Risk Factors for Common Mental Disorder in Caregiving and Bereavement

Transcription

1 844 Journal of Pain and Symptom Management Vol. 40 No. 6 December 2010 Original Article Risk Factors for Common Mental Disorder in Caregiving and Bereavement Marcia Kapari, BSc, MSc, PhD, Julia Addington-Hall, BA, PhD, HonMFPH, and Matthew Hotopf, BSc, MBBS, MSc, PhD, MRCPsych Centre for Behavioural Medicine (M.K.), The School of Pharmacy, London; School of Health Sciences (J.A.-H.), University of Southampton, Southampton; and Institute of Psychiatry (M.H.), King s College London, London, United Kingdom Abstract Context. Longitudinal studies that have described the trajectory of familial caregiving and bereavement outcomes have often failed to include a comprehensive range of carer and caregiving variables or any patient assessments, and only a few have used a recognized structured interview for psychiatric disorders. Objectives. To address these limitations, this study aimed to establish links between symptoms of CMD in carers of patients (with advanced disease) during the caregiving phase and their subsequent bereavement. Methods. To identify the risk factors for poor caregiving and bereavement outcomes, we assessed patients and caregivers using a wide range of measures in a prospectively acquired sample. The main outcome, CMD, was measured through the use of a standardized interview (Revised Clinical Interview Schedule). One hundred carers of patients with advanced disease (more than 95% of patients had a cancer diagnosis) were interviewed shortly after the patient was referred to a U.K. hospice. Interviews were repeated at three and six months after the death of the patient. Results. Multivariate analyses revealed that carers who perceived their caring experience as more burdening had more symptoms of CMD while caring for their loved one. Carer mental health during the caregiving experience was predictive of their mental health at three and six months after death. No relationships were observed between family relations, levels of social support, levels of religious or spiritual beliefs, carers coping strategies, quality of death in the patient, and caregiving and bereavement outcomes. Conclusion. Our findings suggest that much psychological distress detected during caregiving continues into bereavement. J Pain Symptom Manage 2010;40:844e856. Ó 2010 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Address correspondence to: Marcia Kapari, BSc, MSc, PhD, Centre for Behavioural Medicine, The School of Pharmacy, BMA House, Tavistock Square, Ó 2010 U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. London WC1H 9JP, United Kingdom. marcia.kapari@pharmacy.ac.uk Accepted for publication: March 2, /$ - see front matter doi: /j.jpainsymman

2 Vol. 40 No. 6 December 2010 Caregiving and Bereavement 845 Key Words Palliative care, cancer, carers, informal caregiving, common mental disorder (CMD), bereavement Introduction The majority of deaths among older people are preceded by an extended period of informal familial caregiving. 1 There is a strong consensus that caring for a terminally ill relative is stressful and places the carer at a high risk of poor social, psychological, and physical health outcomes. 2e4 The provision of familial caregiving has potential benefits for health and social services, as family involvement may reduce the use of long-term residential care, lessen hospitalization, and reduce the use of community services. 5 Depression and impaired physical health in carers may reduce their ability to function and threaten their ability to provide informal care. Policy makers have identified the needs of carers, particularly in recognizing and managing carer distress, as a priority. For example, in the United Kingdom, the National Institute for Health and Clinical Excellence has endorsed the development of palliative care services that will directly attend to the needs of informal carers. 6 Although the negative outcomes of informal caregiving and their harmful effect on the carer s psychological and physical health have been extensively investigated, 7e11 there is limited information about how the prolonged experience of caregiving can affect bereavement outcomes in the carer. The literature on caregiving outcomes and bereavement outcomes has developed in relative isolation of one another. Two alternativedalthough not mutually exclusivedframeworks for understanding the bereavement process after the provision of informal caregiving, based on the stress and coping paradigm, have been suggested. The first, known as the relief model, suggests that caregivers who suffer increased strain will experience relief after the death of the care recipient. 12 As the caregiving role ends, it is assumed that the caregiving strain vanishes at the same time. The individual is relieved of former responsibilities related to the caregiving role and is enabled to return to other precaregiving roles such as employment, leisure, and so on. As a result, the vanished stresses of caregiving provide a sense of relief, which alleviates the bereavement process in turn. The second framework, known as the complicated grief model, proposes that increased caregiver role strain predicts poorer bereavement adjustment. It argues that caregiver stress increases over time, leaving the individual with weaker coping resources to adjust to the bereavement experience. 12 A number of studies have used a retrospective design to relate self-reported caregiving experiences with bereavement outcomes. These studies have mostly found that a positive caregiving experience is protective against negative bereavement outcomes. More precisely, positive caregiving appraisals were associated with better bereavement adjustment, 13 whereas highly burdening and distressing caregiving scenarios were associated with worse bereavement outcomes. 14e16 Contrasting evidence has been reported in a study comparing the outcomes of noncaregivers, caregivers who reported no strain, and strained caregivers. 17 These authors found that carers who were strained before the death of the care recipient reported better health outcomes in bereavement. These studies are, however, limited, as appraisals of the caregiving situation are based on retrospective accounts and could, therefore, be subject to recall bias. Individuals who are distressed during bereavement are likely to have more negative perceptions of prior experiences. This limitation has been overcome in a number of longitudinal studies that have identified carers during the caregiving phase and have followed them up during bereavement describing the continuous trajectory of familial caregiving and bereavement outcomes. 18e22 Such studies have produced inconsistent findings in heterogeneous groups of patients. Most longitudinal studies that have examined the effect of caregiving strain on bereavement outcomes in groups of carers of cancer patients have concluded that greater psychological and physical strain during caregiving is predictive of worse bereavement outcomes. 12,19,23 This relation has been described among carers of patients with other diagnoses such as HIV/AIDS, 24 dementia, 18,25 and patients from diagnostically heterogeneous groups. 26,27 Contrasting evidencedsuggesting

3 846 Kapari et al. Vol. 40 No. 6 December 2010 that caregivers who suffer from increased caregiver strain will be relieved after the care recipient s deathdalso has been reported. 1,22 These studies have found that the transition from caregiving to bereavement provides relief rather than posing mental health risks. Other patient, carer, and family variables can affect bereavement outcomes and should be studied in addition to caregiver strain to provide a more comprehensive picture of the transition from caregiving to bereavement. A number of prospective studies have also highlighted the protective role of a good family relation on bereavement outcomes. 20,23,28 For instance, one prospective study reported that family tension during the caregiving phase was predictive of poorer mental health during bereavement. 23 The carer s coping mechanisms during caregiving also have been found to have an impact on the bereavement adjustment process. Moskowitz et al. 29 found that active problem-solving and positive reappraisals of the caring situation during the caregiving phase were associated with improved mood during the bereavement phase, whereas selfblame and behavioral avoidance coping mechanisms at the caregiving phase were associated with worsened mood during bereavement. Not surprisingly, social support during the caregiving experience also has been found to be protective against poor bereavement outcomes. 26 Religiosity also has been related with better caregiving outcomes, 30 and higher levels of religious and spiritual beliefs during the caregiving phase have been associated with better bereavement outcomes. 20 The quality of death in the patient also has been found to affect bereavement outcomes. In a large study, 31 hospice use and a good death in the patient (described, by the authors, as painless, anticipated, and not too burdening on family members) were associated with significantly lower mortality rates in the grieving spouse. There is also evidence that one s place of death will influence the survivor s bereavement. For instance, death in a hospital setting might provide less opportunity for emotional closeness resulting in worse bereavement outcomes. A death at home, by contrast, may lead to increased distress as the home can become a constant reminder of the loss. 32 Lauer et al. 33 examined the effects of home vs. hospital deaths and found that the former were related to better bereavement outcomes. Contrasting evidencedarguing that home deaths are associated with poor outcomes in the bereaved relativesdalso has been reported. 34 Previous studies that have described the transition from caregiving to bereavement have focused on diverse caregiving, patient, and family variables. Few are comprehensive in terms of the domains of risk factors assessed. Most of these studies focused on caregiver characteristics and did not include any assessments of the patient s state, such as quality of life or psychiatric morbidity. Similarly, the patient s quality of death, which elsewhere has been found to play an important role in bereavement adjustment, 31 has been left largely unexplored. Finally, few studies have used recognized structured interviews for psychiatric disorders. The present study aimed to identify the risk factors for poor caregiving and bereavement outcomes by assessing both patients and caregivers on a range of measures in a prospectively acquired sample. The following hypotheses were tested: 1. Symptoms of common mental disorder (CMD) in caregivers will be associated with more negative experiences of caregiving, poorer family relations, and lower levels of support during the caregiving phase. 2. Poorer bereavement outcome (defined as more symptoms of CMD and higher scores on the Bereavement Phenomenology Questionnaire [BPQ] during bereavement) will be associated with more symptoms of CMD during caregiving, a poorer experience of caregiving, poorer family relations, lower levels of social support, and a poorer quality of death in the patient. 3. Caregivers who report higher levels of religious or spiritual beliefs will experience more positive caregiving outcomes and better bereavement outcomes. 4. Caregivers who use more problem-focused strategies will have better outcomes in both caregiving and bereavement. Methods Setting The research took place in two hospices in South London, St. Christopher s Hospice and

4 Vol. 40 No. 6 December 2010 Caregiving and Bereavement 847 Trinity Hospice. These are voluntary sector organizations that provide specialist palliative care to defined catchment areas, usually, although not exclusively, to patients with cancer. Patients are cared for in the community by multidisciplinary teams, usually with clinical nurse specialists (CNS) as their main contact. Both hospices have inpatient and daycare facilities. Design This was a prospective cohort study following carers of patients receiving palliative care from the point shortly after the patient was referred to a hospice to six months after the death of the patient. Interviews and questionnaires were administered to participants at three time points; approximately two to six months before the anticipated death of the patient (T1), three months after the death (T2), and six months after the death (T3). A brief questionnaire on quality of death was completed by the CNS caring for the dying patient approximately one week after the death (Fig. 1). Inclusion/Exclusion Criteria Family members of newly referred patients were invited to participate. To be included, the family member had to be identified by the patient as his/her main carer. This study recruited those patients with a life expectancy of more than two months but less than six months as assessed by the clinical team. Carers and patients who had problems understanding English or limited literacy skills (as assessed by their home care nurse) were excluded. Questionnaires/Interviews Table 1 contains a summary of the measures included in each time interval of the present study. All measures have satisfactory reliability and validity. The Hospital Anxiety and Depression Scale (HADS) was used to gain a more accurate picture of patients psychological well-being. This scale consists of 14 items within two subscales, anxiety and depression. Higher scores suggest more symptoms of anxiety and/or depression. A total score of $11 indicates significant anxiety and/or depression. This cutoff point is widely used in palliative care patients. 35 Fig. 1. Study design. The Revised Clinical Interview Schedule (CIS-R) has been widely used in both hospital and community settings as a means of evaluating CMD. Caseness, according to the CIS-R, has been defined as a score of $12, which indicates that the individual is displaying symptoms of CMD. 36 Procedure Baseline Interviews. Potentially eligible participants were informed about the study by their CNS. At that time, patients and their carers were given the opportunity to opt out. Carers and patients who agreed to participate were then contacted by the researcher within one week and asked to give written informed

5 848 Kapari et al. Vol. 40 No. 6 December 2010 Table 1 Measures Used in All Time Intervals Participant Time Point Instrument(s) and Referencesdn of Items Variable Under Measure Patient Two to six months before Patient s General Interview Sociodemographics death (T1) MQOL 38 d16 Patient quality of life HADS 39 d14 Patient mood The ECOG Performance Status Patient functional status Rating Scale 40 d1 Carer Two to six months before CIS-R 36 d14 sections Caregiver psychiatric assessment the patient s death (T1) Caregiving Burden Caregiver burden Interview 41 d22 WSAS 42 d5 Caregiver functional impairment SOS 43 d8 Caregiver social support FACES-III 44 d20 Family relations Brief COPE 45 d28 Caregiver coping styles Royal Free Interview for Spiritual and Religious Beliefs 46 d20 Caregiver religious/spiritual beliefs CNS One week after the patient s death QOD 47 d13 Patient s QOD Carer Three months after the patient s CIS-R 36 d14 sections Caregiver psychiatric assessment death (T2) QOD 47 d13 Patient s quality of death BPQ 48 d22 Caregiver bereavement outcome Carer Six months after the patient s CIS-R 36 d14 sections Caregiver psychiatric assessment death (T3) BPQ 48 d22 Caregiver bereavement outcome MQOL ¼ McGill Quality of Life Questionnaire; WSAS ¼ Work and Social Adjustment Scale; SOS ¼ Significant Others Scale; FACES-III ¼ Family Adaptability and Cohesion Evaluation Scales; QOD ¼ quality of death; BPQ ¼ Bereavement Phenomenology Questionnaire. consent. Interviews were arranged to take place at the participants own homes. Follow-Up Interviews. If a patient died, the carer was sent a brief letter approximately one month after the patient s death, asking whether he/she would mind being seen by the researcher again. Carers who agreed to be seen were asked to give written informed consent for the second part of the study, which involved two interviews (three and six months after the death). Interviews took place at the carers homes. Statistical Analyses The statistical analyses used to examine the hypotheses in question included the following: 1. Prevalence estimates for CMD (with 95% confidence intervals) were calculated at three time points (T1, T2, and T3). 2. Multiple regression analysis was conducted to model the relationship between independent variables on two main outcomes: CIS-R score (measuring symptoms of anxiety and depression) before bereavement and at three and six months; and bereavement symptoms measured on the BPQ, measured at three and six months after bereavement. 3. A set of models was tested for each outcome, where we entered independent variables in blocks. The approach we took was hierarchical, where we assumed that some independent variables were relatively immutable (e.g., age, sex, and marital status of the carer), and these should be entered before those related to the patient s disease (e.g., diagnosis, presence of depression in the patient). These variables might, in turn, impact on variables associated with the carer s perceived burden or coping mechanisms that were entered next. Finally (for the post bereavement outcomes), we entered variables related to the nature of the death (quality and place of death) and the carer s mental state during the bereavement phase. The precise variables entered in each model are described in the Results section. Because the sample size was modest and we wanted to present the most parsimonious models, variables were only carried over from one model to the next if they were associated with the dependent variable, with a P-value set arbitrarily at 0.2.

6 Vol. 40 No. 6 December 2010 Caregiving and Bereavement 849 Results Recruitment Shortly after being referred to St. Christopher s or Trinity Hospice, patients and their carers were invited to participate in this study by their CNS. The researcher (M.K.) had no access to the lists of newly referred patients at that time; therefore, recruitment was led and influenced by the nurse specialists. A retrospective estimation of all potentially eligible participants was carried out for a six-month period during which recruitment was taking place. Although there were no statistically significant differences with regard to sociodemographic variables between potentially eligible participants who were invited to participate in the study and those who were not, we cannot be certain that our sample is unbiased. Characteristics of Sample One hundred and one potential participants were referred to the study, and 100 agreed to participate and were recruited to the study, yielding a sample of 100 patients and carers. The mean age of the caregiving group was 65.3 years and the patient group was 71.2 years. Eighty-five percent of the carers were the spouse or partner of the person for whom they were caring. The majority of caregivers were female (75%), whereas most patients were male (63%). Of the 30% of carers who were employed, 11% were on leave because of their relative s condition. The majority of both the caregivers and patients (89% and 90%, respectively) described their ethnicity as white British. Clinical Characteristics The clinical characteristics of the patient group are shown in Table 2. More than 95% had a cancer diagnosis with lung cancer being the most frequent diagnosis (22%). Patients functional status varied from normal activity to being bedridden. A small proportion of patients (8%) were recognized to be cognitively impaired. Patient and Carer Assessments Patients and carers were interviewed two to six months before the anticipated death of the patient (T1). One hundred carer-patient dyads were recruited, with 78 patients dying within the time frame of the study. The remaining 22 Table 2 Patient Clinical Characteristics Diagnosis n (%) Cancer 96 (96) Breast 5 (5) Ovarian 5 (5) Colon 5 (5) Prostate 12 (12) Bowel 5 (5) Bladder 2 (2) Lung 22 (22) Other cancer 40 (40) Noncancer 4 (4) Motor neuron disease 2 (2) Chronic obstructive pulmonary disease 1 (1) Liver failure 1 (1) Cognitive impairment (as assessed by CNS) Cognitively impaired 8 (8) No cognitive impairment 92 (92) Performance status (ECOG) Normal activity (0) 3 (3) Symptomatic but nearly fully ambulatory (1) 27 (27) Needing some extra bed time (2) 47 (47) In bed more than 50% of the time (3) 15 (15) Unable to get out of bed (4) 8 (8) patients were still alive at the end of the study (11 patients were still under hospice care and 11 patients had been discharged). Of the 78 bereaved carers invited to participate in the second wave of the study (T2), 22 (28%) declined. Of the 56 carers who agreed to the follow-up at T2 (72%), 10 requested to be interviewed at a later date, resulting in a total of 46 participants (59%) in the second wave of the study. Of the 46 carers who were interviewed at T2, six could not be contacted at T3. The remaining 40 carers agreed to be interviewed in the third wave of the study. Another 10 carers who had not been willing to see the researcher at T2 agreed to be interviewed at T3, resulting in a total of 50 participants (64%) at the six-month follow-up (T3). Other than the 22 (28%) bereaved carers who refused further participation in the study at T2 and the six bereaved carers (8%) who could not be contacted at T3, all remaining bereaved carers (64%) agreed to be interviewed at T3. There were no significant differences in terms of age, sex, and predeath mental health (CIS-R scores at T1) between participants who declined the follow-up at T2 and T3. Patient Measures at T1 Of the 100 patients recruited to the study, 75 were able to complete the baseline assessment. There was a statistically significant difference between the patient performance status

7 850 Kapari et al. Vol. 40 No. 6 December 2010 (Eastern Cooperative Oncology Group [ECOG]) and ability to complete the assessment (c 2 ¼ 5.1, P ¼ 0.02). Patients who were unable to complete the assessment had poorer performance status. In this sample, the mean (standard deviation [SD]) anxiety score on the HADS was 5.11 (4.43), and the mean (SD) depression score was 7.12 (3.61); 10.7% of patients were classified as suffering from anxiety and 16.0% of patients were classified as suffering from depression according to their HADS scores. The McGill Quality of Life Questionnaire was used to assess the patients overall well-being. Despite their severe disease, a surprisingly high proportion (approximately 40%) described their quality of life as very good or excellent; 10.9% of the patients described their quality of life as very bad. Carer Measures at T1 CIS-R as the Main Outcome. At T1, 35% of carers were cases according to their CIS-R scores when using a cutoff point of $12. CIS-R scores ranged from 0 to 39. The mean (SD) score was 9.89 (8.70). To examine the association between carer sociodemographic, patient and caregiving and family variables and CIS-R scores at baseline, several linear regressions were conducted, with CIS-R total scores as the dependent variable. Independent variables were included in further linear regression models if they had a univariate association of P < 0.2 with the dependant variable. Different models included different independent variables. The first set of variables in the multivariable analysis consisted of carer sociodemographic variables (i.e., age, sex, marital status, ethnicity, socioeconomic status, employment status, and relationship to patient). Model 2 included patient illness characteristics (i.e., HADS, diagnosis, cognitive impairment performance status (ECOG), and quality of life). Model 3 included caregiving variables that are likely to have resulted from responses to the patient s illness characteristics (i.e., caregiving burden, strength of religious/spiritual beliefs, social support, and carer coping mechanisms). In the final model, carers overall levels of daily impairment were included (Table 3). The final model revealed the following associations: 1. Levels of caregiving burden were positively associated with higher CIS-R scores. 2. Carers who professed stronger religious/ spiritual beliefs had better mental health outcomes. 3. Active coping in carers was associated with more symptoms of CMD, whereas substance abuse was associated with fewer symptoms of CMD. 4. Daily life functional impairment (work and social adjustment) was associated with the CIS-R outcome. Carers who reported that their daily life was more impaired because of their caregiving role experienced more symptoms of CMD. Carer Measures at T2 CIS-R as the Main Outcome. At T2, 32.6% of carers were cases according to their CIS-R scores when using a cutoff point of $12. CIS-R scores ranged from 0 to 39. The mean (SD) score was 8.91 (9.39). To examine the association between carer sociodemographic, patient, caregiving, family, and death variables, and CMD (CIS-R scores) three months after the death of the patient, several linear regressions were carried out, with CIS-R scores as the dependent variable. Independent variables were included in further linear regression models if they had a univariate association of P < 0.2 with the dependent variable. As before, Model 1 included carer sociodemographic characteristics. Model 2 included patient illness characteristics, which are outside the carer s control. Model 3 included caregiving and family variables that are the result of the caregiving experience. Finally, Model 4 included variables related to the death of the patient. The final model (Table 3) revealed the following relation: caregivers mental health at baseline was a predictor of their mental health at three months after death. Carers who reported more symptoms of CMD during the caregiving phase were more likely to do the same at T2. BPQ as the Main Outcome. At T2, BPQ scores ranged from 1 to 62. The mean (SD) score was (13.09). Using an identical analytic strategy as for the CIS-R multivariable modeling, multivariable associations between BPQ scores and carer, patient, caregiving, family, and death variables (Table 4) revealed that carers who

8 Vol. 40 No. 6 December 2010 Caregiving and Bereavement 851 Table 3 Associations of CIS-R Scores at T1, T2, and T3: Final Model a Regression Coefficients (95% CI) T1 T2 T3 Variables n ¼ 94; R 2 ¼ 50% n ¼ 34; R 2 ¼ 62% n ¼ 37; R 2 ¼ 82% Carer age (1) 0.11 ( 0.26, 0.04) d d Socioeconomic status (1) 0.50 ( 0.22, 1.22) d d Cancer diagnosis (2) d d 1.33 ( 7.29, 9.95) Caregiving burden (3) 0.21 (0.08, 0.34) 0.04 ( 0.21, 0.29) 0.04 ( 0.10, 0.03) Religion (strength of belief) (3) L0.07 (L0.14, L0.01) d d Coping mechanismsdactive coping (3) 1.44 (L0.05, 2.94) d 3.63 (1.86, 5.41) Coping mechanismdsubstance abuse (3) L1.86 (L3.43, L0.30) d d Coping mechanismsdbehavioral d d 1.52 ( 0.41, 3.46) disengagement (3) Discrepancy between perceived and ideal levels 0.35 ( 0.16, 0.86) 0.90 ( 1.94, 0.13) L0.68 (L1.21, L0.15) of practical support (3) Discrepancy between perceived and ideal levels d 0.09 ( 1.38, 1.56) d of emotional support (3) Level of daily life impairment (4) 0.26 (0.08, 0.45) d 0.08 ( 0.24, 0.08) Carer mental health at baseline Not applicable 0.74 (0.44, 1.04) 0.70 (0.50, 0.90) (CIS-R scores at T1) (4) Death away from home (4) Not applicable 0.91 ( 4.36, 6.18) 1.25 ( 2.00, 4.49) QOD (carer assessment) (4) Not applicable 0.03 ( 0.06, 0.12) 0.06 ( 0.01, 0.13) CI ¼ confidence interval; QOD ¼ quality of death. Predictor variables significant at P<0.05 level are highlighted in bold. a See text for an explanation of the variables entered into each model. (1) Model 1 variables (sociodemographic characteristics of the carer). (2) Model 2 variables (disease characteristics of the patient). (3) Model 3 variables (caregiver burden, social support, and coping). (4) Model 4 variables (level of impairment, baseline CIS-R, and death-related variables). experienced more symptoms of mental disorder at baseline were likely to have worse bereavement outcomes three months after the death of their loved one. Carer Measures at T3 CIS-R as the Main Outcome. At T3, 11 (22%) carers met the criteria for CIS-R caseness when using a cutoff point of $12. CIS-R scores ranged from 0 to 31. The mean (SD) score was 7.02 (7.99). Multivariable associations between CIS-R scores and carer, patient, caregiving, family, and death variables (using the same analytic strategy as described earlier) revealed the following relations (Table 3): 1. Caregivers who had reported more symptoms of CMD at baseline were likely to do the same at six months after the death of their loved one. 2. Carers who had reported high levels of active coping mechanisms during their Table 4 Associations of BPQ-R Scores at T2 and T3: Final Model a Regression Coefficients (95% CI) T2 T3 Variables n ¼ 34; R 2 ¼ 55% n ¼ 37; R 2 ¼ 51% Religion (strength of belief) (2) d 0.10 ( 0.02, 0.21) Family type (2) d 2.03 ( 1.09, 5.14) Discrepancy between perceived and ideal levels of practical support (3) 0.94 ( 2.41, 0.53) L1.07 (L2.06, L0.08) Discrepancy between perceived and ideal levels of emotional support (3) 0.76 ( 1.15, 2.67) d Carer mental health at baseline (CIS-R scores at T1) (4) 0.90 (0.52, 1.28) 0.79 (0.43, 1.15) Death away from home (4) 0.75 ( 6.47, 7.98) 0.15 ( 5.86, 6.17) QOD (carer assessment) (4) 0.07 ( 0.19, 0.06) 0.04 ( 0.16, 0.09) CI ¼ confidence interval; QOD ¼ quality of death. Predictor variables significant at P<0.05 level are highlighted in bold. a See text for an explanation of the variables entered into each model. (1) Model 1 variables (sociodemographic characteristics of the carer). (2) Model 2 variables (disease characteristics of the patient). (3) Model 3 variables (caregiver burden, social support, and coping). (4) Model 4 variables (level of impairment, baseline CIS-R, and death-related variables).

9 852 Kapari et al. Vol. 40 No. 6 December 2010 caregiving experience had worse mental health outcomes during their bereavement. 3. Carers whose practical support needs were not met while caring for their loved one had lower CIS-R scores. 4. Carers whose patients had undergone a better death had higher CIS-R scores at six months after death (although this association was only marginally significant). BPQ as the Main Outcome. At T3, BPQ scores ranged from 7 to 48. The mean (SD) score was 26.2 (9.9). Multivariable associations between BPQ scores and carer, patient, caregiving, family, and death variables revealed the following relations (Table 4): 1. Carer mental health at baseline was a significant predictor of the BPQ outcome six months after the death of their loved one. Carers who had experienced more symptoms of CMD while caring for their loved one had higher BPQ scores. 2. Carers who had reported a higher discrepancy between their actual and ideal levels of practical support while caring for their loved one had lower BPQ scores. Discussion This study sought to describe trajectories from caregiving to bereavement by following informal carers of patients receiving palliative care from shortly after the patient was referred to a specialist service to six months after the death of their loved one. Our results accorded with previous work showing a significant association between caregiver burden and poorer mental health at baseline, with carers who perceived their caregiving experience as burdensome suffering from more symptoms of CMD. The relationship between caregiver burden and symptoms of CMD was not maintained in the bereavement phase. However, our results indicated a significant association between carers symptoms of distress during the caregiving experience and their mental health at three and six months after death. Carers who suffered from more symptoms of CMD while caring for their loved ones were likely to do so during the first six months of their bereavement, suggesting that the complicated grief model is a better description of our data than the relief model. During the caregiving phase, we found that active coping was significantly associated with more symptoms of CMD, whereas substance abuse was significantly associated with fewer symptoms of CMD. Similarly, higher levels of active coping were associated with more symptoms of CMD at six months after death. One way of explaining this surprising association is that high levels of active coping might be an indication of a demanding and stressful experience during which the caregiver feels the need to take immediate action to make the situation better. Although this mechanism might be effective in dealing with the practical elements of the caregiving situation, it is not necessarily beneficial when coping with more uncontrollable elements. It can, therefore, be argued that problem-focused strategies can be more effective in dealing with elements of the caring situation that are within the carer s control, whereas emotion-focused coping strategies are probably more effective when dealing with more uncontrollable elements of the situation. We found no evidence for the effect of different types of belief systems on bereavement outcomes. Our findings, however, confirmed a positive relation between the strength of religious/spiritual beliefs and caregiver mental health at baseline. Carers who professed strong religious/spiritual beliefs reported fewer symptoms of CMD while caring for their loved one. We found no association between symptoms of CMD and the quality of family relations during the caregiving phase. In the same way, we report no evidence for the protective role of good family relations on bereavement outcomes. This finding is not consistent with other prospective studies that have highlighted the protective role of good family relations on caregiving and bereavement outcomes. One way of explaining this difference in findings is the nature of the present sample. Recruitment issues might have affected the internal validity of our findings. A total of 76.5% of the participants in this study described their families as moderately balanced or balanced. One might, therefore, argue that the nurse specialists might have been more likely to introduce our research to seemingly balanced couples and families that were more

10 Vol. 40 No. 6 December 2010 Caregiving and Bereavement 853 approachable. As a result, we might have failed to uncover a relation between dysfunctional family relations and symptoms of CMD in both caregiving and bereavement as our sample might be mostly represented by welladjusted and connected families. The protective role of social support while caring for an ill relative has been repeatedly highlighted in the caregiving literature. By measuring both actual and ideal levels of emotional and practical support, we were able to examine carers unmet needs for support at the same time. No significant association was found between actual and ideal levels of practical or emotional support and symptoms of CMD during the caregiving phase. Our results suggest a significant negative relation between differing levels of actual and ideal levels of practical support and bereavement outcomes at six months after death. Carers who had reported higher discrepancies between their actual and ideal levels of practical support, while caring for their relative, experienced better bereavement outcomes. One could argue that a high discrepancy between actual and ideal levels of practical support would indicate a demanding caregiving experience. The termination of the caregiving experience might have signaled the end of a demanding situation, and carers might have, therefore, felt relieved and experienced better bereavement outcomes in turn. We also hypothesized that a good death in the patient would be protective against poorer bereavement outcomes in the carer. Our findings did not confirm this hypothesis. On the contrary, we found a positive association between a good death in the patient and more symptoms of CMD at six months after death. We had limited statistical power to examine this hypothesis. Christakis and Iwashyna, 31 who reported a positive association between a good death in the patient and carer bereavement outcomes, studied a representative national cohort of patients using a large, well-powered sample and long followups. Our findings were based on a small number of observations and could, therefore, be imprecise. Our findings provide evidence for the depletion model advocating that carers mental health at T1 is predictive of their mental health at T2 and T3. In other words, carers who experienced more symptoms of CMD while caring for their relative were more likely to do the same during their bereavement. At the same time, we found some evidence for the relief model. Carers who had reported higher discrepancies between their actual and ideal levels of practical support while caring for their relative had better bereavement outcomes. One could argue that these carers might have undergone a much more demanding and stressful experience, and its termination might have relieved them. Strengths and Limitations The longitudinal design of this study overcame the methodological limitations of crosssectional studies and established links between predeath caregiving variables and bereavement outcomes. A particular strength of this study was the inclusion of a wide range of variables known to be important in determining levels of CMD in both caregiving and bereavement. Follow-up rates during bereavement were satisfactory, with a participation rate of 59% at T2 and 64% at T3. This acceptance rate is comparable to that of other studies in bereavement. 22 The most serious limitation of this study is the small sample size, which, given attrition and a proportion of carers whose relatives had not died over the study period, meant that we had limited statistical power to detect key associations. As discussed in the results section, there was no information about nonparticipation for the people who were not referred in the study. Such a limitation in the recruitment process can affect both the external and internal validity of our findings. Future studies in the field of palliative care should address these sampling issues and be cautious of gate-keeping. Implications for Practice During the caregiving experience, 32% of carers were cases according to their CIS-R scores. This figure demonstrates that there is substantial psychological morbidity in the caregiving population. The figure fell after bereavement. Our findings suggest that carers who suffer from significantly more symptoms of CMD while caring for their relative are likely to have worse bereavement outcomes (defined as more symptoms of CMD and intense bereavement phenomena). This

11 854 Kapari et al. Vol. 40 No. 6 December 2010 relation suggests that services to identify carers at a high risk might be able to alter subsequent grief reactions. Implications for Research This study further demonstrates the feasibility of prospective studies in palliative care research. Although health professionals may be reluctant to refer palliative care patients and their carers for research studies (given the patient s poor physical and often emotional condition and the carer s added stresses), patients and carers themselves are often willing to take part. 37 Additionally, our follow-up rates validate that a cohort study in the field of palliative care was well received by the participants. Even in this sensitive area, it is possible to recruit carers and patients to participate in research studies and many expressed appreciation that they had been included in the research. Conclusion The U.K. National Institute of Health and Clinical Excellence 6 guidelines acknowledge the vital role informal carers play in caregiving contexts as they allow terminally ill patients to remain in the community. Those guidelines now emphasize the importance of recognizing their needs and treating them alongside those of patients. In accordance with other caregiving studies, our findings confirmed that there is significant psychiatric morbidity within the caregiving population. Our findings revealed that carers who perceived their caregiving experience as considerably burdening had more symptoms of CMD while caring for their loved one. Carer mental health baseline was predictive of their mental health at both three and six months after death. An important limitation of our findings is the lack of power in our sample that might render our results less valid. A larger prospective study is needed to replicate and extend these findings. Disclosures and Acknowledgments The authors thank the patients, carers, and staff of St. Christopher s and Trinity Hospices for their commitment to this research and generosity in participating. Prof. Hotopf is supported by the National Institute for Health Research Biomedical Research Center for Mental Health at the South London and Maudsley National Health Service Foundation Trust and Institute of Psychiatry, Kings College London. The authors declare no conflicts of interest. References 1. Schulz R, Mendelsohn AB, Haley WE, et al. Endof-life care and the effects of bereavement on family caregivers of persons with dementia. N Engl J Med 2003;349:1936e Schulz R, Visintainer P, Williamson GM. Psychiatric and physical morbidity effects of caregiving. J Gerontol 1990;45:181e Schulz R, O Brien AT, Bookwala J, Fleissner K. Psychiatric and physical morbidity effects of dementia caregiving: prevalence, correlates, and causes. Gerontologist 1995;35:771e Schulz R, Beach SR. Caregiving as a risk factor for mortality: the Caregiver Health Effects Study. JAMA 1999;282:2215e Department of Health. Caring about carers: a national strategy for carers. London, UK: Department of Health, National Institute for Clinical Excellence. Improving supportive and palliative care for adults with cancerdthe manual. London, UK: National Institute for Clinical Excellence, Aranda SK, Hayman-White K. Home caregivers of the person with advanced cancer: an Australian perspective. Cancer Nurs 2001;24:300e Baumgarten M, Battista RN, Infante-Rivard C, et al. The psychological and physical health of family members caring for an elderly person with dementia. J Clin Epidemiol 1992;45:61e Carter PA, Chang B. Sleep and depression in cancer caregivers. Cancer Nurs 2000;23:410e Coope B, Ballard C, Saad K, et al. The prevalence of depression in the carers of dementia sufferers. Int J Geriatr Psychiatry 1995;10:237e Williamson GM, Schulz R. Coping with specific stressors in Alzheimer s disease caregiving. Gerontologist 1993;33:747e Bernard LL, Guarnaccia CA. Two models of caregiver strain and bereavement adjustment: a comparison of husband and daughter caregivers of breast cancer hospice patients. Gerontologist 2003; 43:808e Koop PM, Strang VP. The bereavement experience following home-based family caregiving for persons with advanced cancer. Clin Nurs Res 2003; 12:127e144.

12 Vol. 40 No. 6 December 2010 Caregiving and Bereavement Brazil K, Bedard M, Willison K. Correlates of health status for family caregivers in bereavement. J Palliat Med 2002;5:849e Almberg BE, Grafstrom M, Winblad B. Caregivers of relatives with dementia: experiences encompassing social support and bereavement. Aging Ment Health 2000;4:82e Wyatt GK, Friedman L, Given CW, Given BA. A profile of bereaved caregivers following provision of terminal care. J Palliat Care 1999;15:13e Schulz R, Beach SR, Lind B, et al. Involvement in caregiving and adjustment to death of a spouse. JAMA 2001;285:3123e Boerner K, Sculz R, Horowitz A. Positive aspects of caregiving and adaptation to bereavement. Psychol Aging 2004;19:668e Ferrario SR, Cardillo V, Vicario F, Balzarini E, Zotti AM. Advanced cancer at home: caregiving and bereavement. Palliat Med 2004;18:129e Kelly B, Edwards P, Synott R, et al. Predictors of bereavement outcome for family carers of cancer patients. Psychooncology 1999;8:237e Kurtz ME, Kurtz JC, Given CW, Given B. Predictors of postbereavement depressive symptomatology among family caregivers of cancer patients. Support Care Cancer 1997;5:53e Li LW. From caregiving to bereavement: trajectories of depressive symptoms among wife and daughter caregivers. J Gerontol B Psychol Sci Soc Sci 2005;60:190e McHorney CA, Mor V. Predictors of bereavement depression and its health services consequences. Med Care 1988;26:882e Rosengard C, Folkman S. Suicidal ideation, bereavement, HIV serostatus and psychosocial variables in partners of men with AIDS. AIDS Care 1997;9:373e Aneshensel CS, Botticello AL, Yamamoto- Mitani N. When caregiving ends: the course of depressive symptoms after bereavement. J Health Soc Behav 2004;45:422e Bass DM, Bowman K. The transition from caregiving to bereavement: the relationship of care-related strain and adjustment to death. Gerontologist 1990;30:35e Beery LC, Prigerson H, Bierhals AJ, et al. Traumatic grief, depression and caregiving in elderly spouses of the terminally ill. Omega 1997;35: 261e Kristjanson LJ, Sloan JA, Dudgeon D, Adaskin E. Family members perceptions of palliative cancer care: predictors of family functioning and family members health. J Palliat Care 1996; 12:10e Moskowitz JT, Folkman S, Collette L, Vittinghoff E. Coping and mood during AIDSrelated caregiving and bereavement. Ann Behav Med 2008;18:49e Folkman S, Chesney M, Collette L, Boccellari A, Cooke M. Post-bereavement depressive mood and its pre-bereavement predictors in HIVþ and HIV gay men. J Pers Soc Psychol 1996;70:336e Christakis NA, Iwashyna TJ. The health impact of health care on families: a matched cohort study of hospice use by decedents and mortality outcomes in surviving, widowed spouses. Soc Sci Med 2003;57: 465e Houts PS, Lipton A, Harvey HA, Simmonds MA, Bartholomew MJ. Predictors of grief among spouses of deceased cancer patients. J Psychosoc Oncol 1989;7:113e Lauer ME, Mulhern RK, Schell MJ, Camitta BM. Long-term follow-up of parental adjustment following a child s death at home or hospital. Cancer 1989;63:988e Addington-Hall J, Karlsen S. Do home deaths increase distress in bereavement? Palliat Med 2000;14:161e Hotopf M, Chidgey J, Addington-Hall J, Ly KL. Depression in advanced disease: a systematic review. Part 1. Prevalence and findings. Palliat Med 2002; 16:81e Lewis G, Pelosi AJ, Araya R, Dunn G. Measuring psychiatric disorder in the community: a standardized assessment for lay interviewers. Psychol Med 1992;22:465e Henderson M, Addington-Hall J, Hotopf M. The willingness of palliative care patients to participate in research. J Pain Symptom Manage 2005;29: 116e Cohen SR, Mount BM, Strobel MG, Bui F. The McGill Quality of Life Questionnaire: a measure of quality of life appropriate for people with advanced disease. A preliminary study of validity and acceptability. Palliat Med 1995;9:207e Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983; 67:361e Zubrod CG, Schneiderman M, Frei E, et al. Appraisal of methods for the study of chemotherapy of cancer in man: comparative therapeutic trial of nitrogen mustard and triethylene thiophosphoramide. J Chronic Dis 1960;11:7e Zarit SH, Reever KE, Bach-Peterson J. Relatives of the impaired elderly: correlates of feelings of burden. Gerontologist 1980;20:649e Mundt JC, Marks IM, Shear MK, Greist JM. The work and social adjustment scale: a simple measure of impairment in functioning. Br J Psychiatry 2002; 180:461e Power MJ, Champion LA, Aris SJ. The development of a measure of social support: the Significant

13 856 Kapari et al. Vol. 40 No. 6 December 2010 Others Scale (SOS). Br J Clin Psychol 1988;27: 349e Olson D. Circumplex Model VII: validations studies and FACES-III. Fam Process 1986;25:337e Carver CS. You want to measure coping but your protocol s too long: consider the brief COPE. Int J Behav Med 1997;4:92e King M, Speck P, Thomas A. The Royal Free Interview for religious and spiritual beliefs: a development and standardization. Psychol Med 1995;25:1125e Wallston KA, Burger C, Smith RA, Baugher RJ. Comparing the quality of death for hospice and non-hospice cancer patients. Med Care 1988;26: 177e Byrne GJ, Raphael B. A longitudinal study of bereavement phenomena in recently widowed elderly men. Psychol Med 1994;24:411e421.