Psycho-Oncology Psycho-Oncology 17: 300303 (2008) Published online 23 May 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/pon.

1218

Brief report

Assessing pre-death grief in cancer caregivers using the MarwitMeuser Caregiver Grief Inventory (MM-CGI)
Samuel J. Marwit1*, John T. Chibnall2, Rebecca Dougherty3, Chrystal Jenkins2 and Jenifer Shawgo2
1 2

Department of Psychology, University of Missouri}St. Louis, St. Louis, MO, USA Department of Psychiatry, Saint Louis University School of Medicine, St. Louis, MO, USA 3 Saint Louis University Cancer Center, St. Louis, MO, USA * Correspondence to: 4540 Laclede Ave., #306, St. Louis, MO 63108, USA. E-mail: smarwit@yahoo.com

Abstract
The MarwitMeuser Caregiver Grief Inventory (MM-CGI) was developed and initially validated with caregivers of persons with Alzheimers disease, and subsequently evaluated with caregivers of persons with brain injury. The present study examined MM-CGI psychometric and validity properties in 75 caregivers of persons with cancer. As with previous studies, high internal consistency reliability of MM-CGI Total Grief and subscale scores (Personal Sacrice and Burden, Heartfelt Sadness and Longing, and Worry and Felt Isolation) was demonstrated. Construct validity of the subscale scores was supported by dierential associations with other caregiver measures of depression, strain, well-being, and family support. Based on these preliminary results, the MM-CGI appears useful for assessing pre-death grief in caregivers of persons with cancer. Copyright # 2007 John Wiley & Sons, Ltd.
Keywords: cancer; oncology; caregivers; grief; assessment

Received: 1 November 2006 Revised: 28 March 2007 Accepted: 29 March 2007

Introduction
Historically, grief research has focused on loss associated with death. Recently, grief has been considered within the emotional reactions of predeath loss, particularly among caregivers of people with chronic/debilitating and life-threatening conditions, including dementia, where a series of cognitive, emotional, and social losses precede death [13]. Caregiver grief is considered a unique aect, similar to bereavement [4], discriminable from depression and anxiety [57], and associated negatively with caregiver health, social relations, and post-death bereavement [8]. Eorts to measure caregiver grief include the Inventory of Complicated Grief (ICG), a pre-loss version of which has been used extensively to study complicated (or traumatic) pre-death grief in caregivers of persons with cancer, other terminal illnesses, and dementia [914]. A more recent grief measure, the MarwitMeuser Caregiver Grief Inventory (MM-CGI), was initially developed and psychometrically evaluated with independent samples of dementia caregivers [6], and was subsequently evaluated with caregivers of persons with acquired brain injury (ABI; [15]), with both studies focusing on pre-death grief. Results for the dementia and brain injury samples were very similar and supported the reliability and construct validity (in relation to caregiver depression, strain, well-being, and family support) of the MM-CGI

(and its subscales of Personal Sacrice and Burden, PSB; Heartfelt Sadness and Longing, HSL; Worry and Felt Isolation, WFI). The results of the previous MM-CGI research [6,15] prompted the present study, which sought to extend MM-CGI application to pre-death grief in cancer caregivers. Cancer was selected because its lifethreatening nature is high, psychological and behavioral toll on caregivers is great, and existing research using the ICG [1014] and other grief measures [16] suggests that the MM-CGI may have direct applicability to the pre-death grief experiences of cancer caregivers. Although the study was exploratory, results were expected to parallel those from previous MM-CGI studies, including high internal consistency reliabilities. We expected MM-CGI Total Grief and PSB to correlate moderately with strain and wellbeing; HSL to correlate moderately with depression and strain; and WFI to correlate with depression, well-being, and family support.

Methods
The study was approved by the Saint Louis University institutional review board.

Participants
Fifty-two (69.3%) female and 23 (30.7%) male caregivers participated, including 67 (89.3%)

Copyright # 2007 John Wiley & Sons, Ltd.

MM-CGI in cancer caregivers

301

Caucasians and eight (10.7%) African Americans, with a mean age of 52.8 years (SD 12:8). Education was 4high school for 27 caregivers (36.0%), some college/technical school for 26 (34.7%), and college degree for 22 (29.3%). Religion was Christian for 69 (92.0%), Buddhist for one (1.3%), and none for ve (6.7%). Sixtyseven (89.3%) caregivers were married (or equivalent). The person with cancer was a spouse (or equivalent) to 42 (56.0%) caregivers, a parent to 18 (24.0%), and other (e.g. sibling, friend) to 15 (20.0%). Forty-two (56.0%) caregivers lived with the person with cancer. Caregiving occurred 7 days/week for 49 (65.3%) and less often for 26 (34.7%). On caregiving days, 37 (49.3%) caregivers spent 512 h with the person, while 38 (50.7%) spent 512 h. Forty-nine (65.3%) caregivers always attended medical appointments and 26 (34.7%) less often. Nineteen caregivers rated illness severity of the person with cancer as mild (25.3%), 35 as moderate (46.7%), and 21 (28.0%) as serious/ grave. (No objective corroboration of severity was obtained.) Caregiving had been occurring for a median of 12 months (Q1 6; Q3 25).

Materials and procedure

Participants completed ve self-report scales}all with documented reliability and validity}in this order: Center for Epidemiological Studies Depression Scale (CESD), 060 range [17]; Marwit Meuser Caregiver Grief Inventory (MM-CGI), Total Grief (range 502250); Personal Sacrice and Burden (PSB; range 18290); Heartfelt Sadness and Longing (HSL; range 15275); Worry and Felt Isolation (WFI; range 17285) [6]; Caregiver Well-Being Scale (Basic Needs) (CWBSBN), 22100 range [18]; Caregiver Strain Index (CSI), 013 range [19]; Perceived Social Support Questionnaire (Family) (PSSQ-F), 20100 range [6,20]. Data were collected at the Cancer Center of Saint Louis University Health Sciences Center in clinics treating cancers of the skin, blood, lung, liver, breast, gastrointestinal/urological systems, and head/neck (JuneJuly, 20052006). Although caregivers were not systematically sampled, an attempt was made to visit all cancer clinics during each week of data collection. Only caregivers who accompanied the person with cancer to an appointment were recruited. Participants were approached in the clinics by a research assistant, where the study was explained and materials completed. Of caregivers approached, $90% agreed to participate.

Results
MM-CGI internal consistent reliabilities (Cronbachs coecient alpha) ranged from 0.90 for WFI,
Copyright # 2007 John Wiley & Sons, Ltd.

0.94 for HSL, 0.95 for PSB, and 0.96 for Total Grief. These values were nearly identical to those reported previously [6,15]. The other measures used also had adequate internal consistency reliability: CSI 0:80; CWBS-BN 0:89; CESD 0:91; and PSSQ-F 0:95: Descriptive data are in Table 1, with comparisons to previous MM-CGI studies. Cancer caregivers reported lower levels of grief than dementia or ABI caregivers. Also in Table 1 are univariate construct validity correlations between MM-CGI scores and other caregiver measures. MM-CGI scores were signicantly correlated with CESD, CWBS-BN, CSI, and PSSQ-F, with the exception of HSL with PSSQ-F. Correlations were comparable across MM-CGI studies. Multiple linear regression was used to account for shared variance among the caregiver measures and demographic associations with MM-CGI scores. Four mixed-entry, hierarchical regressions were computed, one for each MM-CGI score. In Step 1 of the regressions, demographics (see Participants subsection above) were considered for inclusion in a forward-entry fashion. (Marital status, race, and religion were not considered due to limited numbers.) In Step 2, CESD, CWBS-BN, CSI, and PSSQ-F were entered simultaneously. Table 2 displays regression results. For Total Grief, the Step 1 regression equation (demographics) was signicant, F2; 72 4:5; p50:05; R2 0:11: More education predicted lower Total Grief, while higher perceived illness severity predicted higher Total Grief. The regression equation following Step 2 (entry of CESD, CWBS-BN, CSI, and PSSQ-F) was also signicant, F4; 68 24:2; p50:001; R2 change 0:52: Greater depression symptoms (CESD) and strain (CSI) predicted higher Total Grief. For MM-CGI PSB, none of the demographics was signicant in Step 1. The Step 2 regression equation was signicant, F4; 70 25:5; p50:001; R2 0:59: Greater strain (CSI) predicted higher PSB. For MM-CGI HSL, regression Step 1 was signicant, F2; 72 5:6; p50:01; R2 0:13: More education predicted lower HSL, while higher illness severity predicted higher HSL. The Step 2 equation was also signicant, F4; 68 12:8; p50:001; R2 change 0:37: Greater depression symptoms (CESD) and strain (CSI) predicted higher HSL. Regarding MM-CGI WFI, regression Step 1 was signicant, F1; 73 5:7; p50:05; R2 0:07: More education predicted lower WFI. The Step 2 equation was again signicant, F4; 69 32:7; p 50:001; R2 change 0:61: Greater depression (CESD) and strain (CSI), and lower well-being (CWBS-BN) and family support (PSSQ-F), predicted higher WFI.
Psycho-Oncology 17: 300303 (2008) DOI: 10.1002/pon

302

S. J. Marwit et al.

Table 1. Descriptive statistics and univariate correlations between MM-CGI scores and other caregiver measures (present study and other MM-CGI studies)
Measures Caregiver type Mean (SD) CESD/BDI MM-CGI Total Grief Cancer Dementiaa ABIc Cancer Dementiaa ABIc Cancer Dementiaa ABIc Cancer Dementiaa ABIc Cancer Dementiaa ABIc Cancer Dementiaa ABIc Cancer Dementiaa ABIc Cancer Dementiaa ABIc 121.0 (34.6) 144.0 (31.6) 168.0 (35.0) 41.9 (13.7) 54.3 (14.1) 63.1 (13.8) 41.1 (13.1) 48.2 (11.1) 53.1 (13.1) 38.0 (11.1) 40.6 (12.0) 51.8 (12.8) 15.7 (11.4) 9.3 (6.7)b 16.4 (11.5)b 79.1 (11.9) 80.0 (14.8) 72.3 (14.0) 5.7 (3.4) 6.5 (2.2) 8.8 (2.7) 71.6 (15.2) 72.0 (15.6) 66.4 (19.2) 0.61*** 0.76**,b 0.72**,b 0.41*** 0.72**,b 0.63**,b 0.60*** 0.59**,b 0.57**,b 0.69*** 0.66**,b 0.74**,b } } } } } } } } } } } } Pearson r CWBS-BN 0.44*** 0.66** 0.64** 0.35** 0.61** 0.71** 0.32** 0.33** 0.37 0.57*** 0.72** 0.63** } } } } } } } } } } } } CSI 0.66*** 0.66** 0.58** 0.74*** 0.73** 0.65** 0.53*** 0.43** 0.42* 0.54*** 0.51** 0.48* } } } } } } } } } } } } PSSQ-F 0.26* 0.36** 0.26 0.30* 0.33** 0.22 0.05 0.08 0.06 0.39*** 0.61** 0.42* } } } } } } } } } } } }

MM-CGI PSB

MM-CGI HSL

MM-CGI WFI

CESD/BDI

CWBS-BN

CSI

PSSQ-F

ABI, acquired brain injury; BDI, Beck Depression Inventory; CESD, Center for Epidemiological Studies Depression Scale; CWBS-BN, Caregiver Well-Being Scale (Basic Needs); CSI, Caregiver Strain Index; HSL, Heartfelt Sadness and Longing; MM-CGI, MarwitMeuser Caregiver Grief Inventory; PSB, Personal Sacrifice and Burden; PSSQ-F, Perceived Social Support Questionnaire (Family); WFI, Worry and Felt Isolation. *p50.05; **p 50.01; ***p50.001. a Marwit and Meuser [6]; p-values not reported beyond p50.01. b BDI was used, rather than CESD; BDI score range is 063, CESD range is 060. c Marwit and Kaye [15].

Discussion
Results support the reliability and construct validity of the MM-CGI in a sample of caregivers of persons with cancer. More generally, they support the idea of multi-faceted, pre-death grief in caregivers of persons who are still living, but dying in other ways [2]. The ndings were very similar to those reported for dementia [6] and ABI [15] caregivers. Regarding univariate correlations, slightly stronger associations were found in dementia and ABI samples, relative to the cancer sample, for Total Grief with well-being, and PSB with depression and well-being. The correlation of WFI with family support was slightly stronger in the dementia sample, relative to the other two. Overall, however, the similarity across samples argues for MM-CGI stability in various caregiver types.
Copyright # 2007 John Wiley & Sons, Ltd.

This study had the advantage of multivariate analysis. After correcting for signicant demographics, measures of depression, strain, wellbeing, and/or family support explained 3761% of the variance in MM-CGI scores. The dierential associations of these measures with MM-CGI scores (regression results) add strength to the validity evidence. Total Grief and HSL were predicted by depression and strain, but not wellbeing or family support. Since HSL measures internalized grief over loss of the way things used to be, its associations with depression and demands of the way things are now are consistent. PSB was predicted by caregiver strain only, given that PSB measures grief over the loss of ones personal life due to caregiving. All caregiver measures explained variance in WFI, an indicator of apprehension-related grief regarding uncertainty about the future and being alone, higher levels of
Psycho-Oncology 17: 300303 (2008) DOI: 10.1002/pon

MM-CGI in cancer caregivers

303

Table 2. Association of other caregiver measures and demographics with MM-CGI scores: multiple regression results
MM-CGI criterion Step Total Grief 1 2 Predictors Caregiver educationa Illness severityb CESD CWBS-BN CSI PSSQ-F CESD CWBS-BN CSI PSSQ-F Heartfelt Sadness and Longing 1 Caregiver educationa Illness severityb CESD CWBS-BN CSI PSSQ-F Caregiver educationa CESD CWBS-BN CSI PSSQ-F b R2

References
1. Loos C, Bowd A. Caregivers of persons with Alzheimers disease: some neglected implications of the experience of personal loss and grief. Death Stud 1997;21:501514. 2. Meuser TM, Marwit SJ, Sanders S. Assessing grief in family caregivers. In Living with Grief: Alzheimers Disease, Doka K (ed.). Hospice Foundation of America: Washington, DC, 2005; 169196. 3. Sanders S, Adams KB. Grief reactions and depression in caregivers of individuals with Alzheimers disease: results from a pilot study in an urban setting. Health Soc Work 2005;30:287295. 4. Sanders S, Corley CS. Are they grieving? A qualitative analysis examining grief in caregivers of individuals with Alzheimers disease. Soc Work Health Care 2003;37:3553. 5. Walker RJ, Pomeroy EC. Depression or grief? The experience of caregivers of people with dementia. Health Soc Work 1996;21:247254. 6. Marwit SJ, Meuser TM. Development and initial validation of an inventory to assess grief in caregivers of persons with Alzheimers disease. Gerontologist 2002;42:751765. 7. Adams KB, Sanders S. Alzheimers caregivers dierences in experience of loss, grief reactions, and depressive symptoms across stage of disease: a mixedmethod approach. Dement Int J Soc Res Pract 2004; 3:195210. 8. Robinson-Whelen S, Tada Y, MacCallum RC, McGuire L, Kiecolt-Glaser JK. Long-term caregiving: what happens when it ends? J Abnorm Psychol 2001;110:573584. 9. Prigerson HG, Maciejewski PK, Reynolds III CF et al. Inventory of Complicated Grief: a scale to measure maladaptive symptoms of loss. Psychiatry Res 1995;59:6579. 10. Beery LC, Prigerson HG, Bierhals AJ et al. Traumatic grief, depression and caregiving in elderly spouses of the terminally ill. Omega 1997;35:261279. 11. Van Doorn C, Kasl SV, Beery LC, Jacobs SC, Prigerson HG. The inuence of marital quality and attachment styles on traumatic grief and depressive symptoms. J Nerv Ment Dis 1998;186:566573. 12. Chen JH, Bierhals AJ, Prigerson HG, Kasl SV, Mazure CM, Jacobs S. Gender dierences in the eects of bereavement-related psychological distress in health outcomes. Psychol Med 1999;29:367380. 13. Prigerson HG, Cherlin E, Chen JH, Kasl SV, Hurzeler R, Bradley EH. The Stressful Caregiving Adult Reactions to Experiences of Dying (SCARED) Scale. Am J Geriatr Psychiatry 2003;11:309319. 14. Hebert RS, Dang Q, Schulz R. Preparedness for the death of a loved one and mental health in bereaved caregivers of patients with dementia: ndings from the REACH study. J Palliat Med 2006;9:683693. 15. Marwit SJ, Kaye PN. Measuring grief in caregivers of persons with acquired brain injury. Brain Injury 2006;20:14191429. 16. Kelly B, Edwards P, Synott R, Neil C, Baillie R, Battistutta D. Predictors of bereavement outcome for family caregivers of cancer patients. Psycho-Oncology 1999;8:237249. 17. Ensel WM. Measuring depression: the CES-D Scale. In Social Support, Life Events, and Depression, Lin N, Dean A, Ensel WM (eds). Academic Press: New York, 1986; 5170. 18. Berg-Weger M, Rubio DM, Tebb SS. The Caregiver WellBeing Scale revisited. Health Soc Work 2000;25:255263. 19. Robinson BC. Validation of a caregiver strain index. J Gerontol 1983;38:344348. 20. Procidano ME, Heller K. Measures of perceived social support from friends and from family: three validation studies. Am J Community Psychol 1983;11:124.
Psycho-Oncology 17: 300303 (2008) DOI: 10.1002/pon

0.30* 0.11* 0.24* 0.33*** 0.52*** 0.11 0.48*** 0.07 0.13 0.06 0.66*** 0.11 0.33** 0.13** 0.27* 0.41*** 0.37*** 0.06 0.37*** 0.10 0.27* 0.07* 0.59***

Personal Sacrifice and Burdenc

Worry and Felt Isolation

1 2

0.44*** 0.61*** 0.21* 0.27*** 0.18*

CESD, Center for Epidemiological Studies Depression Scale; CWBS-BN, Caregiver Well-Being Scale (Basic Needs); CSI, Caregiver Strain Index; MM-CGI, MarwitMeuser Caregiver Grief Inventory; PSSQ-F, Perceived Social Support Questionnaire (Family). *p50.05; **p50.01; ***p50.001. a College degree 1 vs 5college degree 0. b Grave/serious 1 vs moderate/mild 0. c No demographic variable was significant in Step 1.

which were reected in depressed aect, high strain, unmet needs (e.g. companionship, future planning), and less family support. These dierential patterns support the unique characteristics of the MM-CGI as a multi-faceted measure of pre-death grief. More MM-CGI research is needed in cancer caregivers. Future studies would benet from systematic, representative sampling of more caregivers, in contrast to our small convenience sample. Longitudinal studies, unlike our cross-sectional design, would provide validity for MM-CGI grief assessment across illness stages. Larger samples would also allow cross-validation of MM-CGI factor structure [6], precluded here by insucient sample size. Despite limitations, this study provides preliminary, yet compelling, support for MM-CGI use with caregivers of people with cancer, and for continued research on this measure. Conict of interest: None of the authors of this manuscript has any conict of interest}nancial, personal, or otherwise}that could potentially bias this work.
Copyright # 2007 John Wiley & Sons, Ltd.