Health Coaching Improves Outcomes of Informal Caregivers of Adults With Chronic Heart Failure: A Randomized Controlled Trial

Abstract

BACKGROUND:

Caring for someone with heart failure takes an emotional and physical toll. Engaging in self-care may decrease stress and improve the health of informal caregivers. We conducted a randomized controlled trial testing the efficacy of a virtual health coaching intervention, compared with health information alone, on the self-care, stress, coping, and health status of heart failure caregivers.

METHODS:

We enrolled 250 caregivers providing care at least 8 hours/week, reporting poor self-care, and able to use technology. All received a tablet device programmed with websites providing vetted information on heart failure and caregiving. Half were randomized to also receive 10 synchronous support sessions virtually with a health coach over 6 months. Data on self-care, stress, coping, and health status were collected at baseline and 3 and 6 months. Linear mixed-effects models were used to assess the interaction between time and treatment group.

RESULTS:

The sample was majority female (85.2%), White (62.2%), spouses (59.8%), and aged 55±13.6 years. Many were employed full time (41.8%). They had been caring for the patient 8 hours/day for a median of 3.25 years. In the intention-to-treat analysis, caregivers who received the health coach intervention had statistically and clinically greater improvement across 6 months compared with the control group in the primary outcome of self-care maintenance (5.05±1.99; P=0.01) and stress (−4.50±1.00; P<0.0001). Self-care neglect declined significantly (−0.65±0.32; P=0.04), but the difference between the treatment arms disappeared when the results were adjusted for multiple comparisons. Mental health status improved statistically but not clinically (3.35±1.61; P=0.04). Active coping improved in both groups but not significantly more in the intervention group (P=0.10). Physical health status was unchanged (P=0.27).

CONCLUSIONS:

This virtual health coaching intervention was effective in improving self-care and stress in heart failure caregivers.

WHAT IS NEW?

• Patients with heart failure are living longer and remaining in the community with support from caregivers. Caregiving for a loved one is difficult. Few interventions have been developed to support the caregivers of these patients. We developed a virtual support intervention delivered by health coaches. All caregivers were provided a tablet with vetted websites addressing heart failure and caregiving. In those randomized to the intervention, a health coach worked with each individual caregiver, discussing caregiving demand, stress, self-care, sleep, thought distortions, automatic thoughts, relaxation techniques, and confidence. At 6 months, caregivers in the intervention group had improved self-care and decreased stress.

WHAT ARE THE CLINICAL IMPLICATIONS?

• Caregiving for a loved one is stressful, especially when the caregiving role is lengthy. Caregivers who are stressed risk their own health because of poor self-care. Clinicians can use these results to support caregivers and minimize the potential for burnout. Just acknowledging the demands of caregiving communicates support. Encourage caregivers to attend to their own self-care by seeking support from others and formal respite care. Emphasize the importance of sleep, exercise, a healthy diet, and relaxation.

More than 53 million unpaid informal caregivers provide regular care or assistance to a friend or family member with a health problem, limitation, or disability,^1,2 carrying the bulk of responsibility for the care of these individuals.³ Most patients with chronic heart failure (HF) remain in the community through the end of their lives, depending on informal caregivers to assist them.⁴ HF is extremely common, with a prevalence of 6.7 million US adults, which is anticipated to rise to 8.5 million by 2030.⁵ The trajectory of illness in HF is highly variable, which limits the use of palliative care and respite services.⁶ Serving as the primary caregiver can take an emotional and physical toll. These informal caregivers are commonly overwhelmed and face emotional, psychosocial, and financial risks that result in a decline in their own health and wellness.⁷ The estimated cost of replacing informal caregivers by unskilled and skilled paid caregivers was $221 and $642 billion in 2015.⁸ Clearly, these caregivers need to be supported because, without them, the burden on the health care system would be overwhelming and unsustainable.

HF caregivers manage complex medication regimens, assist with symptom monitoring and response, motivate physical activity, reinforce dietary restrictions, and coordinate health care visits.⁹ These responsibilities often cause the caregivers to neglect their own personal health and self-care needs, and many report feeling left alone to cope with problems.¹⁰ Engaging in self-care, defined as behaviors undertaken to maintain health and manage illness,¹¹ has been shown to improve mental and physical well-being in patients with chronic illness.¹² Yet, few caregiver interventions specifically support the self-care of caregivers.

Caregiver interventions typically offer education and skills training on patient disease and care management.¹³ A systematic review of interventions for HF caregivers showed limited effectiveness.¹⁴ With only half impacting caregiver burden and well-being, alternative approaches are imperative. A 2022 meta-analysis revealed that despite some positive primary study outcomes, the overall effects on caregiver depression and quality of life were not significant.¹⁵ In contrast, dementia caregiver interventions seem more successful, with over half showing beneficial outcomes, particularly those using cognitive behavioral therapy.¹⁶

We tested the efficacy of a virtual support intervention, virtual caregiver coach for you (ViCCY), designed to promote self-care of HF caregivers.^17,18 Performing self-care was anticipated to decrease stress and improve health status. All caregivers received an enhanced standard of care with health information provided through vetted websites, and those randomized to the intervention received individual health coaching sessions. We hypothesize that 6 months after enrollment, caregivers randomized to ViCCY would have greater improvement in self-care, stress, coping, and perceived mental and physical health status compared with those receiving only health information.

METHODS

Trial Design and Oversight

The study adheres to Transparency and Openness Promotion Guidelines, and the report is formatted using the CONSORT (Consolidated Standards of Reporting Trials) statement for randomized trials of nonpharmacologic treatments.¹⁹ The data that support the findings of this study are available from the corresponding author upon reasonable request. We used a randomized controlled trial design, enrolling a consecutive sample of 250 informal HF caregivers between 2019 and 2022. After providing informed consent, we collected baseline data and caregivers were block-randomized 1:1 to the intervention or health information-only control group. We stratified randomization by caregiver sex and relationship to the patient (eg, spouse) because these factors have been shown previously to influence caregiving burden, perceived stress, and receptivity to interventions.^20,21 An independent data and safety monitoring committee providing oversight recommended also stratifying by race. The study protocol was approved by the institutional review board of the University of Pennsylvania.

Setting and Identification of Participants

Caregivers were enrolled primarily from the Heart and Vascular Center, clinical practices, and the acute care hospital affiliated with a major regional referral center in the northeastern United States. The COVID pandemic began during the first year of the study. Recruitment occurred in person before the pandemic began, but, afterward, recruitment was predominately virtual using monthly information sessions offered to caregivers referred by clinical staff. In-person enrollment resumed 4 months into the pandemic although access to caregivers was limited for ≈1 year. The Project Manager and Clinical Research Coordinator screened caregivers and enrolled all study participants.

Inclusion Criteria

Caregivers were eligible to participate in the randomized controlled trial if the person they cared for had HF, regardless of treatment. This included caregivers of patients who were managed medically or surgically, including those with mechanical circulatory support, or cardiac transplantation. Caregivers had to score ≥2 on the health self-care neglect (HSCN) scale²² and provide ≥8 hours of support to the patient with HF each week.²³ Trained research staff judged the caregiver’s ability to complete the protocol based on hearing, vision, and English language skills. The only change in the planned inclusion criteria was the need to live within 50 miles of the research office in case a home visit was needed to facilitate the use of the technology. Early experience with the COVID pandemic taught us that this restriction was not necessary because we were able to enroll caregivers virtually and mail the tablets.

Exclusion Criteria

Caregivers were assessed for adequacy of cognition with the Telephone Interview for Cognitive Status; those with a score <25 were excluded.²⁴ Caregivers participating in another support trial, those with untreated major psychiatric illness (eg, schizophrenia) or self-reported inability or discomfort in using technology, were excluded. If the patient was expected to die imminently (ie, in hospice), the caregiver was excluded but, if already enrolled, caregivers were invited to stay in the study if the patient entered hospice or died.

Sample Size

To detect significant differences in primary outcome of self-care, we estimated a sample size of 250 (125 per study arm).²⁵ Our pilot studies informed estimates, predicting that the intervention would result in greater improvements in self-care over 6 months. We expected a 20% dropout rate, as previously observed. With 3 assessment points—baseline and 3 and 6 months—and 100 participants per group postattrition, we would achieve 99% power to evaluate the between-group effect, 99% power for the time effect, and 90% power to assess the interaction between group and time, using a 5% significance threshold and no interim analyses.

Randomization

A biostatistician independent of the study investigators generated the randomization sequence using a randomly permuted blocks algorithm to ensure equal distribution of caregiver sex, relationship to the patient (spouse versus nonspouse), and race (White versus other) in each study arm. After a new participant was fully enrolled and had completed the baseline survey, the project manager accessed sequentially numbered containers to allocate them to the appropriate group. Then, she notified each participant by email to tell them of their group assignment. If the participant was randomized to the intervention group, a health coach was assigned and prompted to initiate contact.

Blinding

The principal investigator, most coinvestigators, and staff collecting data were blinded to study participants’ group assignments. Only the project manager, health coaches, and coinvestigators responsible for intervention fidelity were unblinded because they needed to have direct contact with caregivers or review recorded intervention sessions. Study participants were not blinded to their group assignment because it was obvious to them when they were contacted by a health coach. Caregivers were, however, told that both study arms may provide some benefit to them.

Intervention

The intervention is described in detail in the published protocol.²⁵ In brief, the intervention was virtual, delivered in a home environment where distractions are predictable and privacy can be optimized. While much of the US population, including older adults, has access to a computer and broadband Internet,²⁶ to make sure that all participants had equal access, study participants were provided with Samsung tablets with mobile connectivity, an embedded camera capable of full 2-way duplex video, and real-time audio transmissions. All tablets were preprogrammed with vetted websites addressing HF and caregiving. Participants were encouraged to spend ≥30 minutes each week reviewing the content on the websites, with reminders provided monthly to do so during the 6 months of their active participation. Websites’ links were checked quarterly for problems and updates pushed out remotely or fixed when tablets were returned before redistribution. At 3 and 6 months, caregivers were asked if they were spending at least 30 minutes weekly on the websites, but we could not confirm these reports objectively and were not able to use this in the analysis. A newsletter was also prepared and emailed quarterly to all caregivers to maintain engagement. The newsletter featured recipes, caregiver tips, and local events.

In addition to health information, caregivers randomized to the intervention were assigned a health coach who offered 10 sessions of synchronous virtual health coaching over a 6-month period. The intervention was provided by video instead of telephone because video offers important visual cues that can promote the therapeutic alliance.²⁷ Sessions were designed to provide support for coping with the stress of caregiving by encouraging the caregivers to engage in self-care. Session content addressed caregiving demand, stress, self-care, sleep, thought distortions, automatic thoughts, relaxation techniques, and confidence.²⁸ Each session was described in detail in a treatment manual provided to the health coaches. Caregivers also received a manual by email with suggested homework for each session. Health coaches were allowed to modify intervention sessions if unplanned subject content came up during the discussion with the caregiver (eg, COVID challenges, anticipatory grief). If a caregiver was not interested in addressing the content planned for a specific session, the health coach did not insist and instead addressed the caregiver’s concerns. The pandemic did not influence intervention delivery because it was always intended to be virtual. A single health coach provided all sessions for their assigned caregivers. Three early sessions completed by each health coach were reviewed by 2 team members to identify protocol deviations.

Five experienced health coaches with graduate degrees in nursing or public health were hired and trained to provide the ViCCY intervention sessions. Four had formal training in health coaching, and 1 had extensive experience in health coaching. Three of these health coaches were trained in person during a 2-day session at the start of the study, before the pandemic. Two more were trained virtually when staff turnover and caregiver volume required more staff. The training focused on the content described in the intervention manual. The content of the training was the same for all 5 health coaches, but the 3 trained in-person were able to tour the Heart and Vascular Center and hospital units in addition to meeting the cardiologist coinvestigator and the social worker.

Health coaches were asked to contact new participants within 2 days of receiving the assignment and schedule the first session within 1 week of the call although there was often a delay in reaching caregivers. Although we intended the sessions to be front-loaded, with early sessions occurring weekly and decreasing in frequency over time, session timing varied based on caregiver availability. All sessions had to be finished within the 6-month intervention period, so some sessions were combined to complete the specified content within that period.

Intervention fidelity was addressed through study design, planning for implementation setbacks, health coach training, and monitoring of intervention delivery and enactment. Three investigators held virtual sessions with the health coaches twice monthly where issues (eg, technology challenges,²⁹ caregiver engagement, intervention fidelity) were discussed. All intervention sessions were audio-recorded and reviewed by unblinded coinvestigators to monitor intervention enactment, and 21% of recordings were coded for fidelity. Overall intervention fidelity was achieved 93% of the time (session range, 85%–99%).

Outcomes and Measurement Procedures

Data were collected at baseline and 3 and 6 months. The 3-month session was intended to keep participants engaged in the study, and they only completed one of the primary outcome measures of self-care during that brief contact. Self-care was assessed with the maintenance scale of the self-care inventory (SCI)³⁰ and the HSCN scale.³¹ These 2 measures are related, but they do not measure the same construct, as self-care reflects a priority given to specific behaviors, while neglecting self-care reflects a choice that may be intentional or due to deprioritizing one’s own health while caregiving. Higher self-care maintenance scale scores indicate improvement in health-promoting behaviors. Lower HSCN scale scores indicate less neglect of oneself. Secondary outcomes assessed were perceived stress, coping, and health status. Perceived stress was assessed with the perceived stress scale, with scores ranging from 0 to 40 and higher scores indicating higher perceived stress.³² Scores between 0 and 13 are considered low stress, scores between 14 and 26 reflect moderate stress, and scores between 27 and 40 reflect high perceived stress. Coping was assessed with the ways of coping questionnaire, short-form, with scales measuring active, avoidance, and minimization coping.³³ Increasing active coping scores and decreasing avoidance and minimization coping scores indicate improvement. Health status was assessed with the original short-form 36 scale.³⁴ Increasing short-form 36 scores indicates improvement. In addition to these measures, caregiver burden,³⁵ sociodemographic, and clinical characteristics were assessed at baseline and used to describe the sample of caregivers. A half SD change was considered clinically relevant for all scores.³⁶ Patients’ sociodemographic and clinical information was collected from the medical record.

Data were collected virtually. Staff arranged follow-up appointments and then called the caregiver and pushed the survey through REDCap to the caregiver’s email. Caregivers entered responses directly into REDCap while research staff remained on the call and were available to address questions while the survey was completed. In a few instances (≈5×), when caregivers had problems accessing technology, staff interviewed the caregivers and completed the surveys in REDCap.

Statistical Methods

Using intention-to-treat analysis, participants’ clinical and sociodemographic characteristics were compared between treatment groups using the Fisher exact tests and 2-sample t tests for categorical and continuous measures. To assess intervention efficacy, linear mixed-effects models were constructed for each outcome in which a group-by-time interaction was evaluated. All models included a random intercept at the participant level to account for repeated measures across time. To assess the impact of attrition on study results, participant characteristics were compared between those who withdrew and those who completed the 6-month time point. Additionally, χ² was used to determine whether the treatment group was associated with caregivers’ transition to a lower perceived stress scale category over time.³²

To test the significance of our 2 primary outcomes, we applied the Bonferroni correction and considered P<0.05/2=0.025 to be statistically significant. When the 2 primary and 6 secondary outcomes were considered simultaneously, we used a false discovery approach to adjust for multiple comparisons. Missing data were minimal, so no data were imputed. An α level of 0.05 was used for statistical significance. All analyses were performed using the original assigned groups. Analyses were conducted using SAS, version 9.4, for Windows.

RESULTS

A total of 667 caregivers were assessed for eligibility; 250 were enrolled and randomized to the ViCCY intervention (n=125) or the control group (n=125; Figure 1). The primary reasons for ineligibility were an HSCN score that was too low (n=124); the caregiver was not supporting the patient ≥8 hours each week (n=21) or discomfort with technology (n=13). Those who were eligible and declined to participate commonly felt that the intervention was too time-consuming (n=123) or said that they did not feel the need for support (n=87). Of those enrolled and assigned to the intervention group, completion rates varied (mean, 6±4 sessions; median, 9; 21% did not complete any formal sessions).

Figure 1.

CONSORT (Consolidated Standards of Reporting Trials) diagram. The analytic sample size is shown in Table 4. Note that 8 of 125 (6%) caregivers withdrew from the virtual caregiver coach for you (ViCCY) intervention but stayed in the study.

Attrition was 22%, with 56 caregivers withdrawing from the study within 6 months. In the intervention group, 25% withdrew or were lost to follow-up. In the control group, 20% withdrew or were lost to follow-up. Most of these participants dropped out within 3 months. There was only 1 significant difference at baseline between those who subsequently dropped out: 48.2% of those who subsequently dropped out rated their quality of social support as good, while only 31.6% of those who completed the study rated their support as good (P=0.04).

The full sample of 250 was predominately female (85.2%), White (62.2%), married (71.9%), middle aged (55.3±13.6 years), college educated (45.8%), employed full or part time (58.3%), and reported having sufficient financial resources (82.3%; Table 1). Most participants were spouses (59.8%) who cared for their partner 8 hours per day and had been doing so for a median of 3.25 years. Most caregivers reported having no chronic conditions (53.4%). Ten patients in each group died during the 6-month follow-up. At baseline, only 1 significant group difference was found in caregivers (avoidance coping) and no differences were found in patients (Table 2), so neither sociodemographic nor scale scores were adjusted in the analysis.

Table 1.

Sociodemographic Characteristics and Baseline Instrument Scores of Caregivers Measured at Enrollment

Table 2.

Sociodemographic and Clinical Characteristics of Patients Measured at Enrollment

Six-Month Outcomes

A significant group-by-time interaction was observed for the primary outcome, SCI maintenance scale (Figure 2). The mean SCI maintenance scale score of caregivers in the intervention group was estimated to increase by 5.05 (95% CI, 1.12–8.98) units more than that of caregivers in the control group from baseline to 6 months (P=0.01; Table 3). Across 6 months, the mean SCI maintenance score increased by 10.01 (95% CI, 7.18–12.84) units in the intervention group but only by 4.96 (95% CI, 2.23–7.69) in the control group. The increase was both statistically significant (P=0.01) and clinically relevant.

Figure 2.

Changes in self-care over time and by group. A, Change in the self-care inventory maintenance scale (standardized), over time and by group. B, Change in the health self-care neglect scale over time and by group.

Table 3.

Six-Month Effect Size Estimates: ViCCY Intervention Compared With Health Information Control

A significant group-by-time interaction was also observed for the HSCN scale. Specifically, mean HSCN scale scores of participants in the intervention group were estimated to decrease by 0.65 (95% CI, −1.27 to −0.03) units more than that of participants in the control group from baseline to 6 months.

Without adjustment for multiple comparisons, there was a statistically significant difference between the 2 treatment groups for both self-care maintenance (P=0.012) and health self-care neglect (P=0.041). After adjustment for multiple comparisons, the difference between the 2 treatment arms remained statistically significant for self-care maintenance (P=0.012<0.025) but not for health self-care neglect (0.041>0.025).

For the secondary outcomes, there was a significant group-by-time interaction for the perceived stress scale score with the intervention group decreasing 4.5 (95% CI, −6.48 to −2.52) points more than the control group. The decrease in stress was statistically significant (P<0.0001) and clinically relevant in the intervention group. Furthermore, at 6 months, 44.6% (n=41) of caregivers in the intervention group transitioned to a lower stress category, while only 22% (n=22) of the control group improved (P=0.0009). Coping scores did not exhibit a significant group-by-time interaction. Caregivers’ mental health status improved 3.35 points more than the control group (P=0.04), but this improvement was not clinically relevant. No effect was seen for the short-form 36 physical health score. After adjustment, we observed statistically significant differences between treatment groups for the primary outcome self-care maintenance (positive false discovery rate, 0.048) and the secondary outcome of stress (positive false discovery rate, 0.0001), whereas without this adjustment, self-care maintenance (P=0.012), self-care neglect (P=0.041), stress (P<0.0001), and mental well-being (P<0.0388) were statistically significant.

Change Over Time

As shown in Table 4, both groups improved significantly over time in the SCI maintenance scale score although only the improvement in the intervention group was clinically meaningful. Both groups also improved in self-care neglect and active coping over time. Only the intervention group improved in perceived stress and mental health status over time. As this study took place during the COVID pandemic, we compared change over time in the 40 participants recruited before to those recruited after the pandemic began. No differences in outcomes were observed between these cohorts.

Table 4.

Sample Mean Change in Instrument Scores Over Time in Caregivers Randomized to the Intervention and Control Groups

DISCUSSION

The purpose of this study was to test the efficacy of a virtual support intervention provided by health coaches, ViCCY, designed to improve self-care of caregivers of adults with chronic HF. At 6 months, the intervention effectively improved self-care and stress, and the extent of improvement was significantly greater than the benefit seen with health information alone. An important finding was the clinically meaningful degree of improvement in self-care seen in the caregivers randomized to ViCCY. Notably, stress was the outcome that improved the most.

An interesting finding was the number of caregivers who met eligibility but declined participation, reporting that the intervention was too time-consuming, and others who enrolled and were randomized to the intervention but did not fully engage with the health coach. We designed the intervention to provide a series of sessions so that the health coach could develop a relationship with the caregiver and achieve a lasting effect.¹⁷ In preliminary studies, we found a direct relationship between the number of sessions provided and the effect obtained, with 10 sessions of health coaching a sufficient dose. More recent studies of dementia caregiver interventions provide support for low-intensity interventions lasting no more than 6 sessions.³⁷ Considering that the average number of sessions completed by these caregivers was 6, a low-intensity approach may also be useful in HF caregivers but will need to be tested. Another interesting finding was that caregivers who withdrew from the study rated the quality of their social support better than those who remained in the study. Others have found that caregivers without strong social support are more likely to feel burdened and, thus, perceive more of a need for a support intervention.^38–40 Future investigators may want to add inadequate social support as a study inclusion criterion to assure that participants are able to benefit from the intervention. We specified that caregivers needed to be caring for the patient for at least 8 hours weekly. Caregiving can impose both subjective and objective burdens due to individual perception, multifaceted strain, and change over time.⁴¹ We assumed that 8 hours of caregiving weekly would be burdensome; however, other factors such as insufficient financial resources, multiple responsibilities, and lack of social activities might be considered as inclusion criteria in future studies.

We were surprised that the ViCCY intervention did not significantly improve coping. One reason for the lack of effectiveness could be the Web-based delivery; a recent meta-analysis of Web-based interventions for caregivers of people with dementia demonstrated a poor effect of virtual interventions on coping.⁴² It is important to note, however, that only 2 Web-based interventions were included in that meta-analysis. Our intervention addressed active coping strategies such as problem-focused coping, which has been effective in dealing with specific problems such as caregiver burden.⁴³ A recent systematic review of coping interventions for caregivers found the most effective interventions to be psychoeducation, cognitive behavioral therapy, and resilience-building activities,⁴⁴ all of which were included in this intervention. Another reason for lack of effectiveness could be the measure used for coping, which assesses coping style. Measures of coping are known to have many limitations.⁴⁵ Furthermore, coping can be influenced by cognitive behavioral therapy but may not be responsive to health coaching. Further research is needed to determine if a health coaching intervention can influence coping.

Strengths and Limitations

Limitations include enrollment from a single center although the center receives referrals from a diverse northeastern US region. Attrition was larger than anticipated, and intervention engagement was variable. Although the sample enrolled was predominately White, female, and well-educated, it was large and diverse in many ways; we enrolled a racially diverse sample of almost 40% from groups who did not identify as White and almost 15% men. Another strength was the strong study design and robust attention to intervention fidelity.

A review of mHealth interventions for caregivers of older adults identified that social determinants of health influence accessibility of support resources for caregivers.⁴⁶ Although our sample was racially diverse, 85% of the total sample reported having sufficient financial resources, so the sample was not economically diverse. The intervention may need to be modified to be more focused on equity to minimize attrition and assure accessibility by caregivers who need support.

Conclusions

The ViCCY intervention tested in this randomized controlled trial was highly effective in improving self-care and stress in this sample of caregivers of patients with HF. Future research is needed to modify the intervention to improve coping and mental health status and improve generalizability by making the intervention more appealing to a diverse population. Future research might target populations with less educational access and lower economic stability. If this intervention is shown to be cost-effective, currently in analysis, a pragmatic trial may be considered as the next step.

ARTICLE INFORMATION

Acknowledgments

The authors are grateful to the health coaches who delivered the intervention, the staff who enrolled and followed participants for data collection, and the clinic and hospital staff who referred caregivers to the study. The authors are especially grateful to Debbie Gordon, LCSW, who facilitated enrollment and supported the caregivers and the health coaches throughout the study period.

Sources of Funding

The research reported in this publication was supported by the National Institute of Nursing Research of the National Institutes of Health (grant R01 NR-018196). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Disclosures

None.