2003) Caregiver unmet needs, burden, and satisfaction in symptomatic advanced cancer patients at a Veterans Affairs (VA) medical center. Palliative & Supportive Care 1 (4), 319–329. Kapari M., Addington-Hall J. & Hotopf M. (2010) Risk factors for common mental disorder in caregiving and bereavement. Journal of Pain and Symptom Management 40 (6), 844–856. Knight B., Robinson G., Flynn Longmire C., Chun M., Nakao K. & Kim J. (2002) Cross cultural issues in caregiving for persons with dementia: do familism values reduce burden and distress? Ageing International 27 (3), 70–94. Lee K.C., Chang W., Chou W., Su P.J., Hsieh C.H., Chen J.S. & Tang S.T. (2013) Longitudinal changes and predictors of caregiving burden while providing end-of-life care for terminally ill cancer patients. Journal of Palliative Medicine 16 (6), 632–637. McNamara B. & Rosenwax L. (2010) Which carers of family members at the end of life need more support from health services and why? Social Science & Medicine 70 (7), 1035–1041. Nijboer C., Triemstra M., Tempelaar R., Sanderman R. & van den Bos G.A. (1999) Determinants of caregiving experiences and mental health of partners of cancer patients. Cancer 86 (4), 577–588. Papastavrou E., Charalambous A. & Tsangari H. (2009) Exploring the other side of cancer care: the informal caregiver. European Journal of Oncology Nursing 13 (2), 128–136. Park C., Shin D.W., Choi J.Y. et al. (2012) Determinants of the burden and positivity of family caregivers of terminally ill cancer patients in Korea. Psycho-Oncology 21 (3), 282–290. Pottie C.G., Burch K.A., Montross Thomas L.P. & Irwin S.A. (2014) Informal caregiving of hospice patients. Journal of Palliative Medicine 17 (7), 845–856. Raudenbush S.W. & Bryk A.S. (2002) Hierarchical Linear Models, 2nd edn. SAGE, Thousand Oaks, CA. Rhee Y.S., Yun Y.H., Park S. et al. (2008) Depression in family caregivers of cancer patients: the feeling of burden as a predictor of depression. Journal of Clinical Oncology 26 (36), 5890–5895. Schulz R. & Beach S.R. (1999) Caregiving as a risk factor for mortality: the caregiver health effects study. Journal of the American Medical Association 282 (23), 2215–2219. Schulz R., Hebert R. & Boerner K. (2008) Bereavement after caregiving. Geriatrics 63 (1), 20–22. Severens J.L., Mulder J., Laheij R.J. & Verbeek A.L. (2000) Precision and accuracy in measuring absence from work as a basis for calculating productivity costs in The Netherlands. Social Science & Medicine 51 (2), 243–249. Shieh S., Tung H. & Liang S. (2012) Social support as influencing primary family caregiver burden in Taiwanese patients with colorectal cancer. Journal of Nursing Scholarship 44 (3), 223–231. Singer J.D. (1998) Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics 23 (4), 323–355. Singer J.D. & Willet J.B. (2003) Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. Oxford University Press, New York, NY. Song J.I., Shin D.W., Choi J.Y. et al. (2011) Quality of life and mental health in family caregivers of patients with terminal cancer. Supportive Care in Cancer 19 (10), 1519–1526.

437

D. Guerriere et al.

Teixeira R.J. & Pereira M.G. (2013) Psychological morbidity, burden and the mediating effect of social support in adult children caregivers of oncological patients undergoing chemotherapy. Psycho-Oncology 22 (7), 1587–1593. Thornton M. & Travis S.S. (2003) Analysis of the reliability of the modified caregiver strain index. The Journals of Gerontology. Series B: Psychological Sciences and Social Sciences 58 (2), S127–S132. Towns K., Dougherty E., Kevork N. et al. (2012) Availability of services in Ontario hospices and hospitals providing inpatient palliative care. Journal of Palliative Medicine 15 (5), 527–534.

438

Utne I., Miaskowski C., Paul S.M. & Rustoen T. (2013) Association between hope and burden reported by family caregivers of patients with advanced cancer. Supportive Care in Cancer 21 (9), 2527–2535. Williams A.L. & McCorkle R. (2011) Cancer family caregivers during the palliative, hospice, and bereavement phases: a review of the descriptive psychosocial literature. Palliative & Supportive Care 9 (3), 315–325. Yoon S.J., Kim J.S., Jung J.G., Kim S.S. & Kim S. (2014) Modifiable factors associated with caregiver burden among family caregivers of terminally ill Korean cancer patients. Supportive Care in Cancer 22 (5), 1243–1250.

© 2015 John Wiley & Sons Ltd