Advance Care Planning Impact on Caregivers and End-of-life Care in Advanced Cancer

Abstract

Background:

Advance care planning (ACP) may improve outcomes for patients with advanced cancer and their family caregivers, but the optimal approach is not known. This study compared the impact of facilitated vs patient-directed ACP on caregiver psychological symptoms and perceptions of goal-concordant EOL care.

Methods:

Adult patients with advanced solid tumors and their caregivers were recruited from 8 oncology clinics in western Pennsylvania. Participants were randomly assigned to complete facilitated ACP with a trained nurse facilitator or patient-directed ACP using written and web-based tools. Caregivers were followed through bereavement and completed surveys assessing depression and anxiety symptoms (HADS, range 0-21, scores >7 indicative of significant symptoms), PTSD symptoms (IES-R, range 0-88, scores > 24 considered clinically significant), and validated measures of goal-concordant EOL care.

Results:

Among 400 enrolled patients, 272 (68%) had an enrolled caregiver. Caregivers were predominantly female (73%), Caucasian (95%), and were spouses/partners (64%) of the patient with cancer. Among bereaved caregivers (n=98), those in the facilitated ACP arm reported significantly fewer post-traumatic stress symptoms compared to the patient-directed arm (mean IES-R scores 23.9 vs 31.5, p=0.01). Both depression and anxiety symptoms remained similar between arms (depression: 5.34 vs 5.87, p=0.50; anxiety: 6.56 vs 6.72, p=0.84) and low overall. Caregiver-reported goal-concordant care was higher in the facilitated ACP arm compared to the patient-directed arm (95.8% vs 75.5%, p=0.01).

Conclusion:

In this randomized trial comparing facilitated vs patient-directed ACP, a facilitated approach was associated with lower post-traumatic stress symptoms among bereaved caregivers and higher rates of goal-concordant EOL care.

Introduction

Despite unprecedented advances in cancer treatment in the last decade, it is expected that in 2025, over six hundred thousand patients in the United States will die from cancer¹. Advance Care Planning (ACP) represents a structured process that enables patients to understand, articulate, and document their values, life goals, and preferences for medical care². ACP interventions typically fall into two distinct categories: those using trained facilitators and patient-directed approaches using various media (such as videos, web-based applications, or printed materials). Although ACP is recommended by clinical guidelines for patients with advanced cancer ^3–7, the most effective ACP approach is unclear^{4, 6, 8}.

Supporting family caregivers and surrogate decision-makers is widely recognized as an important ACP outcome ². Caregivers often shoulder the weighty responsibility of medical decision-making, a role that can drive significant psychological trauma⁹. This burden is starkly illustrated in a study of intensive care unit caregivers, where post-traumatic stress disorder was observed in 33% of all caregivers nine months after patient death or discharge, rising to 81.8% among caregivers who shared in end-of-life (EOL) decisions¹⁰. Caregiving for patients with cancer has been described as intense and episodic, with 62% of cancer caregivers in a high-burden situation¹¹. The prevalence of post-traumatic stress disorder (PTSD) symptoms among cancer caregivers varies considerably, ranging from 3.4% to 36.8%, influenced by multiple factors including cancer type, caregiver sex, PTSD measures used, and timing of assessment relative to diagnosis^12–14.

A number of studies and meta-analyses have found that ACP improves multiple outcomes impacting caregivers^15–17, including improved alignment between caregiver and patient EOL preferences^17–19 and better bereavement and mental health outcomes^{17, 20–22}. For example, two studies exploring ACP, one in end-stage renal disease patients and another in older adults, demonstrated reduced PTSD symptoms among caregivers who engaged in ACP ^{17, 20}. In the oncology setting, data on the impact of ACP on caregiver outcomes is less robust and focuses on caregivers’ assessment of their care partner’s treatment. A systematic review by Levoy et al found that compared to no ACP, all levels of ACP were significantly associated with caregivers’ positive perceptions of cancer decedents’ EOL experiences ²³. The impact of ACP on PTSD in bereaved caregivers of patients with cancer is unknown, and no studies were identified that have examined the impact of different ACP methods on these caregiver outcomes.

The PEACe (Patient Centered and Efficacious Advance Care Planning in Cancer - PEACe-compare) trial compared the effects of facilitated ACP delivered via the Respecting Choices^® trained facilitator model with a patient-directed program delivered via the PREPARE for Your Care website and written materials. The patient outcomes from this study have been previously published²⁴. This analysis focuses on caregiver outcomes, comparing symptoms of PTSD, anxiety, and depression, as well as caregiver-reported quality and goal concordance of EOL care between facilitated and patient-directed ACP approaches.

Methods:

Study design:

PEACe-compare was a single-blind, patient-level, randomized comparative effectiveness trial of facilitated vs. patient-directed ACP between August 21, 2019, and November 15, 2023. Results are reported following CONSORT guidelines. Details of the study protocol and primary patient outcomes have been published previously ²⁴. Here we report our main caregiver outcomes: symptoms of PTSD, anxiety, and depression, as well as caregiver-reported quality and goal concordance of EOL care. All study procedures were approved by the University of Pittsburgh Institutional Review Board (STUDY19080337), and the trial was registered on ClinicalTrials.gov (NCT03824158).

Study setting and participants:

Patients and caregivers were enrolled from eight oncology clinics within the University of Pittsburgh Medical Center Hillman Cancer Center Network in western Pennsylvania. Eligible patients were adults (18 years or older) with solid tumor malignancies for whom the oncologist would not be surprised” if the patient died within the next year ^{25, 26}. Patient sex and race were self-reported. Patients with prior advance directive documentation were not excluded. Eligible caregivers were at least 18 years old and identified by patients as a family member or friend caregiver who is most involved in their healthcare and who is capable of participating in the study. Patients who were unable to identify a caregiver were not excluded from the study. All caregivers provided informed consent. Enrolled caregivers were paid $25 each upon completion of the 12-week follow-up assessment and an additional $25 each upon completion of the bereavement assessment.

Randomization and enrollment

Clinical team members reviewed daily schedules to identify eligible patients and obtained permission for study team contact. Study staff explained the study, verified eligibility, and obtained written consent during clinic visits or via mail. Consenting patients identified potential caregivers, who provided consent in-person, electronically, or by phone. Upon completion of baseline patient and caregiver interviews, patient-caregiver pairs were randomized 1:1 at the patient level to receive either facilitated or patient-directed ACP.

Facilitated ACP Intervention

Facilitated ACP was delivered via the Respecting Choices^® model^{19, 20, 27–30}. Facilitators were nurses who were not members of the patients’ clinical care team who completed a 3-day Respecting Choices^® training and were certified by Respecting Choices^® faculty. Sessions were audio recorded, and Respecting Choices^® faculty reviewed one recording per month to ensure fidelity and adherence to the Respecting Choices conversation. Patients chose the mode of delivery that worked best for them: in person, telephone, or video conference. Structured ACP discussions followed the First Steps^® ACP Conversation Guide. The First Steps^® discussion is designed to take place in a single session, but allows for up to three visits per patient to allow patients time to discuss care preferences. Patients were encouraged (but not required) to attend facilitator sessions with their medical decision maker and other family members.

Facilitators offered patients assistance with completing a Pennsylvania Advance Health Care Directive and provided a copy to clinical staff when appropriate. Participants were encouraged to share completed directives with caregivers and decision makers to reinforce patient decision-making.

Patient-directed ACP

The patient-directed ACP intervention was delivered using PREPARE for Your Care. PREPARE consists of an hour-long, 5-step ACP process delivered through interactive online content and videos. The program includes easy-to-read patient materials, advance directives, and a pamphlet outlining the PREPARE website. All patients received an introductory packet containing ACP step-by-step instructions, PREPARE website access information, a blank Pennsylvania Advance Health Care Directive form designed for readability, and guidance for incorporating the completed directive into their medical record.^{31, 32}. They were directed to review the website in their own time via a study-specific PREPARE account, allowing us to track website usage activity. Research staff made up to three follow-up calls to encourage patients to review the PREPARE materials and access the website, offering technical support as needed. Patients could also access the PREPARE website using a study-provided tablet at their oncology clinic.

Data collection and outcomes

Caregivers completed telephone- or paper-based surveys at baseline and 12 weeks, administered by research assistants trained in study procedures and blinded to the intervention group. We collected sociodemographic information, including age, sex, race/ethnicity, partner status, education level, relationship with the patient, type of caregiving support, time spent caregiving weekly, employment, and income security. Caregivers of patients who died during the study period also completed assessments 12 weeks or later after the patient’s death.

Caregiver symptoms of depression and anxiety were measured at baseline, 12 weeks, and at bereavement using Hospital Anxiety and Depression Scale (HADS), which is a widely-used 14-item instrument measuring symptoms of depression and anxiety³³. A score of ≥ 7 on either domain indicates significant symptoms of depression or anxiety with good sensitivity and specificity³⁴. It has been validated as a tool for cancer patients³⁵ and caregivers of patients admitted to an intensive care unit³⁶. Post-traumatic stress disorder (PTSD) symptoms among caregivers were measured at bereavement using the Impacts of Events scale-Revised (IES-R). The IES-R scale is a 22-item self-report instrument for PTSD symptoms with three subscales measuring intrusion (8 items), avoidance (8 items), and hyperarousal (6 items)³⁷. The revised version of the Impact of Event Scale (IES-R) has seven additional questions and a scoring range of 0 to 88³⁸. A score of 24 or higher is considered clinically significant.

For patients who died during the study period, we recorded caregiver-reported measures of goal-concordant EOL care and a caregiver-reported measure of quality of EOL care. Caregiver-reported measures of goal-concordant EOL care were assessed by asking caregivers about the extent to which patients’ wishes were followed in medical care during the last month of life, and by comparing patients’ preferred and actual places of death, with questions separated throughout the survey to minimize conscious comparison. Quality of EOL care was assessed using the Caregiver Evaluation of Quality of End-of-Life Care (CEQUEL) scale which is a 13-item instrument (range of scores 13-26) that includes items that measure caregiver evaluations of four domains: prolongation of death, perceived suffering, shared decision making, and preparation for the death. It has demonstrated reliability and convergent validity among cancer caregivers³⁹. Higher CEQUEL scores have been positively associated with the patient-physician therapeutic alliance and hospice enrollment and negatively associated with bereaved caregiver regret and a diagnosis of PTSD ³⁹.

Statistical analysis

Baseline characteristics for enrolled patients with caregivers and enrolled caregivers were summarized overall and across the two study intervention groups. Continuous variables were reported as means and standard deviations; categorical variables were reported as frequencies and percentages. Patient survival probability by intervention group was estimated using the Kaplan-Meier method and compared using the log-rank test. Changes from baseline to 12 weeks within each group were evaluated using paired t-tests. Differences between the two study intervention groups at 12 weeks were assessed using t-tests or Mann-Whitney U tests for continuous variables and chi-square tests for categorical variables. Caregiver outcomes measured at baseline, 12 weeks, and 12 weeks post-death were analyzed using mixed-effects models, with baseline measures and intervention group included as fixed effects and clinics treated as a random effect.

Regarding missing data, given that outcome measures were collected only at baseline and at 12 weeks, we performed a complete case (CC) analysis, excluding participants with missing outcome data at either time point. We chose the CC approach for its clarity and straightforward interpretation, especially with only two measurement occasions. To address potential bias from missing data and to confirm the robustness of our findings, we then performed sensitivity analyses using multiple imputation (MI). We implemented MI using the Fully Conditional Specification method, and we assumed missingness to be at random (MAR). Separate imputation models were applied for all enrolled caregivers and for eligible bereaved caregivers. Thirty imputed datasets were generated with both baseline patients and caregiver characteristics, and survey item responses were imputed individually rather than imputing total or composite scores. All statistical analyses were conducted using SAS v9.4. The statistical significance was set at p<0.05.

Results:

Caregiver baseline characteristics

Among 400 enrolled patients, 272 (68%) had an enrolled caregiver (133 in the patient-directed ACP arm and 139 in the facilitated ACP arm, as depicted in Figure 1). Demographic information for patients with enrolled caregivers is depicted in Table 1, and that of enrolled caregivers is included in Table 2. Additional demographic information is included in Tables S1 and S2. Caregivers were predominantly female (199, 73.2%), Caucasian (257, 94.5%), and married (218, 80.2%), with a mean age of 62 (SD 13.5) years. Most were spouses/partners (175, 64.3%) or adult children (49, 18%) of the patients, and 192 (71.1%) lived with the patient. The majority were either working full-time (86, 31.6%) or retired (121, 44.5%). Educational attainment varied, with 25.7% having a high school diploma or general education development diploma (GED), 25% having a bachelor’s degree, and 20.1% having advanced degrees. Most caregivers (188, 69.1%) reported having enough money or that money was not a problem. Almost all caregivers (241, 88.6%) were actively providing care at baseline, averaging 5.46 days per week (SD=2.3) and 7.03 hours per day (SD=8.0) of care. The most common caregiving tasks included providing emotional/social support (227, 83.5%), medical care (211, 77.6%), household chores (206, 75.7%), and transportation (183, 67.3%). Additionally, 60 caregivers (22.5%) reported providing care for someone else besides the patient. Complete demographic characteristics, caregiving responsibilities are illustrated in Table 2 and Table S2.

Figure 1:

Consort Flowchart

Table 1:

Baseline Characteristics of Patients with Enrolled Caregivers, Overall and by Intervention Group.

	Overall	Patient-Directed ACP	Facilitated ACP	P-value
N(%)	272(100)	133(48.9)	139(51.1)
Age	68.2 (10.9)	68.7 (10.4)	67.9 (11.3)	0.54
Sex
Female	121 (44.5)	57 (42.9)	64 (46)	0.87
Male	151 (55.5)	76 (57.1)	75 (54)
Race
Asian	2 (0.7)	1 (0.8)	1 (0.7)	0.96
Black	14 (5.1)	8 (6.0)	6 (4.3)
Other	1 (0.4)	1 (0.8)	0 (0)
White	255 (93.8)	123 (92.5)	132 (95.0)
Ethnic
Non Hispanic	269 (98.9)	131 (98.5)	138 (99.3)	1.0
Hispanic	2 (0.7)	1 (0.8)	1 (0.7)
ECOG Performance Status
0=Fully Active	85 (31.3)	37 (27.8)	48 (34.5)	0.74
1=Restricted in physically strenuous activity but ambulatory and able to carry out work of a light or sedentary nature	156 (57.4)	82 (61.7)	74 (53.2)
2=Ambulatory and capable of all selfcare but unable to carry out any work activities. Up and about more than 50% of waking hours	31 (11.4)	14 (10.5)	17 (12.2)
Completion of a living will or advance directive
No	134 (49.3)	67 (50.4)	67 (48.2)	0.94
Yes	138 (50.7)	66 (49.6)	72 (51.8)
Current Marital Status
Never married	12 (4.4)	5 (3.8)	7 (5.0)	0.96
Married	193 (71.0)	99 (74.4)	94 (67.6)
Widowed	36 (13.2)	15 (11.3)	21 (15.1)
Divorced/separated	28 (10.3)	13 (9.8)	15 (10.8)
Decline to answer	1 (0.4)	1 (0.8)	0 (0)
Living situation
In a home I (or my family) own	227 (83.5)	111 (83.5)	116 (83.5)	0.97
In a home I (or my family) rent	34 (12.5)	18 (13.5)	16 (11.5)
In an Assisted Living facility	2 (0.7)	1 (0.8)	1 (0.7)
In a Nursing Home	2 (0.7)	0 (0)	2 (1.4)
Other/unsure	7 (2.6)	3 (2.3)	4 (2.9)
Homemaker (never worked for pay)	5 (1.8)	1 (0.8)	4 (2.9)
Disability	29 (10.7)	12 (9.0)	17 (12.2)
Decline to answer	5 (1.8)	3 (2.3)	2 (1.4)
Other	16 (5.9)	8 (6.0)	8 (5.8)
How long have you been receiving care from your current oncologist?
Less than one month	19 (7.0)	10 (7.5)	9 (6.5)	0.99
≥ 1 month but < 6 months	49 (18)	22 (16.5)	27 (19.4)
≥ 6 months but < 1 year	71 (26.1)	34 (25.6)	37 (26.6)
≥ 1 year but < 2 years	43 (15.8)	23 (17.3)	20 (14.4)
≥ 2 years but < 5 years	50 (18.4)	27 (20.3)	23 (16.5)
≥ 5 years	39 (14.3)	16 (12)	23 (16.5)
Declined to Answer	1 (0.4)	1 (0.8)	0 (0)
Cancer Type
Breast	31 (11.4)	13 (9.8)	18 (12.9)	1.0
Gastrointestinal	55 (20.2)	28 (21.1)	27 (19.4)
Genitourinary	29 (10.7)	13 (9.8)	16 (11.5)
Head and neck	14 (5.1)	7 (5.3)	7 (5.0)
Hepatobiliary	8 (2.9)	5 (3.8)	3 (2.2)
Lung	49 (18)	27 (20.3)	22 (15.8)
Pancreatic	30 (11.0)	14 (10.5)	16 (11.5)
Prostate	33 (12.1)	17 (12.8)	16 (11.5)
Gynecological/Testicular	16 (5.9)	5 (3.8)	11 (7.9)
Other	7 (2.6)	4 (3.0)	3 (2.2)

Table 2:

Baseline Characteristics of Enrolled Caregivers, Overall and by Intervention

	Overall	Patient-Directed ACP	Facilitated ACP	p-value
N (%)	272(100)	133(48.9)	139(51.1)
Age (Years)	62 (13.5)	62.2 (13.6)	61.9 (13.5)	0.94
Sex
Male	73 (26.8)	37 (27.8)	36 (25.9)	0.72
Female	199 (73.2)	96 (72.2)	103 (74.1)
Race				0.71
Asian	2 (0.7)	1 (0.75)	1 (0.7)
Black	11 (4)	6 (4.5)	5 (3.6)
Native American/American Indian/Alaskan Native	1 (0.4)	0	1 (0.7)
Other	1 (0.4)	1 (0.75)	0
Pacific Islander/Samoan/Hawaiian	0	0	0
White	257 (94.5)	125 (94)	132 (95)
Ethnicity				0.98
Non-Hispanic	268 (99.3)	132 (99.25)	136 (99.3)
Hispanic	2 (0.7)	1 (0.75)	1 (0.7)
Current Marital Status				0.51
Never married	18 (6.6)	12 (9)	6 (4.3)
Married	218 (80.2)	105 (79)	113 (81.3)
Widowed	15 (5.5)	6 (4.5)	9 (6.5)
Divorced/Separated	18 (6.6)	8 (6)	10 (7.2)
Declined to Answer	3 (1.1)	2 (1.5)	1 (0.7)
What is your relationship to the patient?				0.38
Spouse/Partner	175 (64.3)	90 (67.7)	85 (61.2)
Child	49 (18)	25 (18.8)	24 (17.3)
Parent	9 (3.3)	5 (3.8)	4 (2.9)
Sibling	18 (6.6)	6 (4.5)	12 (8.6)
Friend	7 (2.6)	2 (1.5)	5 (3.6)
Other	13 (4.8)	4 (3)	9 (6.5)
Declined to Answer	1 (0.4)	1 (0.8)	0
Do you live with the patient? (% answering yes)	192 (71.1)	98 (73.7)	94 (68.6)	0.25
How many other people live in your household?				0.77
0	17 (6.3)	7 (5.3)	10 (7.2)
1	185 (68)	90 (67.7)	95 (68.4)
2	34 (12.5)	17 (12.8)	17 (12.2)
3	28 (10.3)	14 (10.5)	14 (10.1)
4	6 (2.2)	3 (2.3)	3 (2.2)
5	2 (0.7)	2 (1.5)	0
Declined to Answer	0	0	0
Current Employment Status				0.96
Working full-time	86 (31.6)	40 (30.1)	46 33.1)
Working part-time	23 (8.5)	11 (8.3)	12 (8.6)
Unemployed	5 (1.8)	3 (2.3)	2 (1.4)
Retired	121 (44.5)	59 (44.4)	62 (44.6)
Homemaker (never worked for pay)	8 (2.9)	5 (3.8)	3 (2.2)
Disability	10 (3.7)	4 (3)	6 (4.3)
Other	17 (6.3)	10 (7.5)	7 (5)
Declined to Answer	2 (0.7)	1 (0.8)	1 (0.7)
Are you currently a care provider for the patient? (% answered yes)	241 (88.6)	117 (88)	124 (89.2)	0.85
How many average days per week do you provide care?	5.46 (2.3)	5.33 (2.3)	5.59 (2.3)	0.43
How many average hours per day do you provide care?	7.03 (8)	6.92 (8.1)	7.15 (7.9)	0.84
Can you please describe the type of care or support that you provide for the patient? (check all that apply)
Transportation	183 (67.3)	94 (70.7)	89 (64)	0.25
Household chores	206 (75.7)	103 (77.4)	103 (74.1)	0.57
Handling finances	119 (43.8)	65 (48.9)	54 (38.9)	0.11
Caring for other people for whom the patient was providing care prior to his/her illness	34 (12.5)	15 (11.3)	19 (13.7)	0.58
Emotional and/or social support	227 (83.5)	110 (82.7)	117 (84.2)	0.75
Medical care	211 (77.6)	102 (76.7)	109 (78.4)	0.77
Getting around the house	47 (17.3)	30 (22.6)	17 (12.2)	0.03
Bathing, dressing, grooming	53 (19.5)	25 (18.8)	28 (20.1)	0.88
Toileting	18 (6.6)	8 (6)	10 (7.2)	0.81
Using the bathroom	13 (4.8)	7 (5.3)	6 (4.3)	0.78
Walking, transferring, or climbing stairs	47 (17.3)	24 (18.1)	23 (16.6)	0.75
Feeding	10 (3.7)	6 (4.5)	4 (2.9)	0.53
Pet care	85 (31.3)	46 (34.6)	39 (28.1)	0.30
Other	20 (7.4)	13 (9.8)	7 (5)	0.17
Do you currently provide care for anyone else? (% answering yes)	60 (22.5)	29 (21.8)	31 (23.1)	0.88

Bereaved caregivers characteristics

There was no significant difference in length of follow-up between patients in the facilitated and patient-directed arms (505 days (Standard Deviation (SD) 414) versus 500(SD 393) p=0.85). Survival analysis using the Log-Rank test revealed no significant difference in patient survival between treatment arms (χ² = 0.2380, df = 1, p = 0.6256). Among the 98 bereaved caregivers (50 patient-directed, 48 facilitated), demographic and caregiving characteristics were similar to the full cohort. Like the overall sample, bereaved caregivers were predominantly female (73, 74.5%), Caucasian (92, 93.9%), and married (81, 82.7%), with comparable rates of being spouses/partners (62, 63.3%) and providing extensive care (averaging 5.58 days/week and 6.83 hours/day). No significant differences were observed between intervention arms among bereaved caregivers. The full characteristics of bereaved caregivers are presented in Table S3.

Anxiety, depression, and post-traumatic stress symptoms

At baseline, caregivers demonstrated low depressive symptomatology (mean HADS Depression score=4.09, SD=3.48) and slightly higher levels of anxiety (mean HADS Anxiety score=7.22, SD=4.27). These scores were similar between study arms (Table 3). Adjusted analyses using mixed-effects models showed no significant differences between groups in caregiver depression (adjusted mean difference −0.52, p=0.17) or anxiety symptoms (adjusted mean difference −0.72, p=0.11), neither of which increased.

Table 3:

Caregivers symptoms of anxiety and depression by intervention group at baseline and 12 weeks

	Patient-Directed ACP (N=102)	Facilitated ACP (N=108)	Adjusted mean differencea	P value
Depression symptoms (HADS) Mean(SD)
Baseline	3.79 (3.35)	3.63 (3.37)
12 Weeks	4.34 (4.07)	3.69 (3.57)	−0.52	0.17
Anxiety Mean(SD)
Baseline	6.65 (3.75)	7.05 (4.27)
12 Weeks	6.94 (4.6)	6.5 (3.92)	−0.72	0.11

^aAdjusted for baseline measure,

Mental health outcomes among bereaved caregivers (n=98) are depicted in Figure 2. Caregivers in the Facilitated ACP arm reported significantly fewer post-traumatic stress symptoms compared to the Patient-Directed arm (mean IES-R scores 23.9 vs 31.5, p=0.01). Depression (5.34 vs 5.87, p=0.50) and anxiety symptoms (6.56 vs 6.72, p=0.84) remained similar between arms and low overall.

Figure 2: Mental health outcomes among bereaved caregivers.

Symptoms of anxiety and depression (measured using HADS score, score of ≥ 7 on either domain indicates significant symptoms of depression or anxiety) and PTSD symptoms (measured using IES-R, score of 24 or higher is considered clinically significant) among bereaved caregivers.

Goal-concordant care and quality of end-of-life care

As depicted in Figure 3, most bereaved caregivers (83, 85.6%) reported that patient wishes were followed “a great deal” in the last month of life, with higher rates of caregiver-reported goal-concordant care in the facilitated vs patient-directed ACP arm (95.8% vs 75.5%, p=0.01). While most patients preferred to die at home (76.5%), only 54.1% actually did. Overall, 72.5% of patients died in their preferred location, with slightly higher but non-significant rates of patients dying in their preferred location in the Facilitated ACP arm (77% vs 68% p=0.31). Quality of EOL care as measured by the CEQUEL scale was high and similar between arms for all domains (mean overall score of 24.40 (SD 1.93), indicating that caregivers perceived EOL care to be of high quality).

Figure 3: Bereaved caregiver perceptions of the extent to which patient wishes were followed in medical care during the last month of life.

Bereaved caregivers answers to the questions “In your opinion, to what extent were [the patient’s] wishes followed in the medical care received in the last month of life?”

Missing Data

Approximately 23% of all enrolled caregivers had missing follow-up data. Among bereaved caregivers, only 98 of 140 eligible caregivers completed the bereavement assessment. The sensitivity analyses confirmed the complete case results for all outcomes except for the question on goal concordant care question (presented in figure 3). The difference was likely attributable to extremely sparse data in the ‘Not at all’ category, which had only 1 participant in the Patient-directed ACP group and none in the Facilitated ACP group. We then combined the ‘Somewhat’ and ‘Not at all’ categories and reassessed group differences. With this adjustment, the sensitivity analyses aligned with the CC results, indicating that participants in the Facilitated ACP group were more likely to report that the patients’ wishes were followed ‘a great deal’ in the medical care during the last month of life.

Discussion:

In this randomized clinical trial comparing facilitated ACP to patient-directed ACP, we found that a facilitated approach was associated with significantly lower post-traumatic stress symptoms and higher perception of goal-concordant EOL care among bereaved caregivers of patients with advanced cancer. To our knowledge, this is the first study to assess the effects of ACP on PTSD symptoms among bereaved caregivers of patients with cancer.

While patients with cancer undergo significant life changes and hardship post-diagnosis, these effects extend beyond the patient to the family and friends caring for them. A 2016 report by the National Alliance for Caregiving found that, compared to non-cancer caregivers, cancer caregivers provide care for a shorter period of time, approximately two years (although this duration has likely extended as cancer survival continues to increase)¹¹. Despite the short duration, the report found that the burden of caring for someone with cancer was high: 62% of cancer caregivers were in a high-burden situation¹¹. The significant physical, psychological, and financial implications of cancer caregiving have been well documented^40–46.

A particularly stressful aspect of caregiving is the responsibility of medical decision-making^{47, 48}. Caregivers often find themselves navigating complex healthcare choices with limited preparation, potentially leading to psychological distress. Ozdemir et al found that in an oncology setting, higher family involvement in decision-making was associated with higher caregiver burden and psychological distress ⁴⁷. In a more recent study investigating the stresses of surrogate decision-makers, respondents unanimously ranked their experience as a surrogate as one of the most stressful experiences of their lives⁴⁸. Surrogates reported significant stress mitigation when they felt they had been helpful, knew the patient’s wishes (P = .0011), specifically discussed patient wishes, or the patient’s wishes were documented⁴⁸.

Two prior studies in noncancer populations have demonstrated reduced PTSD symptoms using a facilitated ACP approach^{17, 20}. Detering et al investigated the effect of facilitated ACP, using the Respecting Choices model versus usual care in patients eighty years or older and their caregivers²⁰. Compared with the control group, the family members of patients who had died in the intervention group had fewer symptoms of post-traumatic stress as measured using the Impact of Events Scale. Family members of the intervention group were more likely than those of the control group to be very satisfied with the quality of death from both their perspective and the perceived perspective of the patient²⁰. Song et al investigated the effect of a facilitated ACP intervention versus usual care in patients with end-stage renal disease¹⁷. Adjusting for time and baseline scores, they found a reduction in PTSD symptoms in bereaved caregivers who engaged in facilitated ACP. Our study reinforces these results in patients with cancer while also showing that facilitated ACP led to lower PTSD symptoms than the patient-directed ACP arm. We did not find a change in HADS depression and anxiety scores between arms, likely as a result of the low scores of caregivers at baseline.

We also found a higher rate of caregiver-reported goal-concordant care in the facilitated ACP arm compared to the patient-directed arm. Our questions on goal-concordant care were the same as used in the CanCORS study, a study investigating the association of aggressive EOL care with bereaved family members’ perceptions of the quality of EOL care⁴⁹. In the CanCORS study, 81.1% of participants indicated that patients’ EOL wishes were followed a “great deal,” while 18.9% reported that EOL care was “somewhat” or “not at all” consistent with patients’ wishes, similar to our study⁴⁹. In our study, the perception that wishes were followed “a great deal” may provide important psychological protection for bereaved caregivers, potentially contributing to reduced post-traumatic stress. Although we were not powered in this study to examine differences in PTSD scores by levels of perceived goal concordant care, this is an important area for future investigation.

We previously reported that self-reported ACP engagement scores were higher and advance directive completion was more frequent with facilitated compared to patient-directed ACP, but there was no difference in rates of self-reported ACP conversation with caregivers at 12 weeks²⁴. Several potential mechanisms could help explain this difference in caregiver-reported goal-concordant care between arms. First, the interactive nature of facilitated ACP might have encouraged greater caregiver presence and involvement compared to the self-directed paper and web-based format. Participants in the facilitated arm were encouraged verbally during scheduling calls to include caregivers in the interaction. Second, even without necessarily increasing explicit ACP conversations with caregivers, the higher rates of ACP documentation in the facilitated arm possibly allowed caregivers to feel more aware of and confident in understanding patient preferences when making decisions.

Our study population aligned with national cancer caregiver demographics reported by the National Alliance for Caregiving/AARP ^{50, 51}, with most caregivers being female spouses providing intensive care, averaging 5.46 days per week and 7.03 hours per day. However, several limitations warrant consideration. First, our caregiver enrollment rate of 68% may indicate selection bias toward more engaged caregivers. Second, our predominantly homogeneous sample, largely non-Hispanic Caucasian, reflecting the demographic characteristics of participating oncology practices in western Pennsylvania, limits generalizability to more diverse populations. We lack data on caregiver presence during intervention completion, which would have provided valuable context for interpreting our results. Although our results were statistically significant our sample of bereaved caregivers was small. To maximize enrollment, we employed a broad definition of ‘caregiver.’ Additionally, many patients had not designated a surrogate decision maker at study enrollment, preventing us from determining whether the enrolled caregivers ultimately served as surrogate decision makers.

These findings have important implications for clinical practice and healthcare policy. The superior outcomes with facilitated versus patient-directed ACP suggest that providing ACP materials may be insufficient to optimize caregiver outcomes. Instead, dedicated resources for facilitated conversations may be necessary. This has particular relevance given the substantial psychological burden documented among caregivers involved in EOL decision-making. Future research should examine the cost-effectiveness of facilitated ACP, mechanisms in which facilitated ACP may lead to less PTSD among caregivers, and its impact on more diverse populations.

Conclusion:

This randomized clinical trial demonstrated that facilitated ACP, compared to a patient-directed approach, was associated with significantly lower post-traumatic stress symptoms among bereaved caregivers and higher rates of perceived goal-concordant care at the EOL. Given the substantial psychological burden that surrogate decision-making places on caregivers, particularly in the context of intensive cancer caregiving, these findings suggest an important advantage of facilitated over self-directed approaches to ACP. Future research should focus on understanding the specific mechanisms by which facilitated ACP may protect against caregiver post-traumatic stress, examining these effects in more diverse populations, and developing implementation strategies to make facilitated ACP more widely accessible in oncology care settings.

Data availability statement:

Data available upon request