Trajectories of caregiver depressive symptoms while
providing end-of-life care
Siew Tzuh Tang1*, Guan-Hua Huang2, Yu-Chung Wei2, Wen-Cheng Chang3,4, Jen-Shi Chen3,4and Wen-Chi Chou3 1
Graduate School of Nursing, Chang Gung University, Tao-Yuan, Taiwan
2
Institute of Statistics, National Chiao Tung University, Hsinchu, Taiwan
3
Division of Hematology-Oncology, Chang Gung Memorial Hospital, Tao-Yuan, Taiwan
4
College of Medicine, Chang Gung University, Tao-Yuan, Taiwan *Correspondence to:
Chang Gung University, Graduate School of Nursing, 259 Wen-Hwa 1st Road, Kwei-Shan, Tao-Yuan, Taiwan 333. E-mail: sttang@mail.cgu.edu.tw
Received: 30 September 2012 Revised: 12 May 2013 Accepted: 13 May 2013
Abstract
Objective: The course of caregivers’ depressive symptoms may not be homogenous. This study identified trajectories of depressive symptoms among caregivers providing end-of-life care to cancer patients and profiled the unique characteristics of caregivers within each trajectory.
Methods: Trajectories of depressive symptoms were explored in 447 caregivers who completed the Center for Epidemiological Studies Depression Scale over four periods close to the patient’s death (1–30, 31–90, 91–180, and >180 days). Distinct trajectories were identified by latent class analysis.
Results: Four trajectories were identified as endurance, resilience, moderately symptomatic, and chronically distressed and contained 32.0%, 11.4%, 36.9%, and 19.7% of the sample, respectively. Caregivers in the endurance trajectory were relatively well-adjusted individuals with less education, adequatefinancial support, and ample psychological resources but provided care to older patients with greater symptom distress. They perceived less subjective caregiving burden than caregivers with moderate or chronic depressive symptoms. Caregivers in the resilience trajectory were in a more vulnerable position than those in other trajectories when theyfirst transitioned into the caregiving role because they were more likely to be the patient’s spouse, have greater educational attainment and insufficient finances, provide higher intensity assistance to a younger relative, and have weaker psychological resources. However, they were older, reported greater confidence in caregiving, and perceived less caregiving burden than caregivers in other trajectories. The moderately symptomatic and chronically distressed trajectories were differentiated only by the strength of psychological resources.
Conclusions: Caregivers of terminally ill cancer patients follow distinct depressive-symptom trajectories while providing end-of-life care.
Copyright © 2013 John Wiley & Sons, Ltd.
Introduction
Family caregiving at end-of-life (EOL) is essential for patients [1] and society [2] but is a stressful role, especially when death approaches, with deleterious caregiver conse-quences such as increasing depressive symptoms [3,4] and deteriorating quality of life [5,6]. Such effects extend into bereavement, with adjustment difficulties and increased morbidity and mortality [7,8].
Caregivers respond differently to similar events in caregiving. Individual reactions to the possible stressors of caregiving are explained in a widely used framework, the stress-appraisal/coping model [9–11]. A potentially threatening caregiving event cannot be assumed to be stressful per se but is mediated by a complex web of (i) contextual factors, (ii) objective caregiving demands, (iii) appraisal of the caregiving situation, and (iv) available psychological resources. Contextual factors refer to caregivers’ and care recipients’ characteristics, which they bring into the caregiving experience. Caregiving demands
stem from the nature and magnitude of care provided to
meet the dying relative’s needs. Appraisal of the
caregiving situation focuses on how caregivers judge the adequacy of their caregiving performance and the extent of perceived caregiving burden. Psychological resources refer to caregivers’ inner strength (i.e., personal coping capacity).
To better understand caregivers’ different responses
over the course of EOL caregiving, the trajectories of those responses must be examined. Differentiating the caregiver population into such trajectories will identify more precisely care situations most likely to predispose caregiver subgroups to adverse health consequences and resilience factors that may attenuate negative impacts of caregiving [9–11]. Only three studies investigated distinct depressive-symptom trajectories of family caregivers of adults with chronic diseases [12–14], despite longitudinal changes in caregivers’ depressive symptoms having been
well characterized [15]. Four depressive-symptom
with dementia [12] and caregivers of cancer survivors [13], and two trajectories for caregivers of patients with malignant brain tumors [14]. In these studies, some caregivers had adjusted to the caregiving role, whereas others were characterized by a high-distress trajectory. However, depressive-symptom trajectories have never been explored among caregivers of cancer patients at EOL. Therefore, this study aimed to identify trajectories of caregiver depressive symptoms by levels and patterns
of change throughout the patient’s dying process and to
characterize individuals within each trajectory.
Methods
Design and sample
A convenience sample of caregivers was recruited from March 2005 to October 2009 and followed up through February 2010. Caregivers were included if they met these
criteria: (i) identified by patients as the family member
most involved with their care; (ii) care recipients had terminal, advanced cancer as judged by their physician; (iii)>21 years; and (iv) willing to participate and able to communicate with data collectors.
Procedures
Patients recognized by their primary physicians as terminally ill were referred to data collectors who were bachelor-prepared experienced oncology nurses. These data collectors approached patients to explain the study
and ask permission to invite their identified family
caregiver to participate in the study. Identified caregivers who agreed to participate were interviewed in person when they next visited their hospitalized relative (baseline) and approximately every 2 weeks thereafter (when the patient returned for outpatient visits) until they declined to participate or patient death. Information on
patients’ death was obtained by reviewing medical records
or telephoning caregivers. The study sites’ human
research ethics review committees approved the research protocol. All subjects provided written informed consent.
Measures
All instruments were well established [16–22] and had
been used in previous studies [23,24] to investigate caregiving effects on Taiwanese caregivers of terminally ill cancer patients.
Outcome variable
Caregiver depressive symptoms were assessed by the 20-item Center for Epidemiological Studies Depression Scale (CES-D) [16]. CES-D scores range from 0 to 60, with
scores >16 indicating clinical depression (Cronbach’s
alpha = 0.92 in this study).
Independent variables
Contextual factors included caregivers’ demographic
characteristics (gender, age, education [years], financial
status [sufficient income to make ends meet: yes/no], and
relationship with the patient) and patients’ age and symptom
distress. Caregivers’ perceived patient symptom distress
was measured by the 13-item Symptom Distress Scale [18]. Scores range from 13 to 65; higher scores indicate greater distress (Cronbach’s alpha = 0.84 in this study).
Objective caregiving demands included caregiving intensity and time providing care each day (hours). Caregiving intensity was measured by the amount of assistance provided in personal care, homemaking, transportation, and health care on a 4-point scale (1 = none at all to 4 = a lot) [19]. Scores for caregiving intensity range from 4 to 16 (Cronbach’s alpha = 0.88 in this study). Appraisal of caregiving situation: Caregiving was appraised by caregivers’ perceived confidence in caregiving
and perceived caregiving burden. Caregiving confidence
was measured by three items: caregivers’ perceived
confidence in what to expect while the patient is dying, what to do at time of death, and how to care for the patient at home [20] on a 3-point scale (1 = very confident, 2 = fairly confident, 3 = not at all confident). Total scores range from 1 to 9; higher scores indicate greater lack of caregiving confidence (Cronbach’s alpha = 0.69 in this study).
Subjective caregiving burden was measured by the
Caregiver Reaction Assessment [21], withfive subscales:
impact on schedule, caregiver esteem (whether providing care is meaningful and rewarding), lack of family support,
impact on health, and impact on finances. Total scores
range from 24 to 120. Higher scores represent stronger
negative caregiving impact (Cronbach’s alpha = 0.87 in
this study).
Psychological resources of caregivers were measured
by Antonovsky’s 13-item Sense of Coherence (SOC)
scale [22], representing three SOC attributes: meaningful-ness, comprehensibility, and manageability. Total scores range from 13 to 91; higher scores indicate stronger SOC (Cronbach’s alpha = 0.83 in this study).
We hypothesized that SOC and caregiver confidence
protect caregivers from the stress of caregiving and that patient symptom distress, objective caregiving demands, and subjective caregiving burden are risk factors predisposing caregivers to experiencing more depressive symptoms.
Data analysis
Data were first analyzed for variable distributions. To
explore trajectories of caregivers’ depressive symptoms,
time proximity to the patient’s death was defined as the
period between death and day of interview and was
further categorized as 1–30, 31–90, 91–180, and
To identify latent depressive-symptom trajectories
throughout the patient’s dying process, we computed
latent class analysis with continuous latent class indicators by using the software Mplus (Version 4.2, Muthén & Muthén, Los Angeles, CA). This latent class model performed two analyses simultaneously. The goal of the first analysis was to identify discrete trajectories of family caregiver depressive symptoms. Each estimated trajectory is viewed as comprising a homogenous group of individ-uals sharing a distinct pattern of depressive symptoms over four-time proximities to the patient’s death. As
caregivers’ CES-D scores might be influenced by gender,
age, and education, and patients’ symptom distress, these
factors were adjusted in the analysis to obtain groupings
that reflected more accurately the underlying trajectories
of depressive symptoms.
The second analysis of the latent class model involved a multinomial logistic regression to test predictors of
caregivers’ membership within distinct
depressive-symptom trajectories. We modeled the latent trajectories of depressive symptoms with contextual factors, objective
caregiving demands, appraisals of the caregiving
situation, and psychological resources. Except for the time-invariant independent variables of demographics and patient disease-related characteristics, factors used to predict group membership were calculated by averaging all measurements collected during each time interval.
Decisions about the optimal number of trajectories were based on the lowest value of Akaike’s information criterion [25] and Bayesian information criterion [26]. The appropriate shape of each trajectory group (i.e., linear
and quadratic) was further determined byfitting
polyno-mial regressions of CES-D scores on linear and quadratic
terms of time to the patient’s death, on the basis of the
individuals within the given trajectory group. Decision
on trajectory shape was based on statistical significance
(p< 0.05) of linear and/or quadratic terms of time: the qua-dratic shape if the quaqua-dratic term is significant and the linear shape if the quadratic term is not significant, but the linear term is significant. Overall fit of the final model was judged by the entropy measure, which summarizes the distinguish-ability of identified trajectories. Entropy ranges from 0 to 1, with values close to 1 indicating good overallfit.
Results
From 743 potential participants, 659 family caregivers were recruited. Of 659 caregivers recruited, 144 (21.9%) withdrew from follow-ups for various reasons. Among the remaining 515 caregivers, 472 lost their relative to death during the study. Another 25 caregivers were excluded because of missing information on covariates
identified as potential predictors of membership within
distinct depressive-symptom trajectories. Therefore, the final sample comprised 447 caregivers (Figure 1). These
caregivers and those who did not complete follow-ups did not differ significantly at baseline on independent variables. However, participating caregivers perceived that their ill relative suffered from greater symptom distress (mean SD = 32.4 8.5 vs. 26.6 7.8, p < 0.001),
provided more intense patient assistance (12.7 3.4 vs.
10.5 3.8, p < 0.001), and had more depressive symptoms (20.5 11.2 vs. 15.0 10.5, p < 0.001) than those who did not complete follow-ups.
A majority of caregivers were women (67.57%), with a mean age of 48.89 years (range = 21–85) (Table 1). Most
caregivers were married (82.2%), the patient’s spouse
(45.41%) or an adult child (35.57%), and reported making ends meet (76.29%). Caregivers participated in this
study on average for 83.7 88.5 days (range = 1–368;
median = 50) and were interviewed 6.8 6.1 times
(range = 1–30; median = 5.0).
Among patients, over half (57.8%) were men, with a
mean age of 62.58 years (range = 21–94). Their most
com-mon cancer sites were lung (21.2%), stomach (13.8%),
co-lon–rectum (12.9%), liver (10.6%), pancreas (8.3%), and
breast (5.1%). On average, patients had been diagnosed
with cancer for 23.0 29.1 months (range = 1–248,
median = 13.1) at caregivers’ first interview.
Latent trajectories of caregiver depressive symptoms
To identify the optimal number of discrete trajectories of
depressive symptoms, two to eight trajectories werefitted
in latent class modeling. When the number of trajectories
was less thanfive, the latent class model became unstable
and was difficult to converge because of the model
identifiability problem [27]. The Akaike’s information
criterion value decreased from two to eight trajectories, but the Bayesian information criterion value was lowest at three trajectories (Table 2). Four distinct trajectories of
caregiver depressive symptoms were identified on the
basis of the fit statistics, interpretation of class composi-tion, and model identifiability. The entropy value for four
trajectories was 0.852, indicating good overall model fit.
Box plots in Figure 2 show CES-D scores across the four trajectories of caregiver depressive symptoms at four times.
The trajectory characterized by the lowest CES-D
scores over the dying process was labeled “endurance”
and contained 32.0% of the sample. Polynomial
regression results indicated that the most appropriate shape for the endurance trajectory was quadratic.
Depressive symptoms experienced by “endurance”
care-givers accelerated rapidly as the patient’s death
approached, with mean CES-D scores of 6.95, 8.04,
9.45, and 13.34 for>181, 91–180, 31–90, and 1–30 days
from the patient’s death, respectively (Table 1).
Approximately one-tenth (11.4%) of caregivers’
(Figure 2) and exhibited a curvilinear shape. Caregivers in this trajectory experienced a high level of depressive
symptoms (mean CES-D score = 20.90) when the patient’s
death was>180 days away (Table 1). Depressive symptoms
dropped sharply to around the clinical threshold within
91–180 and 31–90 days before the patient’s death (mean
CES-D scores = 16.45 and 16.25, respectively), then increased to 18.38 in the patient’s last month of life.
The final two trajectories were categorized by high
depressive-symptom levels (Table 1). These two trajectories
were named “moderately symptomatic” and “chronically
distressed” and contained 36.9% and 19.7% of the sample,
respectively (Figure 2). Polynomial regression results indicated that the most appropriate shape for these trajecto-ries was linear. The mean CES-D scores increased steadily from 17.32 and 26.86 at>181 days from the patient’s death to 21.45 and 32.18 within the patient’s last month of life for
the“moderately symptomatic” and “chronically distressed”
trajectories, respectively (Table 1).
Predictors of membership in distinct depressive-symptom trajectories
The results of latent class analysis indicated that caregivers
in the“endurance” trajectory were significantly more likely
than those in other depressive-symptom groups to report
less education, indicate making ends meet financially,
provide EOL care to older patients with greater symptom distress, and have stronger psychological resources
(Table 3). Caregivers in the “endurance” trajectory were
also significantly less likely than those in the “moderately symptomatic” and “chronically distressed” trajectories to perceive a higher degree of caregiving burden.
Caregivers in the “resilience” trajectory were
signifi-cantly more likely than those in the“moderately
symptom-atic” and “chronically distressed” trajectories to be older, to
be the patient’s spouse, and to provide greater intensity
assistance to their terminally ill relative, but were signi
fi-cantly less likely to appraise themselves as lacking con
fi-dence in providing EOL care and perceived less subjective
Eligible caregivers (n=743)
Caregivers recruited (n=659)
Caregivers who continually participated (n=515)
Caregivers who lost their relative to death during the
study (n=472)
Final sample: 447 caregivers
Caregivers providing information by time before patient death (days)
1-30 (n=414) 31-90 (n=283) 91-180 (n=151) >181 (n=69) Caregivers refused to participate
1. Too busy caregiving (n=44)
2. Not interested (n=29) 3. Other reasons (n=11)
Caregivers who withdrew (n=144)
Care recipients still alive (n=43)
Participating caregivers missing information on covariates (n=25)
caregiving burden (Table 3). Furthermore, caregivers in the “resilience” trajectory were significantly more likely than
those in the“moderately symptomatic” trajectories to have
more education, report theirfinancial status as not making
ends meet, provide care to younger patients, and have weaker psychological resources.
The two trajectories categorized by more depressive
symptoms (“moderately symptomatic” and “chronically
distressed”) were similar in nearly every aspect of the care-giving-risk and caregiving-protective factors identified in this study (Table 3), except for psychological resources
(SOC score). With each unit increase in SOC score, the
likelihood of being in the“moderately symptomatic” rather
than in the“chronically distressed” trajectory increased by 1.19 (95% confidence interval: 1.06, 1.35).
Discussion
This study contributes to family caregiving research by adding to the few studies identifying latent classes of depressive symptoms in caregivers of adult patients with
chronic diseases. We identified four distinct
depressive-Table 1. Sample characteristics, independent variable measures, and CES-D scores across depressive-symptom trajectories for family caregivers Variable Unit Total (n = 447) Depressive-symptom trajectory Endurance (n = 143) Resilience (n = 51) Moderately symptomatic (n = 165) Chronically distressed (n = 88) Caregiver characteristics Gender: male % 32.43 37.06 41.18 36.36 12.50 Age M(SD) 48.89(12.64) 52.29(11.95) 56.10(12.08) 44.29(11.84) 47.80(12.13) Education (years) M(SD) 11.81(3.94) 11.66(4.24) 10.98(4.00) 12.68(3.66) 10.91(3.62) Relationship to patient Spouse % 45.41 41.26 82.35 30.91 57.95 Adult child % 35.57 35.66 11.76 47.27 27.27 Other % 19.02 23.08 5.89 21.82 14.78 Financial status: not
making ends meet
% 23.71 4.90 52.94 20.00 44.32
Patient demographics and disease-related variables
Symptom distress (SDS score) M(SD) 32.35(7.30) 33.46(6.76) 29.45(6.37) 32.09(6.79) 32.74(7.80) Age (years) M(SD) 62.58(14.17) 68.10(12.46) 58.78(12.41) 60.49(14.80) 59.71(14.13) Objective caregiving demands
Intensity of assistance M(SD) 12.92(2.83) 12.98(3.30) 13.97(1.71) 12.52(2.78) 12.96(2.45) Time providing care (hours) M(SD) 4.33(1.42) 4.07(1.64) 5.11(1.01) 4.17(1.28) 4.59(1.29) Appraisal of caregiving situation
Lack of confidence in EOL caregiving
M(SD) 6.08(1.36) 5.69(1.27) 5.54(1.56) 6.21(1.16) 6.76(1.40) Subjective caregiving
burden (CRA score)
M(SD) 61.29(10.09) 55.80(9.16) 57.83(6.90) 63.13(8.68) 68.76(9.79) Psychological resources
(SOC score)
M(SD) 63.90(15.23) 76.76(8.63) 57.76(14.69) 63.66(11.07) 47.01(11.36)
CES-D score before patient’s death (days)
1–30 M(SD) 20.49(9.83) 13.34(7.13) 18.38(7.17) 21.45(6.93) 32.18(8.49) 31–90 M(SD) 19.04(9.95) 9.45(5.65) 16.25(6.75) 20.25(6.12) 31.00(8.14) 91–180 M(SD) 18.89(8.93) 8.04(4.90) 16.45(5.39) 18.84(5.20) 29.36(4.52) >180 M(SD) 18.14(9.01) 6.95(5.57) 20.90(6.66) 17.32(7.67) 26.86(4.92)
SDS, Symptom Distress Scale; CRA, Caregiving Reaction Assessment; SOC, Sense of Coherence scale; CES-D, Center for Epidemiological Studies Depression Scale.
Table 2. Fit statistics for model selection
No. of groups 2 3 4 5 6 7 8
Akaike’s information criterion 6356.869 6257.233 6241.34 6191.377 6171.443 6149.649 6116.278 Bayesian information criterion 6538.176 6503.387 6561.34 6585.222 6639.134 6691.187 6731.662
symptom trajectories for family caregivers providing EOL care to terminally ill cancer patients. These distinct
trajec-tories varied significantly in magnitude and patterns of
change, consistent with the heterogeneity of depressive-symptom trajectories reported for caregivers of patients
with chronic diseases [12–14].
Approximately one-third of our caregiver subjects were categorized in the“endurance” trajectory. These care-givers thrived in the caregiving experience, considering that their depressive symptoms never reached the clinical CES-D cutoff score throughout the patient’s dying process. This proportion (32.0%) of caregivers who never acknowl-edged depressive symptoms above the clinical threshold was substantially lower than for caregivers of cancer survivors (70.4%) [13], dementia patients (69.8%) [12], and patients with malignant brain tumors (48.9%) [14], indicating the extraordinarily high stress of EOL caregiving. Consistent with the literature on caregivers in general, caregivers of Taiwanese terminally ill cancer patients in
the “endurance” trajectory comprised relatively
well-adjusted individuals with adequate resources for their
caregiving demands, as indicated by adequate financial
support [10,14,28] and ample personal psychological resources [29,30]. Family caregivers in this trajectory perceived less caregiving burden [15,31,32] than those in
both the “moderately symptomatic” and “chronically
distressed” trajectories. They were also more likely to
provide EOL care to older terminally ill cancer patients [3,28]. Aging is recognized in Chinese culture as a natural process [33], and advanced age may be associated with greater acceptance of the evolution of illness and impending mortality [34]. Therefore, Taiwanese care-givers may experience low-level anticipatory grief over the forthcoming loss of their older relative.
However, our finding regarding caregiver educational
attainment as a predictor of membership in the
“endur-ance” trajectory does not agree with the literature. Greater
educational attainment is generally recognized as
associated with low-stable levels of depressive symptoms [10,28,29] because better educated caregivers are more likely to have an advantaged social status and a sense of control over their lives [35]. However, caregivers in the “endurance” trajectory were less educated, consistent with a previous report that less educated caregivers experienced fewer depressive symptoms [15]. Caregiving tasks may be perceived as less rewarding by better educated caregivers familiar with more intellectual activities, whereas less educated caregivers may derive more positive reinforce-ment from caregiving [36]. Furthermore, devotion to caregiving may deprive better educated caregivers of the success and benefits from status achievement in their non-caregiving roles.
Caregivers in the“endurance” trajectory were also more
likely to provide EOL care to a relative with greater symptom distress, in contrast to the commonly reported conclusion that caregivers of cancer patients with greater symptom distress experience more depressive symptoms [31,37]. Caregivers providing EOL care to relatives with greater symptom distress may require increased support
and assistance from healthcare professionals with
concrete symptom-management tasks. Such assistance
may improve confidence in caregiving, reduce subjective
caregiving burden, and alleviate depressive symptoms. However, further study is needed to validate this hypothesis.
Caregivers in the “resilience” trajectory experienced
transient perturbations, with initially moderate depressive symptoms (at a level lower than only that for the
“chronically distressed” trajectory) followed by a sharp decrease to around the clinical threshold. We called this
trajectory “resilience” to indicate “bouncing back” from
stress rather than immunity from an adverse situation [38]. Consistent with the literature, caregivers in the “resilience” trajectory may be in a more vulnerable
posi-tion than those in other trajectories when they first
transitioned into the caregiving role because they are more likely to be the patient’s spouse [32], have more education [15], be unable to make ends meet [10,28], provide more intense assistance [30,32] to a younger terminally ill relative, [3,28] and have weaker personal psychological
resources [29,30]. However, these “resilience”
care-givers are older [28], possibly facilitating their
transi-tion to the caregiving role. Caregiving often conflicts
with the demands of other social roles, for example,
child-rearing, employment, and goal attainment, which are more relevant and stressful roles for younger
caregivers. Furthermore, caregivers in the “resilience”
trajectory more positively appraised the caregiving situation, as indicated by having greater confidence in providing EOL care [15,30] and perceiving less
subjec-tive caregiving burden. These findings may characterize
their capacity to adapt flexibly to the challenges of
EOL caregiving, allowing them to rapidly bounce back to healthy psychological functioning [39].
Caregivers in trajectories characterized by persistently severe (i.e., chronically distressed) and by moderate (i.e., moderately symptomatic) depressive symptoms were differentiated only by weaker psychological resources
[29,30]. Being embedded in difficult caregiving
circum-stances and lacking the benefits of inner strength (i.e.,
Table 3. Significant predictors of individual depressive-symptom trajectories
Potential predictors
Adjusted odds ratio (95% confidence interval)
Endurance vs. Resilience vs. Moderately symptomatic vs. Chronically distressed Resilience Moderately symptomatic Chronically distressed Moderately symptomatic Chronically distressed Caregiver characteristics Age (years) 0.82 1.05 0.99 1.28* 1.21* 0.94 (0.67, 1.00) (0.98, 1.12) (0.92, 1.08) (1.06, 1.55) (1.02, 1.45) (0.89, 1.01) Education (years) 0.41* 0.69* 0.58* 1.69* 1.41 0.84 (0.23, 0.74) (0.51, 0.94) (0.40, 0.83) (1.03, 2.76) (0.95, 2.11) (0.67, 1.04) Relationship to patient Spouse 0.003* 1.75 0.61 612.78* 214.22* 0.35 (0.2 10 4, 0.34) (0.07, 50.00) (0.02, 20.00) (11.33, 33.1 103) (3.11, 14.7 103) (0.07, 1.72) Financial status: not making ends meet 0.7 10 5* 0.02* 0.006* 3238.94* 894.26 0.28
(0.2 10 8, 0.03) (0.0008, 0.65) (0.0002, 0.23) (1.51, 693 104) (0.57, 139 104) (0.06, 1.19) Patient demographics and disease-related variables
Age (years) 1.35* 1.15 1.22* 0.85* 0.90 1.05 (1.12, 1.64) (1.00, 1.33) (1.05, 1.43) (0.76, 0.96) (0.82, 1.00) (0.99, 1.12) Symptom distress (SDS score) 1.96* 1.43* 1.54* 0.74 0.79 1.08
(1.14, 3.33) (1.14, 1.79) (1.21, 1.96) (0.45, 1.21) (0.84, 1.31) (0.97, 1.19) Objective caregiving demands
Intensity of caregiving 0.21* 0.84 0.83 3.99* 3.91* 0.98 (0.08, 0.57) (0.59, 1.21) (0.55, 1.24) (1.73, 9.17) (1.65, 9.24) (0.79, 1.22) Time spent providing care 0.15 0.94 0.93 6.21 6.09 0.98
(0.02, 1.56) (0.48, 1.85) (0.41, 2.08) (0.65, 59.03) (0.68, 54.18) (0.59, 1.64) Appraisal of caregiving situation
Lack of confidence in EOL caregiving 2.27 0.59 0.37* 0.26* 0.16* 0.63 (0.79, 6.67) (0.28, 1.24) (0.15, 0.93) (0.10, 0.65) (0.07, 0.39) (0.35, 1.15) Subjective caregiving burden
(CRA score) 1.67* 0.79* 0.79* 0.48* 0.48* 1.01 (1.02, 2.70) (0.66, 0.94) (0.67, 0.94) (0.30, 0.76) (0.30, 0.77) (0.94, 1.08) Psychological resources (SOC score) 1.75* 1.35* 1.61* 0.77* 0.92 1.19* (1.35, 2.27) (1.12, 1.61) (1.32, 2.00) (0.65, 0.91) (0.80, 1.06) (1.06, 1.35)
SDS, Symptom Distress Scale; CRA, Caregiving Reaction Assessment; SOC, Sense of Coherence scale. *p< 0.05.
SOC) to find meaning in caregiving or clearly compre-hend and appropriately manage caregiving demands probably contributed to the highest depressive-symptom level.
This study’s strengths include the large sample followed prospectively and frequent assessments of caregiver depressive symptoms over the entire patient dying process. Nevertheless, the generalizability of our findings may be limited by convenience sampling from a specific geographic area and by biases due to losses during follow-up. Our study was also limited by assessing de-pressive symptoms with a questionnaire rather than diagnostic interviews and by not including the potential impact of other factors, such as available social support, upholding cultural values, and expectations of caregiving, on depressive-symptom trajectories. Before caregivers enrolled in our study or shouldered the caregiving role, no reliable assessment of their depressive symptoms was available. This information would likely have strongly predicted depressive symptoms while providing EOL care. Therefore, we cannot rule out the possibility that our observations may be anchored in situations that pre-ceded the decades in which these trajectories took shape. In conclusion, our study demonstrates that caregivers of terminally ill cancer patients do not have uniform emotional responses to EOL caregiving; rather, they follow qualitatively distinct depressive-symptom trajecto-ries. These trajectories are linked to their experiences as
caregivers, particularly their objective caregiving
demands, appraisals of the caregiving situation, and psychological resources. Our results on distinct depres-sive-symptom trajectories and factors associated with
different longitudinal patterns of depressive symptoms have strong clinical and policy implications. Because resources for mental health services are limited, interven-tions should be prioritized to interrupting the two unfavor-able high depressive-symptom trajectories and facilitating rapid return to healthy psychological functioning for care-givers in the“resilience” trajectory. Our predictive models may serve healthcare professionals in developing trajec-tory-specific interventions tailored to risk and protective
factors identified for each depressive-symptom trajectory.
Research is urgently needed on caregiver depressive symptoms beyond patient death to understand how these
four distinct trajectories extend into bereavement [40–
42] and how they impact caregivers’ bereavement adjust-ment. This information would guide clinicians in under-standing and relieving the emotional toll exacted by family caregiving before patient death and throughout bereavement.
Acknowledgements
This study is supported by the National Health Research Institute (NHRI-EX94-9806PC and NHRI-EX99-10106PI) and the National Science Council (NSC 96-2314-B-182-029-MY2 and NSC99- 2628-B-182-031-MY2).
Conflict of interest
No financial or other conflict of interest was involved in
this study. None of the funding sources had any role in designing and conducting the study: collecting, managing, analyzing, and interpreting the data; or preparing, reviewing, or approving the manuscript.
References
1. Hebert RS, Schulz R. Caregiving at the end of life. J Palliat Med 2006;9:1174–1186. 2. Van Houtven CH, Ramsey SD, Hornbrook
MC, Atienza AA, van Ryn M. Economic burden for informal caregivers of lung and colorectal cancer patients. Oncologist 2010;15:883–893.
3. Fletcher BAS, Dodd MJ, Schumacher KL, Miaskowski C. Symptom experience of family caregivers of patients with cancer. Oncol Nurs Forum 2008;35:E23–E44. 4. Pitceathly C, Maguire P. The psychological
impact of cancer on patients’ partners and other key relatives: a review. Eur J Cancer 2003;39:1517–1524.
5. Kitrungrote L, Cohen MZ. Quality of life of family caregivers of patients with cancer: a literature review. Oncol Nurs Forum 2006;33:625–632.
6. Tang ST, Li CY, Chen CCH. Trajectory and determinants of the quality of life of family caregivers of terminally ill cancer patients in Taiwan. Qual Life Res 2008;17:387–395.
7. Christakis NA, Allison PD. Mortality after the hospitalization of a spouse. New Engl J Med 2006;354:719–730.
8. Boyle PJ, Feng Z, Raab GM. Does widowhood increase mortality risk?: Testing for selection effects by comparing causes of spousal death. Epidemiol 2011;22:1–5. 9. Lazarus RS, Folkman S. Stress, Appraisal and
Coping. Springer Co.: New York, 1984. 10. Nandi A, Beard JR, Galea S. Epidemiologic
heterogeneity of common mood and anxiety disorders over the life course in the general population: a systematic review. BMC Psychiatry 2009;9:31.
11. Pearlin LI, Mullan JT, Semple SJ, Skaff MM. Caregiving and the stress process: an overview of concepts and their measures. Gerontologist 1990;30:583–594.
12. Taylor DH, Jr., Ezell M, Kuchibhatla M, stbye T, Clipp EC. Identifying trajectories of depressive symptoms for women caring for their husbands with dementia. J Am Geriatr Soc 2009;56:322–327.
13. Lambert SD, Jones BL, Girgis A, Lecathelinais C. Distressed partners and caregivers do not re-cover easily: adjustment trajectories among
partners and caregivers of cancer survivors. Ann Behav Med 2012;44:225–235.
14. Choi CW, Stone RA, Kim KH, et al. Group-based trajectory modeling of caregiver psychological distress over time. Ann Behav Med 2012;44:73–84.
15. Kurtz ME, Kurtz JC, Given CW, Given BA. Depression and physical health among family caregivers of geriatric patients with cancer– a longitudinal view. Med Sci Monit 2004;10: CR447–CR456.
16. Radloff LS. The CES-D Scale: a self-report depression scale for researching the general population. Appl Psychol Measures 1977:1;385–401.
17. Chien CP, Cheng TA. Depression in Tai-wan: epidemiological survey utilizing CES-D. Bull Jpn Soc Neurol Psychiatr 1985;87:335–338.
18. McCorkle R, Young K. Development of a symptom distress scale. Cancer Nurs 1978;1:373–378.
19. Emanuel EJ, Fairclough DL, Slutsman J, Alpert H, Baldwin D, Emanuel LL. Assistance from family members, friends, paid care givers and volunteers in the care of
terminally ill patients. New Engl J Med 1999;341:956–963.
20. Teno JM, Clarridge B, Casey V. Validation of Toolkit After-Death Bereaved Family Member Interview. J Pain Symptom Manag 2001;22:752–758.
21. Given CW, Given B, Stommel M, Collins C, King S, Franklin S. The Caregiver Reaction Assessment (CRA) for caregivers to persons with chronic physical and mental impair-ments. Res Nurs Health 1992; 15:271–283. 22. Antonovsky A. Unraveling the Mystery of
Health. Jossey-Bass: San Francisco, 1987. 23. Tang ST, Li CY, Liao YC. Factors associated
with depressive distress among Taiwanese family caregivers of cancer patients at the end of life. Palliat Med 2007;21:249–257. 24. Tang ST, Li CY. The important role of sense
of coherence in relation to depressive symp-toms for Taiwanese family caregivers of can-cer patients at the end of life. J Psychosom Res 2008;64:195–203.
25. Akaike H. Factor analysis and AIC. Psychometrika 1987;52:317–332.
26. Schwarz G. Estimating the dimension of a model. Ann Stat 1978;6:461–464.
27. Huang GH. Model identifiability. In Encyclo-pedia of Statistics in Behavioral Science, Everitt BS, Howell DC (eds.), Wiley: New York, 2005; 1249–1251.
28. Covinsky KE, Newcomer R, Fox P, et al. Patient and caregiver characteristics associated with depression in caregivers of
patients with dementia. J Gen Intern Med 2003;18:1006–1014.
29. Chumbler NR, Rittman MR, Wu SS. Associa-tions in sense of coherence and depression in caregivers of stroke survivors across 2 years. J Behav Health Serv Res 2008;35:226–234. 30. Nijboer C, Tempelaar R, Triemstra M,
van den Bos GAM, Sanderman R. The role of social and psychologic resources in caregiving of cancer patients. Cancer 2001;91:1029–1039.
31. Braun M, Mikulincer M, Rydall A, Walsh A, Rodin G. Hidden morbidity in cancer: spouse caregivers. J Clin Oncol 2007;25:4829–4834.
32. Schoenmakers B, Buntinx F, Delepeleire J. Factors determining the impact of care-giving on caregivers of elderly patients with dementia. A systematic literature review. Maturitas 2010;66:191–200.
33. Helman CG. Cultural aspects of time and ageing. Time is not the same in every culture and every circumstance; our views of aging also differ. EMBO Rep 2005;6: S54–S58.
34. Surbone A, Kagawa-Singer M, Terret C, Baider L. The illness trajectory of elderly can-cer patients across cultures: SIOG position pa-per. Ann Oncol 2007;18:633–638.
35. Pearlin LI, Nguyen KB, Schieman S, Milkie MA. The life-course origins of mastery among older people. J Health Soc Behav 2007;48:164–179.
36. Nijboer C, Triemstra M, Tempelaar R, Sanderman R, van den Bos GA. Determinants of caregiving experiences and mental health of partners of cancer patients. Cancer 1999;86:577–588.
37. Given B, Wyatt G, Given C, et al. Burden and depression among caregivers of patients with cancer at the end of life. Oncol Nurs Forum 2004;31:1105–1117.
38. Norris FH, Tracy M, Galea S. Looking for resilience: understanding the longitudinal trajectories of responses to stress. Soc Sci Med 2009;68:2190–2198.
39. Diwan S, Hougham G, Sachs G. Strain experienced by caregivers of dementia patients receiving palliative care: findings from the Palliative Excellence in Alzheimer Care Efforts (PEACE) Program. J Palliat Med 2004;7:797–806.
40. Aneshensel CS, Botticello AL, Yamamoto-Mitani N. When caregiving ends: the course of depressive symptoms after bereavement. J Health Soc Behav 2004;45:422–440. 41. Bonanno GA, Wortman CB, Lehman DR,
et al. Resilience to loss and chronic grief: a prospective study from preloss to 18-months postloss. J Pers Soc Psychol 2002;83:1150–1164.
42. Zhang B, Mitchell SL, Bambauer KZ, Jones R, Prigerson HG. Depressive symptom trajec-tories and associated risks among bereaved Alzheimer disease caregivers. Am J Geriatr Psychiatry 2008;16:145–155.
