Family Caregiver Depressive Symptom and Grief Outcomes from the ENABLE III Randomized Controlled Trial

Abstract

Context

Little is known about whether early palliative care (EPC) support for family caregivers (CGs) impacts depressive symptoms and grief after care recipients die.

Objectives

To assess after-death CG depressive symptom and grief scores for early compared to delayed group CGs.

Methods

We conducted a randomized controlled trial (RCT) (10/2010-9/2013) of an EPC telehealth intervention for CGs (n=123) initiated at the time of care recipients' advanced cancer diagnosis (early group) or 12 weeks later (delayed group) in a rural comprehensive cancer center, affiliated clinics, and a Veterans Administration medical center. The ENABLE [Educate, Nurture, Advise, Before Life Ends] CG intervention consisted of three weekly sessions, monthly follow-up, and a bereavement call. CGs completed the Center for Epidemiological Study-Depression scale (CESD) and the Prigerson Inventory of Complicated Grief-Short Form (PG13) 8-12 weeks after care recipients' deaths. Crude and covariate-adjusted between-group differences were estimated and tested using general linear models.

Results

For care recipients who died (n=70), 44 CGs (early: n=19; delayed: n=25) completed after-death questionnaires. Mean depressive symptom scores (CESD) for the early group was 14.6 (standard deviation [SD]=10.7) and for the delayed group was 17.6 (SD=11.8). Mean complicated grief scores (PG13) for the early group was 22.7 (SD=4.9) and for the delayed group was 24.9 (SD=6.9). Adjusted between-group differences were not statistically significant (CESD: d=0.07, P=0.88; PG13: d=-0.21, P=0.51).

Conclusion

CGs' depressive symptom and complicated grief scores 8-12 weeks after care recipients' deaths were not statistically different based on the timing of EPC support. The impact of timing of CG EPC interventions on CGs bereavement outcomes requires further investigation.

Introduction

A family caregiver is an unpaid individual (a spouse, partner, family member, friend, or neighbor) involved in assisting others with activities of daily living and/or medical tasks.¹ An estimated 20 to 50% of family caregivers experience negative psychological aftereffects, including depression and grief, from months to years after a care recipient's death from cancer.^2-4 In one respect, bereaved caregivers are often relieved of burdensome tasks, such as assisting care recipients with activities of daily living and care coordination,⁵ that can consume over eight hours a day.⁶ However, they are also susceptible to a range of potentially long-lasting, negative emotions, including profound loss and emptiness at the absence of a loved one;⁷ regret about unresolved issues or unsaid goodbyes;⁸ anger or shock at the manner in which a person died (e.g., extreme pain, gasping for breath);⁹ guilt about whether decisions they may have made as a surrogate decision-maker were the “right” ones (e.g., withdrawal of life-sustaining treatments);¹⁰ and distress at acclimating to a new lifestyle and daily routine.¹¹

Emerging evidence suggests that specialty palliative and hospice care positively impacts family caregivers after their care recipients die. In a South Australian population-based survey of 9088 caregivers, Abernethy and colleagues¹² found that when care recipients received palliative care prior to death, decedent caregivers were significantly more likely to be able to move on with their lives. Similarly, a study by Ornstein and colleagues² of 1016 family caregivers of hospice patients found significantly lower caregiver depressive symptoms at 1 year post-loss compared to caregivers of non-hospice care recipients. Palliative care, especially as delivered by hospice,¹³ provides specific support to caregivers prior to a care recipient's death and during bereavement period. Though hospice regulations allow for services to be offered for up to 6 months prior to death; on average hospice care is initiated only 23 days before death.¹⁴

Very little is known, about whether offering specialty palliative care to family caregivers soon after an advanced cancer diagnosis may impact caregiver bereavement outcomes. To test if initiating early vs. delayed palliative care support to family caregivers impacts depressive symptoms and grief, we conducted a “fast track”¹⁵ phase III randomized controlled trial (RCT) of a telehealth intervention called ENABLE III (Educate, Nurture, Advise, Before Life Ends). ENABLE III consists of a face-to-face palliative care consultation for patients (caregivers may attend) and manualized coaching sessions and monthly follow-up. Family caregivers of participants with newly diagnosed advanced or recurrent cancer received the intervention either early (within 60 days) of diagnosis or 12 weeks later (delayed group). For patients who died during the study, we hypothesized that bereaved caregiver depressive symptom and grief scores would be lower for early compared to delayed group caregivers.

Methods

Study Design

The ENABLE III study design, methods and patient and caregiver outcomes have been described elsewhere.^16,17 Briefly, from October 11, 2010 to March 5, 2013, patients within 60 days of being newly diagnosed with advanced, recurrent, or progressive cancer with an oncologist-determined prognosis of six to 24 months and their family caregivers signed informed consent and were randomly assigned to receive the ENABLE III intervention either within 60 days of diagnosis (early group) or 12 weeks later (delayed group). Random one-to-one group assignment was computer-generated, stratified by six cancer types and four recruitment sites. Caregivers were assigned to the same group as the patient. The study was approved by the institutional review boards of Dartmouth College and the White River Junction Veterans Affairs Medical Center (VAMC).

Setting and Sample

Patient and caregiver participants were recruited from the Norris Cotton Cancer Center at Dartmouth-Hitchcock Medical Center (DHMC) in Lebanon, New Hampshire, affiliated DHMC outreach clinics, and the VAMC (White River Junction, VT). Patients were identified by research coordinators who reviewed weekly outpatient clinic schedules and tumor board lists. Patient inclusion criteria were: 1) English-speaking; 2) ≥18 years of age; 3) advanced-stage solid tumor or hematologic malignancy (detailed diagnostic criteria included in supplement of patient outcomes paper);¹⁸ and 4) completion of baseline questionnaires. Patient exclusion criteria included: 1) impaired cognition (Callahan score≤4);¹⁹ 2) active Axis I psychiatric (schizophrenia, bipolar disorder) or substance use disorder; 3) uncorrectable hearing loss; and 4) unreliable telephone service. Potential family caregiver participants were identified by asking patient participants to nominate a family member or friend “who knows you well and is involved in your medical care” to consider participation. There were no caregiver exclusion criteria and patient participants were not excluded if they did not have a participating caregiver.

The ENABLE III Caregiver Intervention

The ENABLE III patient^18,20,21 and caregiver^17,22 interventions have been described in detail elsewhere. Briefly, the ENABLE III caregiver intervention was a manualized intervention facilitated by advanced practice palliative care nurse coaches over the phone. Given the largely rural, mountainous Vermont and New Hampshire geography²³ and long travel distances to DHMC, ENABLE pilot work²⁰ demonstrated feasibility and patient preference for a telephone versus in-person approach. Moreover, numerous studies have reported successes of telehealth approaches, including a recent review of telehealth interventions reporting that phone-delivered psychoeducational support had comparable outcomes to in-person care, was cost effective, and improved access for rural populations.²⁴

After completing baseline questionnaires, caregivers were assigned a nurse coach (who was different than the patient nurse coach) and were mailed Charting Your Course-Caregiver (CYC-CG), an investigator-developed guidebook (manual available from authors) that contained resource material supporting the nurse coaching sessions. Caregivers participated in three weekly telephone coaching sessions using CYC-CG. Session 1 focused on “being a caregiver;” the goals of palliative care; an overview of the COPE (Creativity, Optimism, Planning, Expert guidance) attitude;²⁵ and a step-wise framework for engaging in problem-solving coping (derived from the conceptual foundations of problem-solving treatment).^25-27 Session 2 centered on caregiver self-care, addressing topics such as healthy eating and exercise, meditation, spirituality and partnering in a patient's symptom management. Session 3 reviewed communication, decision-making including decision aid tools (e.g., Ottawa Personal Decision Guide), and advance care planning. Caregiver participants then received monthly check-in calls to reinforce prior content and assist with new issues that arose. Follow up calls continued until study completion or care recipients death. Caregivers of care recipients who died, received a bereavement call from nurse coaches approximately 2-4 weeks after the date of death to offer condolences, inquire how they were doing, and ask permission for a data collector to contact them to administer a final survey. To enhance intervention fidelity, nurse coaches followed a script for all CYC-CG sessions, monthly check-in calls, and for the bereavement call.

Seven advanced practice palliative care nurse coaches received approximately 30 hours of training that included self-study of all intervention protocols and scripts; one-on-one didactic education on problem-solving support and decision-aids; review of 12 recorded CYC practice sessions and role playing each session. The PI (MB) provided one-on-one constructive feedback during training. After training, the principal investigator (PI), blinded to group assignment, listened to all caregiver sessions and met weekly with the nurse coaches to discuss any protocol deviations or other issues.

Measures

Data collectors, blinded to group assignment, collected patient and caregiver measures by telephone at baseline, 6, 12, 18, and 24 weeks. After week 24, data collection was every 12 weeks until patients died or the study ended. Baseline family caregiver demographics included age, gender, race, ethnicity, religion, marital status, employment, education, and their relationship to the care recipient. Details of all other patient and caregiver measures are described elsewhere.¹⁷ Eight to 12 weeks after-death, participating caregivers of decedents were asked to complete a measure of depressive symptoms and grief.

Depressive symptoms were measured using the Centers for Epidemiological Studies-Depression (CES-D) scale.²⁸ This 20-item measure has demonstrated very good internal consistency reliability (Cronbach's alpha=0.85) and includes questions about how often one has had symptoms related to depression over the past week, such as appetite loss, feeling blue, restless sleep, loneliness, and crying spells. Response options range from 0 to 3 (0=rarely or none of the time [<1 day]; 1=some or a little of the time [1-2 days]; 2=occasionally or a moderate amount of time [3-4 days]; 3=most of all of the time [5-7 days]). Total scores range from 0 to 60, with higher scores indicating higher depressed mood and scores >16 suggestive of clinical depression.²⁹

Grief was measured using the Prigerson Inventory of Complicated Grief-Short Form (PG13).³⁰ This 13-item short form measure has demonstrated very good internal consistency reliability (Cronbach's alpha=0.94) and includes questions to assess pathological grief over the past month, including yearning for the deceased, acceptance of the deceased death, and feelings of meaninglessness and being able to move on. Responses range from 1 (“never”) to 5 (“always”). Higher scores indicate higher grief and scores >25 indicate clinically significant complicated grief.

Statistical Analysis

Caregiver characteristics (study group, age, gender, race, marital status, education, religious preference, employment status, relationship to patient, patients' primary cancer site, patients' use of hospice prior to death, and baseline CES-D score) were tabulated by completion of after-death questionnaire status (and by study group) and then compared using effect sizes and significance tests. The objective of the analysis was to estimate the baseline to after-death between-group difference (early minus delayed) in CES-D score and the between-group difference in PG13 score. Within-person change in CES-D scores were computed and the between-group mean difference was estimated. The between-group difference in PG13 score was also estimated. These mean differences were compared using effect sizes (Cohen's d) and t-tests. Because the study groups differed on religious preference and patient use of hospice, between-group mean differences in outcomes, adjusted for the probability of group assignment given religious preference and for patient hospice use, were also estimated using general linear models (one covariate at a time because of the small sample size). The probability of group assignment given religious preference was estimated using a logistic regression model fitted with Firth's penalized likelihood method.³¹ The between group difference in PG13 score was also estimated after adjusting for baseline CES-D score. Because of the small sample size, we interpreted our results using both observed effect sizes and significance tests. Analyses were conducted in SAS software v. 9.4 (SAS Institute Inc., Cary, NC).

Results

Sample Characteristics

As shown in Fig. 1, 207 patients were randomized into early and delayed intervention groups. One hundred twenty-four nominated a family caregiver to participate and 122 enrolled and provided data. A total of 70 patient participants with enrolled family caregivers died during the study (early: n=31; delayed: n=39) and 44 caregivers (early: n=19; delayed: n=25) completed after-death questionnaires. Approximately 37% (n=26) of decedents' caregivers did not complete after-death questionnaires. Religious preference and baseline CESD were associated with completion of the questionnaires. Protestants were more likely to complete (22/26, 84.6%), while Catholics were less likely to complete (8/20, 40%; Cramer's V= 0.38, P=0.02). Those who completed the questionnaires had, on average, lower values (fewer depressive symptoms) of baseline CESD (M(SD)=15 (9.3)) compared to those who did not complete (M(SD)=19.4 (12.5); d=0.40, P=0.18).

Fig. 1.

CONSORT Diagram of Family Caregiver Enrollment.

There were no statistically significant differences between decedents' caregiver demographic characteristics by group (Table 1). On average, caregivers were age 62 years, female (84.1%; n=37), white (93.2%; n=41), and married or living with a partner (93.2%; n=41). Forty-three percent (n=19) had completed high school or some college or technical school and 56.8% (n=25) had a college or advanced degree. Approximately 68% (n=30) reported a Christian denomination (Protestant or Catholic), and 18.2% (n=8) reported no religious affiliation. Over half (n=25) were employed full- or part-time and approximately 43% (n=19) were retired or unemployed. Most caregivers (81.8%; n=36) were decedents' spouses or partners. Most cared for patients with lung (45.5%; n=20) and gastrointestinal cancers (31.8%; n=14) and a substantial proportion of patients used hospice prior to death (59.1%, n=26).

Table 1. Family Caregiver Characteristics of Decedent Care Recipients.

Characteristic	All Caregivers (N=44)		Early Group (N=19)		Delayed Group (N=25)		P*	Effect size**
	No.	%	No.	%	No.	%
Age, years
Mean	61.6		62.1		61.2		.77	.09
SD	10.1		11.9		8.6
Female gender	37	84.1	15	78.9	22	88.0	.42	.12
Race
White	41	93.2	18	95.0	23	92.0	.87	.03
Black	0	0	0	0	0	0
Other	2	4.6	1	5.3	1	4.0
Missing	1	2.3	0	0	1	4.0
Marital Status
Never married	2	4.5	1	5.3	1	4.0	.67	.14
Married or living with partner	41	93.2	18	95.0	23	92.0
Divorced or separated	0	0	0	0	0	0
Widowed	1	2.3	0	0	1	4.0
Education
High school or GED, some college or technical school	19	43.2	9	47.4	10	40.0	.63	.07
College graduate or above	25	56.8	10	52.6	15	60.0
Religion
Protestant	22	50.0	11	57.9	11	44.0	.20	.33
Catholic	8	18.2	1	5.3	7	28.0
None	8	18.2	5	26.3	3	12.0
Other	5	11.4	2	10.5	3	12.0
Missing	1	2.3	0	0	1	4.0
Employment Status
Full or part time	25	56.8	9	47.4	16	64.0	.51	.18
Retired	14	31.8	7	36.8	7	28.0
Not Employed	5	11.4	3	15.8	2	8.0
Relationship to patient
Spouse/partner	36	81.8	15	78.9	21	84.0	.81	.17
Other immediate relative (i.e., parent, child)	7	15.9	3	15.8	4	16.0
Other relative or friend	1	2.3	1	5.2	0	0
Decedent cancer diagnosis
Lung	20	45.5	7	36.8	13	52.0	.60	.26
Gastrointestinal	14	31.8	8	42.1	6	24.0
Genitourinary	3	6.8	2	10.5	1	4.0
Hematologic	3	6.8	1	5.3	2	8.0
Other Solid Tumor	4	9.1	1	5.3	3	12.0
Patient hospice use prior to death
Yes	26	59.1	14	73.7	12	48.0	.06	.30
No	17	38.6	4	21.1	13	52.0
Missing	1	2.3	1	5.3	0	0

^*Fisher's exact or Pearson's Chi-square test for categorical variables and t-test for continuous variables

^**Cohen's D for continuous variables and Cramer's V for categorical variables

Depressive Symptoms and Complicated Grief

Table 2 shows the CESD between-group comparisons at baseline, after patient death, and the change from baseline to after-death. Adjusted between-group differences were not statistically significant (P=0.81). The CESD change from baseline for the early group was +3.4 (SD=9.5) and for the delayed group was +2.6 (SD=12.1). The between-group difference in CESD change was of trivial magnitude (d=0.08) even after adjusting for religious preference (d=0.10) or patient hospice use (d=0.07).

Table 2. Comparison of change over time in depression scores, Early vs. Delayed Group.

	All caregivers (N=44)		Early group (N=19)		Delayed group (N=25)		Between-group difference
	Mean	SD	Mean	SD	Mean	SD	Mean	SE	P	d
Baseline	13.2	8.8	11.2	8.2	14.8	9.2	-3.6	2.7	.19	-.41
After care recipient death	16.3	11.3	14.6	10.7	17.6	11.8	-3	3.5	.39	-.34
Change from baseline
Unadjusted	3	10.9	3.4	9.5	2.6	12.1	0.7	3.4	.83	.08
Adjusted for religious preference	-	-	-	-	-	-	0.9	3.6	.81	.10
Adjusted for hospice utilization	-	-	-	-	-	-	0.6	3.6	.88	.07

Notes: Change is follow-up minus baseline. Between group difference is Early minus delayed. P is from general linear models. Cohen's d uses the baseline pooled SD.

Table 3 shows the PG13 between-group grief comparison scores. Again, the adjusted between-group differences were not statistically significant (P=0.51). The unadjusted effect size was moderate (early minus delayed, d=-0.35); however, the effect size decreased to d=-0.27 after adjusting for religious preference. Further adjustment for baseline CESD scores and patient hospice use resulted in a decrease of the effect size difference to d=-0.20 and d=-0.21, respectively (early minus delayed).

Table 3. Comparison of grief scores after patient death, early vs. delayed group.

Group	Raw scores, mean (SD)	Between-group differences
		Crude estimates			Adj. for religious preference			Adjusted for baseline CES-D			Adj. for patient hospice use
		Mean (SE)	P	d	Mean (SE)	P	d	Mean (SE)	P	d	Mean (SE)	P	D
Early	22.7 (4.9)	-2.2 (1.9)	.21	-.35	-1.7 (2)	.40	-.27	-1.21 (1.8)	.51	-.20	-1.33 (2)	.51	-.21
Delayed	24.9 (6.9)

Notes: Between-group difference is early minus delayed. P is from general linear models. Cohen's d uses pooled SD.

Discussion

This trial represents the first attempt to test the effect of early palliative care telehealth support on bereaved family caregivers. We conducted an early versus delayed “fast track” RCT to determine if family caregivers who received the ENABLE III early palliative care telehealth intervention, within 60 days of their care recipients' advanced cancer diagnosis had lower depressive symptom scores and lower grief scores after care recipients' deaths than caregivers who received the intervention 12 weeks later. Our analysis detected no statistical differences on caregivers' after-death depression and grief scores relative to initiation of the intervention.

In our prior report of this trial,¹⁷ a terminal decline analysis of CG outcomes looking backwards in time 36 weeks from care recipient death showed significantly lower depressive symptom scores in early compared to delayed group caregivers. Given that pre-loss caregiver distress has been associated with post-loss outcomes,^3,8,32,33 we were surprised to find that the depressive symptom benefits for early group caregivers did not extend into the period after their care recipients' deaths. Because not all decedent caregivers completed after-death questionnaires, the sample of decedent caregivers was much smaller in this analysis compared to that in the terminal decline analysis, which greatly limited statistical power to detect differences between groups. Despite this limitation, it is worth highlighting that there was an observed small effect size difference in grief scores favoring the early group (early minus delayed; PG13: d=-0.20). We believe this signal of potential efficacy is sufficient to warrant future examination of the impact of early palliative care for caregivers on their after-death outcomes.

There are several explanations for the lack of robust positive results. First, the ENABLE III caregiver intervention did not specifically address death preparedness and coping with the death of a loved one.³⁴ Because ENABLE is an early palliative care intervention, topics were chosen that were most appropriate for the caregiving context while care recipients were still relatively high functioning and with potentially several years to live. Topics such as what to expect and how to care for the actively dying patient, withholding/withdrawing life-sustaining treatment, and funeral care were not discussed by nurse coaches unless the topics were broached by participants themselves. Discussing such topics before death may facilitate better bereavement adjustment.³⁴ Hence, future early palliative care for family caregivers might consider exploring non-threatening ways to integrate end-of-life preparedness topics.

Second, depressed mood and grief scores were collected 8 to 12 weeks after the death of the care recipient. This timeframe falls within the period when normal grief is expected, which is typically up to 1 year.⁷ Evidence suggests that significant differences between normal and prolonged grief are not evident until one to two years after an individual's death.^2,12 For example, Ornstein et al. found that differences in spousal depression in those whose care recipients did and did not use hospice were most pronounced at one year following death. Hence, future studies of early palliative care for caregivers may need to protract the timeline of data collection for up to 2 years after a care recipient's death in order to capture meaningful differences. Though it will be important to control for CGs' use of bereavement services which are covered under Medicare's hospice benefit.¹³

A number of factors have been associated with lower caregiver depression and grief scores that may be impacted by future models of caregiver early palliative care support. Examples include use of hospice,^2,12,35 avoidance of high intensity end-of-life care,^36,37 care recipient's quality of death and dying,³⁶ and the leveraging of social support networks.^3,38 These factors represent prospective mediators of early palliative care support for caregivers that impact after-death outcomes. In combination with evidence showing caregiver underutilization of post-loss bereavement services,³⁹ a potential solution is to educate and encourage caregivers early on in the patient's disease trajectory to leverage these resources if needed in the future. Early palliative care supports family caregivers when they are less entrenched in day-to-day tasks because their care recipients are less ill and have fewer care needs. Hence, caregivers may be more capable during this period of relatively lower stress to obtain knowledge and skills that they will need in the future.

Limitations

Our study has important limitations. First, our trial did not have a usual care comparison group. Both caregiver groups received palliative care much earlier than would be provided through hospice. Therefore, we are unable to determine what after-death differences there might have been between caregivers receiving ENABLE versus those who did not. Second, because we did not collect data on caregivers' use of supportive services such as hospice bereavement programs and counseling, we are not able to account for the potentially mediating impact that use of these services may have had on caregiver after-death outcomes. Third, because data for this analysis could only be collected from caregivers of decedents, the sample size for this analysis was small, thus limiting statistical power and increasing the chances of failing to detect a true relevant effect (Type II error). Fourth, this analysis tested the “timing” of palliative care and did not account for the amount or “dosing” of palliative care one receives before death (e.g., as measured by the number of hours of one-on-one contact). It seems reasonable to hypothesize that the “dose” of palliative care support that caregivers receive pre-loss may predict post-loss depression and grief. However, it remains an unanswered and important research question about whether there are differences between “timing” and “dosing” of specialty palliative care. Finally, the caregiver sample was nearly all white, which is unsurprising given the population demographics of rural New England. Future work in this area would be strengthened by inclusion of racial and ethnic minorities, especially given evidence suggesting non-white caregivers may have higher rates of grief disorders.⁴⁰

Conclusion

Our study did not find statistical differences in depression and grief scores of family caregivers of decedents who received early versus delayed palliative care. However, our small sample size may have limited our ability to detect the impact of ENABLE on caregiver bereavement outcomes. Furthermore, after-death measures were collected within the time frame when normal grief is difficult to distinguish from complicated grief. Given that the observed effect size difference in grief scores favored early group caregivers, the possibility should not be ruled out that early palliative care may positively impact caregivers after patients die. Therefore the impact of early palliative care support for caregivers on bereavement outcomes is still an open and high priority research question.