Abstract
Objectives
The purpose of this study was to test an intervention named ACCESS (Access for Cancer Caregivers to Education and Support in Shared Decision Making). The intervention uses private Facebook support groups to support and educate caregivers, preparing them to participate in shared decision making during web-based hospice care plan meetings. The overall hypothesis behind the study was that family caregivers of hospice cancer patients would experience lower anxiety and depression as a result of participating in an online Facebook support group and shared decision making with hospice staff in a web-based care plan meeting.
Methods
This cluster crossover randomized three-arm clinical trial where one group participated in both the Facebook group and the care plan team meeting. A second group participated only in the Facebook group and the third group was a control group and received usual hospice care.
Results
There were 489 family caregivers who participated in the trial. There were no statistically significant differences between the ACCESS intervention group and the Facebook only or the control group on any outcome. The participants in the Facebook only group, however, experienced a statistically significant decrease in depression compared to the enhanced usual care group.
Conclusions
While the ACCESS intervention group did not experience significant improvement in outcomes, caregivers assigned to the Facebook only group showed significant improvement in depression scores from baseline as compared to the enhanced usual care control group. Further research is needed to understand the mechanisms of action leading to reduced depression.
Keywords: Hospice Care, Family Management, Psychological Care, Social Care
Background
The home of a dying cancer patient is said to become like a hospital room where family members manage all aspects of patient care. Family caregivers (FCGs) provide care without formal healthcare education while isolated from professional healthcare providers who could provide timely care decisions. At the same time, social support from family and friends has been shown to decrease as patients near the end of life.1 Sixty-six percent of hospice care is delivered in the place patients call home.2 More than a third of hospice patients (36%) die from a cancer diagnosis.2 Despite hospice teams’ support, one-third of family caregivers report moderate to severe anxiety and more than half report being depressed.3
Hospice standards consider patients and their family as a single unit of care, and participation in decision-making is consistent with the hospice philosophy. Despite this philosophy, patients and their families are typically absent from meetings where the hospice team makes critical decisions regarding care. Care plan meetings, held in offices up to an hour from patients’ homes, involve discussions of up to 100 patients and can last 3–4 hours. Thus, patient or family participation in these meetings is problematic and uncommon because they have caregiving demands that prevent travel and they would only be engaged with the team for a few minutes.4 As a result, hospice teams have few opportunities to engage patients or family caregivers in a shared decision-making process, and there has been no intervention or evaluation of shared decision-making in the hospice setting.5
Introduction
As early as 1959, physicians discussed the value of active roles for patients in their own care.6 Studies show that 71% of patients prefer to have a role in decision-making.7 Although active roles in decision-making not foreign to oncology, hospice, or palliative care, implementing a shared decision-making intervention by integrating evidence based decision science into hospice care teams is new to hospice.8 A report by the Agency for Healthcare Research and Quality (AHRQ) reviewed the research related to decision-making for individuals facing the end of their life. No hospice intervention studies for shared decision-making were reported.5
Shared decision making (SDM) has produced positive outcomes for oncology patients in non-hospice settings.9,10 These outcomes include improvement in patient anxiety.11 Research suggests that health care providers can improve family caregivers’ anxiety by listening and responding to questions, and by providing information with brochures, videos, and group discussions.12 Providing information and education supports the caregiving role and has the potential to influence family caregivers’ health and satisfaction.12,13 In summary, evidence indicates that providing information and emotional support to family members can enhance shared decision-making, increase knowledge, and reduce anxiety.14–16
Conceptual Model: Team Practice and Shared Decision-making
Figure 1 represents the conceptual model grounding our intervention by combining a model of interdisciplinary team collaboration that includes a role for family17,18 with a SDM framework.19,20 In a collaborative process, providers and FCGs share knowledge, treatment options, and the patient’s values, beliefs, and preferences to arrive at a shared decision.21 The collaborative model, inclusive of patients and family,18 specifies four components: organizational context, team structure, team process, and outcomes, with feedback loops between all components. Integrated into this model, a published literature review found nine critical components of a SDM process. In our study model, the social interaction between the individual factors and communication climate contribute to shared decision-making via the nine critical elements: 1) defining a problem, 2) identifying potential solutions, 3) discussing benefits, risks and costs, 4) clarifying patient/FCG values and preferences, 5) discussing patient/FCG beliefs regarding their ability to follow through with solutions, 6) offering a provider recommendation, 7) clarifying patient/FCG understanding, 8) making a decision, and 9) follow-up.20 ACCESS was designed to add family to the team structure and improve the communication climate through web-conferencing access to the team and using the nine critical elements of SDM.
Figure 1.
Model for Team Meeting Inclusive of Family Members
The Access for Cancer Caregivers to Education and Support for Shared Decision Making (ACCESS) Intervention
While the details of the ACCESS intervention are reported elsewhere,22 in summary, the intervention provided education and social support for family caregivers of hospice cancer patients while using web conferencing to enable them to participate in the bi-weekly hospice interdisciplinary care plan meetings required by Hospice Medicare Conditions of Participation.23 We used a three-arm design (ACCESS, Facebook only, enhanced usual care) with each cluster being randomized to two of three arms separated by a 90 day wash out period. Randomization is detailed elsewhere.22
Family caregivers in the ACCESS intervention group were enrolled in a private, facilitated Facebook group for education and social support, and they participated in care plan meetings with hospice staff. The Facebook facilitator provided education and social support to participants. Additionally, web conferencing connected family members with the hospice team at the bi-weekly care plan meeting, allowing caregivers to engage in SDM with the team. Hospice teams were trained in the nine elements of shared decision making.20 Family members in the Facebook only group did not participate in the care plan meetings, only in the Facebook group. Those in the enhanced usual care group were not involved in an intervention, only their staff were trained in shared decision making.
As shown in the conceptual model supporting the intervention, ACCESS was designed to shape the team process in two ways: 1) using Facebook groups to strengthen the communication climate with information and emotional support, and 2) educating hospice teams to use the nine elements of the SDM process in the meeting. While ACCESS was not designed to directly affect with individual factors outlined in the conceptual model, the factors were assessed as part of the research study, allowing them to later be included in the analysis as covariates. SDM is viewed as a social process in which individual factors of family and team members (e.g., socio-demographics, communication skills) interact with the communication climate during social interaction.19 The communication climate consists of information held by individuals and their expressed emotional state.
Because hospice considers the patient and family as a single unit of care, ACCESS extends social interaction to include family as proxies for patients who are unable to interact themselves. Our premise, based on the literature, was that social media can provide a platform for decision support that facilitates SDM by providing information and emotional support.24 25–27 The Facebook portal has been used as a forum for education and to allow participants to share experiences, information about diagnoses, and disease management advice.20 ACCESS used Facebook groups to provide online information and emotional support to prepare caregivers for participation in the meetings and SDM. Web conferencing served as the social interaction platform used to include family in the SDM process.
Hypotheses
There were two guiding research questions behind the study: 1) What is the effect of the ACCESS intervention on family caregivers? 2) How does the baseline level of anxiety affect the intervention? These questions were operationalized as three hypotheses: 1) Family caregivers participating in both the Facebook only and the ACCESS intervention will report lower levels of anxiety, decreased depression, increased pain knowledge, and lower patient pain compared to the enhanced usual care group at the time of the last follow-up, 2) Family caregivers participating in the ACCESS group (Facebook plus shared decision-making in team meetings) will report lower levels of anxiety, depression, increased levels of pain knowledge, and lower patient pain compared to either the Facebook only or the enhanced usual care group at the time of the last follow-up, 3) The effect of the ACCESS intervention group compared with the usual care will be greater for those with worse anxiety scores at baseline.
Methods
Trial Design, Setting, and Participants
The study took place with caregivers whose patients were enrolled in one of seven participating hospice agencies in the Midwestern state of Missouri in the United States between June 2017 and August 2021. The agencies were chosen based on our existing relationships, their interest and willingness to participate, number of annual admissions, length of stay, and proximity to the Principal Investigator. They represented rural, suburban, and urban populations within one Midwestern state. The intervention was tested with a three-group (ACCESS, Facebook support only, and enhanced usual care) cluster crossover design. Randomization and group assignment are detailed elsewhere.22 In brief, hospice agencies served as clusters and were randomly assigned to two of the three groups. Three team members together computer generated a random number which corresponded to a specific hospice. They first chose the group that would be ACCESS intervention sites, then the Facebook sites with the final groups serving as control. Each hospice agency assignment crossed over every 90 days, rotating the two-group assignment, with a 90 day washout separating the two group assignments. The trial was stopped when the required number of participants had enrolled and completed the study. Table 1 outlines the cluster crossover design by hospice.
Table 1.
Summary of Cluster Crossover Design
| Cluster (number of crossovers) | ACCESS Group Referrals (consents) |
FB Group Referrals (consents) |
Control Group Referrals (consents) |
Total by site Referrals (consents) |
|---|---|---|---|---|
| Hospice I ( 2) | 131 (28) | 327 (94) | - | 458 (122) |
| Hospice W (2) | 183 (47) | - | 54 (28) | 237 (75) |
| Hospice G (2) | 84 (23) | - | 23 (16) | 107 (39) |
| Hospice C (2) | 22 (5) | 135 (57) | 19 (11) | 176 (73) |
| Hospice O (2) | - | 26 (10) | 54 (22) | 80 (32) |
| Hospcie J (2) | - | 38 (11) | 84 (49) | 122 (60) |
| Hospice B (1) | 76 (46) | 125 (42) | - | 201 (88) |
| Total Clusters per group | 5 | 5 | 5 | |
| Total Referrals (consents) per group | 496 (149) | 651 (214) | 244 (126) | 1,391 (489) |
Family caregivers participating in the study were required to be at least 18 years of age, involved in decisions related to the patient’s care, and be a designated family caregiver for a cancer patient enrolled in a participating hospice. They also had to be willing to set up a Facebook account and log in to the account at least once a week. Caregivers without an Internet-enabled device were supplied with an Internet phone and data plan (restricted to Facebook posting and care planning time) for the duration of their participation. Patients of family caregivers had to have a projected life expectancy of more than two weeks as determined by a score of 30 or more on the Palliative Performance Scale28,29 assessed by the hospice nurse.
Internal and External Validity
Treatment fidelity in study design, training, intervention delivery, and treatment receipt was monitored with a series of checklists to ensure appropriate and consistent protocol implementation. A random sample of 25% of the video-conference sessions were recorded and analyzed throughout the study. The Facebook group was regularly monitored by the Co-investigators to ensure compliance with the protocols and Rules of Conduct. Facebook posts were downloaded throughout the study to create a database of qualitative evidence related to the intervention (results published elsewhere).
Measures/ Data Collection/Data Analysis
Based upon our conceptual model, published literature, and our preliminary work, the primary measure for the study was anxiety (General Anxiety Disorder-7).30 The measure ranges from 0 to 21 and has an MCID of 2 or 3, a larger value meaning more severe anxiety (Kroenke et al., 2016) The study was powered at .80 to detect a difference of 4 points in the anxiety measure based on outcomes from our preliminary work in a similar population.31 We also measured depression (PHQ-9),32 which ranges from 0 to 27 and has an MCID of 2 or 3 as well, a larger value meaning more severe depression (Kroenke et al., 2016). In addition to GAD and PHQ, we measured family knowledge of pain (Family Pain Questionnaire)33, and caregiver perception of patient pain (numeric pain scale).34 Data were collected using the Research Electronic Data Capture (REDCap) program with surveys at baseline, 14, 30, 60, and 90 days. The collection schedule reflected the Medicare requirement for required meetings every 14 days and our desire to have a measure close to the patient’s death without unduly burdening the family.
Prior to testing the hypotheses, groups were compared for age, gender, race, ethnicity, marital status, education, employment status, and relationship to patient. The only significant difference between groups was for race, which reflected the demographic differences in the location of the sites. Finding no other significant differences between groups based on these variables, we proceeded with the analysis. Given the longitudinal nature of the data and the number of timepoints, we conducted a mixed model analysis and a last-measure analysis. As there were no striking findings from the mixed effects model, we report only the simpler last-measure analysis.
We first created a variable that contained the value of the last observed score for each case. For instance, excluding baseline, if time 2 contained a value and all subsequent time points did not, the case would be assigned time 2’s value; if time 3 contained a value and subsequent time points did not, the case would be assigned time 3’s value, and so on. We then computed the difference between the last observed measure and the baseline measure for each outcome. The intervention group variable consists of three groups, 0=enhanced usual care, 1=Facebook only, 2=ACCESS, which is the combination of Facebook and in-person meetings.
The clusters in general had very low intra-cluster correlation (ICC) values (ranging from .004 to .05), suggesting that participants were largely independent and were thus treated accordingly. We first examined descriptive statistics for both the patient and their caregiver as well as the caregiver outcomes. For categorical variables, we examined frequency distributions, and for continuous variables, we examined mean, standard deviation, median, and range. We used Chi-square analysis for categorical variables and Kruskal-Wallis for continuous variables as age was not normally distributed.
To test our hypotheses about differences in the change in anxiety and depression between baseline and last measure, we used Kruskal-Wallis as our omnibus test and Dunn tests as post-hoc tests. While Dunn tests are typically only required for significant results, we perform Dunn tests for both outcomes regardless of Kruskal-Wallis result given the validity of these tests regardless of omnibus results and because hypotheses of clinical trials such as this one lend themselves more toward pair-wise methods rather than omnibus tests (Hsu, 1996). For our Kruskal-Wallis tests, we consider results with a p-value <.05 to be statistically significant, and for our pair-wise tests, we consider results with a p-value <.025 to be statistically significant, to adjust for family-wise error (inflated alpha when using multiple t-tests with Dunn test). We then compared the means between the baseline and last measure to better report the between-group differences in change. Missing data were handled using case-wise deletion given the last observation carried forward analysis.
Results
Table 1 outlines the total number of referrals and participants assigned to each group and coming from each cluster (hospice) as well as the number of crossover rounds for each cluster. As shown, we had a total of 5 clusters in each group. All but one cluster crossed over two times. Hospice B crossed over only once because it was in the study for less time. It is noteworthy that Hospice C was originally randomized to the ACCESS group, to be followed by the enhanced usual care group. However, after enrolling five participants in the ACCESS arm, an operational issue resulted in hospice agency’s refusal to participate in the ACCESS group, forcing us to reassign that arm to the Facebook only group. Thus, there were only five participants in the ACCESS arm in that cluster. Figure 2 is a diagram outlining the recruitment journey of participants by group assignment. We had 1391 qualified referrals resulting in 489 consents, an overall consent rate of 35%.
Figure 2.
Diagram by Treatment Group
Sample descriptive caregiver data are presented in Table 2. Group distribution was unequal. Unequal groups were expected due to differing size, attrition, lengths of stay, referral numbers, and patient and hospice characteristics between programs. The mean ages for each group did not significantly differ. Overall, participants were majority female, white, and non-Hispanic. Most caregivers were partnered, had at least a college education, and were employed at least part-time. The majority of caregivers were either the spouse or adult child of the patient. There were no significant differences between groups at baseline on any demographic variables except for race. This is likely due to differences in racial composition between sites. The two urban sites (I and B) had higher proportions of Black/African American caregivers than those in suburban or rural areas. There were no differences between groups in any of our analytic samples except for age in our sample for family knowledge of pain.
Table 2.
Summary of Caregiver Demographics by Group [N (percentage) unless otherwise indicated]
| Control (N=126) |
Facebook only (N=214) |
ACCESS (N=149) |
P-value1 | |
|---|---|---|---|---|
| Age (years) | ||||
| Mean (SD) | 57.8 (12.2) | 55.0 (12.8) | 56.5 (12.0) | .15 |
| Median [Min, Max] | 59.0 [22.0, 90.0] | 56.0 [25.0, 82.0] | 58.0 [22.0, 81.0] | |
| Missing | 1 (0.8%) | 1 (0.5%) | 6 (4.0%) | |
| Gender | ||||
| Male | 39 (31%) | 45 (21%) | 29 (20%) | .05 |
| Female | 87 (69%) | 169 (79%) | 119 (80%) | |
| Missing | 0 (0%) | 0 (0%) | 1 (0.7%) | |
| Race | ||||
| Black/African American | 6 (5%) | 34 (16%) | 16 (11%) | .01 |
| White/ | 118 (94%) | 172 (80%) | 125 (87%) | |
| Other | 2 (2%) | 8 (4%) | 2 (1%) | |
| Missing | 0 (0%) | 0 (0%) | 6 (4.0%) | |
| Ethnicity | ||||
| Non-Hispanic | 125 (99%) | 209 (98%) | 141 (97%) | .49 |
| Hispanic | 1 (1%) | 5 (2%) | 4 (3%) | |
| Missing | 0 (0%) | 0 (0%) | 4 (2.7%) | |
| Marital Status | ||||
| Single, never partnered | 13 (10%) | 35 (16%) | 20 (14%) | .38 |
| Partnered | 99 (79%) | 149 (70%) | 103 (70%) | |
| Divorced/separated | 11 (9%) | 17 (8%) | 15 (10%) | |
| Widowed | 3 (2%) | 7 (3%) | 8 (5%) | |
| Other | 0 (0%) | 5 (2%) | 2 (1%) | |
| Missing | 0 (0%) | 1 (0.5%) | 1 (0.7%) | |
| Education | ||||
| Less than high school | 4 (3%) | 3 (1%) | 6 (4%) | .39 |
| High school or equivalent | 34 (28%) | 37 (17%) | 30 (20%) | |
| Some college/trade school | 40 (33%) | 91 (43%) | 56 (38%) | |
| Undergraduate degree | 29 (24%) | 44 (21%) | 29 (20%) | |
| Graduate/professional degree | 15 (12%) | 36 (17%) | 25 (17%) | |
| Other | 1 (1%) | 2 (1%) | 2 (1%) | |
| Missing | 3 (2.4%) | 1 (0.5%) | 1 (0.7%) | |
| Employment Status | ||||
| Not employed/retired | 63 (51%) | 81 (39%) | 65 (44%) | .10 |
| Employed part-time | 8 (7%) | 29 (14%) | 13 (9%) | |
| Employed full-time | 52 (42%) | 97 (47%) | 70 (47%) | |
| Missing | 3 (2.4%) | 7 (3.3%) | 1 (0.7%) | |
| Income | ||||
| Under $20,000 per year | 15 (14%) | 27 (14%) | 14 (11%) | .80 |
| $20,000 to $39,999 | 23 (22%) | 39 (20%) | 28 (22%) | |
| $40,000 to $69,000 | 31 (30%) | 50 (26%) | 42 (33%) | |
| Over $70,000 | 36 (34%) | 77 (40%) | 44 (34%) | |
| Missing | 21 (16.7%) | 21 (9.8%) | 21 (14.1%) | |
| Relationship to Patient | ||||
| Adult child | 56 (44%) | 96 (45%) | 63 (42%) | .65 |
| Spouse/partner | 44 (35%) | 60 (28%) | 43 (29%) | |
| Other relative | 13 (10%) | 31 (15%) | 26 (17%) | |
| Other | 13 (10%) | 26 (12%) | 17 (11%) | |
| Missing | 0 (0%) | 1 (0.5%) | 0 (0%) |
P values were derived from Chi-square tests for categorical variables and Kruskal-Wallis tests for continuous variables.
As demonstrated by Table 3, similar to the caregiver characteristics, the patient characteristics were not significantly different between the groups. Groups were, in general, balanced on observed characteristics.
Table 3:
Summary of Patient Demographics by Study Group
| Control (N=126) |
Facebook only (N=214) |
ACCESS (N=149)1 |
P-value | |
|---|---|---|---|---|
| Age | ||||
| Mean (SD) | 71.9 (12.4) | 71.9 (14.2) | 71.5 (13.6) | 0.926 |
| Median [Min, Max] | 72.0 [27.0, 97.0] | 73.0 [29.0, 99.0] | 72.0 [25.0, 96.0] | |
| Missing | 8 (6.3%) | 13 (6.1%) | 10 (6.7%) | |
| Gender | ||||
| Male | 65 (52%) | 108 (51%) | 77 (52%) | 0.979 |
| Female | 61 (48%) | 105 (49%) | 72 (48%) | |
| Missing | 0 (0%) | 1 (0.5%) | 0 (0%) | |
| Race | ||||
| Black/African American | 7 (6%) | 33 (15%) | 15 (10%) | 0.0811 |
| White | 115 (93%) | 177 (83%) | 129 (89%) | |
| Other | 2 (2%) | 3 (1%) | 1 (1%) | |
| Missing | 2 (1.6%) | 1 (0.5%) | 4 (2.7%) | |
| Ethnicity | ||||
| Non-Hispanic | 123 (98%) | 213 (100%) | 142 (97%) | 0.189 |
| Hispanic | 3 (2%) | 1 (0%) | 4 (3%) | |
| Missing | 0 (0%) | 0 (0%) | 3 (2.0%) | |
| Marital Status | ||||
| Single, never partnered | 11 (9%) | 26 (12%) | 18 (12%) | 0.698 |
| Partnered | 57 (46%) | 92 (44%) | 59 (40%) | |
| Divorced/separated | 16 (13%) | 32 (15%) | 21 (14%) | |
| Widowed | 38 (30%) | 60 (28%) | 49 (33%) | |
| Other | 3 (2%) | 1 (0%) | 1 (1%) | |
| Missing | 1 (0.8%) | 3 (1.4%) | 1 (0.7%) | |
| Education | ||||
| Less than high school | 14 (11%) | 32 (15%) | 15 (10%) | 0.396 |
| High school or equivalent | 52 (42%) | 78 (37%) | 53 (36%) | |
| Some college/trade school | 30 (24%) | 64 (30%) | 47 (32%) | |
| Undergraduate degree | 15 (12%) | 23 (11%) | 19 (13%) | |
| Graduate/professional degree | 13 (10%) | 13 (6%) | 10 (7%) | |
| Other | 0 (0%) | 1 (0%) | 3 (2%) | |
| Missing | 2 (1.6%) | 3 (1.4%) | 2 (1.3%) |
P values were derived from Chi-square tests for categorical variables and Wilcoxon two-sample tests for continuous variables.
Results of the Kruskal-Wallis tests for depression and anxiety change as well as first and last measures are presented in Table 4. While the result for anxiety is above the threshold for statistical significance, the result for depression is p = .05, indicating that there may be a significant difference between the three groups’ mean changes.
Table 4.
Effect of Facebook online support and ACCESS with caregivers hospice patients our clinical trial
| Enhanced usual care | Facebook only | |||
|---|---|---|---|---|
| z | p | z | p | |
| Anxiety (n=294) | ||||
| Facebook only | 1.60 | .055 | - | |
| ACCESS | −1.37 | .085 | 0.18 | 0.428 |
| Depression (n=294) | ||||
| Facebook only | 2.37 | .009* | ||
| ACCESS | −1.87 | .031 | 0.45 | 0.33 |
| Knowledge of Pain (n=117) | ||||
| Facebook only | 0.39 | .349 | ||
| ACCESS | −0.72 | .236 | −0.38 | .348 |
| Patient Pain (n=96) | ||||
| Facebook only | 1.62 | .053 | - | - |
| ACCESS | −1.18 | .119 | 0.09 | .465 |
Note: Overall significance determined by Kruskal Wallace test, and paired significance determined by Dunn post hoc analysis. P values must be below .025 for significance due to family-wise correction.
The results for Dunn tests for depression and anxiety are presented in Table 4, and the baseline and last measure means as well as the change are reported in Table 5. For anxiety, neither of the pair-wise tests indicate a significant difference between either control and Facebook only or control and ACCESS, whereas for depression, the results indicate a significant difference between control and Facebook only (p < .01).
Table 5.
Mean Differences in Depression Score, by Study Arm
| Study Arm | Baseline Mean (SD) |
Last Measure Mean (SD) |
Mean Difference |
|---|---|---|---|
| Enhanced Usual Care | 7.05 (6.59) | 6.92 (6.45) | 0.14 |
| 7.88 (6.20) | 6.70 (5.46) | 1.18 | |
| ACCESS | 7.87 (6.30) | 6.64 (5.32) | 1.23 |
Discussion
The ACCESS intervention was designed to test the effect of shared decision making between family members and the hospice on caregiver outcomes. We hypothesized that SDM and increased information and emotional support would decrease anxiety and depression for caregivers. We rejected this hypothesis. We also found no differences in our secondary outcome, depression, in the ACCESS group. The Facebook education and social support were designed to play a supportive role to the team care plan meetings as they facilitated information and social support as a foundation for family members participating in the meetings. The explanation for the lack of significant changes in caregiver outcomes can be found in our conceptual model mixed with findings from other published studies from this project.
Our conceptual model noted an interdependency between the hospice organizational context, the team structure, team process, and finally, the decisions resulting from team meetings. We assumed the organizational context in hospice would support family involvement and thus the intervention focused on changing the team structure and team process in an anticipation of a change in outcomes. Using a pragmatic design, we trained the hospice staff in SDM and observed the SDM within their team process; we did not mandate this structure.
Our recent analysis and publication of the care plan meeting recordings found variable use of the 9 elements of SDM.35 The qualitative analysis of SDM in team meetings concluded that the use of all 9 elements of SDM was not feasible given the short time available for discussion and the regulatory requirements related to the content and structure of communication. Therefore, the organizational context was not supportive of SDM, which was contrary to our basic assumption behind the intervention.
Surprisingly, we did find a significantly positive change in caregiver depression in the Facebook only group. This group was designed to be a second comparison group so we could explore the effect of Facebook versus the effect of SDM on caregiver anxiety and depression. In fact, Facebook came to serve as a simpler and more effective intervention for caregiver depression. While the improvement in caregiver depression is not clinically significant, it is statistically significant, which leaves open the possibility to further explore Facebook support and strengthen its potency. If we can further understand the mechanisms of action addressing the caregiver stressors and strains leading to the change in depression, it is possible the intervention can be customized to target and strengthen those mechanisms of action to provide a clinically significant effect.
The literature shows that social support offered in online support groups can serve as a mechanism of change and improve depression.36–39 Theories of social support point to four types: esteem/emotional, informational, instrumental, and social companionship support.39 In this study, without intent, our Facebook only group participants received most of these types of social support. In future research we could adapt the online group educational material and structure to focus on the types of social support and measure the effect of each type of support on caregiver depression.
While this clinical trial involved a large sample and met the majority of its goals, it does have some limitations. One cluster had one fewer crossover due to its delayed participation in the project. Additionally, one cluster randomization had to be changed due to unexpected operational issues following 5 participant consents. We were unable to include dosage (measured by time in our conceptualization) due to the nature of the simplified analysis; understanding and better measuring dosage levels would also further this aim. 40 Finally, while our sample aligns with the national demographic data for hospice utilization, it is not as ethnically or racially diverse given the majority of study participants were white non-Hispanic.
Conclusion
The use of online platforms to provide support to family caregivers of hospice patients with cancer is feasible. The brief interaction time for families and team members in the hospice meeting resulted in the inconsistent inclusion of 9 the nine elements of SDM. The social interaction in the team meetings did not create a climate conducive for SDM. We conclude that due to the organizational issues (time and regulations) restricting the team process and the inability to leverage the elements necessary for SDM, changes in the team structure alone (adding family input) did not result in shared decisions nor improved caregiver outcomes. Future research should explore ways to improve the organizational context through meeting efficiency and tools to facilitate SDM. Additionally, the social support offered in the Facebook support groups needs further investigation given the promise of statistically lower caregiver depression in these data.
Box 1. Key Messages.
What is known:
Family caregivers experience great burden, anxiety, and depression.
Few interventions for hospice caregivers have been tested.
What study adds:
A clinical trial to improve family caregiver outcomes
Evidence to improve caregiver depression.
How can this study affect research and practice:
Hospice should explore offering online support groups to family caregivers.
Research needs to identify mechanisms of action driving the change in outcome.
Acknowledgements
Research reported in this publication was supported by the National Cancer Institute under award number R01CA203999 (Parker Oliver). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Partial support was also received from the Barnes Jewish Hospital Foundation, St. Louis, Missouri.
Footnotes
Conflict of Interest Statement: No authors report any conflict of interest with this study
Clinical Trial registration: NCT02929108
Contributor Information
Debra Parker Oliver, Division of Palliative Medicine, Department of Medicine, Washington University in St. Louis, Goldfarb School of Nursing, 4590 Children’s Place, Mailstop 90-29-931, St. Louis, MO. 63110.
Karla T. Washington, Division of Palliative Medicine, Department of Medicine, Washington University in St. Louis.
Jacquelyn Benson, Division of Palliative Medicine, Department of Medicine, Washington University in St. Louis.
Robin L. Kruse, Department of Family Medicine, University of Missouri, Columbia, Missouri.
Lori Popejoy, Sinclair School of Nursing, University of Missouri, Columbia, Missouri.
Jingxia Liu, Division of Public Health Sciences, Department of Surgery, Washington University in St. Louis.
Jamie Smith, University of Missouri, Department of Family and Community Medicine, Columbia, Missouri.
Kyle Pitzer, Division of Palliative Medicine, Department of Medicine, Washington University in St. Louis.
Patrick White, Division of Palliative Medicine, Department of Medicine, Washington University in St. Louis.
George Demiris, Department of Biobehavioral and Health Sciences, School of Nursing and Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania.
References
- 1.Albright D, Parker Oliver D, Washington K, Lewis A, Kruse R, Demiris G. The social support convoy for caregivers over the course of hospice. J Pain & Symptom Manage. 2016;51(2):213–219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.NHPCO Facts and Figures 2020. 2020. https://www.nhpco.org/wp-content/uploads/NHPCO-Facts-Figures-2020-edition.pdf
- 3.Parker Oliver D, Washington K, Smith J, Uraizee A, Demiris G. The Prevalence and Risks for Depression and Anxiety in Hospice Caregivers. J Palliat Med. Apr 2017;20(4):366–371. doi: 10.1089/jpm.2016.0372 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Parker-Oliver D, Bronstein LR, Kurzejeski L. Examining variables related to successful collaboration on the hospice team. Evaluation Studies. Health Soc Work. Nov 2005;30(4):279–86. [DOI] [PubMed] [Google Scholar]
- 5.Abendroth M, Lutz BJ, Young ME. Family caregivers’ decision process to institutionalize persons with Parkinson’s disease: A grounded theory study. Article. International Journal of Nursing Studies. 2012;49(4):445–454. doi: 10.1016/j.ijnurstu.2011.10.003 [DOI] [PubMed] [Google Scholar]
- 6.Menzel H, Coleman J, E. K Dimensions of being modern in medical practice. J Chronic Dis. 1959;9:20–40. [DOI] [PubMed] [Google Scholar]
- 7.Chewning B, Bylund CL, Shah B, Arora NK, Gueguen JA, Makoul G. Patient preferences for shared decisions: a systematic review. Patient Educ Couns. Jan 2012;86(1):9–18. doi: 10.1016/j.pec.2011.02.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bakitas M, Kryworuchko J, Matlock DD, Volandes AE. Palliative medicine and decision science: the critical need for a shared agenda to foster informed patient choice in serious illness. J Palliat Med. Oct 2011;14(10):1109–16. doi: 10.1089/jpm.2011.0032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Greenfield S, Kaplan S, Ware JE Jr. Expanding patient involvement in care. Effects on patient outcomes. Ann Intern Med. Apr 1985;102(4):520–8. [DOI] [PubMed] [Google Scholar]
- 10.Jefford M, Tattersall MHN. Informing and involving cancer patients in their own care. Lancet Oncol. Oct 2002;3(10):629–637. doi:Doi 10.1016/S1470-2045(02)00877-X [DOI] [PubMed] [Google Scholar]
- 11.Parker Oliver D, Porock D, Demiris G, Courtney K. Patient and family involvement in hospice interdisciplinary teams. Research Support, N.I.H., Extramural Research Support, Non-U.S. Gov’t. J Palliat Care. Winter 2005;21(4):270–6. [PubMed] [Google Scholar]
- 12.Yeh PM, Wierenga ME, Yuan SC. Influences of Psychological Well-being, Quality of Caregiver-patient Relationship, and Family Support on the Health of Family Caregivers for Cancer Patients in Taiwan. Asian Nurs Res (Korean Soc Nurs Sci). Dec 2009;3(4):154–66. doi: 10.1016/S1976-1317(09)60027-X [DOI] [PubMed] [Google Scholar]
- 13.Ryff CD. Happiness is everything, or is it? Explorations on the meaning of psychological well-being. J Pers Soc Psychol. 1989;57:1069–1081. [Google Scholar]
- 14.Kessler D, Peters TJ, Lee L, Parr S. Social class and access to specialist palliative care services. Palliat Med. Mar 2005;19(2):105–10. doi: 10.1191/0269216305pm980oa [DOI] [PubMed] [Google Scholar]
- 15.Hall KL, Vogel AL, Stipelman B, Stokols D, Morgan G, Gehlert S. A Four-Phase Model of Transdisciplinary Team-Based Research: Goals, Team Processes, and Strategies. Transl Behav Med. Dec 01 2012;2(4):415–430. doi: 10.1007/s13142-012-0167-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gattellari M, Butow PN, Tattersall MH. Sharing decisions in cancer care. Soc Sci Med. Jun 2001;52(12):1865–78. [DOI] [PubMed] [Google Scholar]
- 17.Saltz C Promoting skills in evaluating interdisciplinary geriatric teams. Gerontology & Geriatrics Education. 1996;16(4):79–90. [Google Scholar]
- 18.Saltz CC, Schaefer T. Interdisciplinary teams in health care: integration of family caregivers. Soc Work Health Care. 1996;22(3):59–70. doi: 10.1300/J010v22n03_04 [DOI] [PubMed] [Google Scholar]
- 19.Siminoff LA, Step MM. A communication model of shared decision making: accounting for cancer treatment decisions. Health Psychol. Jul 2005;24(4S):S99–S105. doi: 10.1037/0278-6133.24.4.S99 [DOI] [PubMed] [Google Scholar]
- 20.Makoul G, Clayman ML. An integrative model of shared decision making in medical encounters. Patient Educ Couns. Mar 2006;60(3):301–12. doi: 10.1016/j.pec.2005.06.010 [DOI] [PubMed] [Google Scholar]
- 21.Towle A, Godolphin W. Framework for teaching and learning informed shared decision making. BMJ. Sep 18 1999;319(7212):766–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Oliver DP, Washington KT, Demiris G, White P. Challenges in Implementing Hospice Clinical Trials: Preserving Scientific Integrity While Facing Change. J Pain Symptom Manage. Feb 2020;59(2):365–371. doi: 10.1016/j.jpainsymman.2019.09.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Medicare and Medicaid Programs: Hospice Conditions of Participation (42 CFR 418) (2008). [PubMed]
- 24.Berg J. Surrogate decision making in the Internet age. Am J Bioethics. 2012;12(10):28–33. [DOI] [PubMed] [Google Scholar]
- 25.Berg J. Response to open peer commentaries on “Surrogate decision making in the internet age”. Am J Bioethics. 2012;12(10):W1–2. doi: 10.1080/15265161.2012.712417 [DOI] [PubMed] [Google Scholar]
- 26.Pope TM. Facebook Can Improve Surrogate Decision Making. American Journal of Bioethics. 2012;12(10):43–45. doi: 10.1080/15265161.2012.708095 [DOI] [PubMed] [Google Scholar]
- 27.Steehler KR, Steehler MK, Pierce ML, Harley EH. Social media’s role in otolaryngology-head and neck surgery: informing clinicians, empowering patients. Otolaryngol Head Neck Surg. Oct 2013;149(4):521–4. doi: 10.1177/0194599813501463 [DOI] [PubMed] [Google Scholar]
- 28.Brown E, Morrison RS, Gelfman LP. An Update: NIH Research Funding for Palliative Medicine, 2011–2015. J Palliat Med. Feb 2018;21(2):182–187. doi: 10.1089/jpm.2017.0287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Downing M, Lau F, Lesperance M, et al. Meta-analysis of survival prediction with Palliative Performance Scale. Meta-Analysis Research Support, Non-U.S. Gov’t. J Palliat Care. Winter 2007;23(4):245–52; discussion 252–4. [PubMed] [Google Scholar]
- 30.Lowe B, Decker O, Muller S, et al. Validation and standardization of the Generalized Anxiety Disorder Screener (GAD-7) in the general population. Med Care. Mar 2008;46(3):266–74. doi: 10.1097/MLR.0b013e318160d093 [DOI] [PubMed] [Google Scholar]
- 31.Parker Oliver D, Demiris G, Washington K, Kruse RL, Petroski G. Hospice Family Caregiver Involvement in Care Plan Meetings: A Mixed-Methods Randomized Controlled Trial. Am J Hosp Palliat Care. Nov 2017;34(9):849–859. doi: 10.1177/1049909116661816 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Kroenke K, Spitzer RL. The PHQ-9: A new depression diagnostic and severity measure. Psychiat Ann. Sep 2002;32(9):509–515. [Google Scholar]
- 33.Ferrell B, Rhiner M, Rivera LM. Development and evaluation of the Family Pain Questionnaire. Journal of Psychosocial Oncology. 1993;10(4):21–35. [Google Scholar]
- 34.Hawker GA, Mian S, Kendzerska T, French M. Measures of adult pain: Visual Analog Scale for Pain (VAS Pain), Numeric Rating Scale for Pain (NRS Pain), McGill Pain Questionnaire (MPQ), Short-Form McGill Pain Questionnaire (SF-MPQ), Chronic Pain Grade Scale (CPGS), Short Form-36 Bodily Pain Scale (SF-36 BPS), and Measure of Intermittent and Constant Osteoarthritis Pain (ICOAP). Arthritis Care Res (Hoboken). Nov 2011;63 Suppl 11:S240–52. doi: 10.1002/acr.20543 [DOI] [PubMed] [Google Scholar]
- 35.Oliver DP, Washington KT, Pitzer K, et al. Mixed methods analysis of hospice staff perceptions and shared decision making practices in hospice. Support Care Cancer. Mar 2022;30(3):2679–2691. doi: 10.1007/s00520-021-06631-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Angelle D, Rose CL. Familiarity And Prevalence of Facebook Use For Social Networking Among Individuals With Traumatic Brain Injury. Health Serv Manage. 2011;28(2):35040. [DOI] [PubMed] [Google Scholar]
- 37.Bender JL, Jimenez-Marroquin MC, Jadad AR. Seeking support on facebook: a content analysis of breast cancer groups. J Med Internet Res. Feb 04 2011;13(1):e16. doi: 10.2196/jmir.1560 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Greene JA, Choudhry NK, Kilabuk E, Shrank WH. Online social networking by patients with diabetes: a qualitative evaluation of communication with Facebook. J Gen Intern Med. Mar 2011;26(3):287–92. doi: 10.1007/s11606-010-1526-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Benson JJ, Oliver DP, Washington KT, et al. Online social support groups for informal caregivers of hospice patients with cancer. European journal of oncology nursing : the official journal of European Oncology Nursing Society. Feb 2020;44:101698. doi: 10.1016/j.ejon.2019.101698 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Parker Oliver D, Demiris G, Washington KT, Pitzer K, Ulrich C. The Effect of Digital Literacy on Participation in Social Media Clinical Trials in Cancer: Tailoring an Informed Consent Process. Telemed J E Health. Mar 21 2022;doi: 10.1089/tmj.2021.0555 [DOI] [PMC free article] [PubMed] [Google Scholar]