depression and grief reactions in hospice caregivers: from pre-death to 1 year afterwards

Journal of Affective Disorders 69 (2002) 53–60www.elsevier.com/ locate/ jad

Research report

Depression and grief reactions in hospice caregivers: from pre-deathto 1 year afterwards

a a b bYulia Chentsova-Dutton , Stephen Shucter , Susan Hutchin , Linda Strause ,c d d d ,*Kathleen Burns , Laura Dunn , Marci Miller , Sidney Zisook

aDepartment of Psychiatry, University of California at San Diego and VA Healthcare Systems, 0603-R, 9500 Gilman Drive, La Jolla,

CA 92093-0603,USAbSan Diego Hospice, San Diego, CA, USA

cStevens Cancer Center, Scripps Hospital, San Diego, CA, USAdUniversity of California at San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0603,USA

Received 22 August 2000; accepted 15 November 2000

Abstract

Background: As the US population ages, more and more individuals will find themselves facing the demanding task ofcaring for terminally ill family members. Yet strikingly little is known about the emotional toll such caregiving exacts fromcaregivers, or how the stresses and strains of caregiving affect later grief reactions. This study examines the emotionaladjustment and grief intensity of bereaved caregivers from their prebereavement (caregiving) baseline through the first yearafter the death and compares the effects of caregiving and subsequent bereavement on spouses and adult children.Methods:Forty-eight adult children and spousal caregivers of hospice patients and 36 controls were evaluated shortly before the deathsof their loved ones and again at 2, 7, and 13 months after their deaths. All subjects were administered the Hamilton RatingScale for Depression, Brief Symptom Inventory, and the Texas Revised Instrument of Grief.Results: Depression and otherindices of psychological distress are highest during the caregiving period and during the first few months after the death,before decreasing over the duration of the first year. Many symptoms of grief remain prominent as long as 13 months afterthe death of a parent or a spouse. There were no differences in intensity of grief, depression or other indices of distressbetween bereaved children and bereaved spouses.Conclusions: The magnitude of the stress of caregiving may beunderestimated. Depression is at least as likely to emerge in the context of caregiving as it is in the postbereavement period.Therapeutic interventions may need to take into consideration the expected distress associated with caregiving and thechronicity of grief reactions.Limitations: The large dropout rate, reliance on self-report ratings and demographicallyhomogeneous sample may limit generalizability of findings. 2002 Elsevier Science B.V. All rights reserved.

Keywords: Hospice; Caregivers; Grief; Depression

*Corresponding author. Tel.:11-858-552-8585; fax:11-858-822-0231.E-mail address: [email protected] (S. Zisook).

0165-0327/02/$ – see front matter 2002 Elsevier Science B.V. All rights reserved.PI I : S0165-0327( 00 )00368-2

54 Y. Chentsova-Dutton et al. / Journal of Affective Disorders 69 (2002) 53–60

1. Introduction groups reporting substantially higher incidences ofdepression than controls. The latter group of inves-

There is abundant literature on grief and bereave- tigators demonstrated that approximately one-fourthment but strikingly less on the prebereavement of bereaved caregivers continued meeting the criteriachallenges of family members facing the impending for Major Depressive Episodes throughout the first 3death of a loved one or on how prebereavement years of bereavement. These figures are somewhatfactors relate to postbereavement adjustment. These higher than those obtained from the general bereavedare important issues in middle and late life as there population that have found rates as high as 16–17%are a growing number of families in this country at 1 year after the loss (Bornstein et al., 1973; Zisookfacing the demanding tasks of caring for a terminally et al., 1997).ill family member (Bliss et al., 1994). It is not clear whether adult children and spouses

Caregivers, in general, are at risk for substantial differ in their adjustment to bereavement. One reportpsychological distress. For example, significant found substantial levels of psychological distress inlevels of depression have been reported in approxi- approximately one-third of spouses and one-fourth ofmately 8–46% of family members caring for in- adult children caregivers (Kissane et al., 1994),dividuals with dementia (Gallagher et al., 1989). while two studies found that the rates of depressionLess well studied have been other groups of caregiv- did not differ significantly for spouse and offspringers; however, caring for a terminally ill relative has caregivers (Gallagher et al., 1989; Bodnar andbeen found to be associated with high levels of stress Kiecolt-Glaser, 1994). Examining the prevalence ofand psychological morbidity (Kissane et al., 1994). ‘core bereavement phenomena’ among parent,The consequences of caregiving can include symp- spouse and adult children caregivers, Middleton et al.toms of depression and anxiety, medical problems, (1998) found that adult children tended to experienceand deteriorating social and occupational function- fewer grief symptoms than spouses. Thus, whethering. We have recently reported that hospice caregiv- adult children or spouses have a more difficult timeers experience higher levels of depression, anxiety, with their grief and bereavement reactions is still ananger, and health problems than demographically open question.matched, non-caring controls. In addition, we found The purpose of this study is to examine thethat adult children and spouse caregivers experienced emotional adjustment and grief intensity amongsimilar levels of psychological and physical morbidi- hospice caregivers from the prebereavement baselinety (Chentsova-Dutton et al., 2000). through the first year after the death. This study also

The damage does not stop when caregiving ends. compares the effects of caregiving and bereavementEven after several months, or in some cases, years, on spouses and adult children of dying patients. Weof providing care for their dying relatives, hospice hypothesize that:caregivers have to face yet another significant stres-sor, the death of that family member. The death of a 1. survivors will report higher levels of depressionloved one is painful and disrupting under any and psychopathology during the first 13 monthscircumstance. But for a hospice caregiver, bereave- of bereavement compared to non-care giving,ment can add to already high levels of chronic non-bereaved controls;distress. There is no consensus yet on how the death 2. surviving spouses will experience more intenseaffects prebereavement levels of pain and suffering. and prolonged grief reactions than adult children.Mullen (1992) and Folkman et al. (1996) have foundsupport for the hypothesis that levels of depressionincrease among recently bereaved caregivers, sub-2. Methodssequently returning to prebereavement levels (whichare elevated in comparison with population norms). 2.1. SubjectsOn the other hand, Bodnar and Kiecolt-Glaser(1994) found that levels of distress do not differ for Adult children and spousal caregivers or ‘spousalbereaved and continuing caregivers, with the two equivalents’ (common law spouses) of hospice pa-

Y. Chentsova-Dutton et al. / Journal of Affective Disorders 69 (2002) 53–60 55

tients were recruited for the study from San Diego from all subjects prior to initiation of all protocol-Hospice. Control subjects were volunteers recruited related activities. Each control subject also gaveinto the study from the community through public written informed consent to participate in this study.relations at San Diego Hospice and the University of Fewer than 10% of interviewed subjects ultimatelyCalifornia at San Diego. They were recruited by refused to participate in the study, usually because ofword of mouth, by asking research and hospice staff uneasiness with research, difficulty making a timeto inform friends and neighbors of the study, and by commitment, or the wish that the study was linkedposting notices at the Hospice and UCSD Medical more closely to treatment.Center. The control subjects were included into the The initial session (T1) included a structuredstudy only if they had not experienced the death of a interview and various self-administered scales, in-first-degree relative or a friend within the past 2 cluding the HRSD and BSI. The visit was to takeyears. All subjects were excluded from the study if place no more than 2 weeks before the expectedthey were younger than 18 years of age, unable to death; a new assessment was administered andcommunicate in English, or unwilling to sign an considered T1 if the dying relative lived more than 1informed consent form. No other exclusions for month. The interview and self-report questionnaires,caregivers or controls were utilized. along with the TRIG for survivors, were adminis-

tered again at 2 (T2), 7 (T3), and 13 months (T4)2.2. Materials after the death of a relative. Similarly, for controls,

T2 always occurred 8–10 weeks after T1. These timeThe following instruments were chosen for assess- points were selected for several reasons. We utilized

ment of depression, psychological symptomatology 2 months for T2 so that data would be comparable toand grief: other bereavement studies (Zisook et al., 1997;

Zisook and Shucter, 1993) and because the American1. a self-administered demographical bereavement Psychiatric Association’s Diagnostic and Statistical

questionnaire; Manual considers 2 months the cut-off between2. the 28-item Hamilton Depression Rating Scale normal grief (i.e., ‘bereavement’) and Major Depres-

(HDRS) (Williams, 1988) administered by sion. Seven months was selected because it fallstrained research assistants; between 2 and 13 months; 13 months was selected

3. the 53-item Self-Rated Brief Symptom Inventory rather than 1 year to avoid anniversary reactions.(BSI) (Derogatis and Spencer, 1982);

4. the self-administered 2-scale, Texas Revised In-ventory of Grief (TRIG) (Faschingbauer et al., 3. Results1987).

3.1. Demographics2.3. Design and procedures

One hundred and twelve caregivers were admittedThe study was approved by the San Diego Hos- into the study initially. Of the initial caregiver

pice Internal Review Board (IRB). All families sample, 63 (56%) dropped out of the study beforeaccepted into the hospice program were invited to T4. The rates of major depressive disorder and theparticipate in this study. A social worker or hospice levels of depressive symptoms did not differ sig-nurse explained the study to the hospice caregivers at nificantly for caregiver completers and noncomple-an initial home visit and received written permission ters. Nor were completers and noncompleters demo-from the caregivers to be contacted by a member of graphically different. Similarly, 47 (67%) controlsthe research team. Prior to contacting the caregiver, dropped out before T4. Forty-eight caregivers and 36research assistants obtained additional verbal consent controls completed the interview and self-reportfrom the hospice nurses or social workers. questionnaires at 2 (T2), 7 (T3), and 13 months (T4)

Subjects were interviewed in their own homes. after the death of a relative.Witnessed informed consent forms were obtained The completer caregiver sample included 18


(38%) adult child caregivers and 30 (62%) spouse controls (F(1, 56)5 11.20, P , 0.01). The time bycaregivers. There were 11 (31%) adult children and group interaction was significant (F(3,56)59.4, P,

25 (69%) spouses among the completer controls. 0.01). Caregivers reported significantly higher levelsDemographic characteristics of the two samples are of depression than controls at T1 (F(74)519.98,presented in Table 1. Caregiver and control groups P,0.01) and T2 (F(72)532.00,P,0.01), but not atwere similar in most respects. However, caregivers T3 and T4. No main effect of family role and notended to be older than controls (t 5 2.57,P , 0.05). interaction between the caregiver-control and familyAs a group, caregivers tended to be less educated role factors were detected at any time point.than controls (t 5 3.1, P ,0.01). When the course of depressive symptoms was

Repeated measures analysis of covariance was examined for caregivers only, the HRSD followed aused in comparing groups to adjust for differences in curvilinear pattern (F(3,117)512.09, P,0.01).age and education between caregivers and controls, Caregivers’ depression did not change significantlyand the difference in ages between spouses and adult between the T1 and T2, but reduced from T2 to T3children. The assumptions of ANCOVA, including (F(1,38)515.45, P,0.01), and from T3 to T4normality, homogeneity of variance, linearity be- (F(1,38)512.00,P,0.01).tween covariates, homogeneity of regression slopes, Twenty-three percent of caregivers had HRSDabsence of multicolinearity between covariates, and scores greater than or equal to 12 (indicating at leastreliability of covariates were adequately met. mild depression), as compared with none of the

controls at T1. The percentage of subjects withHRSD.12 declined to 17% of caregivers (as com-

3.2. Depression pared to none of the controls) at T2, 6% of caregiv-ers and 2% of controls at T3, and 2% of caregivers

2The covariates for this analysis of the HRSD versus none of the controls at T4. Thex values areincluded age and education. The independent vari- not reported for T2–T4 due to cells with few or noable was caregiver-control status. The means for counts.each time are presented in Table 2. Overall, caregiv-ers reported higher levels of depression than did

3.3. Psychological symptomatology

Because few controls completed the BSI ques-tionnaire at T4, only T1–T3 were included in theTable 1following analyses. We conducted two repeatedDemographic characteristics of caregivers and controlsmeasures ANCOVAs, controlling for age and educa-

Demographic characteristics Grouption, with global BSI indices as dependent variables

Caregivers Controls and the caregiver-control groups as a factor. These(n548) (n536)

analyses revealed that caregivers’ mean scores wereGender, male (%) 17 25 significantly higher than those of controls over timeAge 63* 56* on both global indices of psychological symp-Religious preference (%)

tomatology: the Positive Symptom Distress IndexProtestant 58 72(PSDI) (F(1,68)516.17,P,0.01), and the GeneralCatholic 19 11

Jewish 4 3 Severity Index (GSI) (F(1,77)516.88, P,0.01).Other 6 3 When the levels of caregivers’ and controls’ symp-None 13 11 tomatology were compared on the BSI subscales

Race (%)using nine separate repeated measures ANCOVACaucasian 96 100analyses, caregivers’ mean scores were significantlyAfrican-American 2 0

Hispanic 2 0 higher than those of controls over time on theYears of education 14** 16** subscales of anxiety (F(1,79)515.05, P,0.01),

*P,0.05; **P,0.01. hostility (F(1,79)57.35, P,0.01), and depression


Table 2Levels of depression and TRIG scores over time

T1 T2 T3 T4

DepressionMean HRSDBereaved (n548)** 8 8 7 4Controls (n536) 3 3 4 4% HRSD.12Bereaved (n548) 23 17 6 2Controls (n536) 0 0 2 0

Grief (TRIG)TRIG Total Scores, past behaviors (n548) 17 17 16TRIG Total Scores, present feelings (n548) 41 39 38Percentage endorsing specific feelings (n548)Still cry when thinking of the person** 47 33 17Get upset when thinking of the person** 38 24 19Very much miss the person who died 69 80 85Painful to recall memories 36 24 28Preoccupied with thoughts about the person* 41 39 27No one will ever take the place in my life 84 72 66Things and people remind of the person 73 65 73At times I feel the need to cry 69 54 42Functioning about as well as before the death 71 93 88

Note: *P,0.05; **P,0.01, denotes significant changes over time.

(F(1,78)527.77, P,0.01). Caregivers and controls child and spouse caregivers did not differ over timedid not differ significantly over time on the dimen- on the overall levels of psychopathology.sions of somatization, interpersonal sensitivity,phobic anxiety, and paranoid ideation. For the sub-scales of anxiety (F(2,158)53.40, P,0.05) and 3.4. Griefhostility (F(2,158)53.67,P,0.05), significant inter-actions of caregiver-control group and time were As measured by the TRIG, the bereaved caregiv-found. Caregivers’ anxiety and hostility levels di- ers’ reports of past grieving did not significantlyminished over time, whereas controls’ levels of change over time. When the differences betweenanxiety remained relatively stable and low. adult children and spouses and male and female

The time course of psychological symptomatology caregivers were examined, there was no main effectas represented by the BSI subscales and global of family role or gender and no interactions. On theindices was examined separately for caregivers. Of other hand, scores on the subscale of present feelingsthe nine subscales, six (somatization, psychoticism, significantly diminished from T2 to T4 (F(2,58)5paranoid ideation, phobias, interpersonal sensitivity, 5.97,P,0.05). Contrasts showed that the declinesand depression) did not change significantly over from T2 to T3 (F(1,39)56.15, P,0.05) and T3 totime. The levels of anxiety (F(3,135)511.4, P, T4 (F(1,39)54.27,P,0.05) were significant. There0.01), obsessive–compulsive symptoms (F(3,156)5 was no main effect of family role and no interaction.4.6, P,0.01) and hostility (F(3,135)53.9, P,0.05) Males tended to report higher levels of presentsignificantly diminished over time. The GSI global feelings than females at T2–T4; however, this trendindex also showed linear decline over time did not reach significance (F(1,28)53.8, P50.06).(F(3,132)53.03,P,0.05). The main effect of fami- Responses to selected TRIG items are presented inly role was examined for the global indices; adult Table 2.


4. Discussion sample and setting. For example, the deceased wererelatively older than in most previous samples. The

This longitudinal study examined pre- and post- terminal illnesses were of a chronic nature, thebereavement adjustment among a cohort of hospice deaths were anticipated, and most individuals feltcaregivers. The results do not support the hypothesis good about their efforts to help provide care duringthat the death of a family member further increases their parent or spouse’s final weeks. In addition,caregivers’ distress. We found that, across-the-board, Hospice provided a supportive milieu and an oppor-bereaved caregivers’ levels of psychological symp- tunity for both pre- and postbereavement counseling.tomatology at 2 months after the death of a patient Thus, the sample may be at lower risk for untowardwere equal to or less than those reported by the same complications of grief than most other populations,caregivers prior to bereavement. Caregivers’ adjust- accounting for the relatively mild and transient levelsments improved significantly from baseline to 1 year of depressive symptoms most individuals ex-after the deaths of their patients. Furthermore, psy- perienced.chological distress in bereaved caregivers returned to Exploring psychological adjustment beyond de-population control levels by 7 months after the death. pression, we found that global indices of psycho-

Levels of depression were higher for caregivers pathology (BSI) and subscales of anxiety, psycho-than for population controls for the time period of ticism, obsessive–compulsive symptoms and hostili-prebereavement through 2 months post-bereavement. ty also were elevated for hospice caregivers asHowever, by 7 months after the death, levels of compared to controls. However, as with measures ofcaregivers’ depression subsided to equal those of depression, none of the subscales showed an increasepopulation controls. Only 2% of caregivers con- after the death of a relative. One of the global indicestinued reporting mild symptoms of depression at 13 (GSI) as well as measures of anxiety, obsessive–months after the death. While average levels of compulsive symptoms and hostility decreased overcaregivers’ depressions were mild, the literature time, while the PSDI and BSI measures of psycho-suggests that even mild to moderate depressive ticism and depression remained stable throughout thesymptomatology can be debilitating, impacting psy- study. These findings provide some evidence, albeit,chosocial functioning and quality of life (Well et al., limited, that distress in caregivers may persist for1989). The comparisons between prebereavement some individuals throughout at least the first year ofand postbereavement (2 months) levels of depression bereavement.revealed that the death of a family member was not The intensity of grief feelings was highest at 2associated with a further increase in depressive months and lowest at 1 year after the death of asymptomatology. Therefore, unlike Mullen (1992) relative. The items most commonly endorsed byand Folkman et al. (1996), we found no support for caregivers included believing that no one will takethe hypothesis that the death of a relative further the place of the deceased person, noticing that thingsincreases caregivers’ depressions. If such an increase and people remind them of the deceased person,does happen, it is brief and temporary, occurring feeling a need to cry, and missing the person whoaround the time of death and disappearing by 2 died ‘very much’. While many symptoms (e.g.,months after bereavement. crying when thinking about the person, getting upset

The longitudinal data on depression is different when thinking about the person and feeling pre-than previously reported in other bereaved popula- occupied with thoughts about the person) diminishedtions (Bornstein et al., 1973; Zisook et al., 1997). significantly from 2 to 13 months after the death,While the percentage of subjects with depression at 2 other symptoms (e.g., being reminded of the de-months is similar to previous reports, most other ceased person by things and people and missing thestudies have reported more sustained levels of de- person) remained stable through the first year ofpression throughout at least the first year after the bereavement. Again, this data supports the notiondeath of a loved one (Zisook et al., 1997; Zisook and that the intensity of grief decreases over time but thatShucter, 1993). The reason for this apparent dis- certain aspects of grief remain an ongoing source ofcrepancy may be related to characteristics of the pain for many, if not most, individuals over time.


This study supports previous findings (Kissane et to subjects’ evolving circumstances and to recruit aal., 1994; Folkman et al., 1996) that spouses and demographically diverse group of subjects and con-adult children caregivers do not differ in their pre- or trols.postbereavement levels of psychopathology or grief. In summary, this study found that depression andContrary to our expectations, as well as the pub- psychological distress peak before and shortly afterlished findings of some others (Middleton et al., the death of a loved one; and that most, but not all,1998) our data show that caring for and losing a symptoms of grief tend to subside in intensity duringparents appears to be every bit as debilitating as the the first year of bereavement. This study also lendstrauma experienced by bereaved spouse caregivers. support to the notion that adult children are as likelyPerhaps the predeath caregiving experiences bring as spouses to experience symptoms of depression,terminally ill parents and their children caregivers anxiety, stress, and grief. The next logical studycloser together, making the loss more painful than would be approaches at therapeutic interventions tootherwise expected. It is also possible that the child help attenuate some of the more painful aspects ofclosest to the dying parent is more likely to assume both caregiving and bereavement. Given the highcaring responsibilities when the parent becomes ill. levels of distress associated with caregiving, it isIn any case, adult children were every bit as likely to likely that such interventions should begin earliergrieve intensely as were spouses. than heretofore appreciated.

There are a number of design limitations that canpotentially compromise the validity of the presentedresults. These limitations include a small sample size Acknowledgementsdue to the high attrition rate of subjects and controlsby the end of the study, ethnic homogeneity, age and All research conducted through the Center foreducational differences between caregiver and con- Palliative Studies is made possible through a gener-trol groups, and excessive reliance on self-report ous grant ([901152) from the W.M. Keck Founda-measurements. In particular, the dropout rate was tion.much higher than anticipated. Subjects often ex-pressed discomfort anticipating post-death inter-views; this, coupled with the staff’s attitude of being Referencesextra cautious to be non-intrusive and sensitive to thesurvivor’s comfort, played a role. In addition, many

Bliss, M.E., Watson, P.E, Given, B., 1994. Health care reform:of the caregivers returned to work and re-engaged in positioning nursing for survival and success. In: The 1994other roles and obligations, finding less time to Annual Congress-Symposia Highlights. Philadelphia, Medical

Associations and Oncology Nursing Society, pp. 10–15.participate in research. Even though the completerBodnar, J.C., Kiecolt-Glaser, J.K., 1994. Caregiver depressionand drop out groups of caregivers did not differ at

after bereavement: chronic stress isn’t over when it’s over.baseline, it is possible that subjects who were morePsychol. Aging 9 (3), 372–380.

healthy and less depressed were less likely to drop Bornstein, P.E., Clayton, P.J., Halcias, J.A., Maurice, W.L.,out of the study. Also, despite the efforts of the San Robins, E., 1973. The depression of widowhood after thirteen

months. Br. J. Psychiatry 122, 561–566.Diego Hospice staff to enroll a diverse group ofChentsova-Dutton, T., Shucter, S., Hutchin, S., Strause, L., Burns,subjects, the subjects in this study were almost

K., Zisook, S., 2000. The psychological and physical health ofexclusively Caucasian, mirroring the demography ofhospice caregivers. Ann. Clin. Psychiatry 12 (1), 19–27.

San Diego Hospice clients, where the study was Derogatis, L.R., Spencer, P.M., 1982. In: BSI Administration anddone. For controls, the sites of recruitment (Hospice Procedures Manual. Clinical Psychometric Research, Balti-and UCSD) yield primarily Caucasian volunteers more, MD.

Faschingbauer, T.R., Zisook, S., DeVaul, R., 1987. The Texasunless special efforts are made to reach out toRevised Inventory of Grief. In: Zisook, S. (Ed.), Biopsychoso-minority groups. It would benefit future investiga-cial Aspects of Bereavement. American Psychiatric Press,

tions to use more clinician-rated measurements to Washington, DC, pp. 109–128.validate subjects’ self-reports, to improve study Folkman, S., Chesney, M., Collette, L., Boccellari, A., Cooke, M.,retention rates through close follow-up and attending 1996. Postbereavement depressive mood and its bereavement


predictors in HIV1 and HIV2 men. J. Pers. Soc. Psychol. 70 Well, K.B., Stewart, A., Hays, R.D., Burnam, M.A., Roger, W.,(2), 336–348. Daniels, M., Berry, S., Greenfield, S., Ware, J., 1989. The

Gallagher, D., Rose, J., Rivera, P., Lovett, S., Thompson, L.W., functioning and well being of depressed patients. Results from1989. Prevalence of depression in family caregivers. Geron- the Medical Outcomes study. J. Am. Med. Assoc. 262, 914–tologist 29 (4), 449–456. 919.

Kissane, D.W., Bloch, S., Burns, W.I., 1994. Psychological mor- Williams, J.B.W., 1988. A structured guide for the Hamiltonbidity in the families of patients with cancer. Psychooncology Depression Rating Scale. Arch. Gen. Psychatry 45, 742–747.3, 47–56. Zisook, S., Paulus, M., Shucter, S.R., Judd, L.L., 1997. The many

Middleton, W., Raphael, B., Burnett, P., Martinek, N., 1998. A faces of depression following spousal bereavement. J. Affect.longitudinal study comparing bereavement phenomena in re- Disord. 45, 85–95.cently bereaved spouses, adult children and parents. Austr. NZ Zisook, S., Shucter, S.R., 1993. Major depression associated withJ. Psychiatry 32, 235–241. widowhood. Am. J. Geriatr. Psychiatry 1, 316–326.

Mullen, J.T., 1992. The bereaved caregiver: a prospective study ofchanges in well being. Gerontologist 32 (5), 673–683.

depression and grief reactions in hospice caregivers: from pre-death to 1 year afterwards

Documents