Dyadic versus individual delivery of a yoga program for family caregivers of glioma patients undergoing radiotherapy: Results of a 3‐arm randomized controlled trial

Abstract Background Despite their significant distress, supportive care interventions for caregivers of glioma patients are generally lacking. And, whether caregivers are more likely to benefit from interventions targeting patient‐caregiver dyads or caregivers individually is unknown. This pilot randomized controlled trial compared the feasibility and preliminary efficacy of a dyadic yoga (DY) versus an individual caregiver yoga (CY) intervention as a supportive care strategy for family caregivers. Methods Patient‐caregiver dyads were randomized to a DY, CY or usual care (UC) arm. DY and CY interventions were delivered over 15 sessions. Caregivers completed assessments of their depressive symptoms, quality of life (QOL), and caregiving reactions at baseline, 6 weeks, and 12 weeks, and a subset completed qualitative interviews at 12 weeks. Results With a consent rate of 63%, 67 dyads were randomized. Attendance in the DY was higher than in the CY group (session means, DY = 12.23, CY = 9.00; p = 0.06). Caregivers (79% female; 78% non‐Hispanic White; mean age, 53 years) reported significantly more subjective benefit in the CY arm than in the DY arm (d = 2.1; p < .01), which was consistent with the qualitative assessment. There were medium effect sizes for improved mental QOL (d = 0.46) and financial burden (d = 0.53) in favor of the CY over the UC group. Caregivers in the CY group reported more caregiving esteem (d = 0.56) and less health decline (d = 0.60) than those in the DY group. Conclusion Individual rather than dyadic delivery may be a superior supportive care approach for this vulnerable caregiver population. A larger, adequately powered efficacy trial is warranted.


| INTRODUCTION
Because of the debilitating sequelae, patients with primary brain tumors, particularly those with high-grade gliomas, often depend on their family members for care and support early in the disease course. [1][2][3][4][5][6] Although caregivingrelated activities are generally taxing, family caregivers of glioma patients face unique challenges related to patients' cognitive decline, neurological and motor deficits, and personality changes due to the disease and its treatment. 3,[5][6][7][8][9][10][11][12][13][14][15] In contrast to caregivers of patients with other diseases, caregivers of glioma patients manage both the oncological and neurological concerns of the patients. 16 Given the typically rapid progression of the disease, caregivers of glioma patients must assist patients with all activities of daily and instrumental living while having to cope with the emotional burden of the uncertainty of the disease course. 1 Considering the extensive care demands and poor prognosis of glioma patients, it is unsurprising that caregivers of these patients report high rates of psychological distress, fatigue, and sleep disturbances, any of which can undermine the quality of care they are able to provide. 16,17 In fact, the anxiety and depression levels that caregivers experience may exceed those experienced by glioma patients themselves. 8 Despite their significant distress, caregivers of glioma patients report that their supportive care needs are often unaddressed by multi-disciplinary oncology teams. 18 Moreover, the neuro-oncology literature generally lacks reports of evidence-based supportive care interventions for this vulnerable caregiver population. 18 There are a few exceptions; for instance, one randomized controlled trial showed that cognitive behavioral therapy may increase caregiving mastery and quality of life (QOL) for caregivers of patients with high-grade gliomas. 19 In addition, a recent pilot study suggests that an intervention seeking to help caregivers of primary brain tumor patients identify social support networks may improve caregivers' depressive symptoms. 20 To address the supportive care needs of both patients and their caregivers, we developed a dyadic yoga intervention that integrates targeted physical exercise with stress-management techniques to manage symptoms and health-related QOL. 21,22 A behavioral intervention that includes both psychological and physical components appears to be promising given the multifaceted needs of families coping with glioma. [23][24][25] While yoga as a supportive care strategy addressing both physical and psychological/cognitive treatment side effects has been widely studied in patients with cancer (particularly breast cancer), its application to support the wellbeing of their caregivers is limited. [26][27][28] For caregivers of patients with dementia, a previous trial revealed that yoga significantly improved QOL outcomes. 29 Whether caregivers are more likely to benefit from a yoga intervention targeting patient-caregiver dyads or one targeting only caregivers is unknown, as few studies have directly compared the feasibility and efficacy of these types of behavioral interventions. Studies revealing that family caregiver and patient distress and QOL are interdependent seem to support the use of a dyadic intervention, which could create a positive shared experience in the midst of ongoing challenges. [30][31][32][33][34] However, most dyadic interventions are patient-focused and enroll caregivers predominately to assist patients or to learn skills to better care for patients. Thus, an individual intervention targeting caregivers' own symptoms and needs may provide greater benefit to caregivers.
Building on our previous pilot studies, we sought to address these important gaps in the caregiver literature by comparing the feasibility and preliminary efficacy of a patient-caregiver dyadic yoga (DY) intervention and an individual caregiver yoga (CY) intervention relative to a usual care (UC) group. On the basis of previous trials enrolling caregivers of glioma patients, we hypothesized that at least 50% of eligible dyads would consent to participate; at least 70% of enrolled participants would complete the post-treatment follow-up assessment; and that participants would attend at least 2 of every 3 (i.e., 10 of 15) sessions. 19, 35 We also hypothesized that, compared with UC, the CY and DY interventions would have at least medium effect sizes regarding improvements in caregiver depressive symptoms, QOL, and caregiver reactions (i.e., schedule disruption, lack of support, health declines, financial burden, and esteem).

| Participants
Patients were asked to identify their primary family caregiver (e.g., spouse/partner, parent, adult child). Both the patient and the caregiver had to be willing to participate in this study and both had to be at least 18 years old; able to read and speak English; and able to provide informed consent. In addition, the patient had to be diagnosed with primary malignant glioma; be scheduled to undergo standard radiotherapy for at least 5 weeks; and have a Karnofsky Performance Status (KPS) score of at least 80 at the time of recruitment. We excluded dyads whose caregiver or patient reported regularly (self-defined) practicing yoga or whose patient had cognitive deficits that the clinical team deemed would impede the completion of the self-report instruments.

| Procedures
The MD Anderson Institutional Review Board approved all procedures prior to participant enrollment. Research staff used the electronic clinic appointment system to identify potential participants. Before the COVID-19 pandemic (i.e., prior to March 2020), we approached patients and caregivers during clinic visits and confirmed their eligibility and obtained their written informed consent prior to data collection; if caregivers were not present, we asked patients for their permission to contact their caregivers to obtain consent. During the pandemic (March 2020 onward), all participants were contacted via phone, and consent was obtained electronically. Participants completed survey measures prior to randomization at baseline and completed follow-up assessments in REDCap at the completion of patients' radiotherapy (6 weeks) and at the 3-month follow-up (12 weeks). To gain further insights into caregivers' experiences with study participation, a subset of participants (n=15 caregivers) completed individual, qualitative interviews at the 12-week assessment. (A description of the qualitative interview methods along with participant characteristics is given in the Supplementary S1). Participants received a $20 gift card for each assessment they completed. Although both patients and caregivers completed assessments, the present study focused on caregivers' outcomes only.

| Randomization
After baseline data collection, patient-caregiver dyads were randomized to the DY, CY, or UC arm through minimization, a form of covariate-adaptive randomization that ensured that the groups were balanced in terms of patient and caregiver sex and age and patient tumor grade and Karnofsky Performance Status score. 36

| Intervention groups
Participants in the DY and CY intervention groups attended 15 sessions (45 minutes each) over the course of the patients' 6-week standard radiotherapy (2 or 3 sessions per week). This dose was selected based on prior yoga efficacy trials that included women with breast cancer undergoing RT where the sessions were provided 2-3 times a week for 6 weeks. 37,38 For participants enrolled before the pandemic, sessions 1-4 were delivered in person at the hospital, and sessions 5-15 were delivered in-person or via Zoom, a HIPAA-compliant videoconferencing platform. For participants enrolled during the COVID-19 pandemic, all sessions were delivered via Zoom. Three yoga instructors certified by the International Association of Yoga Therapists delivered both interventions.

| DY intervention
Patients and caregivers attended DY sessions together. Session content was developed with consideration of the needs and limitations of brain tumor patients as described previously. 39 Briefly, each DY session included a brief introduction to the session followed by individual and dyadic mind-body techniques. In accordance with universal Hatha Yoga practices, each session had 4 main components: (a) joint loosening with mindfulness training; (b) postures (asanas) with relaxation techniques; (c) breathing exercises (pranayama); and (d) meditation. The DY sessions highlighted communal coping and interconnectedness within the family, included dyadic postures, and focused the meditations on the family context (e.g., gratitude and lovingkindness mediations were focused on the couple/dyad). Participants were encouraged to offer mutual support during the practice (e.g., stabilizing support during balance poses). Because of contraindications in the patient population, common inversion yoga postures (e.g., standing forward bend, downward-facing dog) were not used.

| CY intervention
The structure and content of the CY sessions were as similar as possible to those of the DY sessions (e.g., no additional poses were included); however, the CY sessions were only attended by the caregiver and did not include a dyadic focus and the partner exercises. Patients in the CY group received UC care as described below. To avoid contamination bias, we asked CY group participants not to share study content with patients for the duration of the study.

| Quality control
All yoga sessions were audio-and video-recorded with the participants' permission, which was obtained during the informed consent process. We reviewed the recordings for treatment fidelity on an ongoing basis using a fidelity checklist.

| Control group
Patients in the UC group received usual care from their health care team. Patients and caregivers were invited to attend the monthly Brain Tumor Support Group and had access to a clinical social worker if needed.

| Demographic and medical data
Demographic data (e.g., age, race/ethnicity) were collected with questionnaires at baseline.
Patients' medical data were extracted from their electronic medical records.

| Feasibility data
We documented rates of consent (including refusal reasons), session attendance (including in person and via videoconference), and assessment completion (including attrition reasons). Participants randomized to the DY or CY intervention completed a program evaluation using Likert scale items (see Table 2 for items including scale anchors) that we had developed for our previously published yoga trials. 38,40

| Intervention outcome measures
Depressive symptoms were assessed with the Centers for Epidemiological Studies-Depression (CES-D), 41 which consists of 20 items focused on the affective component of depression. Overall QOL was assessed with the Medical Outcomes Study 36-item short-form survey (SF-36), which has 8 distinct domains (physical functioning, physical impediments to role functioning, pain, general health perceptions, vitality, social functioning, emotional impediments to role functioning, and mental health) and yields both a mental composite summary score and a physical composite summary score. 42 Caregiving reactions were assessed with the Caregiver Reaction Assessment (CRA), a 24-item measure assessing four negative caregiving domains (lack of family support, financial strain, schedule disruption, and health problems) and one positive caregiving domain (esteem). 43

| Data analysis
To determine feasibility, we calculated descriptive statistics for consent, session attendance, assessment completion, and program satisfaction. We included the attendance data of all participants who were randomized to the DY and CY arms regardless of their assessment completions. We used t-tests to compare session attendance and program satisfaction between the DY group and CY group. To examine preliminary intervention efficacy for caregivers, we performed multilevel modeling (MLM) using PROC MIXED (SAS, version 9.4). We controlled for the baseline level of the given outcome, included assessment time as a categorical variable, and used CONTRAST statements within the procedure to test for group differences (i.e., DY vs UC, CY vs UC, and DY vs CY). Because the current study was a feasibility study and not adequately powered for definitive efficacy testing, we supplemented the inferential statistics with effect sizes (Cohen's d) associated with each between-group comparison and interpreted each effect size as small (d = 0.2), medium (d = 0.5), or large (d = 0.8). 44 A medium effect size was considered hypothesis-confirming.

| Recruitment and retention
A CONSORT flow diagram illustrating participant recruitment and retention is shown in Figure 1. We approached 128 dyads and obtained consent from 76 (59%). The main reasons dyads declined participation were "not interested" (n = 31; 60%) and "too busy" (n = 14; 27%). Of the dyads who consented to participation, 3 became ineligible because the patient transferred to hospice, 1 patient died,  Drop-out was due to high patient symptom burden/disease progression (n = 7, 10%), patient death (n = 6; 14%), and loss to follow-up (n = 9; 13%), particularly during the pandemic. Attrition was not a function of group assignment (p = 0.68); however, patients in the dyads that were not retained had lower KPS scores at baseline (p = 0.09).

| Session attendance and acceptability
The mean number of sessions the DY participants attended was 12. However, compared with those in the DY arm, caregivers in the CY arm indicated deriving significantly greater benefit from the intervention (t = 3.5; p = 0.01; d = 2.1) and rated the program as more useful (t = 3.6; p = 0.002; d = 1.8). Caregivers in the CY arm also reported having higher expectations regarding the lasting benefit of the intervention than those in the DY arm (t = 2.4; p = 0.03; d = 1.1). The means of the program evaluation variables are given in Table 2. We found a medium effect in favor of in-person delivery over videoconference delivery (d = 0.76), but this effect was not significant (p = 0.10). Themes and subthemes from the qualitative interviews are presented in Table 3.

| Preliminary efficacy
The raw means of the outcome measures at each assessment point by group are given in Table 4. Least square means (LSMs) from the MLM analyses (controlling for baseline level of the outcome and assessment point) are presented below.

T A B L E 2 Means of the program evaluation variables
T A B L E 3 Quotes from caregivers illustrating themes related to feasibility and acceptability of the yoga intervention

| DISCUSSION
The goal of this pilot randomized controlled trial was to assess the feasibility and preliminary efficacy of a yoga intervention for caregivers of glioma patients undergoing radiotherapy that was delivered to either patient-caregiver dyads or caregivers individually. Our results revealed that the intervention is feasible, as our a priori criteria regarding consent, retention, and attendance were met in both the CY and DY groups. Although the DY group had a higher numeric attendance rate, both the DY and CY groups reached the a priori attendance rate benchmark of 67%. It appears that delivering the interventions concurrently to patients' radiation treatment plan facilitates session attendance regardless of modality (i.e., in-person at the hospital or via videoconference). The overall 3-month

Theme Quotes
Yoga may be helpful for the patient (caregiver yoga only arm) I think doing yoga would be really good for her. It was for me. Female, 60 yrs old, caring for daughter, CY264 I would go back and do yoga more with the patient and the caregiver. I think that would be a good connection for the two, and I think it would benefit the patient. I thought the yoga program was wonderful. I thought it was a great way for the hospital to reach out past the patient. But I do think it needs to be patient and caregiver if both feel up to it and are able to do it. Female, 66 yrs old, caring for daughter, CY270 Theme category: acceptability of virtual yoga intervention Virtual setting was convenient for caregivers participating in the yoga intervention It was actually easier because I didn't have to leave work early or anything like that. So that was good. I was even able to do it while on vacation, so that worked as well. It wasn't challenging at all. Female, 44 yrs old, caring for mother, CY266 I liked the fact that it was remote because I didn't have to spend time going somewhere, leaving somewhere, or being away from the patient. Female, 46 yrs old, caring for sister, CY271 Caregivers may have enjoyed an in-person yoga intervention I wish we would have been able to meet in person and actually be in front of the instructor. But it was a good experience overall.

Female, 43 yrs old, caring for father, DY263
We did yoga by video; maybe if I would have been able to do it in person, that might have been a little better, but I don't think I missed anything by doing it through a Zoom meeting.

Female, 60 yrs old, caring for daughter, CY264
Theme category: continued use of yoga following study participation Continuing to practice yoga after study participation And even now sometimes, when [the patient] is on chemo and he is not feeling well, or he is just feeling anxious, we will do some of the relaxation yoga that we have available still. Female, 43 yrs old, caring for father, DY263 Well, we have continued that. I mean, we both stretch. We stretch each other now. We are doing a little bit of that before, but we do a lot of things. But he has gotten so much better. follow-up retention rate was relatively low (62%), and attrition was mainly due to patients' high symptom burden, disease progression, and death-a challenge that is frequently encountered in studies that include caregivers of patients whose expected survival time is short. Moreover, with the onset of the COVID-19 pandemic, caregivers' access to the hospital was restricted, and participants received follow-up care from their medical team mainly via telemedicine. Thus, attrition may have also been a result of less frequent clinic visits.
We were particularly interested in determining whether caregivers favor and benefit dyadic or individual delivery of the intervention. Although dyadic delivery as a care model has increasingly gained attention in psycho-oncology and supportive care, our results revealed that caregivers in the CY arm found the intervention significantly more useful and beneficial than those in the DY arm. These quantitative results are supported by the qualitative accounts highlighting that caregivers in the CY arm enjoyed practicing self-care and having time to relax and escape their caregiving role. Although caregivers in the DY arm mentioned benefits to the patient and benefits to their relationship with the patient, they did not mention the benefits of respite and relaxation. The dyadic setting may not allow caregivers to fully relax and focus on themselves as they remain aware of the needs of the patient. This finding may be particularly true for those caring for patients with cognitive deficits. Lastly, although the transition to videoconference delivery with the onset of the pandemic did not impact attendance, our effect size results indicate that caregivers favored in-person delivery. However, caregivers also acknowledged the benefits of remote delivery in the qualitative interviews.
Our findings also suggest that, even with the lower attendance rate of the CY group, the CY intervention revealed stronger evidence for preliminary efficacy than the DY intervention in terms of improved mental QOL and caregiver reactions. Although the trial was not adequately powered to definitively assess efficacy, our hypotheses were supported by the medium effect sizes for improved mental QOL and financial burden, the latter of which could be driven by the larger proportion (albeit not statistically significant) of caregivers in the CY group who were employed full-time. As such, group differences regarding financial burden need to be interpreted with caution particularly as the intervention did not target financial distress. Yet, as the subscale assessing financial burden measures perceptions of financial distress rather than objective financial toxicity/hardship, aspects of the intervention such as gratitude and lovingkindness meditations may alter participants' perceptions of financial burden. Further investigations are needed to corroborate our findings.
Moreover, we would like to note that there was not only a lack of evidence regarding the benefits for the DY group relative to the UC group, caregivers in the DY group also reported worse outcomes for the esteem and health decline subscales of the CRA. We found no evidence of preliminary efficacy for depressive symptoms and physical QOL in either the DY or CY group, which is surprising considering that caregivers, particularly those in the CY arm, reported subjective emotional, cognitive, and physical wellbeing benefits on the program evaluations. However, the caregivers were experiencing low levels of depressive symptoms at study entry, with only 49% meeting the minimum threshold for clinical screening criteria for depression, suggesting a potential floor effect may have influenced study findings. And, the relatively low assessment completion rates, even among participants with high intervention attendance (i.e., ≥10 sessions), may have contributed to a failure to capture all intervention effects. Our findings generally agree with those of the few previous studies in this population. For instance, a Dutch study found that a 6-session, psychologist-led cognitive behavioral therapy intervention had a small effect on mental QOL. 19 In addition, a recent study showed that a cognitive behavioral therapy-based intervention had a medium effect on caregiving-specific distress but failed to yield significant group differences for depressive symptoms. 35 Another study showed that an electronic social network intervention may improve depressive symptoms but may not have an effect on caregiver burden or other psychosocial outcomes. 20 Given the findings of the present study, further investigation of the CY intervention is warranted. Our findings suggest that, for caregivers of patients with primary brain tumors, a caregiver-focused supportive care delivery is preferable to dyadic approaches. However, further study that includes the direct comparisons of dyadic and individual delivery is needed along with models that more precisely establish settings in which a dyadic, individual or a combination of dyadic and individual components as in a "hybrid" (or even group-based) format is advantageous should be developed. For instance, caregivers of patients who have early-stage disease and/or less disease burden may respond to dyadic interventions, whereas those facing a high caregiving load may benefit from individual interventions.
The findings of the present study should be interpreted with caution. The objective of this trial was to examine the feasibility and preliminary efficacy of the DY and CY interventions. The trial was not designed to confirm efficacy, as it was underpowered. The a priori benchmark regarding a medium effect size was rather conservative; however, its use is justifiable, as we did not control for non-specific intervention effects, such as attention, in the UC group. Moreover, our study is limited by the participants' fairly homogenous characteristics, particularly race/ethnicity and socio-economic status.
In summary, the findings of the present study suggest that, although both the CY and DY interventions are feasible, the CY intervention is subjectively more beneficial to caregivers of glioma patients. This study also revealed that the use of a videoconferencing platform (Zoom in this case) is an acceptable approach for the delivery of interventions to caregivers of glioma patients. Finally, the findings of this trial suggest that relative to the DY intervention and UC, the CY intervention has greater preliminary efficacy in terms of improving mental health aspects of QOL as well as caregiving-related esteem, health decline, and financial burden. An adequately powered efficacy trial of the CY intervention with an attention control group is warranted.

ACKNOWLEDGMENT
This study was funded in part by a grant from the National Cancer Institute at the National Institutes of Health (R21 CA218309), Milbury, PI.

CONFLICT OF INTEREST
None.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

ETHICS STATEMENT
The MD Anderson Institutional Review Board approved all procedures prior to participant enrollment.