Research Recherche Family caregiver burden: results of a longitudinal study of breast cancer patients and their principal caregivers Eva Grunfeld, Doug Coyle, Timothy Whelan, Jennifer Clinch, Leonard Reyno, Craig C. Earle, Andrew Willan, Raymond Viola, Marjorie Coristine, Teresa Janz, Robert Glossop See related article page 1811 DOI:10.1503/cmaj.1031205 Abstract Background: The vital role played by family caregivers in supporting dying cancer patients is well recognized, but the burden and economic impact on caregivers is poorly understood. We prospectively examined the psychosocial, occupational and economic impact of caring for a person with a terminal illness. Methods: We studied 89 caregivers of women with advanced breast cancer receiving care at either the Ottawa or Hamilton regional cancer centres in Ontario. Patients were followed until their death or study completion at 3 years. Patients identified a principal caregiver to participate in the study. The Karnofsky Performance Status (KPS) index, the Medical Outcomes Study 36-item Short Form (SF-36), the Hospital Anxiety and Depression Scale, the Zarit Burden Inventory, FAMCARE and the Medical Outcomes Study Social Support Survey were administered during follow-up. Economic data were collected by means of a questionnaire administered by an interviewer. Assessments were conducted every 3 months during the palliative period (KPS score > 50) and every 2 weeks during the terminal period (KPS score 50). Results: Over half of the caregivers were male (55%) and the patient's spouse or partner (52%), with a mean age of 53 years. At the start of the palliative period, the caregivers' mean physical functioning score was better than the patients' (51.3 v. 35.1, 95% confidence interval [CI] 13.3-20.0); there were similar mean mental functioning scores (46.6 and 47.1 respectively); similar proportions were depressed (11% and 12%); and significantly more caregivers than patients were anxious (35% v. 19%, p = 0.009). More caregivers were depressed (30% v. 9%, p = 0.02) and had a higher level of perceived burden (26.2 v. 19.4, p = 0.02) at the start of the terminal period than at the start of the palliative period. Burden was the most important predictor of both anxiety and depression. Of employed caregivers, 69% reported some form of adverse impact on work. In the terminal period 77% reported missing work because of caregiving responsibilities. Prescription drugs were the most important component of financial burden. Interpretation: Caregivers' depression and perceived burden increase as patients' functional status declines. Strategies are needed to help reduce the psychosocial, occupational and economic burden associated with caregiving. CMAJ 2004;170(12):1795-801 I n Canada about 65 000 people die each year as a result of cancer.1 The vital role played by family caregivers in supporting dying cancer patients is well recognized.2-7 With health care services being restructured and more cancer patients dying outside of hospital,8 the reliance on family caregivers to support patients with terminal illness at home is growing.9,10 However, the concern is that these caregivers may themselves be elderly or ill, or may be adult children with many other family and work responsibilities.11 To inform the growing public policy debate about family caregivers,2,9 we need a better understanding of who they are and how caregiving affects them. To this end we followed family caregivers of patients with advanced breast cancer and caregiver-patient pairs prospectively from the beginning of the patient's life-limiting illness through to terminal illness. We measured the psychosocial, occupational and economic impact of caregiving over the course of the patient's life-limiting illness, and compared the psychosocial impact of life-limiting illness on specific patient and caregiver pairs. Methods We selected patients with hormone refractory advanced breast cancer who were receiving care at either the Ottawa or Hamilton regional cancer centres and their principal family caregivers. The regional cancer centres are principal providers of tertiary cancer services in their respective regions. We enrolled 130 prospective eligible patients within 3 months of diagnosis of visceral metastases or hormone refractory uncontrollable local recurrence or bone metastases. Prospective eligible patients were identified by 2 methods: clinicians informed the research nurse when a patient meeting the eligibility criteria was seen in an outpatient clinic, and all outpatient clinic dictation notes were reviewed after each breast cancer clinic to identify eligible patients who may have been missed by the first method. Assessments were initially made every 3 months, and increased to every 2 weeks when the patients' functional status deteriorated (Karnofsky Performance Status [KPS] score 50).12-16 Data are reported for the palliative period (KPS score > 50) for all patients, and the terminal period for patients whose KPS score fell to 50 or lower. Participating patients were asked to identify a principal family caregiver, defined as a family member or friend who would be most responsible for ongoing caregiving. Of the 130 patients enrolled in the study, 89 identified a caregiver who consented to participate in the study. CMAJ JUNE 8, 2004; 170 (12) 2004 Canadian Medical Association or its licensors 1795 Grunfeld Consenting patients and their caregivers were followed prospectively until the patient's death or study completion, whichever came first. The study took place from October 1997 to October 2000. In this article we report the results for the 89 family caregivers who participated in the study. Ethical approval to conduct the study was obtained from the Ottawa Hospital and the McMaster University research ethics boards. Caregiver and patient sociodemographic data were collected at the beginning of the study by means of a self-administered questionnaire. As summarized in the Appendix 1, standardized validated instruments were used to measure all domains. Caregiver general health status was measured by means of the Medical Outcomes Study 36-item Short Form (SF-36).17,18 Psychological distress was measured by means of the Hospital Anxiety and Depression Scale (HADS).19-21 Family caregiver burden was measured by the Zarit Burden Inventory.22,23 Satisfaction with care was measured using the FAMCARE instrument, developed specifically to measure family satisfaction with palliative care.24,25 The Medical Outcomes Study (MOS) Social Support Survey was used to measure patient's social support.26 When the caregiver and patient pairs were compared, or caregivers were compared at 2 different time points (beginning of palliative period compared with end of palliative period; beginning of palliative period compared with beginning of terminal period), a matched pair t test was used to compare means on quantitative variables, and McNemar's test was used to compare proportions on categorical variables. Fig. 1 summarizes the number of patients, caregivers and patient-caregiver pairs at each time point. Enrolled in study 130 patients 89 caregivers 59 patients continued in the palliative period Yes Entered terminal period 51 patients 36 caregivers (36 pairs, 15 solo patients) 20 patients died precipitously Completed HADS 27 patients 24 caregivers (15 pairs, 12 solo patients, 9 solo caregivers) Fig. 1: Flow of patients and caregivers through study. HADS = Hospital Anxiety and Depression Scale, KPS = Karnofsky Performance Status. 1796 Table 1: Characteristics of caregivers at the start of the palliative period (n = 89) Characteristic Age; mean (SD, range), yr Male Married Education College or university No postsecondary Employment status Employed, full- or part-time Retired Stopped working because of caregiving Other Missed work because of caregiving Occupation Professional Managerial Other Involved in religious activities Participated in support group Had 1 dependent at home Relationship to patient Spouse or partner Sister or daughter Friend Other Proximity to patient Lives in same house Lives nearby Completed HADS 125 patients 85 caregivers (84 pairs, 41 solo patients, 1 solo caregiver) KPS score 50 All caregivers with 2 or more assessments in the palliative period (n = 67) had their first and last burden (Zarit Burden Inventory) and satisfaction with care (FAMCARE) scores compared using a paired t test. Since the time interval between the 2 visits varied considerably (61-1023 days), which could have influenced the extent of change, a correlation coefficient was calculated between the change JAMC 8 JUIN 2004; 170 (12) Lives > 1 h drive away Health status, mean score (SD) Social functioning Physical functioning Role functioning, physical Role functioning, emotional General health Pain Vitality Mental health Satisfaction with care, mean score (SD) Caregiver burden, mean score (SD) Psychological distress, mean score (SD) Anxiety Depression No. (%) of caregivers* 52.8 (15.2, 19-82) 49 (55) 66 (74) 57 (64) 27 (30) 50 (56) 24 (27) 2 (2) 15 (17) 40 (69) 35 (39) 18 (20) 36 (40) 36 (40) 4 (5) 34 (38) 46 (52) 26 (29) 6 (7) 11 (12) 59 (66) 10 (12) 20 (23) 87.1 (18.5) 87.5 (17.8) 77.1 (37.6) 70.0 (38.6) 70.3 (21.4) 75.9 (18.5) 57.0 (21.0) 68.5 (19.4) 79.3 (14.9) 18.3 (11.6) 8.3 (4.3) 5.1 (4.0) Note: SD = standard deviation. *Unless stated otherwise. Percentages may not equal 100 because of rounding or \"other\" category not reported. n = 58 (employed or resigned/retired after becoming caregiver). See Appendix 1 for details on scores. Family caregiver burden scores of each scale and the visit intervals: neither correlation was significant (r = 0.06, p = 0.63, and r = 0.05, p = 0.67, respectively). We used stepwise multiple regression analyses to construct 2 models for the prediction of caregiver anxiety and depression as measured by the HADS subscales at the start of the palliative period. Candidate predictor variables for caregiver anxiety were patient's KPS score, HADS anxiety and depression scores, SF-36 physical and mental component scores, caregiver's age, Zarit Burden Inventory score and MOS emotional support subscale. For caregiver depression, all the same variables were included, except the caregiver's SF-36 physical component score was excluded and the caregiver's participation in religious or spiritual activities was included. Only variables showing a univariate association with p less than or equal to 0.15 were used in the regression analysis. Direct financial burden to the patient or caregiver consisting of both direct treatment and nontreatment costs was obtained through interviewer-administered questionnaires at each assessment. We assumed that patients with extended health coverage (88/130 [68%]) were not responsible for expenses such as prescription drugs or private hospital rooms. The cost of individual resource items were either obtained directly from the participant or through standard resource valuation measures.27 Financial burden is presented as both a weekly estimate for the palliative and terminal periods and as a cumulative burden over the entire study period. Results Caregivers were, on average, younger than the patients (52.8 years [range 19-82] v. 56.2 years [range 26-83]). Just over half were male (55%) and the patient's spouse or partner (52%) (Table 1). The rest were female relatives (29%) or friends (7%). There were no significant differences between the patients with a caregiver enrolled in the study and the 41 patients who did not have a caregiver enrolled in the study with respect to baseline characteristics (age, education, income, marital status or living alone) (data not shown). A comparison of patient and caregiver pairs at the start of the palliative and terminal periods is shown in Table 2. For the palliative period, the mean caregiver's physical functioning score was better than the mean patient's physical functioning score (51.3 v. 35.1; 95% CI 13.3-20.0, p < 0.001). Similar proportions of caregivers (11%) and patients (12%) were depressed. Significantly more caregivers than patients were anxious, as reflected by both mean scores (8.4 v. 7.3, p = 0.03) and proportion of cases (35% v. 19%, p = 0.009). The findings were similar for the terminal period. Significantly more caregivers than patients were anxious, as reflected by both mean scores (9.9 v. 6.8, p = 0.01) and proportion of cases (40% v. 27%, p = 0.4), although cases did not reach significance. Psychological impact At the start of the palliative period, the mean score of caregiver burden was 18.3 (Table 1) and did not change significantly during the entire palliative period (p = 0.45). Satisfaction with care was high and increased significantly during the palliative period, from a mean score of 79.3 (standard deviation [SD] 14.9) to 83.4 (SD 11.2) (p = 0.04). The impact over time on the caregivers of 24 patients who reached the terminal period is presented in Table 3. Table 2: Comparison of patient and caregiver psychological distress, paired samples Characteristic Start of palliative period (n = 84)* Physical summary score, mean (SD) Mental summary score, mean (SD) Anxiety, Mean score (SD) No. of cases (%) Depression, Mean score (SD) No. (%) of cases Start of terminal period (n = 15)* Anxiety Mean score (SD) No. (%) of cases Depression Mean score (SD) No. (%) of cases Caregivers 51.3 (9.7) 46.6 (12.0) p value Patients Difference (95% CI) 35.1 (12.4) 47.1 (11.2) -16.2 (-20.0 to 13.3) 0.5 (-2.5 to 3.1) < 0.001 0.8 8.4 (4.3) 7.3 (4.1) -1.1 (-2.2 to 0.1) 0.03 29 (35) 16 (19) - 0.009 5.1 (3.9) 9 (11) 5.2 (4.2) 10 (12) 0.12 (-0.94 to 1.18) - 0.82 0.79 9.9 (3.0) 6 (40) 6.8 (4.4) 4 (27) -3.05 (-5.3 to -0.8) - 0.01 0.4 7.7 (4.3) 4 (27) 8.1 (3.4) 4 (27) 0.4 (-1.9 to 2.8) - 0.7 1.0 Note: CI = confidence interval. *Complete data were available for 84 patient-caregiver pairs in the palliative period and 15 in the terminal period. Medical Outcomes Study 36-item Short Form (SF-36) physical and mental component summary scores. See Appendix 1 for details on scores. Significant correlation between patient and caregiver (p 0.009). CMAJ JUNE 8, 2004; 170 (12) 1797 Grunfeld Similar proportions of these caregivers were anxious at the start of the terminal period as at the start of the palliative period (39% v. 35%, p = 0.18). More of these caregivers were depressed (30% v. 9%, p = 0.02) and experienced a higher level of burden (mean score 26.2 v. 19.4, p = 0.02) at the start of the terminal period than at the start of the palliative period. Predictors of caregiver anxiety and depression at the start of the palliative period are shown in Table 4. Caregiver burden was the most important predictor of both anxiety and depression. Occupational impact At the start of the palliative period, about half of the caregivers (56%) were in full- or part-time employment and about one quarter (27%) were retired. In the paired comparison between employed caregivers who moved from the palliative period to the terminal period (n = 17), there was no change in employment status: they did not quit, lose or change their jobs (data not shown). However, more caregivers missed work in the terminal period than in the palliative period (77% v. 53%, p = 0.04). Caregivers also reported an increased inability to work regular hours and increased hours of work lost in the terminal period because of care provided (Table 3). Economic impact Prescription drugs were, on average, the most significant component of financial burden (Table 5). The average financial burden born by families (either the caregiver or pa- tient) during the patient's illness was higher for those without extended health insurance (n = 42, $8292 [SD $7235]) than for those with extended health insurance coverage (n = 88, $5765 [SD $8926]) (p = 0.04). Interpretation In our longitudinal study, we found that caregivers experienced substantial psychological morbidity (anxiety and depression) at the onset of the patient's palliative illness and a substantial increase in caregiver burden and depression when the patient reached a terminal stage of the illness. In addition to psychological morbidity, caregivers bore both economic and occupational burdens. Our finding that caregiver's psychological morbidity was equal to or greater than the patient's is consistent with other research.28-35 The finding that a large proportion of caregivers suffered from both anxiety and depression suggests that there are unmet needs that are measurable at an early stage of the patient's life-limiting illness, and points to the need for early psychological assessment and potential intervention.36 Our research with a subset of these same caregivers during their bereavement suggests that they are so focused on their caregiving role that their own medical needs go unattended,37 as has been found in other studies.38 Our findings indicate that, although patient's physical and emotional factors predict caregiver distress, perceived burden is the strongest predictor of caregiver anxiety and depression. In previous research feelings of burden have been found to be associated with anxiety in daughters of patients with cancer39 and caregivers of patients with dementia.35 Table 3: Change in caregiver burden over time (n = 24)* Start of palliative period Start of terminal period Difference (95% CI) 10.2 ( 3.2) 9.0 (39.1) -1.3 (-3.2 to 0.5) No. (%) of cases Depression Mean score (SD) No. (%) of cases Caregiver burden, mean score (SD) Satisfaction with care, mean score (SD) 8.8 (4.1) 8 (35) 0.15 0.18 5.2 (3.8) 2 (9) 19.4 (12.9) 82.1 (9.7) 8.1 ( 4.9) 7 (30) 26.2 (12.4) 80.6 (10.8) -2.9 (-4.6 to -1.2) -6.8 (-12.7 to -0.9) 1.5 (-3.1 to 6.1) 0.002 0.02 0.02 0.7 Impact on work (n = 17) Missed work, no. (%) Left work for appointments, no. (%) Received interrupting phone calls, no. (%) Used holidays or special leave, no. (%) Decreased work hours, no. (%) Hours of work lost, mean (SD) 9 (53) 8 (47) 4 (24) 6 (35) 4 (24) 6.1 (12.4) 13 (76) 9 (53) 7 (41) 7 (41) 9 (53) 11.4 (15.9) - - - - - -5.4 (-10.0 to -0.7) 0.04 0.65 0.08 0.56 0.02 0.03 Characteristic Anxiety Mean score (SD) *Results are for the 24 caregivers of patients who entered the terminal period. See Appendix 1 for details on scores. Cases of anxiety and depression are defined as scores 11. Of the 24 caregivers, 17 were in full- or part-time paid employment. In the preceding 4 weeks for palliative period; preceding 2 weeks for terminal period. 1798 JAMC 8 JUIN 2004; 170 (12) p value Family caregiver burden Our finding that a majority of caregivers experienced an adverse impact on their employment, particularly during the terminal period, is consistent with the results of the 1996 General Social Survey40 and other studies of family caregivers of cancer patients.14,41-44 In our study, 5% of caregivers had quit their job or declined advancement, Table 4: Predictors of caregiver's psychological distress at the start of the palliative period Predictor For caregiver anxiety Caregiver burden Patient's anxiety Patient's physical functioning For caregiver depression Caregiver burden Patient's emotional support Raw regression () coefficient* p value 0.16 0.23 -0.07 < 0.001 0.03 0.03 0.2 -1.06 < 0.001 0.03 * coefficients indicate the extent to which anxiety or depression increase (positive coefficients) or decrease (negative coefficients) by an increase of 1 unit in the predictor (e.g., if the Zarit burden score increases by 1, the caregiver's anxiety will increase by 0.16, all other predictors remaining constant). As measured by the SF-36 Physical Component Score. Patient's worsening physical functioning is a predictor for caregiver anxiety. and a large proportion lost work hours or used special leave or holidays to fulfill their caregiving responsibilities. The Senate report on end-of-life care2 recommends income security and job protection for family members who care for people who are dying. Our finding underscores the importance of implementing this recommendation. Similarly, our finding of significant costs of prescription drugs particularly for those without extended health insurance supports the need for a policy that addresses the growing impact of drug costs on patients and their families, as recommended by the National Forum on Health45 and the Romanow report.46 Our study has limitations. The complexity and challenges of conducting palliative care research are well recognized.47,48 Although we began with 89 caregivers at the outset of the palliative period, only 36 had patients who entered the terminal period, either because the patient died before we were able to document (KPS score 50) or because the patient's functional status remained high (KPS score > 50) throughout the study. This might have limited our power to detect real but small changes over time. A further limitation of the study is that it involved an urban, relatively well-educated group of caregivers. Few patients in this study lived alone, and even those who did Table 5: Economic burden per patient, by period Weekly burden, mean (SD), $ Expense Prescription drugs IQR Home help IQR Food IQR Travel IQR Furniture and modifications IQR Alternative medicines IQR Alternative therapies IQR Over-the-counter medicines IQR Accommodation IQR Private nurse IQR Child care IQR Other expenses IQR Total Palliative period Terminal period n = 130 n = 51 Cumulative burden, mean (SD), $ n = 130 25.5 (60.1) 0-18 11.1 (46.4) 0-0 13.0 (18.5) 0-18 9.6 (8.6) 3-14 7.8 (41.4) 0-0 11.0 (40.0) 0-5 11.5 (41.9) 0-3 1.6 (2.2) 0-2 1.4 (8.1) 0-0 0.9 (6.3) 0-0 0.7 (5.2) 0-0 11.5 (41.8) 0-4 25.0 (45.2) 0-38 5.3 (26.8) 0-0 16.5 (25.3) 0-26 14.9 (18.8) 1-18 25.7 (92.1) 0-0 2.8 (13.6) 0-0 2.8 (11.8) 0-0 2.1 (4.0) 0-3 29.1 (157.3) 0-0 1.0 (7.5) 0-0 1.9 (9.7) 0-0 14.3 (60.9) 0-0 1402.4 (3298.3) 0-1227 946.2 (4931.5) 0-0 841.0 (1289.9) 36-1085 661.7 (703.9) 176-1023 611.1 (2795.4) 0-83 536.1 (1345.2) 0-322 531.6 (1468.4) 0-190 116.0 (166.5) 16-135 140.1 (815.6) 0-0 64.9 (400.8) 0-0 46.0 (365.0) 0-0 670.9 (2097.1) 0-389 105.8 (138.56) 27-114 141.6 (215.0) 12-134 6581.6 (8472.5) 1879-7317 Note: IQR = interquartile range CMAJ JUNE 8, 2004; 170 (12) 1799 Grunfeld live alone had a caregiver. In fact we found no differences on baseline variables between patients who had a caregiver enrolled in the study and those who did not. Thus, it is likely that all of the patients in the cohort had a caregiver involved in their care, despite the enrolment of only 68% of the caregivers. Although there were no differences between patients with and those without an enrolled caregiver, we are unable to determine whether there were any systematic differences between caregivers who consented to participate in this study and those who did not. We have documented substantial psychological, occupational and economic burdens associated with caregiving at the outset of life-limiting illness that increase as patients' functional status declines and death approaches. Caregiver burden was the most important predictor of both caregiver anxiety and depression. We have reported the positive aspects of caregiving expressed by a subset of these caregivers.37 In that report we stated that the caregiver's commitment to the patient had primacy over his or her own occupational and health concerns. The changes to Canada's health care system fewer acute care and palliative care in-patient beds8 will result in an increased dependence on family caregivers to support dying cancer patients. Strategies to support caregivers, as recommended by the Senate report on end-of-life care,2,49 would reduce caregiver burden, potentially prevent psychosocial distress and allow caregivers to focus on their most immediate concern, caring for the dying patient.37 This article has been peer reviewed. From the Ottawa Regional Cancer Centre and Division of Medical Oncology, University of Ottawa (Grunfeld), the Clinical Epidemiology Program, Ottawa Health Research Institute, Ottawa Hospital (Grunfeld, Coyle, Clinch, Viola, Coristine, Janz), the Institute of Palliative Care, Elizabeth Bruyre Health Centre (Viola) and the Vanier Institute of the Family (Glossop), Ottawa, Ont.; the Supportive Cancer Care Research Unit (Grunfeld, Whelan, Reyno, Willan), and the Hamilton Regional Cancer Centre and Division of Radiation Oncology (Whelan), McMaster University, and the Center for the Evaluation of Medicine, St. Joseph's Hospital (Willan), Hamilton, Ont.; the Division of Medical Oncology, Dalhousie University, Halifax, NS (Grunfeld, Reyno); and the Center for Outcomes and Policy Research and Department of Adult Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, Mass. (Earle) Contributors: Eva Grunfeld was principally responsible for the development of the protocol, all aspects of ongoing study management, analysis of the data and writing of the manuscript. Doug Coyle, Timothy Whelan, Leonard Reyno, Craig Earle and Raymond Viola contributed to the development of the protocol, interpretation of the analysis and review of the manuscript. Marjorie Coristine contributed to the study design and ongoing study management. Doug Coyle was responsible for all aspects of the economic analysis. Andrew Willan was responsible for construction of the database. Andrew Willan and Jennifer Clinch oversaw the analysis. Jennifer Clinch and Teresa Janz conducted the analysis. Robert Glossop contributed to the development of the protocol, interpretation of the results and review of the manuscript. All authors reviewed the manuscript and approved the final version. Competing Interests: None declared. Acknowledgements: We would like to acknowledge with gratitude the women and family caregivers who participated in this study. Their dedication to the study and willingness to be interviewed under very stressful circumstances, for the sake of helping other women with breast cancer and those who care for them, were inspiring. The study was funded by the Canadian Breast Cancer Research Alliance (grant 008026) with support from the Ontario Ministry of Health and Long-term Care (grant 04171). 1800 JAMC 8 JUIN 2004; 170 (12) References 1. National Cancer Institute of Canada. Canadian Cancer Statistics 2002. Toronto: National Cancer Institute of Canada; 2002. p. 1-96. 2. Subcommittee of the Standing Senate Committee on Social Affairs, Science and Technology. Quality end-of-life care: the right of every Canadian. Ottawa: The Senate of Canada; 2000. Available: www.parl.gc.ca/36/2/parlbus/commbus /senate/Com-e/upda-e/rep-e/repfinjun00-e.htm (accessed 2004 Feb 17). 3. McWhinney IR, Bass MJ, Orr V. Factors associated with location of death (home or hospital) of patients referred to a palliative care team. CMAJ 1995; 152(3):361-7. 4. Townsend J, Frank AO, Fermont D, Dyer S, Karran O, Walgrove A, et al. Terminal cancer care and patients' preference for place of death: a prospective study. BMJ 1990;301(6749):415-7. 5. Howarth G, Willison KB. Preventing crises in palliative care in the home. Role of family physicians and nurses. Can Fam Physician 1995;41:439-45. 6. Chappell NL. Home care research: What does it tell us? Gerontologist 1994; 34:116-20. 7. Zimmer JG, Groth-Juncker A, McCusker J. A randomized controlled study of a home health care team. Am J Public Health 1985;75:134-41. 8. Burge F, Lawson B, Johnston G. Trends in the place of death of cancer patients, 1992-1997. CMAJ 2003;168(3):265-70. 9. Keating NC, Fast JE. Bridging policy and research in eldercare. Can J Aging 1997;16:S22-S41. 10. Houts P, Nezu AM, Maguth Nezu C, Bucher JA. The prepared family caregiver: a problem-solving approach to family caregiver education. Patient Educ Couns 1996;27:63-73. 11. Grunfeld E, Glossop R, McDowell I, Danbrook C. Caring for elderly people at home: the consequences to caregivers. CMAJ 1997;157(8):1101-5. 12. Morris JN, Sherwood S. Quality of life of cancer patients at different stages in the disease trajectory. J Chronic Dis 1987;40:545-53. 13. Mor V. Assessing patient outcomes in hospice: what to measure? Hosp J 1986; 2:17-35. 14. Addington-Hall JM, MacDonald LD, Chamberlain J, Freeling P, Bland JM, Raferty J. Randomized controlled trial of effects of coordinating care for terminally ill cancer patients. BMJ 1992;305:1317-22. 15. Mor V, Laliberte L, Morris JN, Wiemann M. The Karnofsky Performance Status Scale. An examination of its reliability and validity in a research setting. Cancer 1984;53:2002-7. 16. Morris JN, Suissa S, Sherwood S, Wright SM, Greer D. Last days: a study of the quality of life of terminally ill cancer patients. J Chronic Dis 1986;39:47-62. 17. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 health survey manual and interpretation guide. Boston: The Health Institute, New England Medical Center; 1993. 18. Ware JE, Kosinski M, Keller SD. SF-36 Physical and Mental Health Summary Scales: a user's manual. Boston: The Health Institute, New England Medical Center; 1994. 19. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361-70. 20. Zung WWK. A rating instrument for anxiety disorders. Psychosomatics 1971; 12:371-9. 21. Lewis G, Wessely S. Comparison of the General Health Questionnaire and the Hospital Anxiety and Depression Scale. Br J Psychiatry 1990;157:860-4. 22. Zarit SH, Reever KE, Bach-Peterson J. Relatives of the impaired elderly: correlates of feelings of burden. Gerontologist 1980;20:649-55. 23. Zarit SH, Todd PA, Zarit JM. Subjective burden of husbands and wives as caregivers: a longitudinal study. Gerontologist 1986;26:260-6. 24. Kristjanson L. Validity and reliability testing of the FAMCARE scale: measuring family satisfaction with advanced cancer care. Soc Sci Med 1993;36:693-701. 25. Kristjanson LJ. Indicators of quality of palliative care from a family perspective. J Palliat Care 1986;1:8-17. 26. Sherbourne CD, Stewart AL. The MOS Social Support Survey. Soc Sci Med 1991;32:705-14. 27. Institute of Health Economics. A national list of provincial costs for health care: Canada, 1997/8. Available: www.ihe.ca/download/national_costlist.pdf (accessed 2004 Feb 17). 28. Weitzner MA, McMillan SC, Jacobsen PB. Family caregiver quality of life: differences between curative and palliative cancer treatment settings. J Pain Symptom Manage 1999;17:418-28. 29. Cassileth BR, Lusk EJ, Strouse TB, Miller DS, Brown LL, Cross PA. A psychological analysis of cancer patients and their next-of-kin. Cancer 1985;55:72-6. 30. Baumgarten M. The health of persons giving care to the demented elderly: a critical review of the literature. J Clin Epidemiol 1989;42:1137-48. 31. McMillan SC. Quality of life of primary caregivers of hospice patients with cancer. Cancer Pract 1996;4:191-8. 32. Nijboer C, Tempelaar R, Sanderman R, Triemstra M, Spruijt RJ, van den Bos GA. Cancer and caregiving: the impact on the caregiver's health. Psychooncology 1998;7:3-13. 33. Hinton J. Can home care maintain an acceptable quality of life for patients with terminal cancer and their relatives? Palliat Med 1994;8:183-96. Family caregiver burden 34. Hinds C. The needs of families who care for patients with cancer at home: are we meeting them? J Adv Nurs 1985;10:575-81. 35. Chappell NL, Penning M. Behavioural problems and distress among caregivers of people with dementia. Ageing Soc 1996;16:57-73. 36. Kelly B, Edwards P, Synott R, Neil C, Baillie R, Battistutta D. Predictors of bereavement outcome for family carers of cancer patients. Psychooncology 1999; 8:237-49. 37. Coristine M, Crooks D, Grunfeld E, Stonebridge C, Christie A. Caregiving for women with advanced breast cancer. Psychooncology 2003;12:709-19. 38. Schulz R, Beach SR. Caregiving as a risk factor for mortality: the caregiver health effects study. JAMA 1999;282:2215-9. 39. Raveis VH, Karus D, Pretter S. Correlates of anxiety among adult daughter caregivers to a parent with cancer. J Psychosoc Oncol 1999;17:1-26. 40. Cranswick K. Canada's caregivers. Canadian social trends backgrounder. Ottawa: Statistics Canada; 2001. Available: www.hc-sc.gc.ca/seniors-aines/pubs /unsorted/survey.htm (accessed 2004 Feb 17). 41. Stommel M, Given CW, Given BA. The cost of cancer home care to families. Cancer 1993;71:1867-74. 42. Hayman JA, Langa KM, Kabeto MU. Estimating the cost of informal caregiving for elderly patients with cancer. J Clin Oncol 2001;19:3219-25. 43. Philp I, McKee KJ, Meldrum P. Community care for demented and nondemented elderly people: a comparison study of financial burden, service use, and unmet needs in family supporters. BMJ 1995;310:1503-6. 44. Covinsky KE, Goldman L, Cook EF. The impact of serious illness on patients' families. JAMA 1994;272:1839-44. 45. National Forum on Health. Canada Health Action: Building on the legacy. Volume 1. The Final Report. Ottawa: Health Canada; 1997. Available: www.hc-sc .gc.ca/english/care/health_forum/forum_e.htm (accessed 2004 Feb 17). 46. Health Canada. Commission on the Future of Health Care in Canada. Final Report. Ottawa: Health Canada; 2002. Available: www.hc-sc.gc.ca/english/care /romanow/hcc0086.html (accessed 2004 Feb 17). 47. McWhinney IR, Bass MJ, Donner A. Evaluation of palliative care service: problems and pitfalls. BMJ 1994;309:1340-2. 48. Cohen SR, Bultz BD, Clarke J, Kuhl DR, Poulson MJ, Baldwin MK, et al. Well-being at the end of life: Part 1. A research agenda for psychosocial and spiritual aspects of care from the patient's perspective. Cancer Prev Control 1997;1(5):333-42. 49. Chochinov HM. The Senate report on end-of-life care: the ball is in our court. CMAJ 2001;164(6):794-5. Correspondence to: Eva Grunfeld, Cancer Care Nova Scotia, 1278 Tower Rd., Bethune Building, Halifax NS B3H 2Y9; Eva.Grunfeld@ccns.nshealth.ca Appendix 1: Measurement instruments Domain and instrument Functional status Karnfosky Performance Status15,16 General health status SF-3617,18 Psychological distress Hospital Anxiety and Depressions Scale19-21 Caregiver burden Zarit Burden Inventory22,23 Satisfaction with care FAMCARE24,25 Properties (score range) Administered (0-100) Higher score means better performance status Patient Every 3 mo in palliative period Every 2 wk in terminal period (0-100) Higher score means better health 8 subscales Physical Component Score Mental Component Score Caregiver and patient Start of palliative period (0-21) Caregiver and patient 2 subscales Start of palliative period Start of terminal period Clinical anxiety (HADS score 11) Clinical depression (HADS score 11) (0-88) Higher score means greater burden Caregiver Every 3 mo in palliative period Every 2 wk in terminal period (20-100) Higher score means more satisfied Caregiver Every 3 mo in palliative period Every 2 wk in terminal period Social support MOS Social Support Survey26 (0-100) 4 subscales (e.g., emotional support) Higher score means greater support Patient Start of palliative period CMAJ JUNE 8, 2004; 170 (12) 1801 Copyright 7986 by The Cerontological Society of America Investigated were changes over time for caregivers of dementia patients, with a focus on factors associated with nursing home placement. Caregivers were all spouses, and at the time of the initial interview were caring for the patient at home. Subsequent nursing home placement was more strongly associated with subjective factors, particularly caregivers' perceived burden, than with objective indicators of the severity of dementia. Although wives initially reported more burden than husbands, no differences were found at the follow-up. Key Words: Caregivers, Dementia, Institutionalization Subjective Burden of Husbands and Wives as Caregivers: A Longitudinal Study1 Steven H. Zarit, PhD,2 Pamela A. Todd, MSG,3 and Judy M. Zarit, PhD4 1 This research was supported in part by NIMH research grant (5 RO1 MH 34507), Family Interventions for Senile Dementia, and by a grant from the Lexington Fund, Andrus Gerontology Center. A version of this paper was presented at the 37th annual scientific meeting of the Cerontological Society of America. The authors wish to thank Dr. Margaret Gatz for her critical comments on both the original and follow-up study. Associated Professor of Gerontology and Psychology, Leonard Davis School of Gerontology, Andrus Gerontology Center, University of Southern California, University Park, Me 0191, Los Angeles, CA 90089-0191. 'Director of Resident Services, Silvercrest Residency, Santa Monica, \\JA. Director, Center for Adult Development, Pasadena, CA. 260 who were placed. Similarly, studies of caregivers of dementia patients living in the community have found that severity of symptoms is not strongly associated with their subjective feelings of burden (Zarit et al., 1980; Zarit, 1982). While the patient's impairment created the situation in which burden was experienced, how much burden caregivers felt also depended on how they responded to the patient's disabilities, that is, whether they coped well or poorly with specific problems, and the resources they had available, especially for obtaining assistance and occasional relief from caregiving tasks. Whether the caregiver is a man or woman may also make a major difference. Caregiving is usually defined as a woman's role, an expansion of the traditional responsibilities of a wife or daughter. There has been some suggestion based on qualitative studies that the caregiving role conflicts with the decisions many women are making to take more control of their lives and not be locked into traditional roles. This may be particularly the case for middle-aged women who are emerging from their child-rearing responsibilities only to have their steps toward more personal fulfillment threatened by a parent's disability (Brody, 1985). Many older women also look to their later years as a time for more personal opportunity and growth and may resent becoming caregivers to their husbands. There has been less attention given to male caregivers even though samples of caregivers seen in clinic settings have been approximately one-third male (Zarit et al., 1980; Zarit, 1982). Some preliminary research indicates that complaints are often expressed differently by husbands and wives (Steinberg & Shulman, 1981). Wives find their husband's dependency upsetting while husbands as caregivers are more likely to have problems assuming household responsibilities and to be distressed because their wife can no longer interact with them. The present study is a 2-year follow-up of a sample of 64 caregivers of dementia patients who participated in research on the experience of burden. The The Gerontologist Downloaded from http://gerontologist.oxfordjournals.org/ at Ohio University on November 5, 2016 A 2-year follow-up was conducted of a sample of husbands and wives who were caregivers for a spouse with senile dementia to identify factors affecting the course of caregiving, especially the decision to institutionalize, and to determine if differences between husbands and wives found at the initial testing remained stable over time. An important issue in research on caregivers is their variability how they differ from one another in their response to caregiving demands. Clinical observations suggest that some are quickly and decisively overwhelmed, and others make quiet, stubborn efforts to care for their relative at home. Research that identifies sources of this variability can be useful in planning programs that reduce the stress on family members and prevent premature or unnecessary institutionalization. While there has been little theory to guide research, it is frequently assumed that the caregiver's burden is strongly related to the severity of the patient's disabilities. From this perspective, burden is inseparable from the demands that are placed on the caregiver, and when these demands become too great, placement will occur. Some studies, however, suggest that the relation between severity of the patient's impairments and the caregiver's subjective burden may be complex. Lowenthal and her associates (1967), for example, reported that placement was associated with a breakdown of the family support system, and not deterioration of the patient's condition. Patients who were maintained in the community often had impairments as severe as those initial study was concerned with identifying potential mediators of the caregiver's sense of burden as well as differences between husbands and wives who were caregivers (Zarit, 1982). The 2-year follow-up was conducted to explore whether initial findings predicted subsequent changes for caregivers, specifically: 2) is the caregiver's initial level of burden associated with institutionalization; 2) is the subjective experience of burden mediated by similar variables at 2 years as at the initial interview; and 3) do the differences found initially between husbands and wives remain constant over time? Table 1 . Major Demographic Characteristics of Husbands and Wives Who Were Caregivers Age Mean* s.d. Religion Jewish Catholic Protestant Other None Education Mean years s.d. Methods and Procedures Sample Procedures Caregivers were contacted after an approximately 2-year interval and were asked about the current status of the patient. Except in cases where the dementia patient was deceased, the caregiver was Vol.26, No. 3,1986 261 Wives (n = 33) 72.27 (6.72) 63.37 (7.48) 11 5 11 1 3 11 10 9 1 2 14.46 (4.81) 14.00 (3.42) Occupation3 Higher executive or professional Business manager, lesser professional Administrative, minor professional Clerical, sales, secretarial Skilled manual Semi-skilled Housewives 4 1 8 12 4 3 0 0 7 6 6 1 1 11 Monthly income Under $400 $400 $700 $700 $1,000 $1,000 $1,300 $1,300 $1,600 $1,600 and over 1 2 5 4 4 15 0 2 7 7 4 13 42.60 (6.96) 35.69 (12.26) Years married Mean s.d. "Hollingshead Index (Hollingshead & Redlich, 1958). *p < .05. asked if he or she was willing to be interviewed again. Interviews usually took place in the caregiver's home. The interviewer had not participated in the time 1 study and did not know the results of the time 1 interviews. In addition, all patients who were still at home were given the clinical screening measures (MSQ and Face-Hand test). Because of limited time, there was no attempt to re-test patients who were now in nursing homes. Measures The interview was comprised of measures given to respondents at time 1, along with a series of questions designed for caregivers who had placed their relative. These questions inquired about factors affecting their decision. Burden. The focus of the first study was on the caregiver's burden. Burden was defined as the extent to which caregivers perceived their emotional or physical health, social life, and financial status as suffering as a result of caring for their relative. A 20item Burden Interview was devised, using items from an earlier version (Zarit etal., 1980). Caregivers were asked to indicate the extent to which each item described how they felt, with responses ranging Downloaded from http://gerontologist.oxfordjournals.org/ at Ohio University on November 5, 2016 The sample in the original study was comprised of 33 wives who were caregivers for their husbands and 31 husbands who were caring for wives. At the time of the initial study, all patients were living at home with their spouses. The sample was drawn from two sources: patients seen in a clinic offering counseling and support groups for caregivers, and from membership lists from an Alzheimer's Disease advocacy group. Similar proportions of husbands and wives came from each source. Subjects were recruited consecutively from the 2 groups until 31 husbands and 33 wives had been interviewed. There was no selection factor into the sample other than having a dementia patient at home. Table 1 shows characteristics of caregivers at the time of the first interview. The major difference that emerged between husbands and wives was that husbands were older (t (62) = 4.36, p < .01) and had been married longer (t (62) = 3.07, p, .01), probably reflecting a tendency for men to marry younger women. As part of the initial study, all patients were evaluated and met the DSM 111 criteria for dementia. About one-third had clinical characteristics suggesting a multi-infarct dementia, while two-thirds had symptoms and histories indicating probable senile dementia of the Alzheimer type. Although caregivers reported similar durations of the illness for male and female patients, women had significantly more cognitive impairment, as indicated by 2 clinical screening measures, the Mental Status Questionnaire (MSQ) and the Face-Hand Test (Kahn et al., 1960). Male patients' scores at time 1 were 4.70 correct (out of 10) on the MSQ compared to 1.97 correct for women (t (62) = 3.83, p < .05). On the Face-Hand test, husbands who were patients averaged 8.27 correct trials (out of 16) compared to 4.03 for wives (t (59) = 2.36, p < .05). Table 2 presents characteristics of the dementia patients. Husbands (n = 31) Table 2. Major Demographic Characteristics of Husbands and Wives Who Were Dementia Patients Age Mean s.d. Religion Jewish Catholic Protestant Other None Education Mean years s.d. Years of memory loss Mean s.d. Wives with dementia (n = 31) 69.30 (7.22) 70.53 (7.12) 9 11 7 14 1 2 6 13 0 15.43 (4.72) 12.77 (2.28) 6 1 8 5 7 4 6 1 1 0 0 11 2 1 0 11 5.60 5.87 Social support. Caregivers were asked to estimate the frequency of social support from informal and formal sources. Informal sources included how often caregivers had contact with family and friends, and what assistance, if any, they received. Formal supports were the frequency of services provided by agencies or other paid helpers. A combined total was obtained for the weekly frequencies of contact with informal or formal help. Caregivers were also asked to subjectively rate the quality of the support they were receiving on a 4-point scale ranging from 4 = "I get most of the help I need" to 1 = "I feel overwhelmed and do not know where to turn." 1 Quality of the relationship. Following Bengtson (personal communication), quality of the prior marital relationship was estimated by asking caregivers to respond to a 10-point scale with choice points ranging from 1 = poor to 10 = excellent. Mental status questionnaire items correct Mean s.d. 4.70 (2.61) 1.97 (2.52) The placement decision. Caregivers who had placed their relative were asked to respond to a series of open-ended questions about what had led to their decision. Face-Hand test Mean correct* s.d. 8.27 (7.51) 4.03 (5.75) Findings at Time 1 a Hollingshead Index (Hollingshead & Redlich, 1958). *p < .05. from "not at all" to "extremely." A total score was obtained by summing the items endorsed. Adequate reliability and validity were found using the 20-item version (Zarit, 1982). Severity of impairment. Severity of impairment was measured in 2 ways by the clinical screening tests mentioned earlier (MSQ and Face-Hand test) and with an inventory of dementia-related problems, the Memory and Behavior Problem Checklist (Zarit et al., 1980; Zarit, 1982). The Memory and Behavior Problem Checklist is comprised of 28 problems that caregivers are likely to encounter on a day-to-day basis. Caregivers were asked to indicate how frequently each problem had occurred during the past week, with responses ranging from 0 = "never occurred" to 4 = "occurs daily or more often." Using an approach suggested by Sanford (1975), subjects were also asked their responses to the problems they reported, specifically, how well they were able to tolerate the problem. Tolerance ratings ranged from 0 = "I can tolerate this behavior when it occurs" to 4 = "I can no longer tolerate this behavior and have to do something about it." Three scores from the checklist were obtained: (1) total frequency of mem262 In the initial study subjective factors were more important than the severity of the illness in determining burden (Zarit, 1982). Severity of the symptoms as measured by the MSQ or Face-Hand test or the frequency of memory and behavior problems did not account for a significant proportion of variance in burden scores. Husbands who were caregivers reported less burden than wives. This difference appeared to be related to husbands' greater tolerance of memory and behavior problems, although the possibility that men under-reported and women over-reported their subjective burden could not be ruled out. For wives, both the quality of the past relationship and the cross-product score for the Memory and Behavior Problem Checklist (frequency x tolerance) were significantly associated with burden, while for husbands, only the cross-product score was a significant predictor. Data Analysis at Time 2 The data analysis proceeded in 3 steps. First, t-tests were used to compare time 1 scores on burden and other variables for caregivers who placed their relatives and for those whose spouse was still at home. Second, correlations were made between burden and predictor variables used in the time 1 study. Third, changes in scores of men and women caregivers from time 1 time 2 were evaluated with paired The Gerontologist Downloaded from http://gerontologist.oxfordjournals.org/ at Ohio University on November 5, 2016 Occupation 3 Higher executive or professional Business manager, lesser professional Administrative, minor professional Clerical, sales, secretarial Skilled manual Semi-skilled Unskilled Housewives Husbands with dementia (n = 33) ory and behavior problems; (2) total tolerance; and (3) the sum of the cross product of frequency and tolerance ratings for each of the 28 items. The use of a cross-product score is based on the work of Lewinsohn and MacPhillamy (1974), who found that the impact of behaviors could be more accurately evaluated by adding a subjective dimension. fs. Multivariate statistics were not used due to the small sample size. Results at 2 Years Table 3 summarizes the current status of patients. Over half of those for whom status was determined were still at home and only 11 patients were in nursing homes, a small number given the severity of symptoms of the sample at time 1. Of the 15 patients who died, 9 had spent some time before their deaths in a nursing home, ranging from a few weeks to 6 months. The status of 5 patients could not be determined either because the caregiver could not be located (n = 3) or because the caregiver had requested complete anonymity at time 1 (n = 2). In addition, one caregiver was now deceased. There were no gender differences in outcome except for a higher death rate among male patients. Similar numbers of husbands and wives (5 and 6, respectively) were placed in nursing homes. To highlight differences between caregivers who kept their relatives at home and those who did not, the deceased group will not be included in the subsequent analyses and discussion. As the first step in the data analysis, time 1 burden scores were compared for subjects who subsequently placed their spouse in a nursing home and those who did not. As shown in Table 4, initial burden was significantly higher for those caregivers who had placed (t = 2.28, d.f. 42, p ^ .05). Other variables were then examined for contributing to placement. Neither the initial mental status score nor the total frequency of memory and behavior problems differed significantly at time 1 between those who placed and those who did not. Table 3. Two Year Status of Dementia Patients Patients outcome At home Nursing home Patient deceased Caregiver deceased Unknown Husbands with dementia (n = 33) Wives with dementia (n = 31) 15 5 11 0 2 17 6 4 1 3 Table 4. Means for Measures at Time 1 and Time 2 According to Time 2 Outcome At home Burden Mental status questionnaire Frequency of memory and behavior problems Cross product of memory and behavior problems Formal/informal social supports Subjective social supports Quality of prior relationship3 Deceased Timel (n = 32) Time 2 (n = 29) Time 1 (n = 11) Time 2 (n = 9) Timel (n = 15) 33.59 3.54 45.72 47.76 12.25 2.22 8.35 29.17 2.08 44.55 19.06 11.70 2.72 43.54 2.36 48.00 67.36 11.36 1.64 7.64 23.44 * 44.50 35.25 12.71 .75 39.33 2.87 51.93 65.40 11.00 1.87 6.67 a Patients in nursing homes were not tested at Time 2. Vol. 26, No. 3,1986 Nursing home 263 Downloaded from http://gerontologist.oxfordjournals.org/ at Ohio University on November 5, 2016 While total severity of symptoms was the same in both groups, analysis of individual items of the Memory and Behavior Problems Checklist revealed a few differences. Of the 28 items, patients who were placed had higher frequencies of difficulty doing simple tasks, being unable to dress or bathe self, and incontinency. Some behaviors often mentioned in the literature as leading to institutionalization occurred in similar frequency in both groups, such as waking the caregiver at night, being constantly restless or talkative, and hearing voices. There were other large differences at time 1 between those who subsequently placed their relative and those who did not, particularly on the crossproduct score of the Memory and Behavior Problem Checklist and their subjective rating of social support; the differences, however, were not significant, perhaps due to the small sample size. Another noteworthy finding in Table 4 is that burden scores decreased for both groups as did the cross-product score on the Memory and Behavior Problem Checklist. The frequency of problems reported, however, stayed the same over time. To replicate in part the original analysis of mediators of burden at time 1, zero-order correlations were made between major study variables and burden, using responses given by caregivers at time 2. (The time 1 analysis of predictors of burden was made with multiple regressions, but the smaller sample size at time 2 precluded that approach). The findings on caregivers with patients at home and in nursing homes were combined in this analysis. Although sources of burden may change with placement, there also may have been some relation between factors associated with burden and the placement decision. Use of the entire sample therefore provides for better comparability with the original results. As shown in Table 5, 2 variables at time 2 were related to burden: the Mental Status Questionnaire and the cross-product score on the Memory and Behavior Checklist. As mentioned earlier, the crossproduct score was the major contributor to burden at time 1. The correlation between the mental status score and burden is in the opposite direction than expected; when patients made more correct responses, caregivers reported more burden. As noted caregivers remained constant over time. As shown in Table 6, burden was higher for women than men at time 1. At time 2, average scores for men and women were about the same. Paired f-tests were used to examine the changes in scores for men and women, respectively, from time 1 to time 2. Women's burden showed a significant decrease from time 1 scores (f = 4.31, d.f. 16, p < .001). This decrease was especially large for those women who had placed their spouse and whose initial burden at time 1 was higher. Men and women reported significantly more tolerance for memory and behavior problems as indicated by their lower cross-product scores (t for men = 2.87, d.f. 19, p < .010; f for women = 4.07, d.f. 16, p < .001). earlier, time 2 mental status scores were obtained only for subjects still at home. The third focus of the study was whether the differences found at time 1 between men and women Table 5. Correlations of Patient and Caregiver Variables With Burden Timel (n = 64) Variables .15 .36* .02 -.001 .46* .24 .15 -.43* -.45* .38* .01 .11 -.06 -.02 Discussion This study clarifies the relation of the caregiver's subjective burden to the decision to institutionalize, and also provides information on differences between husbands and wives and other factors associ- *p < .05. Table 6. Description of Subpopulations by Criterion Variables for Husbands and Wives as Caregivers Husbands as caregivers Mean n (standard deviation) Burden Frequency of memory and behavior problems Cross product of memory and behavior problem Formal/informal social supports Subjective social supports Wives as caregivers Mean n (standard deviation) Time 1 Time 2 Timel Time 2 At home 26.52 (17) (10.81) 25.81 (16) (9.53) 41.60 (15) (13.09) 33.31 (13) (13.63) Nursing home 28.00 (6) (10.53) 26.60 (5) (3.51) 50.20 (5) (3.90) 19.40 (4) (4.73) All subjects 29.48 (23) (10.31) 26.00 (21) (8.67) 43.53 (20) (12.20) 30.06 (17) (13.42) At home 51.94 (17) (20.71) 52.63 (16) (10.95) 38.67 (15) (18.73) 34.62 (13) (13.80) Nurisng home 56.67 (6) (8.66) 47.40 (4) (12.45) 38.60 (5) (18.08) 41.40 (4) (5.57) All subjects 54.05 (23) (18.65) 51.60 (11.12) 38.82 (20) (19.60) 36.25 (17) (12.56) At home 41.24 (17) (29.93) 21.44 (16) (27.23) 55.40 (15) (43.12) 16.15 (13) (15.96) Nursing home 81.17 (6) (56.19) 58.75 (5) (43.25) 50.80 (5) (25.93) 11.74 (4) (19.12) All subjects 52.90 (23) (37.82) 28.90 (21) (33.40) 54.69 (20) (41.86) 15.12 (17) (16.23) At home 12.53 (17) (5.29) 12.86 (14) (4.13) 11.93 (15) (4.25) 10.46 (13) (2.26) Nursing home 9.17 (6) (6.01) 11.33 (3) (5.51) 14.00 (5) (4.64) 13.75 (4) (5.06) All subjects 12.35 (23) (6.20) 12.59 (17) (4.24) 12.59 (20) (4.50) 11.24 (17) (3.27) At home 2.41 (17) (.80) 2.69 (16) (.48) 2.00 (15) (1.07) 2.77 (13) (1.17) Nursing home 1.50 (6) (1.09) 1.20 (5) (1.30) 1.80 (5) (1.10) .00 (4) All subjects 2.24 (23) (.83) 2.33 (20) (.97) 2.00 (20) (1.03) 2.25 (17) (1.53) 264 The Gerontologist Downloaded from http://gerontologist.oxfordjournals.org/ at Ohio University on November 5, 2016 Mental status questionnaire Frequency of memory and behavior problems Cross-product score memory and behavior problems Formal social supports Informal social supports Subjective social supports Quality of relationship before onset Time 2 (n = 36) Vol. 26, No. 3,1986 265 longer manifested by many patients at time 2. In contrast, deficits in activities of daily living were more prominent at the 2-year follow-up. This observation points out that care demands shift over time, but do not necessarily increase as patients move from a moderate to severe level of dementia. The possibility that the presence of specific symptoms led to institutionalization cannot be ruled out. Four of the 28 problems on the Memory and Behavior Problems Checklist were found to be significantly higher at time 1 for patients who were subsequently institutionalized. Because multiple f-tests were used in conducting this portion of the data analysis, it is possible that these findings are due to chance. On the other hand, three of the problems identified, difficulties bathing, dressing, and incontinence, are frequently cited in the literature as leading to the decision to place a patient. As has been emphasized, however, some caregivers were able to develop effective management strategies for these and other troubling problems. The availability of social supports was not a factor in the placement decision. The amount of assistance that families received, both from informal and formal sources, was relatively low. In response to openended questions, both those who placed a patient and those who still had their spouse at home expressed a need for respite services, which they had generally been unable to obtain or afford. The absence of a relation of social supports to burden and to the placement decision may therefore be due to the fact that most caregivers were receiving only a small amount of assistance. One other finding has to do with the need for support after the placement decision. As shown in Table 6, wives who placed their husbands all reported minimal subjective support. While the number of cases is small, the prospect that some caregivers remain isolated and unsupported after placing a spouse should be considered. The difference in subjective burden between husbands and wives at time 1 were no longer apparent in the 2-year follow-up. It is possible that wives changed their response style on the self-report Burden Interview, emphasizing their distress initially and then adopting a more stoical attitude at followup. Another possibility is that there was a change in strategies for coping with everyday problems. At time 1, husbands were often observed to adopt an instrumental approach to daily prdblems, while wives had difficulty maintaining the emotional distance necessary to consider alternative strategies for managing problems (Zarit, 1982). Interviewers at time 2, however, felt that many wives had adopted a similar coping style, which was reflected in their lower cross-product scores on the Memory and Behavior Problems Checklist. Another factor that may account for wives' initial burden scores is that they experienced conflicting demands or resentment at having been placed again in a caregiving role, but this point was not addressed in the current- study. These differences between husbands and wives at relatively early stages of their spouse's illness have Downloaded from http://gerontologist.oxfordjournals.org/ at Ohio University on November 5, 2016 ated with burden. It should not be surprising that the caregiver's feelings of burden are associated with nursing home placement. As in the research of Lowenthal and her associates (1967), however, severity of the patient's symptoms did not differentiate caregivers who placed their relative from those who did not. This is not to say that behavior problems did not play a role in the placement decision. In fact, caregivers who placed their spouse responded to open-ended questions about their decision by indicating that they felt unable to cope with daily tasks. Other caregivers, however, did not report high levels of burden when faced with similar caregiving demands. Rather than viewing specific tasks (e.g., caring for an incontinent patient) as overwhelming, the impact appears variable. The dementing illness brings about the changes that cause caregivers to feel burdened and to consider institutionalization among their care alternatives, but the experience of greater or lesser degrees of burden at any point in the patient's disease may be related to many factors, and not just severity of symptoms. The findings in this study are consistent with general models of stress, which propose that the impact of harmful events is mediated by whether subjects actually perceive them as harmful and by respondents' own coping responses (Lazarus, 1966; Lazarus & Folkman, 1984). The variable that had the strongest association with burden both at time 1 and in this follow-up was the cross-product score for the Memory and Behavior Checklist. In other words, caregivers feel burdened when the patient manifests deficits in behavior and caregivers have difficulty tolerating those behaviors. This finding underscores the point that caregivers react differently to problem behaviors and vary in their skills for managing these problems. Further, all caregivers do not find the same problems to be troublesome. While some become upset by incontinence, others may have more difficulty with repetitive behaviors, and a few are able to remain calm in the face of extreme behavioral deficits. This variability in caregivers' reactions is, in fact, of major importance for planning interventions to alleviate their stress (Zarit et al., 1985). The longitudinal analysis in this study suggests that caregivers' ability to tolerate problem behaviors actually increases, even as the disease progresses. This was indicated by the lower cross-product scores at the 2-year follow-up (Table 6). Caregivers frequently explained their responses by stating that they had learned to manage problems more effectively, or they simply did not let problems bother them any more. Many of those providing home care in this sample genuinely seemed to have established a daily routine that while not without its stresses, also was not excessively demanding or burdensome. Another factor that should be considered is that the total frequency of memory and behavior problems did not increase over the 2-year period. As the patient's dementia worsened, some behaviors diminished in frequency or ceased altogether. Some of the most troublesome behaviors, including wandering, paranoid accusations, and restlessness, were no now been replicated in a new study (Boutselis & Zarit, 1984). Finally, a major implication of these findings is that carefully planned interventions may effectively relieve some of the burden caregivers experience. Programs that focus on improving coping with everyday problems, that provide opportunities for respite, and that give special attention to early interventions with wives, may have considerable impact on subjective burden. References Bengtson, V." L. (1981). Personal communication. Boutselis, M., & Zarit, S. H. (1984, November). Burden and distress of dementia caregivers: Effects of gender and relationship. Paper presented at the meeting of the Cerontological Society of America, San Antonio, TX. Brody, E. M. (1985). Parent care as a normative family stress. The Cerontologist, 25,19-29. Kahn, R. L , Coldfarb, A. I., Pollack, M., & Peck, R. (1960). Brief objective measures for the determination of mental status in the aged. American Journal of Psychiatry, 117, 326-328. springer publishing comp