Longitudinal Changes in and Modifiable Predictors of the Prevalence of Severe Depressive Symptoms for Family Caregivers of Terminally Ill Cancer Patients over the First Two Years of Bereavement

Abstract

Background:

Bereaved families endure tremendous grief. However, few studies have longitudinally investigated caregivers' bereavement grief for more than one year postloss and none is from family-oriented Asian countries.

Objectives:

We explored longitudinal changes in and modifiable predictors of severe depressive symptoms for Taiwanese family caregivers of terminally ill cancer patients over the first two years postloss.

Design:

For this descriptive, longitudinal study, severe depressive symptoms (Center for Epidemiological Studies Depression Scale score ≥16) were measured 1, 3, 6, 13, 18, and 24 months postloss. To identify modifiable factors (preloss depressive symptom levels, subjective caregiving burden, objective caregiving load, preparedness for patient death, and postloss social support) associated with postloss severe depressive symptoms, we used multivariate logistic regression modeling with the generalized estimating equation.

Setting/Participants:

Bereaved caregivers (N = 285) were recruited by convenience from a medical center in northern Taiwan.

Results:

Bereaved family caregivers' prevalence of severe depressive symptoms peaked one month postloss (73.3%) and significantly decreased to 15.2% 24 months after the patient's death. After adjusting for confounders, bereaved caregivers were more likely to have severe depressive symptoms if they had heavy objective caregiving load and higher preloss depressive symptom levels. However, their likelihood of severe depressive symptoms was buffered by being better prepared for the patients' death and having greater social support.

Conclusions:

Bereaved family caregivers' severe depressive symptoms decreased significantly over the first two years postloss. Healthcare professionals should appropriately assess at-risk bereaved caregivers when patients are still alive and provide effective interventions to facilitate caregivers' return to normal life.

Introduction

Family members in Asian cultures feel obligated by filial piety to assume caregiving roles sometimes without being prepared,¹ which can lead to bodily and emotional distress. This distress does not end with the relative's death, which initiates a grieving process with physiological, emotional, social, and spiritual suffering.² Although most bereaved caregivers gradually adjust to the loss,³ some cannot cope with their relative's death and continue suffering from depression/depressive symptoms. Indeed, the prevalence of bereaved families' severe depressive symptoms or a major depressive disorder in longtidinal studies was 35%–59% one month postloss,^3–6 9%–47% two to four months postloss,^3,7–9 5.7%–42% six to nine months postloss,^3,6–11 6.9%–55% after one year,^{3–5,8,9,12,13} and 14%–16% after two years.^5,10,11 Severe postloss depressive symptoms and their consequences^2,5 excessively burden both families and society.^14,15

Postbereavement psychological adjustment is proposed to be affected by caregiving situation, death circumstances, and intrapersonal as well as interpersonal factors.¹⁶ Modifiable factors include caregiving burden, preloss depressive symptom levels, preparedness for patient death, and postloss social support, but evidence for their influence is conflicting. Bereaved caregivers' severe depressive symptom prevalence was reported to (1) decrease,^3,17 increase,^13,18 or remain unaffected³ by greater subjective/objective caregiving burden; (2) increase^3,11–13 or remain unaffected⁴ by higher preloss depressive symptom levels; (3) decrease^7,19,20 or remain unaffected^7,21 by better preparedness for the patient's death; and (4) decrease,^3,18 increase,²² or remain unaffected¹² by social support. The inconclusive effects of these factors on severe postloss depressive symptoms make clinical applications difficult, highlighting the importance of clarifying the roles of these modifiable predictors.

Most longitudinal studies on bereaved caregivers' severe depressive symptoms were conducted over a decade ago^{5–8,10,11,17,18} and may no longer reflect present-day greater use of palliative care²³ that emphasizes facilitating family caregivers' adjustment to losing a loved one throughout bereavement.²⁴ Furthermore, in some studies^6,12,13,21 data were collected at preloss and only once postloss, precluding the ability to determine longitudinal changes in severe depressive symptoms at different times during the bereavement process. Most longitudinal studies followed bereaved caregivers for only 12–13 months postloss.^{3–5,7,8,12,13} However, 10%–20% of bereaved caregivers with severe depressive symptoms did not return to their normal/preloss levels during the first year of bereavement.²⁵ Bereaved caregivers should ideally be followed for at least two years postloss to differentiate between those with chronic grief and chronic depression.²⁶ Furthermore, caregivers in both Taiwan³ and the United States²⁷ are primarily middle-aged women (65%–69.4%, mean age = 47–55 years), and patients' spouses (45.7%–66%) or adult children (17%–37.6%). However, caregivers in previous studies on severe postloss depressive symptoms were primarily patients' spouses^{4–6,10,11,21} and the elderly.^{4–7,10–12,18,19,20}

Furthermore, bereavement grief reactions may differ between Asian and Western cultures.²⁸ For example, compared to bereaved American caregivers, bereaved Chinese caregivers experience higher psychological distress, which decreases rapidly after 18 months postloss.²⁸ However, most studies on severe postloss depressive symptoms were conducted in the United States^{4–13,17–20} with only three^3,29,30 from Asian countries. The single, longitudinal Asian study³ only collected data for one year postloss. Therefore, we conducted this longitudinal study to explore changes in and modifiable predictors of the prevalence of severe postloss depressive symptoms in family caregivers of terminally ill Taiwanese cancer patients over their first two years of bereavement.

Methods

Study design and sample

For this descriptive, longitudinal study, a convenience sample of Taiwanese family caregivers of terminally ill cancer patients was recruited from March 2009 to December 2012 and followed through July 2015. Adult cancer patients were referred by their oncologist who declared them terminally ill when their disease continued to progress and was unresponsive to curative treatments. Caregivers identified by patients as most involved with their care without payment and who signed informed consent to participate were interviewed in person by experienced oncology nurses. The study site's ethics committee approved the study.

Measures

Outcome variable

Caregivers' severe depressive symptoms were measured with the 20-item Center for Epidemiological Studies Depression Scale (CES-D),³¹ the most commonly used instrument for measuring bereaved caregivers' depressive symptoms.^{3,6,8,12,19,20} The CES-D has four subscales: positive emotions, depressive emotions, physical activities, and social difficulties. Each item (symptom) is rated for its frequency during the past week on a scale from 0 to 3. Total scores range from 0 to 60, with scores ≥16 indicating severe depressive symptoms. The CES-D has high internal consistency (Cronbach's alpha = 0.90); 4- and 12-week test–retest reliabilities of 0.67 and 0.32, respectively; and adequate construct validity.³¹ The Chinese version CES-D was translated by two psychiatrists,³² who established the scale's internal consistency, test–retest reliability, and validity of case identification.

Independent variables

Subjective caregiving burden was measured by the 24-item Caregiver Reaction Assessment (CRA) scale³³ with five subscales: impact on schedule, health, and finances, caregiver esteem in providing care (rewarding or causing resentment), and lack of family support. Items are rated on a 5-point scale from strongly agree to strongly disagree. Total scores range from 24 to 120. Higher scores represent stronger negative caregiving impact. Psychometric properties of the Chinese version CRA have not been formally tested in Taiwan, but its scores in family caregivers of terminally ill cancer patients were negatively associated with scores for quality of life³⁴ and social support.³⁵ Cronbach's alpha for the CRA in this study was 0.89.

Objective caregiving load included care tasks, level of care, and time spent in caregiving. Care tasks and level of care were determined by rating the amount of assistance provided in personal care, homemaking, transportation, and healthcare on a 4-point scale from none at all to a lot.³⁶ Time spent in caregiving was measured on a 6-point scale for <2, 3–5, 6–8, 9–12, 13–16, and 17–24 hours/day. A composite score was calculated for these five items. Scores range from 5 to 22; higher scores indicate greater objective caregiving load.

Social support was measured with the 19-item Medical Outcomes Study Social Support Survey (MOS-SSS),³⁷ which measures emotional, informational, tangible, and affectionate support, as well as positive social interaction. Each item is scored from 1–5, with higher scores indicating a better support system. Total scores are computed, and raw scale scores are transformed to a scale of 0 to 100. The Taiwanese version MOS-SSS was shown to have a two-factor structure and good construct validity by its scores being positively associated with Spiritual Well-Being Scale scores.³⁸ Cronbach's alphas in this study were 0.94–0.96.

Caregivers' preparedness for the patients' death was measured by a single item: “How prepared were you for your relative's death?” The item is assessed on a 7-point scale from not at all prepared to very well prepared.

Confounding factors

To determine the unique associations of modifiable independent variables with the likelihood of severe depressive symptoms, we controlled for demographic characteristics (e.g., age, gender, and educational attainment), financial status, and caregivers' relationship to the patient (i.e., spouse, child, and other) as suggested.¹⁶

Data collection

Data on preloss depressive symptoms, subjective caregiving burden, and objective caregiving load were collected approximately every two weeks postenrollment while the patient was alive. Preparedness for the patients' death was assessed only at one month postloss. Postloss depressive symptoms and social support were measured 1, 3, 6, 13, 18, and 24 months postloss to reflect the most stressful times of the bereavement experience.^4,7,10 The 13-month data collection time was chosen to avoid an anniversary effect on caregivers' grief reactions. To evaluate the association of independent variables with severe postloss depressive symptoms, however, the only data used for analyses were preloss caregiving experiences and depressive symptoms measured just before patient death,³⁹ depressive symptoms and social support at 1, 3, 6, 13, 18, and 24 months postloss, and preparedness for the patient's death.

Data analysis

Data were first descriptively analyzed to check distributions. Baseline characteristics of caregivers who did and did not participate in postloss follow-ups were compared by chi-square and t tests. Bereaved caregiver participants who withdrew before and completed postloss surveys were compared at each bereavement assessment for the likelihood of severe depressive symptoms before withdrawal.

Changes in the prevalence of severe postloss depressive symptoms were examined using a simple logistic regression model with the generalized estimating equation (GEE).⁴⁰ Factors associated with the likelihood of caregivers' severe postloss depressive symptoms were identified by multivariate logistic regression modeling with the GEE. The GEE uses robust standard error estimates to account for within-subject correlations of severe depressive symptoms during the follow-up period and to allow the timing and number of repeated assessments to differ across participants. Missing data were not imputed in our study. The GEE assumes data are missing due to covariate dependence.^41,42 In other words, missingness is explained by observed model covariates, for example, caregiving burden, preloss depressive symptoms, and differences between caregivers who completed and witdrew before postloss surveys. These covariates are also hypothesized to be associated with participants' severe depressive symptoms in their first two years of bereavement. Under the assumption of covariate-dependent missingness, the GEE method handles missing data by list-wise deletion (or complete case analysis), thus excluding subjects with missing data from the analysis.^41,42 The effect of individual explanatory variables on the outcome variable was measured by adjusted odds ratio (AOR) with 95% confidence intervals (CI).

Results

Participants' characteristics

Of 392 caregivers recruited, 387 were bereaved by the patient's death during the study. Of these bereaved caregivers, 344 completed the last survey before the patient's death (88.9% participation rate), 59 withdrew from the study postloss, and 22 missed the first postloss assessment only, for 82.9% follow-up for bereavement assessments (Fig. 1). The primary reasons for bereaved caregivers declining participation in postloss surveys were difficulty adjusting to postloss grief and being reminded of the deceased. Caregivers who completed and declined postloss surveys were similar in demographic characteristics and independent variables (data not shown) except for three variables. Those who completed postloss surveys had heavier objective caregiving load (16.82 vs. 14.36, p < 0.01) and subjective caregiving burden (66.26 vs. 62.02, p = 0.02), higher depressive symptom levels (22.80 vs. 19.37, p = 0.04), and perceived less financial sufficiency (73.0% vs. 83.1%, p = 0.04) than caregivers who declined participation.

FIG. 1.

Study process and times of data collection.

For demographic characteristics of the 285 bereaved caregivers who completed follow-up surveys (including 22 bereaved caregivers who missed the one-month postloss data collection but returned for later assessments; Fig. 1), see Table 1. At 1, 3, 6, 13, 18, and 24 months postloss, 263, 246, 240, 224, 203, and 190 caregivers, respectively, completed follow-up surveys, with participation rates of 82.9%, 86.3%, 84.2%, 78.6%, 71.2%, and 66.7%, respectively (Fig. 1). Moreover, the 18- and 24-month postloss data collections were not due at the end of data collection for 10 and 15 bereaved caregivers, respectively. Caregivers who completed and withdrew from the study did not differ significantly at any time in the last prevalence of severe depressive symptoms before the latter caregivers withdrew (Table 2).

Table 1.

Participants' Baseline Characteristics

Characteristic	n (%)
Gender
Male	109 (38.2)
Female	176 (61.8)
Married
Yes	226 (79.3)
No	59 (20.7)
Relationship with patient
Spouse	150 (52.6)
Other	134 (47.4)
Educational level
≤ Junior high school	114 (40.0)
High school	103 (36.1)
≥ College degree	68 (23.9)
Financially sufficient
Yes	208 (73.0)
No	55 (19.3)
Refused to answer	22 (7.7)
Comorbidity
No	190 (66.7)
Yes	95 (33.3)
Age (years)
Mean (SD)	48.33 (11.99)
Range	20–77
Subjective caregiving burden
Mean (SD)	66.26 (12.28)
Range	38–104
Objective caregiving load
Mean (SD)	16.82 (4.83)
Range	5–22
Preloss CES-D score
Mean (SD)	22.78 (12.27)
Range	0–54

CES-D, Center for Epidemiological Studies Depression Scale; SD, standard deviation.

Table 2.

Prevalence of Severe Depressive Symptoms for Bereaved Caregivers Who Completed and Withdrew from the Study at Diffrent Postloss Times

Time after loss, months	Severe depressive symptoms ^a	Completed %	Withdrew %	χ²	p
3				0.03	0.86
	Yes	72.6	70.6
	No	27.4	29.4
6				3.05	0.08
	Yes	50.8	25.0
	No	49.2	75.0
13				<0.01	0.99
	Yes	40.1	40.0
	No	59.9	60.0
18				0.86	0.35
	Yes	24.0	36.4
	No	76.0	63.6
24				0.18	0.67
	Yes	17.3	20.8
	No	82.7	79.2

Severe depressive symptoms were measured by Center for Epidemiological Studies Depression Scale scores ≥16.

Changes in the prevalence of severe postloss depressive symptoms

Caregivers' severe postloss depressive symptom prevalence peaked one month after the patient's death and subsided subsequently (Fig. 2). Simple logistic regression modeling with the GEE indicated that bereaved caregivers' likelihood of severe depressive symptoms declined significantly from 3 to 24 months postloss (Table 3).

FIG. 2.

Longitudinal changes in bereaved family caregivers' prevalence of severe depressive symptoms.

Table 3.

Changes in the Prevalence of Severe Depressive Symptoms Relative to Time of Patient Death by Simple Logistic Regression Analysis with Generalized Estimating Equation

		95% confidence interval
Time relative to loss, months	Odds ratio	Lower	Upper	Wald χ²	p
Intercept	2.72	2.06	3.58	50.63	<0.001
24	0.07	0.04	0.10	145.84	<0.001
18	0.08	0.05	0.12	157.15	<0.001
13	0.11	0.08	0.16	137.71	<0.001
6	0.22	0.16	0.31	85.38	<0.001
3	0.36	0.25	0.52	29.98	<0.001
1	Reference

Factors associated with severe postloss depressive symptoms

After adjusting for age, gender, educational attainment, financial sufficiency, and relationship with patient, the prevalence of severe depressive symptoms decreased significantly over time after the patient's death (Table 4). Bereaved caregivers were significantly more likely to have severe postloss depressive symptoms if they reported higher preloss objective caregiving load and more preloss depressive symptoms. However, the likelihood of severe depressive symptoms decreased significantly by 0.96 (95% CI = 0.94–0.97, p < 0.001) and 0.87 (95% CI = 0.79–0.95, p = 0.003) with each unit increase in scores for perceived postloss social support and preparedness for the patient's death, respectively.

Table 4.

Multivariate Analysis of Factors Associated with the Prevalence of Postloss Severe Depressive Symptoms in Family Caregivers after the Patient's Death

		95% confidence interval
Variable	AOR ^a	Lower	Upper	Wald χ²	p
Intercept	44.83	8.01	250.82	18.74	<0.001
Objective caregiving load	1.04	1.00	1.09	3.83	0.050
Subjective caregiving burden^b	0.99	0.97	1.00	2.57	0.109
Preloss depressive symptoms^c	1.02	1.00	1.04	4.18	0.041
Postloss social support^d	0.96	0.94	0.97	66.36	<0.001
Death preparedness	0.87	0.79	0.95	8.62	0.003
Time since loss (months)
24	0.06	0.04	0.09	145.46	<0.001
18	0.06	0.04	0.10	157.09	<0.001
13	0.08	0.05	0.13	124.63	<0.001
6	0.21	0.15	0.31	62.19	<0.001
3	0.33	0.21	0.52	23.09	<0.001
1	Reference

Multivariate logistic regression analysis after adjustment for age, gender, educational level, financial sufficiency, and relationship with patient.

Caregiver Reaction Assessment Scale score.

Center for Epidemiological Studies Depression Scale score.

Medical Outcomes Study Social Support Survey score.

AOR, adjusted odds ratio.

Discussion

The prevalence of severe depressive symptoms among bereaved Taiwanese caregivers of terminally ill cancer patients decreased significantly over the first two years of bereavement (Fig. 2), consistent with the literature.^3,10,11 Caregivers' tendency to have severe postloss depressive symptoms was facilitated by having higher preloss depressive symptom levels and a heavier objective caregiving load when the patient was still alive, whereas this tendency was buffered by better preparedness for the relative's death and greater perceived postloss social support.

The prevalences of bereaved Taiwanese caregivers' severe depressive symptoms were generally similar to those in other longitudinal studies using the CES-D, except for our participants' higher one-month (73.3%) and three-month (49.6%) postloss prevalences. The severe depressive symptom prevalences in other studies were 58%–59%,^3,6 17%–47%,^3,8 6%–42%,^3,6 2%–55%,^3,8,12 and unmeasured at 1, 2–4, 6–9, 12–13, and 24 months postloss, respectively.

The higher prevalence of severe depressive symptoms in bereaved Taiwanese caregivers at one and three months postloss may be due to their lack of prognostic awareness, cultural grieving practices, and our possible overestimation due to caregiver dropout. Prognostic awareness gives families sufficient time to prepare for their relative's death. Cognitive and emotional preparedness for an imminent death lowers the risk of severe depressive symptoms.^7,19,20,43 However, the rate of prognostic disclosure for family caregivers of terminally ill Taiwanese cancer patients was merely 66.7%.⁴⁴ Without appropriate prognostic disclosure,⁴⁵ terminally ill cancer patients were shown to often overestimate their life expectancy and to be unprepared for their forthcoming death as indicated by preferring futile life-extending therapy.⁴⁶ Similarly, physicians' prognostic nondisclosure and inaccurate prognostic awareness may lead to Taiwanese caregivers commonly being unprepared for an imminent death and subsequently having a higher prevalence of postloss severe depressive symptoms,⁴⁴ especially during the early postloss months. Physicians' accurately disclosing the patient's prognosis and timely end-of-life care discussion may prepare family caregivers for the forthcoming death, thereby decreasing their likelihood of severe depressive symptoms during bereavement.^43,47

Second, in Taiwanese culture, funeral ceremonies occur 7 days, 7 weeks, and 100 days after a relative's death, during which the bereaved are encouraged to publicly display their sorrow and grief,^48–50 potentially magnifying the prevalence of severe depressive symptoms. Third, we may have overestimated the prevalence of severe depressive symptoms due to caregiver dropout. Caregivers who completed postloss surveys exhibited higher role strain than caregivers who withdrew beforehand, as indicated by their heavier objective/subjective caregiving burden, higher preloss depressive symptom levels, and lower perceived financial sufficiency. Indeed, caregivers with a heavy caregiving burden tended to suffer more postloss¹⁷ because they may have exhausted all available resources before the patient's death.

Taiwanese caregivers' early high prevalence of severe depressive symptoms quickly subsided at six months postloss. This quick resolution of postloss depressive symptoms may be related to cultural grieving practices and the strong social network of Taiwanese families. In Asian cultures, people are expected not to immerse themselves in grief after the funeral ceremonies (up to 100 days postloss).⁴⁹ The bereaved may conceal their emotions and avoid expressing grief in public.^30,49,50 Similarly, in Western culture, bereaved older adults may be concerned about independence from family members and believe that if they display emotional distress to them, they will also become upset.⁵¹ Therefore, Western bereaved people may also avoid expressing their grief feelings, leading to similar observed prevalence of severe depressive symptoms.^3,6,8 Furthermore, family members in family-oriented Asian cultures may help each other adjust to their loss by providing stronger support, both instrumentally and psychologically, than those from individual-oriented Western cultures.^52,53 Such support may have attenuated the high prevalence of severe depressive symptoms in Taiwanese caregivers' first one to three months postloss, leading to their comparable severe depressive symptom prevalence 6–24 months postloss.

We found a higher prevalence of severe depressive symptoms in caregivers with higher objective caregiving burden, as reported.^13,18 Caregivers with arduous caregiving work may deplete all available resources, affecting their postloss adjustment.¹⁷ Moreover, caregivers who provide substantial care to their dying relative may find positive values in caregiving.⁵⁴ When patients eventually die, the loss of these positive caregiving rewards might increase the likelihood of severe postloss depressive symptoms.

In line with the literature, Taiwanese caregivers reported experiencing severe depressive symptoms significantly more often during bereavement if they had more preloss depressive symptoms^3,11–13 and were poorly prepared for their relative's death.^7,19,20 Higher preloss depressive symptom levels suggest that bereaved caregivers had fewer coping capabilities well before the loss,² whereas less cognitive or emotional preparedness for the forthcoming death may reflect lack of appropriate awareness and acceptance of a loved one's death,^19,20,55 predisposing bereaved caregivers to a higher probability of severe depressive symptoms.

Our results indicate that greater perceived social support protected caregivers from severe depressive symptoms during bereavement, as reported.^3,18 Emotional, informational, and instrumental support makes bereaved caregivers feel loved and valued, which can help them adjust to the patient's death and buffer the resulting emotional distress, lowering the prevalence of severe postloss depressive symptoms.

This study had some limitations. First, we recruited bereaved caregivers of cancer patients from northern Taiwan, limiting the generalizability of our results to those who receive care in other settings. Second, a remarkable proportion of caregivers (17.1%) declined to participate in the bereavement surveys and endured lower caregiving burden and lower depressive levels before their relative's death, preventing extension of our findings to those bereaved caregivers with certainty. Third, our findings warrant validation for bereaved caregivers whose data collection was not yet due. Fourth, our findings from bereaved Taiwanese caregivers of cancer patients should be replicated for bereaved family caregivers of patients with other diseases (e.g., dementia and cardiovascular disease) in other countries where cultural/societal characteristics may substantially differ. Fifth, preparedness for death is a multifaceted concept, but we measured it with only one item. Our findings should be validated by assessing preparedness with a multi-item tool. Finally, bereaved caregivers' likelihood of severe depressive symptoms may have been affected by factors not measured, for example, death circumstances, caregivers' coping capabilities, and their endorsement of Taiwanese family values.

In conclusion, bereaved caregivers' severe postloss depressive symptom prevalence peaked one month postloss and subsided subsequently. Healthcare professionals should increase their ability to identify factors predisposing family caregivers to suffer from severe postloss depressive symptoms starting when patients are still alive. At-risk caregivers should be provided effective interventions to reduce caregiving load, manage depressive symptoms, and facilitate preparedness for their relative's forthcoming death. Healthcare professionals should also follow at-risk caregivers, particularly one and three months postloss, to extend high-quality EOL care into bereavement by providing adequate support and appropriate psychological counseling or referrals to alleviate their severe postloss depressive symptoms.

Footnotes

Acknowledgments

This work was supported by the National Science Council (NSC 96-2314-B-182-029-MY2 and MOST 104-2314-B-182-027-MY3), National Health Research Institutes (NHRI-EX105-10208PI), and Chang Gung Memorial Hospital (BMRP888), Taiwan.

Author Disclosure Statement

No competing financial interests exist.

References

Lai

: Filial piety, caregiving appraisal, and caregiving burden. Res Aging, 2010; 32:200–223.

Stroebe

, Schut

, Stroebe

: Health outcomes of bereavement. Lancet, 2007; 370:1960–1973.

Ling

, Chen

, Li

, et al.: Trajectory and influencing factors of depressive symptoms in family caregivers before and after the death of terminally ill patients with cancer. Oncol Nurs Forum, 2013; 40:32–40.

Bornstein

, Clayton

, Halikas

, et al.: The depression of widowhood after thirteen months. Br J Psychiat, 1973; 122:561–566.

Clayton

, Herjanic

, Murphy

: Mourning and depression: Their similarities and differences. Can Psychiatr Assoc J, 1974; 19:309–312.

Harlow

, Goldberg

, Comstock

: A longitudinal study of the prevalence of depressive symptomatology in elderly widowed and married women. Arch Gen Psychiat, 1991; 48:1065–1068.

Barry

, Kasl

, Prigerson

: Psychiatric disorders among bereaved persons: The role of perceived circumstances of death and preparedness for death. Am J Geriat Pychiat, 2002; 10:447–457.

Chentsova-Dutton

, Shucter

, Hutchin

, et al.: Depression and grief reactions in hospice caregivers: From pre-death to 1 year afterwards. J Affect Disord, 2002; 69:53–60.

Guldin

, O'Connor

, Sokolowski

, et al.: Identifying bereaved subjects at risk of complicated grief: Predictive value of questionnaire items in a cohort study. BMC Palliat Care, 2011; 10:9.

10.

Zisook

, Shuchter

: Major depression associated with widowhood. Am J Geriat Psychiat, 1993; 1:316–326.

11.

Zisook

, Paulus

, Shuchter

, et al.: The many faces of depression following spousal bereavement. J Affect Disord, 1997; 45:85–95.

12.

Allen

, Haley

, Small

, et al.: Bereavement among hospice caregivers of cancer patients one year following loss: Predictors of grief, complicated grief, and symptoms of depression. J Palliat Med, 2013; 16:745–751.

13.

Kris

, Cherlin

, Prigerson

, et al.: Length of hospice enrollment and subsequent depression in family caregivers: 13-month follow-up study. Am J Geriat Psychiat, 2006; 14:264–269.

14.

King

, Vasanthan

, Petersen

, et al.: Mortality and medical care after bereavement: A general practice cohort study. PLoS One, 2013; 8:e52561.

15.

Stephen

, Macduff

, Petrie

, et al.: The economic cost of bereavement in Scotland. Death Stud, 2015; 39:151–157.

16.

Stroebe

, Folkman

, Hansson

, et al.: The prediction of bereavement outcome: Development of an integrative risk factor framework. Soc Sci Med, 2006; 63:2440–2451.

17.

Bass

, Bowman

: The transition from caregiving to bereavement: The relationship of care-related strain and adjustment to death. Gerontologist, 1990; 30:35–42.

18.

: From caregiving to bereavement: Trajectories of depressive symptoms among wife and daughter caregivers. J Gerontol B Psychol Sci Soc Sci, 2005; 60:190–198.

19.

Schulz

, Boerner

, Klinger

, et al.: Preparedness for death and adjustment to bereavement among caregivers of recently placed nursing home residents. J Palliat Med, 2015; 18:127–133.

20.

Hebert

, Dang

, Schulz

: Preparedness for the death of a loved one and mental health in bereaved caregivers of patients with dementia: Findings from the REACH study. J Palliat Med, 2006; 9:683–693.

21.

Hauksdóttir

, Steineck

, Fürst

, et al.: Long-term harm of low preparedness for a wife's death from cancer—A population-based study of widowers 4–5 years after the loss. Am J Epidemiol, 2010; 172:389–396.

22.

Stroebe

, Zech

, Stroebe

, et al.: Does social support help in bereavement?. J Soc Clin Psychol, 2005; 24:1030–1050.

23.

Lynch

, Connor

, Clark

: Mapping levels of palliative care development: A global update. J Pain Symptom Manage, 2013; 45:1094–1106.

24.

Institute of Medicine. Dying in America: Improving Quality and Honoring Individual Preferences Near the End of Life. Washington, DC: The National Academies Press, 2014.

25.

Bonanno

, Kaltman

: The varieties of grief experience. Clin Psychol Rev, 2001; 21:705–734.

26.

Galatzer-Levy

, Bonanno

: Beyond normality in the study of bereavement: Heterogeneity in depression outcomes following loss in older adults. Soc Sci Med, 2012; 74:1987–1994.

27.

Kim

, Kashy

, Kaw

, et al.: Sampling in population-based cancer caregivers research. Qual Life Res, 2009; 18:981–989.

28.

Bonanno

, Papa

, Lalande

, et al.: Grief processing and deliberate grief avoidance: A prospective comparison of bereaved spouses and parents in the United States and the People's Republic of China. J Consult Clin Psychol, 2005; 73:86–98.

29.

Song

, Shin

, Choi

, et al.: Quality of life and mental health in the bereaved family members of patients with terminal cancer. Psychooncology, 2012; 21:1158–1166.

30.

Tanimukai

, Hitoshi

, Hirai

, et al.: Association between depressive symptoms and changes in sleep condition in the grieving process. Supportive Care Cancer, 2015; 23:1925–1931.

31.

Radloff

: The CES-D scale: A self-report depression scale for research in the general population. Appl Psych Meas, 1977; 1:385–401.

32.

Chien

, Cheng

: Depression in Taiwan: Epidemiological survey utilizing CES-D. Seishin Shinkeigaku Zasshi, 1985; 87:335–338.

33.

Given

, Given

, Stommel

, et al.: The caregiver reaction assessment (CRA) for caregivers to persons with chronic physical and mental impairments. Res Nurs Health, 1992; 15:271–283.

34.

Tang

, Li

, 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.

35.

Lee

, Chang

, Chou

, et al.: Longitudinal changes and predictors of caregiving burden while providing end-of-life care for terminally ill cancer patients. J Palliat Med, 2013; 16:632–637.

36.

Emanuel

, Fairclough

, Slutsman

, et al.: 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.

37.

Sherbourne

, Stewart

: The MOS social support survey. Soc Sci Med, 1991; 32:705–714.

38.

Shyu

YIL

, Tang

, Liang

, et al.: Psychometric testing of the Social Support Survey on a Taiwanese sample. Nurs Res, 2006; 55:411–417.

39.

, Snow

, Heckert

, et al.: Latent Markov model for analyzing temporal configuration for violence profiles and trajectories in a sample of batterers. Sociol Methods Res, 2010; 39:222–255.

40.

Burton

, Gurrin

, Sly

: Extending the simple linear regression model to account for correlated responses: An introduction to generalized estimating equations and multi-level mixed modelling. Stat Med, 1998; 17:1261–1291.

41.

Nakai

, Ke

: Review of the methods for handling missing data in longitudinal data analysis. Int J Math Analysis, 2011; 5:1–13.

42.

, Raina

, Beyene

, et al.: Comparison of population-averaged and cluster-specific models for the analysis of cluster randomized trials with missing binary outcomes: A simulation study. BMC Med Res Methodol, 2013; 13:9.

43.

Wright

, Zhang

, Ray

, et al.: Associations between end-of-life discussions, patient mental health, medical care near death, and caregiver bereavement adjustment. JAMA, 2008; 300:1665–1673.

44.

Tang

, Liu

, Lai

, et al.: Congruence of knowledge, experiences and preferences for disclosure of diagnosis and prognosis between terminally-ill cancer patients and their family caregivers in Taiwan. Cancer Invest, 2006; 24:360–366.

45.

Lamont

, Christakis

: Prognostic disclosure to patients with cancer near the end of life. Ann Intern Med, 2001; 134:1096–1105.

46.

Weeks

, Cook

, O'Day

, et al.: Relationship between cancer patients' predictions of prognosis and their treatment preferences. JAMA, 1998; 279:1709–1714.

47.

Christakis

. Death Foretold: Prophecy and Prognosis in Medical Care. Chicago, IL: University of Chicago Press, 2001.

48.

Chu

: The funeral custom of Taiwan by “Seven days in heaven.”. Zhishan, 2011; 10:5–34.

49.

Lin

: Social culture and bereaved reaction. J Life Death Stud, 2005; 2:107–127.

50.

Tsai

: Pattern of grief expression in Chinese families. Taiwan Counsel Quar, 2012; 4:16–38.

51.

Moss

, Moss

: Widowhood in old age: Viewed in a family context. J Aging Stud, 2014; 29:98–106.

52.

Chow

, Chan

, Ho

: Social sharing of bereavement experience by Chinese bereaved persons in Hong Kong. Death Stud, 2007; 31:601–618.

53.

Asai

, Akizuki

, Fujimori

, et al.: Psychological states and coping strategies after bereavement among spouses of cancer patients: A quantitative study in Japan. Support Care Cancer, 2012; 20:3189–3203.

54.

Boerner

, Schulz

, Horowitz

: Positive aspects of caregiving and adaptation to bereavement. Psychol Aging, 2004; 19:668–675.

55.

Buckley

, Spinaze

, Bartrop

, et al.: The nature of death, coping response and intensity of bereavement following death in the critical care environment. Aust Crit Care, 2015; 28:64–70.