Hospice Social Work Preferences for the Delivery of Facebook Support Groups: A Discrete Choice Experiment

Introduction

The National Cancer Institute (NCI) estimates that more than 1.8 million new cases of cancer were diagnosed in the United States during 2020; worldwide, that figure is expected to rise to 24 million by 2030.¹ Each patient has at least one friend or family member who, not always by choice, must help care for them as they face the disease, often for years. This paper defines family caregiver as a non-paid relative or friend who takes the responsibility to care for the hospice patient. Family is not defined by genetics but rather socially defined by the patient and identified by the hospice agency.

While research as shown caregiving to have some positive benefits, the overwhelming findings have been that the demands of caregiving someone exact a toll on the physical, emotional, and mental health of such caregivers.² The burdens are even greater for informal family caregivers of persons with advanced cancer.³ As many as one-third of hospice family caregivers of patients with advanced cancer experience high levels of depression and anxiety.⁴ Ample research demonstrates that, as cancer patients near death, their caregivers’ psychosocial burden can even exceed that of the patients.⁴ This burden not only diminishes caregivers’ ability to provide quality care but also increases their own risk of mortality.² Our previous work with family caregivers in hospice indicates that those caring for patients with advanced cancer lack emotional and social support as well as sufficient information on how to care for their patients.^5,6 They report a range of anxieties including frustrations communicating with the patient, family members and their providers, uncertainty about disease progression such as symptoms and pain management, as well as negative emotions as they balance the demands of caregiving with other responsibilities and roles.^{7 8} Our team has recently completed a clinical trial to support caregivers of hospice cancer patients.

Access for Cancer Caregivers to Education and Support for Shared Decision-making (ACCESS)

In a clinical trial, Access for Cancer Caregivers to Education and Support for Shared Decision-making (ACCESS) enrolled 489 participants and found that those who participated in a private hidden Facebook educational and support group had significantly lower depression than those who received usual hospice care. The online Facebook intervention was feasible, caregivers were satisfied, and identified positive benefits.⁹

Like many interventions sponsored by the National Institutes of Health, despite its successful outcomes, ACCESS faces barriers in practice that make its translation into local hospices challenging. Research has shown that change becomes more manageable if those subject to it have an opportunity to participate in the decision making and express their preferences and concerns.¹⁰ Despite the efficacy shown by Facebook groups in our clinical trial, hospice staff have shared numerous barriers to their facilitation of such groups. Staffing shortages, financial pressures, tight productivity expectations, and administrative policy issues related to online interventions are potential barriers to hospice agency interest in offering this evidence-based intervention. Given these challenges, understanding the preferences of hospice social workers toward the delivery of evidence based online support groups is an important consideration to it’s widespread translation.

The objective of this study was to identify preferences for the nationwide dissemination of private Facebook support groups for caregivers of hospice patients. This project was grounded in the research question; Would hospice social workers prefer to facilitate online support groups themselves or refer to a trusting external resource? A second related set of questions asked; What are the perceived challenges for social workers facilitating online support groups and What would be the perceived challenges to referring family caregivers to others who provide online support groups?

Methods

The study was approved by the Washington University Institutional Review Board. We chose to address the research questions using a Discrete Choice Experiment (DCE).

Discreet Choice Experiment (DCE)

Discrete Choice Experiments are surveys designed to solicit preferences using hypothetical scenarios and asking participants to choose one over another. Each scenario is based on attributes that contribute to the ultimate choice. We used a three-step process to create a DCE for hospice social workers. First, we informally interviewed 3 hospice social workers asking them the general question, “What is your reaction to offering online Facebook groups to caregivers of your hospice patients.” Their responses provided us guidance on finalization of our DCE question and the attributes that would influence response choice.

The DCE offered two alternatives for translation of the online group into practice. The first DCE option was stated,” You and your social work colleagues would be trained to facilitate the group for your hospice families.” The second option stated, “You and your social work colleagues would refer caregivers to an external group (such as the state hospice association or your corporate office) who would facilitate the group for families in several hospice programs.” Each alternative had five attributes involving three level of choices for each attribute. Using an online Qualtrics survey, the social workers were asked to make a choice six different times, based on the random combinations of the three levels within the five attributes. See the one sample scenario on the DCE survey in Figure 1. Finally, as a last step prior to distribution we had 5 social workers review the DCE and provide feedback.

Figure 1. DCE Survey With One sample Block Scenario.

A convenience sample and snowball sampling method were used to identify and recruit hospice social workers for participation. The survey was mentioned three times on one list serve for hospice and palliative care social workers. Additionally, emails were sent to several individual hospice agencies who agreed to forward the opportunity to their social workers. Each hospice social worker responding to the survey received $50 for their participation.

The Experimental Design and Questionnaire

For this study, we developed a labelled binary, forced-response design. This type of design has two choices for each scenario, and the choices are labeled, i.e., they represent two different alternatives rather than the same alternative. For example, this means that the choice was between conducting the support group in-house (A) versus referring the service out (B) rather than two of the same alternatives (conducting the support group in-house) with varying attributes. It was also forced-response because there was no opt-out or “neither” option. We designed this DCE using the rotation method, which generated 18 total choice sets. To reduce the cognitive load on participants, participants were randomized to one of three blocks with 6 scenarios each. Each choice set had the same 5 attributes: 1) hours of work, 2) cost to agency, 3) caregiver mental health benefit, 4) change in caseload, and 5) change in pay. The levels for hours of work were 1) no change, 2) 4 hours increase, or 3) 8 hours increase. The levels for cost to agency were 1) $5 per caregiver, 2) 10$ per caregiver, and 3) $20 per caregiver. The levels for mental health benefit were 1) no decrease in depression, 2) 10% decrease in depression, and 3) 20% decrease in depression. The levels for change in caseload were 1) stay the same, 2) 5% lower caseload, and 3) 10% lower caseload. Finally, the levels for change in pay were 1) no change, 2) 5% increase, and 3) 10% increase.

In addition to the DCE scenarios, the survey also contained demographic variables including age, education, race, and ethnicity and context-specific variables such as respondent training and responsibilities, duration of work in hospice, and social media usage. Finally, it had 4 questions regarding the difficulty of completing the exercise as well as perceptions of the respondent about the value of Facebook support groups and their agency’s capacity and willingness to implement such groups, and two open ended questions asking about the perceived challenges with facilitating a Facebook support group and the challenges in referring to a Facebook support group.

Sample and Data Collection

Participants had to be or had to have been social workers in a hospice agency. Palliative care social workers were excluded. The DCE survey was deployed via Qualtrics.

Data Analysis

Prior to examining the overall preference between conducting support groups in-house versus referring to an external resource, we looked at descriptive statistics for the sample. Since we only use categorical variables, we examined frequencies and percent distribution. To examine difference in preference between facilitation within the agency and referral to a third party for Facebook support groups, we used a binary choice model, which is essentially a logit model where the outcome is the binary choice and the independent variables are dummy variables for each attribute level. In a labelled design, effects cannot be attributed completely to the attributes, and are more likely a function of the label itself¹¹. Due to this assumption regarding labelled designs, we primarily examined the estimate and significance level for the alternative specific constant (ASC) of the referral method to determine whether there was a difference in preference between within-agency facilitation and referral of Facebook support groups. To determine this ASC, the ASC for in-agency facilitation was fixed while the model estimated the effect of the ASC for referral compared to in-agency facilitation. In addition to our model results, we also fielded feedback from hospice social workers regarding the DCE, perceived value of support groups, and their agency’s resources and willingness to implement Facebook support groups. We examined these results with descriptive statistics, specifically, frequencies and percent distribution. We only used responses from participants that completed all 6 scenarios in their randomized block. Missing data were handled using case-wise deletion.

The open-ended question responses were reviewed by two authors. Clear themes were identified and then broken into sub-themes based on the very short responses. The coded utterance themes were then counted and frequencies created. When more than one challenge was provided by a participant, the first mentioned challenge was the one coded and counted.

Results

Preferences of Hospice Social Workers

Two hundred and fourteen hospice social workers participated in the survey. Forty-one of these respondents did not complete all of their assigned scenarios resulting in a total valid sample size of 173 hospice social workers for analysis. The sample demographics appear in Table 1. This sample represented social workers in hospices from 34 states, the largest percentage were from Missouri (17%) and Ohio (14%) with 22% not providing a state. A majority of these hospice social workers (87%) provided clinical care to patients and families. Over 50% of these social workers were between 40 and 60, 88% were white, and 88% had graduate degrees. Over 50% had 5+ years working in hospice. In terms of social media use, 93% of respondents had a Facebook account, and most used this account daily or several times per day (68%).

Table 1.

Respondent Characteristics

	Overall (N=214)
Responsibilities
Administration or management of staff and oversight of agency policy	19 (9%)
Clinical care of patients and families	186 (87%)
Other	8 (4%)
Missing	1 (0.5%)
Age
29 years or younger	23 (11%)
30–39 years	57 (27%)
40–49 years	67 (32%)
50–59 years	44 (21%)
60 or more years	20 (9%)
Missing	3 (1.4%)
Education
Bachelor’s degree	19 (9%)
Graduate degree	185 (88%)
Other	6 (3%)
Missing	4 (1.9%)
Years in Hospice
0–1 years	22 (10%)
1–5 years	64 (30%)
5–10 years	52 (25%)
More than 10 years	72 (34%)
Missing	4 (1.9%)
Race
White	183 (88%)
Black or African American	18 (9%)
Other	7 (3%)
Missing	6 (2.8%)
Hispanic or Latino?
No	190 (91%)
Yes	18 (9%)
Missing	6 (2.8%)
Have a Facebook Account
No	14 (7%)
Yes	196 (93%)
Missing	4 (1.9%)
Facebook Usage
Several times per day	49 (23%)
Daily	95 (45%)
Weekly	34 (16%)
Monthly or less	14 (7%)
Never	18 (9%)
Missing	4 (1.9%)

The largest proportion of respondents in a single category felt that the questions were neither difficult nor easy (45%), with a relatively even distribution in the easy and difficult categories. In terms of value to patients, the majority of hospice social workers (73%) felt that Facebook support groups would be very or extremely valuable. Only 55% of social workers felt their agency had the resources to provide this service, and 58% felt that their agency was unwilling to offer this service.

The 173 social work participants responded to a total of 1038 scenarios. The question response choice of referral to an outside provider was chosen 604 times (58%) as compared with facilitating Facebook support groups within the agency which was chosen 434 times (42%). The results for the binary choice model are presented in Table 3. While this table also includes estimates for choice attributes, as mentioned in the description of our analysis in the methods section, these are difficult to interpret given the complexity of a labelled design. Therefore, we pay primary attention to the ASC estimate and significance level to answer our research question and urge caution to readers when interpreting results for the attributes. Although referral was chosen more than facilitating Facebook support groups within the agency, the binary choice model required a significance level of .05 and the p value for the alternative specific constant (referring out compared to facilitating within the agency) was only .093, indicating there was not a significant difference between these choices.

Table 3.

Binary Choice Model Results

Variables	Estimate	SE	P value
Within Agency	-	-	-
Referral	−0.615	0.465	0.093
Time
No Additional Hours	-	-	-
4 additional hours	0.483	0.187	0.005
8 additional hours	0.186	0.174	0.142
Cost
No cost	-	-	-
$10/caregiver	0.227	0.182	0.106
$20/caregiver	−0.166	0.176	0.174
Mental Health
No change
10% decrease in anxiety and depression	−2.223	0.178	0.000
20% decrease in anxiety and depression	0.029	0.180	0.436
Caseload
No change	-	-	-
5% lower caseload	−0.272	0.172	0.056
10% lower caseload	0.365	0.185	0.024
Pay
No change in pay	-	-	-
5% increase in pay	−0.321	0.175	0.033
10% increase in pay	0.182	0.182	0.160

Open Ended Responses

The majority of respondents responded to the open-ended questions. There were 166 participants (77%) listing challenges to facilitating the groups and 153 (71%) identifying challenges with referring to others for facilitation of the online group. The replies were short, usually only a few words or at most one sentence.

Tables 4 and 5 outline the themes, sub themes, examples, and frequencies of the responses to the open-ended questions. The most frequently identified challenge for social workers in facilitating a Facebook group was related to the technology issues (39%). Specifically, issues related to privacy, HIPAA regulations and web security were 66% of the technology concerns. Organizational issues (34%) were the second biggest challenge. Specifically, staff time and work load were more than half (53%) of the organizational issues mentioned.

Table 4:

Categories of Responses to Open Ended question asking the challenges of facilitating a Facebook support group for family caregivers

General Category	Sub-Categories	Exemplars	Frequency
Organizational issue 57 (34%)	Policy/Administration/Government	We are part of a hospital and therefore regulated by it. I have attempted to do our own social media but it has been declined in the past.(81)	21 (37% of cat) (13% of total)
	Cost	Money (3)	2
	Staff Time or Availability	Not enough time to set up Facebook support group and maintain caseload on own (199)	30 (53% of category) (18% of total)
Technology Issue 66 (39%)	Privacy HIPAA Security	Privacy and confidentiality of pt/family information. Facebook is a public place and not guaranteed that pt/family information will remain private/protected. (137)	44 (66% of category and 27% total)
	Internet or Device Access	not all caregivers have a FB account or access to computer/phone with internet	11 (17% category; 7% total)
Practioner Issue 22(13%)	Monitoring of the group comments	the ability to provide 24/7 support and responding to post or questions that could come in at any time.(191)	11 (50% of category; 17% of total)
	Practioner Boundary Issue	Blurry lines with using Facebook with caregivers, personal vs. professional (60)	5 (23% of category; 3% total)
	Training	How to start, manage the content, & valid results (86)	6 27% of category; 3% total)
Caregiver Issue 25(15%)	Family Boundary Issue	Families using it as a place to complain about the agency instead of bringing to the attention of administration (152)	3 (12% of category 2% total)
	Digital literacy	While most older adults are growing to be more technologically savvy, most of the older adults I have worked with that were interested in support groups shared that they would be uncomfortable with an online group (in any capacity) as they don’t feel as comfortable on the computer/tablet/etc. (92)	22 (88% of category; 13% of total)
Other 11(7%)	Other	Getting them to warm up to the use of technology especially if the caregiver is a spouse.	11

Table 5:

Categories of Responses to Open Ended question asking the challenges of referring to an external resource for a Facebook support group for family caregivers

General Category	Sub-Categories	Exemplars	Frequency
Referral organization issue 44 (29%)	Finding an organization to refer to	Finding agencies willing to do this (103)	5
	Quality and Trust	It feels a bit like offloading what could be an integral offering by our team if there was time for it. It may also be easier to inspire participation if it’s internal and the team has connections with those caregivers. When we refer out, we don’t always know what type of experience they'll have	39 (89% of category)
Technology Issue 23 15%	Privacy	lack of confidentiality	5
	Technology Access	Internet connectuon in our area and inabiluty for our caregivers to understand or have access to a computer.	18
Practioner Issue 6 (4%)	time	I wouldn’t be able to provide them much tech support if they have difficulty connecting with/using the group.	4
	Remembering	Remembering to do so, as we have many other priorities.	1
	Not being in control	The challenge would be, not being involved and not knowing what support caregivers need.	1
Caregiver Issue 50 (33%)	Buy in of family	The biggest challenge would be getting people to "take the first step" and actually join/participate.	24
	Age/Digital literacy	Assuring that all caregivers have equal access as not all of them have access or knowledge of using social media	26
Local Hospice Issue 10 (7%)	Hospice looks bad	Perception that my agency is not equipped to provide adequately for the needs of the families we serve (39)	3
	Cost	Unsure my agency would or would be able to pay for this although this is preferred for me	5
	Organizational acceptance	Organizational acceptance	2
No issue 15 (10%)	No problem referring	I see it as no different than all of the other referrals I make to other agencies for services.	15
Other 5 (3%)		Not knowing enough info about the group to prepare the caregivers	5

The most frequently mentioned challenge in referring caregivers to an externally facilitated group involved caregiver issues (33%). Challenges were split nearly evenly between, the buy in and follow up of caregivers to referral and concerns regarding digital literacy of family caregivers. Referral organizational issues were the next largest challenge (29%) and nearly all (89%) of the referral organizational challenges were concerns about the quality of the services and the ability to establish trust in the external provider.

Discussion

The purpose of this paper was to identify the preferences of hospice social workers in facilitating a Facebook support group for caregivers. We also sought to identify perceived challenges and benefits of such support groups. While online support groups in health care are not uncommon, ^12–14 the details on the background of the facilitator and protocols on how they are facilitated are not well documented. Within the practice of hospice these details are even more rare, outside the work of this intervention.¹⁵

The evidence has found that family caregiver depression can be lessened with participation in a Facebook support group.⁹ Likewise, the overwhelming perceptions of social workers in this sample is that Facebook online support groups would be at least somewhat beneficial for hospice caregivers (96%). However, the specifics on how to offer this service are not as clear. Social Workers were nearly evenly divided in their perception the agency had the resources to provide groups, the majority did not believe the agency was willing to invest in the groups. When presented with two options for delivery, facilitating the group themselves or referring to an external resource, there was no statistically significant difference in preference. There were however, consistent themes in the challenges of both options. It is these themes that provide us the strongest support for a path forward.

When exploring issues related to social workers themselves facilitating Facebook groups within their specific hospice agency, the biggest concerns were related to the privacy and security of Facebook. Social workers were also clear that they currently lacked the time to take to facilitate and manage an online group. The challenges related to privacy and security are easily addressed with education.

Our experiences have been the private Facebook group settings provide adequate privacy and security in that the posts are not available to members outside the group and are not searchable. In addition, they allow the facilitator to control the account, enforce the agreed upon guidelines, and if necessary eliminate inappropriate posts.¹⁵ It is important to remember that the information which is shared by caregivers is their information to share thus, while important to protect the identity of those in the group, protecting the sharing of private information is at the discretion of the caregiver owning that information.

Expanding caseloads and documentation demands have created staffing issues for hospice social workers. In addition to being an effective intervention approach, group work has traditionally been a tool social workers have used to meet staffing challenges, helping many people at one time verses individual sessions. This tool has not been available in the hospice setting due to the homebound status of caregivers dealing with dying patients. However, especially following a pandemic where telehealth became normal, online groups can still offer the same time savings. However, many social workers have not been trained in the unique aspects of online group work, thus social workers comfort and skill level in this setting are not equivalent to in person groups.¹⁶ Despite the recent efforts of four social work organizations to update their guidelines for practice with electronic technology, specific guidelines and skill building with online support groups were not a part of that document.¹⁷ One of the perceived challenges identified by these social workers was the 24 -hour monitoring of the group. Our team has addressed this with a clear understanding with participants that the response time should not be expected in less than 24 hours. Setting expectations is key to the success of these groups.

Another perceived challenge for social workers is a concern that caregivers are unable and/or unwilling to participate in these groups. It is important to acknowledge the groups may not be for every family caregiver however, that does not mean they are not a valuable service. It has been shown that older persons who realize the benefits in technology and who have the support for its use are willing to learn how to use it.¹⁸ In the same manner, we have found that if seniors realize the benefit such groups offer, they are often willing to learn. AARP reports that in 2018 Facebook use by those over 65 years of age had grown to 33%. Similary, the PEW center reported 34% of seniors used Facebook and seniors were the fastest growing demographic on social media. ¹⁹ The Pew found that 70% of seniors on the Internet visited Facebook every day.²⁰ An important factor to widespread implementation of Caregiver online groups will be the assurance that social workers have the skills required to facilitate and older family caregivers have the knowledge and interest necessary to join and participate in them.

Equally enlightening were the challenges identified in the option of referring family caregivers to groups offered by an external provider. In this option, we found the two most significant perceived challenges to be the relationship with the referral organization, the concerns related to the participation of the family, and the concerns regarding the digital literacy of family caregivers. The trust and accountability in a referral source, the ease of use and ability to use that resource is always a concern when you stake your reputation on the line and refer someone to a resource. This is no different. The success of this delivery is dependent upon how the referral source manages the group and their relationship with the social worker, as well as how the coach and help the participants who struggle with the technology.

These data demonstrate that to implement online Facebook groups in local hospice agencies there is a need to educate agency personnel on the privacy and security of the private online support groups. Additionally, it is important to train social work staff to facilitate and monitor groups in the online setting. If on the other hand these groups are offered by a larger external organization outside the local agency, it will be important that social workers understand that, like any resource referral, the follow through by family will be variable. This can be assisted with follow through and strong encouragement by the social worker as well as assistance when possible. It will be important that social workers reach out to these referral groups to establish relationships with the leaders. They should find ways to receive feedback and review the quality of the service, thus establishing trust in the operations and content being given to caregivers. Likewise, these referral organizations will need to have strong resources available to address digital literacy concerns of participants.

While we had a strong nationally representative sample, this study generalization was limited by its lack of racial and ethnic minorities. While this is representative of the lack of diversity in hospice staffing and patients, it does present a bias opinion of a white population. Likewise, we are limited by the lack of a comparison group and a convenience and snowball sample. Finally, the type of DCE we designed did not allow us to determine the influence of the designated attributes on the ultimate choice made by the participant. Despite the limitations, we were able to obtain valuable feedback from hospice social workers and have direction for further study.

There is a range of potential possibilities for external groups to provide online Facebook groups. State Hospice Associations are struggling to find new roles and opportunities and might thus be an appropriate statewide option. ²¹ Another possibility for corporate hospices would be to offer the service from the corporate or regional office. We know that online Facebook groups reduce family caregiver depression ⁹ and we know from these data that 96% of social workers believe they are valuable for family members. Therefore, it is important to commit to providing this service as an option for family caregivers, even if the group is not an alternative for every caregiver. These groups offer a relatively low cost option that can yield high benefit for many family members. Further research is needed to understand potential implementation and operational strategies.

Table 2:

Post DCE Questions

	Overall (N=214)
	Overall (N=214)
DCE Difficulty
Very easy	6 (3%)
Fairly easy	22 (13%)
Easy	25 (15%)
Neither difficult nor easy	77 (45%)
Quite difficult	33 (19%)
Very difficult	9 (5%)
Missing	42 (19.6%)
Perceived Value of Facebook Support Groups
Extremely valuable	48 (28%)
Very valuable	78 (45%)
Somewhat valuable	39 (23%)
Neutral	5 (3%)
A little valuable	3 (2%)
Not valuable at all	0 (0%)
Missing	41 (19.2%)
Agency Has Resources to Provide
No	77 (45%)
Yes	95 (55%)
Missing	42 (19.6%)
Agency Willing to Provide
No	100 (58%)
Yes	71 (42%)
Missing	43 (20.1%)