The CARING intervention for neuro-oncology family caregivers: Primary outcomes from an abbreviated randomized controlled trial

Abstract

Background

We developed CARING to support neuro-oncology caregivers. CARING includes eSNAP, a web-based tool to identify social support resources, and 8 weeks of individual phone-based non-clinical navigation focused on support identification and coping skills. This study aimed to evaluate the efficacy of CARING on caregiver and patient psychosocial outcomes.

Methods

We compared CARING to a waitlist control (WLC) in an RCT. Neuro-oncology patients and their caregivers were recruited from a National Cancer Institute (NCI)-designated comprehensive cancer center in Florida from February 2020 to June 2024. Eighty-one caregivers were assigned to CARING and 35 to WLC. All participants completed baseline measures, including demographics and health data. Primary outcomes collected at 8 weeks were caregiver support and self-efficacy. Secondary outcomes were caregiver burden, personal gain, and caregiver and patient anxiety and depressive symptoms. General Linear Mixed Models examined changes in outcomes from baseline to 8 weeks by condition.

Results

No significant demographic differences existed between conditions. CARING caregivers reported significantly lower ratings of stress/frustration compared to feeling in control/hope (F(1,92) = 6.19, P = .015) at 8 weeks compared to WLC. There was a significant increase in personal gain (M = 0.7, SD = 2.5, t(63) = 2.13, P = .037) and a marginally-significant increase in self-efficacy (M = 1.6, SD = 6.6, t(61) = 1.88, P = .065) for CARING caregivers at 8 weeks. Although the intervention is targeted only at caregivers, patients of CARING caregivers reported marginally significantly fewer depressive symptoms (F(1,51) = 2.89, P = .096) at 8 weeks compared to WLC.

Conclusion

CARING decreases the ratio of negative to positive emotions and perceptions of personal gain in caregivers at 8 weeks, compared to WLC.

Key Points.

• The CARING intervention may promote caregiver self-efficacy and personal gain.

• Patients of caregivers who receive CARING may be protected from depressive symptoms.

Importance of the Study.

Theoretically guided interventions are needed for family caregivers of patients with brain tumors, who often face challenges in obtaining support. We developed CARING, an 8-week psychosocial intervention consisting of the electronic Social Network Assessment Program, (eSNAP), a web-based tool to identify and organize support resources, and manualized, individual, phone-based navigation sessions for caregivers. We compared CARING to a waitlist control in 116 neuro-oncology patient-caregiver dyads recruited from a single NCI-designated Comprehensive Cancer Center. Despite COVID-related limitations to our sample size, findings from our trial suggest that CARING may increase caregiver self-efficacy, encourage perceptions of personal gain, and protect against patient depressive symptoms. Early integration of caregiver interventions like CARING into routine clinical care could improve caregiver and patient outcomes.

Patients with brain tumors—especially the most common types of malignant brain tumors--often experience rapid, unpredictable disease progression, along with significant physical debilitation and personality and behavioral changes.^1–3 As a result, they often rely on family members or friends who serve as caregivers. While many caregivers report positive aspects, such as meaning and purpose,⁴ this role is also marked by significant burden and distress as well as a need for support across multiple domains.⁵

The stress process model of caregiving⁶ suggests that access to resources for support and coping can help reduce primary and secondary stressors to lower caregiver and patient distress. This can be accomplished by directly addressing specific stressors, such as assistance with care tasks or respite, or through reappraisal and building self-efficacy to manage stressors and find meaning. However, barriers exist to caregivers of patients with brain tumor obtaining support, including difficulty in asking for help and prioritizing the patient’s needs over their own.^7,8 Caregivers are often dealing with new, overwhelming situational circumstances that may limit the amount of bandwidth available to cue important support-seeking behavior⁹, and healthcare systems lack routine, standardized formal services for caregivers, both in general and specifically in neuro-oncology.^10–12 Within the research space, few interventions exist specifically to support neuro-oncology caregivers.¹³ While most of the reporting around RCTs that exist is of high quality, the studies themselves are mostly small-scale and involve heterogenous populations and intervention components and targets, making it difficult to make definitive conclusions about best practices.^13–15

To address the need to support caregivers of patients with primary brain tumor, we developed CARING. CARING is an 8-week psychosocial intervention designed to help caregivers recognize existing social support and improve coping.¹⁶ As described elsewhere, the intervention consists of two components.^16,17 First, caregivers are given access to the electronic Social Network Assessment Program (eSNAP), a web-based tool in which caregivers identify and organize existing support resources from family and friends to create a visualization of available support. An example of a completed visualization is shown in Figure 1. Additionally, eSNAP provides tailored recommendations for formal services aggregated from local community and cancer center resources and national organizations. Second, caregivers receive 8 weekly individual phone-based manualized sessions, each lasting about 30 minutes, from a Caregiver Navigator. Topics are listed in Table 1. Navigators are non-clinical interventionists who receive specific training in the experience of neuro-oncology caregiving and relevant available resources.¹⁸ Based on eSNAP data and ongoing assessment, Caregiver Navigators develop individualized plans to help caregivers take advantage of available social support and formal resources. The goal of CARING is to help caregivers build self-efficacy to better cope with the burden associated with caregiving and other secondary stressors.

Figure 1.

Example completed eSNAP visualization of caregiver social support from family and friends

Table 1.

Manualized Navigation Topics

Topic	Description
Overview & getting social support	Introduction, goal setting. Review existing social support resources in eSNAP. Identify needs, facilitators, and barriers to social support
Problem solving	Reviewing a stepwise process of problem identification and actionable steps
Growing social support network	Discussing potential additional/overlooked support from a variety of sources
How to ask for help	Reviewing best practices and role playing asking for help
Taking care of yourself	Reviewing the importance of self-care and planning how to prioritize self-care activities
Setting limits	Discussing strategies to set boundaries to reduce social stress and role playing scenarios
Positive side of caregiving	Reviewing caregiver growth and discussing strategies to focus on meaning-making and strengths
Review & future planning	Reflection on previous topics, identifying areas for continued work
Bereavement (As needed; ends navigation)	Check-in with caregiver, offer condolences, help connect caregiver to bereavement resources, including social work.

Objective

This study aimed to evaluate the efficacy of the CARING intervention compared to a waitlist control (WLC) on outcomes aligned with the stress process model. Caregiver's primary outcomes were self-reported perceived social support, caregiving self-efficacy, anxiety, depressive symptoms, personal gain, and caregiving burden. Patient primary outcomes were self-reported anxiety and depressive symptoms.

Methods

Study Design

A randomized controlled trial was conducted comparing the CARING intervention to a waitlist control. Although the intervention was designed only for caregivers, both patients and caregivers were required to consent to participate, and data were collected on both dyad members. In cases with more than one caregiver, the one who provided the most care (“primary” caregiver) was selected to participate in the study. The study was initially proposed and funded to enroll 225 dyads in order to analyze data from 160 caregivers (80 per condition: CARING or WLC) who completed the assessment at 8 weeks, providing power ≥ 0.80 to detect a medium effect size.¹⁶ Due to COVID-related issues that limited patient clinical panels and our data collection time, we were required to reduce our enrollment goals. As such, randomization was stopped in April 2023 (with 47 intervention and 35 waitlist dyads) and all subsequent participants were assigned to the intervention condition (n = 34 dyads) to prioritize intervention evaluation during our funding period. The Consolidated Standards of Reporting Trials (CONSORT) checklist is available in Supplementary File 1.

Participants

Patient and caregiver participants were recruited from the neuro-oncology clinic of a large National Cancer Institute (NCI)-designated Comprehensive Cancer Center in Florida from February 2020 to June 2024. Patients were screened for potential eligibility and were approached in clinic or by phone; interested patients were asked to nominate a caregiver for participation, who was also approached in clinic or by phone.

While CARING is targeted only at caregivers, both patients and caregivers were required to enroll together as a dyad. We intended to recruit participants at the first appointment after diagnosis to balance potential distress with early enrollment. Because the intervention was delivered to caregivers, patient participation was discontinued if caregivers dropped the study.

Inclusion criteria were (1) age ≥ 21 years, (2) English-speaking, and (3) able to complete questionnaires (including by proxy). Caregivers self-identify as being a primary caregiver of a patient with a brain tumor. Patients were within 9 months of diagnosis of a new or recurrent brain tumor (primary or secondary), undergoing active treatment (ie, had an ongoing clinical relationship, not a second-opinion consultation), with a prognosis of at least 6 months, and a Karnofsky Performance Status (KPS) score ≥ 50 (determined by oncologist).

Our inclusion criteria were purposely broad for two main reasons. First, the population of patients with brain tumor is relatively small compared to other cancers, and this was a single-institution study. Second, our focus is on the caregiver experience. While burden is often higher in caregivers of patients with higher-grade brain tumors and lower patient function, burden for any type of neuro-oncology care is higher than other types of cancer.¹⁹ Further, burden is multi-faceted,²⁰ and patient clinical characteristics may not be the primary driver of burden in all cases.²¹ Broad clinical inclusion criteria allow for a larger sample of caregivers likely to need support.

Procedures

As described elsewhere,^22,23 all consented participants completed baseline questionnaires, including demographics, upon enrollment. All data collection was done electronically or through mailed surveys and entered by study staff not involved in the intervention to minimize bias. Caregivers were then randomized 1:1 to receive the intervention immediately or to an 8-week waitlist. Randomization was automated through REDCap²⁴ to ensure lack of bias by study staff and was done using an 8-block design stratified by caregiver gender.

Caregivers and patients in both conditions (CARING and WLC) were scheduled to complete primary outcome measures at 8 weeks, at which time waitlist caregivers were offered the intervention. All study procedures were approved by the Advarra Institutional Review Board (Pro00029204). The study was pre-registered on clinicaltrials.gov (NCT04268979).

Measures

All self-report measurements described below were assessed at enrollment and 8 weeks. Participants received up to 3 reminders to complete measures within 2 weeks of the targeted time point.

Caregiver Only

All measures are widely used and validated in cancer caregiver populations.

Social support was assessed using the 8-item Duke/UNC Functional Social Support Questionnaire (DUFSSQ-8).²⁵ Participants rate the amount of different types of support they may receive on a 1 to 5 Likert scale (“Much less than I would like” to “As much as I would like”). Items are summed, and higher scores reflect higher perceived social support. Though no cut-off exists, a score of 30 or lower represents an average report of wanting more support.

Caregiver self-efficacy was measured using two measures. First, the 12-item Communication and Attitudinal Self-Efficacy Scale for Cancer (CASE).²⁶ The scale consists of three domains: understanding and participating in care, maintaining a positive attitude, and seeking and obtaining information. Items are summed; higher scores indicate more self-efficacy. Second, self-efficacy was also measured using a purpose-driven rating of stress, frustration, feeling in control, and feeling hopeful on a scale of 1-5. These items were based on unique quadrants of the interpersonal circumplex, a model widely used in psychology for organizing interpersonal dispositions based on axes of dominance and warmth.²⁷ Our goal was to capture common emotions relevant to their ability to care across various experiences. The sum of stress and frustration was divided by the sum of in control and hopeful for each participant. Higher scores indicate more stress and frustration relative to feeling in control and hopeful—ie, more relative negativity about their experience.

Caregiver burden was assessed using the 12-item Zarit Burden Interview (ZBI) Short Form.²⁸ Responses are summed to calculate a total score, with scores of 10-20 indicating mild to moderate burden, and scores over 20 indicating high burden.

Caregiver personal gain was assessed using the 4-item subscale from the Caregiver Reaction Scale (CRS).²⁹ Items are scored on a 4-point Likert-type scale and averages for each subscale are calculated. Higher mean scores indicating higher levels of gain.

Anxiety was measured using the 7-item Generalized Anxiety Disorder scale (GAD-7). Responses are summed to create a total score ranging from 0 to 21, with a score of 5-9 indicating mild anxiety, 10-14 moderate anxiety, and 15-21 severe anxiety.

Depressive symptoms were measured using the 8-item Patient Health Questionnaire-8 (PHQ-8).³⁰ The PHQ-8 omits a question about suicidality from the PHQ-9. Responses are summed to create a total score. Scores of 5-9 indicate mild depression, 10-14 indicate moderate depression, 15-19 indicate moderately severe depression, and 20-24 indicate severe depression.

Patient Measures

Patients were asked to complete Neuro-QOL cognitive and physical function measures as well as the anxiety and depression measures listed above at baseline and 8 weeks. Neuro-QOL measures are designed to be accessible for people with neurological conditions and the GAD and PHQ are effective measures for people with brain tumors.³¹ If a patient was unable to complete measures at 8 weeks, caregivers were asked to report on patient function by proxy.

Bereavement

If a patient died while on study, caregivers were sent a condolence card, and no questionnaires were sent for two weeks. Bereaved caregivers were asked to complete an abbreviated set of caregiver questionnaires at follow-up (or after 2 weeks; excluding measures relevant to current caregiving). Bereaved caregivers who were actively receiving the intervention at the time of patient death were offered a bereavement navigation session to offer general support and closure prior to discontinuation of navigation.

Analyses

Descriptive statistics were used to summarize demographics and psychosocial variables completed by caregivers and patients at baseline. Group differences were evaluated to determine if primary analysis models required inclusion of covariates (P < .10). For each psychosocial variable, General Linear Mixed Models evaluated change from baseline to 8 weeks by condition using a model with condition, time, and their interaction as the primary predictors. A significant interaction indicated differences in change by condition, and was explored using t-tests evaluating change within each condition. Only data where both baseline and 8-week data was complete for the measure was used. Power for primary analyses was greatly reduced due to the sample sizes attained.

Results

As presented elsewhere,²³ of 502 eligible dyads who were approached, 148 consented to participate (29%), exceeding the median rate (23%) of prior oncology dyadic research.³² The median time from diagnosis to study consent was 97 days. Of those consented, 116 completed baseline surveys and were assigned to a condition; 47 caregivers were randomized; and 34 were assigned to receive the CARING intervention (81 total) and 35 were randomized to a waitlist. After the 8-week study period, 87% of caregivers (83% CARING; 94% WLC) and 77% of patients (72% CARING; 86% WLC) completed the follow-up survey for a total analytic sample of 102 caregivers and 89 patients (21 CARING and 4 WLC patient reports were by proxy). This similarly exceeds the median retention rate (70%) for prior oncology dyadic research.³² The most common reasons for non-participation were lack of interest, being too busy, or feeling overwhelmed. A total of 7 patients died while on study, 2 between consent and allocation, 3 whose caregivers were receiving CARING, and 2 whose caregivers were on WLC. All allocated bereaved caregivers completed 8-week surveys.

Demographic and psychosocial characteristics for caregivers and patients are presented in Table 2. The sample was predominantly non-Hispanic White and generally financially comfortable. Patients were most commonly diagnosed with glioblastoma (66%), and most caregivers were spouses (81%). There were no differences in demographic variables by intervention versus waitlist condition. Table 2 also presents descriptive statistics for baseline measures of patients’ cognitive and physical function. At baseline, both values were about 1SD below the average population. Among those patients who completed the assessment at 8 weeks, average levels physical function decreased significantly for the Intervention condition (M = −3.0, SD = 6.0, t(54) = −3.69, P = .001) and the WLC condition (M = −2.1, SD = 4.8, t(26) = −2.29, P = .030).

Table 2.

Demographic Information

		Caregivers			Patients
		All (n = 116)	Intervention (n = 81)	Waitlist (n = 35)	All (n = 116)	Intervention (n = 81)	Waitlist (n = 35)
Variable	Levels	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)
Age (years)		57.6 (14.7)	58.1 (14.5)	56.3 (15.2)	58.3 (15.1)	59.6 (14.6)	55.5 (15.8)
		N (%)	N (%)	N (%)	N (%)	N (%)	N (%)
Gender	Male	34 (29%)	21 (26%)	13 (37%)	69 (61%)	51 (64%)	18 (55%)
	Female	82 (71%)	60 (74%)	22 (63%)	44 (39%)	29 (36%)	15 (45%)
Race	American Indian or Alaskan Native	2 (2%)	2 (2%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
	Asian	0 (0%)	0 (0%)	0 (0%)	3 (3%)	1 (1%)	2 (6%)
	Black or African American	4 (3%)	2 (2%)	2 (6%)	7 (6%)	4 (5%)	3 (9%)
	Middle Eastern or North African	1 (1%)	1 (1%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
	Native Hawaiian or other Pacific Islander	0 (0%)	0 (0%)	0 (0%)	1 (1%)	1 (1%)	0 (0%)
	White or Caucasian	108 (93%)	77 (95%)	31 (89%)	102 (89%)	74 (93%)	28 (80%)
	Some other race, ethnicity, or origin	4 (3%)	2 (2%)	2 (6%)	2 (2%)	1 (1%)	1 (3%)
	Prefer not to answer	1 (1%)	1 (1%)	0 (0%)	1 (1%)	1 (1%)	0 (0%)
Ethnicity	Non-Hispanic	108 (93%)	76 (94%)	32 (91%)	108 (96%)	76 (95%)	32 (97%)
	Hispanic	7 (6%)	4 (5%)	3 (9%)	5 (4%)	4 (5%)	1 (3%)
	Prefer not to answer	1 (1%)	1 (1%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Education	Less than high school diploma/equivalent	1 (1%)	1 (1%)	1 (1%)	2 (2%)	1 (1%)	1 (3%)
	High school diploma or equivalent	16 (14%)	11 (14%)	5 (14%)	13 (12%)	8 (10%)	5 (15%)
	Some college	36 (31%)	23 (28%)	13 (37%)	44 (39%)	33 (41%)	11 (33%)
	Four-year college degree	32 (28%)	22 (27%)	10 (29%)	27 (24%)	20 (25%)	7 (21%)
	Postgraduate degree	31 (27%)	24 (30%)	7 (20%)	25 (22%)	16 (20%)	9 (27%)
	Other	0 (0%)	0 (0%)	0 (0%)	1 (1%)	1 (1%)	0 (0%)
	Don’t know	0 (0%)	0 (0%)	0 (0%)	1 (1%)	1 (1%)	0 (0%)
Work hours per week	≥35 hours	39 (34%)	27 (33%)	12 (34%)
	<35 hours	6 (5%)	4 (5%)	2 (6%)
	Not working for pay	69 (60%)	48 (60%)	21 (60%)
	Prefer not to answer	2 (2%)	2 (2%)	0 (0%)
Financial situation	Not very good	19 (16%)	11 (14%)	8 (23%)
	Comfortable	67 (58%)	49 (60%)	18 (51%)
	More than adequate to meet my needs	25 (22%)	17 (21%)	8 (23%)
	Prefer not to answer	5 (4%)	4 (5%)	1 (3%)
Relationship to patient	Spouse/partner	94 (81%)	65 (80%)	29 (83%)
	Parent/in-law	7 (6%)	5 (6%)	2 (6%)
	Sibling/in-law	2 (2%)	2 (2%)	0 (0%)
	Child	9 (8%)	6 (7%)	3 (9%)
	Friend/neighbor	2 (2%)	2 (2%)	0 (0%)
	Niece	1 (1%)	0 (0%)	1 (3%)
	Grandson	1 (1%)	1 (1%)	0 (0%)
Length of care for patient	1-3 months	29 (25%)	17 (21%)	12 (34%)
	4-6 months	36 (31%)	26 (33%)	10 (29%)
	7-12 months	16 (14%)	12 (15%)	4 (11%)
	1-2 years	14 (12%)	12 (15%)	2 (6%)
	3-5 years	13 (11%)	8 (10%)	5 (14%)
	More than 5 years	7 (6%)	5 (6%)	2 (6%)
Caregiving hours per week	0	5 (5%)	5 (7%)	0 (0%)
	1-5	20 (18%)	15 (20%)	5 (15%)
	6-15	20 (18%)	11 (14%)	9 (26%)
	16-30	26 (24%)	17 (22%)	9 (26%)
	31-50	18 (16%)	15 (20%)	3 (9%)
	More than 50	21 (19%)	13 (17%)	8 (24%)
Previous caregiving experience	No	73 (63%)	50 (62%)	23 (66%)
	Yes	42 (36%)	31 (38%)	11 (31%)
	Prefer not to answer	1 (1%)	0 (0%)	1 (3%)
					Patients
					All (n = 116)	Intervention (n = 81)	Waitlist (n = 35)
					N (%)	N (%)	N (%)
Patient Diagnosis	Glioblastoma				77 (66%)	49 (60%)	28 (80%)
	Astrocytoma				8 (7%)	5 (6%)	3 (9%)
	Oligodendroglioma				4 (3%)	4 (5%)	0 (0%)
	Glioma NOS				5 (4%)	2 (2%)	3 (9%)
	Meningioma				6 (5%)	6 (7%)	0 (0%)
	Medulloblastoma				1 (1%)	1 (1%)	0 (0 %)
	Brain metastasis				14 (12%)	14 (17%)	0 (0%)
	Leptomeningeal disease				1 (1%)	0 (0%)	1 (3 %)
Recurrent brain tumor	No				98 (84%)	69 (85%)	29 (83%)
	Yes				18 (16%)	12 (15%)	6 (17%)
Patient WHO Grade	1				1 (1%)	1 (1%)	0 (0%)
	2				6 (5%)	4 (5%)	2 (6%)
	3				12 (10%)	9 (11%)	3 (9%)
	4				78 (67%)	50 (62%)	28 (80%)
	Not in medical record/missing				19 (16%)	17 (21%)	2 (6%)
					M (SD)	M (SD)	M (SD)
Patient Cognitive Function (Neuro-QOL: Range 8-40)*					25.8 (9.5)	25.2 (9.2)	27.2 (10.3)
Patient Mobility (Neuro-QOL: Range 8-40)*					32.7 (7.8)	32.4 (8.2)	33.4 (7.1)

* missing data from 5 patients for cognitive functioning and 6 patients for mobility.

Table 3 presents psychosocial outcomes and change scores from Baseline to 8 weeks for each condition. Of note for scales with established cutoffs, average scores at both baseline and 8 weeks are in the mild to moderate range for caregiver burden, mild range for anxiety and depressive symptoms among caregivers, and mild range for depressive symptoms among patients.

Table 3.

Psychosocial Variables at Baseline and 8 weeks by Intervention Group

	CARING Intervention			Wait List Control
Caregiver Variables (Measure, Possible Range)	Baseline M (SD)	8 Weeks M (SD)	Change M (SD), Cohen’s d	Baseline M (SD)	8 Weeks M (SD)	Change M (SD), Cohen’s d
Social Support (Duke/UNC Functional Social Support Questionnaire, 8-40)	31.6 (7.9)	31.4 (8.8)	–0.1 (7.2), 0.01	31.7 (7.8)	32.0 (6.2)	–0.2 (6.3), 0.02
Self Efficacy (Total; Communication and Attitudinal Self-Efficacy Scale for Cancer, 19-76)	61.5 (8.0)	63.0 (8.4)	1.6 (6.6), 0.24	59.4 (8.3)	58.7 (8.5)	–0.8 (6.1), 0.13
Stress & Frustration vs. In Control & Hopeful (4-16)	11.4 (3.3)	11.3 (3.8)	–0.2 (3.0), 0.06	11.1 (3.9)	12.4 (3.2)	1.4 (2.6), 0.54
Burden (Zarit Burden Interview, 0-48)	13.7 (8.7)	13.9 (8.7)	0.3 (7.0), 0.04	11.8 (9.2)	13.5 (10.2)	1.8 (5.4), 0.34
Personal Gain (Caregiver Reaction Scale subscale, 4-16)	11.6 (3.3)	12.3 (3.0)	0.7 (2.5), 0.27	11.2 (3.3)	11.4 (3.3)	0.2 (2.0), 0.12
Anxiety (Generalized Anxiety Disorder Scale-7, 0-21)	8.5 (5.8)	8.6 (6.1)	0.1 (4.0), 0.03	9.4 (6.1)	9.6 (6.0)	0.3 (4.4), 0.06
Depressive symptoms (Patient Health Questionnaire-8, 0-24)	7.0 (5.5)	6.9 (5.6)	–0.1 (4.4), 0.03	7.8 (6.5)	7.5 (5.2)	–0.2 (3.8), 0.05
Patient Variables (Measure, Possible Range)	Baseline M (SD)	8 Weeks M (SD)	Change M (SD), Cohen’s d	Baseline M (SD)	8 Weeks M (SD)	Change M (SD), Cohen’s d
Anxiety (Generalized Anxiety Disorder Scale-7, 0-21)	5.7 (4.9)	5.5 (5.0)	-0.6 (4.2), 0.14	4.5 (4.3)	5.9 (6.0)	1.4 (5.3), 0.27
Depressive symptoms (Patient Health Questionnaire-8, 0-24)	7.6 (5.8)	7.8 (5.5)	0.1 (4.8), 0.02	5.5 (4.8)	8.1 (6.3)	2.5 (5.3), 0.47

1)Caregiver sample sizes for individual measures range from 62-65 for the Intervention group and 27-31 for the Wait List Control group. Patient sample sizes for individual psychosocial measures range from 34-36 for the Intervention group and 22-23 for the Wait List Control group.

Values for Change are based on those who completed the baseline and 8 weeks assessments.

2)Statistically significant (a = 0.05) average Change is in bold.

3)Cohen’s d = M/SD. Typical labels for Cohen’s d are as follows: 0.2 as small, 0.5 as medium, and 0.8 as large.

Among caregivers, feelings of stress and frustration relative to in control and hope showed a significant interaction of condition by week (F(1,92) = 6.19, P = .015). The Waitlist condition had a significant increase in the average from baseline to 8 weeks (t(29) = 2.94, P = .006), whereas the Intervention had no change (t(63) = −0.46, P = .64). Among patients, there was a marginally significant interaction of condition by week for depressive symptoms (F(1,51) = 2.89, P = .096). The average for the Waitlist condition increased (t(21) = 2.13, P = .045), whereas there was no change in the average for the Intervention condition (t(30) = 0.11, P = .91). No other outcomes had changes that significantly differed by condition.

As an exploration of indicators of effectiveness of the CARING, secondary analyses of change in outcome variables were performed within the Intervention condition. There was a significant increase in personal gain (M = 0.7, SD = 2.5, t(63) = 2.13, P = .037) and a marginally significant increase in caregiver self-efficacy (M = 1.6, SD = 6.6, t(61) = 1.88, P = .065). No other outcomes had changes that significantly differed between baseline and 8 weeks.

Discussion

There is a need to develop theoretically guided interventions to support caregivers of patients with brain tumors. Although recruitment was ended early, this is still one of the largest trials of a support intervention for neuro-oncology caregivers in the research literature.¹³ Our data are consistent with prior work documenting the challenges associated with the progressive deterioration of patients over time, requiring more care, resulting in more caregiver need, and heightening distress as the disease progressed.^12,33 We also confirm prior research^20,34,35 demonstrating that neuro-oncology caregivers have high levels of burden and distress even early in the care trajectory, which has been associated with poorer outcomes over time.³⁶

Trends in our data suggest CARING may be protective for caregiver self-efficacy and encourage personal gain, such that those in the intervention condition maintained their scores on these constructs, while those on the waitlist saw reductions in their scores on these constructs. Similarly, trends suggest CARING may protect against patient depressive symptoms over time. These trends should be explored in a larger sample. Other research suggests partner effects exist in caregiver coping strategies and patient well-being³⁷ and self-efficacy specifically may be critical to improving benefit finding in caregivers and clinical outcomes in neuro-oncology patients.^38,39 Caregivers with more self-efficacy may have more confidence and mastery over their role, resulting in better execution of care tasks. In fact, caregiver mastery has been shown to be associated with GBM patient survival.⁴⁰ Given the negative trends of caregiver self-efficacy over time, clinics should consider offering caregivers ongoing opportunities to engage in education and receive affirmation around their skills, particularly as patient function changes and/or caregivers confront new realities. This may be as involved as structured educational programs,⁴¹ or may simply be extra check-ins with a navigator, nurse, or other supportive individual, as is offered in our intervention and others.^35,42,43

We observed extensive variability in caregiver outcomes at each timepoint as well as in change scores, as shown in Table 3. Neuro-oncology caregivers report a variety of unmet needs regarding the patient and their care⁴⁴ as well as family conflict surrounding caregiving⁴⁵ and other unrelated stressors such as work or their own health concerns. However, the timing and details of these needs may vary widely, depending on the patient and caregiver context.⁴⁶ In particular, the patient disease course may impact the caregiver experience. While the majority of patients had glioblastomas, there was variability in tumor type and grade, which likely impacted patient functioning and, in turn, care responsibilities. We may not be capturing all the complexity of caregivers’ contexts that predict our key outcomes, such as the interpersonal relationship or communication between patient and caregiver, or the influence of family or community-level characteristics. Further, this is a multifaceted intervention. As such, CARING is attempting to address multiple different needs and as such may impact different psychosocial outcomes for each caregiver. This may explain why other multi-component support interventions for neuro-oncology caregivers may impact burden or distress constructs, but not necessarily both.^35,42,43 More thorough and frequent assessment of how the intervention addresses fluctuating needs and the resultant change in caregiver psychosocial outcomes could inform the structure or content of the intervention.

Although our goal was to recruit patients and their caregivers as soon after diagnosis as possible to prevent rather than ameliorate burden, data suggest many caregivers in our study had taken on a caregiving role prior to formal diagnosis and already experienced deleterious effects. CARING may be more beneficial if available earlier, potentially from the patient’s diagnosis. Although it can be challenging to enroll participants in research studies during the stress surrounding diagnosis, caregivers value flexible, tailored support that helps them to respond to needs across the illness continuum,⁴⁷ and patients acknowledge the desire for their caregivers to receive support.⁴⁸ Services in the current healthcare system are primarily targeted for patients; there is a need for tailored programs that holistically address patient and caregiver needs and integrate with community resources.⁴⁹ As systems increasingly develop care coordination models to comprehensive, timely, person-centered care for patients with brain tumor,⁵⁰ there is an opportunity to build in programs such as CARING to offer support and coping strategies and help protect caregivers from burden. Clinics can also consider offering opt-out models of evidence-based caregiver support, where norms encourage early participation by default and eliminate some of the stigma that may be associated with accessing support.⁵¹

Limitations

Our findings should be interpreted in light of the COVID-19 pandemic, which may have impacted study outcomes in two ways. First, due to a lower neuro-oncology clinic census,⁵² changing safety protocols, and administrative guidance from funders, we were required to significantly reduce our planned enrollment, greatly reducing statistical power for the primary analyses. Second, restrictions early in the pandemic limited patient and caregiver social interaction for part of the study, and to some extent, social support, one of CARING’s main mechanisms, and added complications and stress to caregiving and other domains.⁵³ A larger study conducted outside of pandemic restrictions, and with a more diverse sample, may enhance power to show stronger effects or determine differences between different types of caregivers. Additionally, though our enrollment rates exceed the average seen in dyadic recruitment in oncology settings,³² allowing caregivers to enroll in CARING independently from patients may encourage more participation and provide more data on CARING’s efficacy. A larger sample size would also allow subgroup analyses to determine particular types of caregivers, including caregivers of patients with a variety of brain tumor diagnoses, who may benefit most from CARING.

Conclusions

The CARING intervention shows promise in addressing the complex and significant needs of caregivers of patients with brain tumors by increasing coping self-efficacy, encouraging perceptions of personal gain, and protecting against burden. Caregiver participation in the intervention also may protect against patient depressive symptoms. Integrating programs such as CARING into routine clinical care early in the neuro-oncology care trajectory could help mitigate the potential negative effects of caregiving in this context.

Supplementary material

Supplementary material is available online at Neuro-Oncology Practice (https://academic.oup.com/nop/).

Funding

This work was supported by the National Cancer Institute, under award R01CA236034. This work was also supported by the Biostatistics and Bioinformatics Shared Resource at the H. Lee Moffitt Cancer Center and Research Institute, a National Cancer Center Institute designated Comprehensive Cancer Center (P30CA76292, PI: Cleveland).

Conflict of interest statement. The authors have no relevant financial or non-financial interests to disclose.

Ethics Approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Advarra Institutional Review Board (Protocol 19731) prior to study implementation. Informed consent was obtained from all individual participants included in the study.

Author Contributions

MR, SKS, MMB, BZ, and KW contributed to study conception and design. Data preparation and review were performed by MR, DW, HC, and SKS. Data analysis was performed by SKS. The first draft of the manuscript was written by MR and SKS. All authors commented on previous versions of the final manuscript. All authors read and approved the final manuscript.

Data Availability

The datasets analyzed during the current study are available from the corresponding author on reasonable request.