Introduction
Providing palliative care in the home is stressful for caregivers and can have detrimental effects on their mental health.
1,2 Palliative care is specialized care which focuses on reducing symptoms and improving quality of life for people with serious chronic illnesses, such as heart failure (HF).
3 While caregiving tasks vary based on disease process and care recipient needs, roughly 40% of caregivers report being in high-stress situations, with caregiving stress increasing as caregiving hours rise.
4 HF caregivers are especially susceptible to chronic stress due to the performance of multifaceted activities that evolve around daily HF demands.
5 Thus, it is not surprising that HF caregivers experience psychological sequalae such as depression, burden, and interference with life due to complex care situations.
6,7
In fact, caregivers experience higher levels of stress and more depression when engaged in complex care versus simplistic care.
8 Individual studies suggest that upwards of 35% of HF caregivers report at least mild symptoms of depression, with family caregivers reporting more depression than the general population.
9–13 Caregiver burden is also common in HF caregivers and encompasses both task frequency and the perceived associated burden.
14–17 Additionally, providing HF care in the home is accompanied by numerous life changes, including decreased time for family and social activities and difficulty coping with stress.
5,7,14,18 Further, research suggests a negative feedback loop, with more negative perceptions of life changes associated with higher levels of depression.
19,20 However, some evidence indicates perceived life changes improve over time, suggesting that caregivers may adapt to their role the longer care is provided.
21 But the mechanisms are currently unclear.
What this role adaptation suggests, however, is that caregivers often tap into or develop coping resources, such as social support, problem-solving, and family function, which ameliorate caregiving stress.
22,23 Adequate amounts of coping resources, such as social support helps buffer stress, while problem-solving helps caregivers manage challenges more effectively.
7,24–26 Family functioning influences individual coping based upon the family's level of cohesiveness, adaptability, and communication.
27–29 More social support and better family function in HF caregivers are linked with less depression, reduced burden, and positive perceptions of life changes.
30–32 Similarly, better problem-solving has been related to less depression, reduced burden, and improved perceived life changes in other caregiver populations.
33–36 Thus, adequate coping resources may influence the adaptation process.
Understanding role adaptation in HF caregivers is important, as caregiver mental health not only influences patient outcomes, but also caregiver physical health, assuring continued engagement in caregiving activities.
7,8,37 Prior analyses, using this same data set, support that social support and problem-solving are important caregiver coping resources, with social support mediating the relationship between caregiver burden, self-care and depression, while problem-solving mediated the relationship between caregiver demands and self-care.
10 Thus, evidence-based interventions to improve coping resources for HF caregivers is vital. Yet, ways in which to classify caregivers as possessing adequate or inadequate coping resources is lacking. Particularly, how to identify those at risk for decompensation due to inadequate resources is needed. Therefore, the purpose of this secondary analysis was to first, identify classes of HF caregivers based upon indicators of coping resources and stress, and then, to examine the relationships between the identified caregiver classes and depression, caregiver burden, and life changes.
Methods
Data in this secondary analysis were collected via an online survey in a cross-sectional, descriptive study. The parent study, published previously, used structural equation modeling to examine whether social support and problem-solving mediated the relationships among caregiver demands and burden, self-care, depression, and life changes.
10 Recruitment was coordinated between researchers at two universities in the southeastern United States. To recruit a diverse sample of HF caregivers, multiple methods were used (eg, flyers, social media, study website).
38,39 Inclusion criteria were designed to obtain a representative sample and limit extraneous variables. Caregivers were included if they were ≥ 19 years old, had provided care for a HF patient ≥ 6 months and could read, write, and speak English. Caregivers who scored ≥ 8 on the 6-Item Cognitive Impairment Test [6CIT] were excluded.
40 We did not use NYHA class as exclusion criteria because palliative care can be initiated at any state of illness and coexist with medical treatment to reduce illness morbidity.
3
Following Institutional Review Board approval and subsequent recruitment, potential participants were directed to an online survey where the informed consent was initially presented, followed by inclusion/exclusion and cognitive screening items. Survey completion represented participants’ consent to participate. The online survey was delivered using Qualtrics, a web-based, HIPPA-compliant online survey platform with multiple data safeguards.
10 Additionally, quality controls for online research were instituted which included individual examination of completed online questionnaires and removal of surveys based upon a set of empirically recommended criteria.
10,41,42 Surveys were concurrently screened for inclusion/exclusion criteria and the data cleaned throughout data collection. Of 768 online surveys completed, 224 surveys were excluded during data cleaning for duplicate responses, completion in less time than anticipated (< 10 minutes) and use of email addresses similar or unlikely to be legitimate.
10 An additional 14 surveys were excluded during cognitive screening.
10 All remaining 530 surveys were included in this analysis.
Measures
In addition to the following self-report questionnaires noted below, caregivers completed a sociodemographic and clinical survey to determine eligibility and assess key caregiver characteristics. Clinical items regarding the HF patient were also answered by proxy, including items regarding the patients’ HF symptom severity and functional ability. These responses were examined by an advanced practice nurse with cardiology expertise, with determination made regarding the NYHA HF Classification based on the caregivers’ responses.
Coping Resources
The Interpersonal Support Evaluation List (ISEL) was used to measure perceived social support.
43,44 This validated 12-item instrument measures perceived belonging, appraisal, and tangible support, with total scores ranging from 0 to 36. Higher scores represent more perceived support, with lower scores indicating less support.
43,44
Problem-solving was measured using the Social Problem-Solving Inventory Revised-Short (SPSIR-S).
45 This validated 25-item survey measures problem orientation and problem-solving style, with scores ranging from 0 to 20. Higher scores suggest better problem-solving, while lower scores indicate poor problem-solving.
44,45
Family function was measured using the 5-item Family APGAR, a validated tool which measures 5 dimensions of family functioning: adaptation, partnership, growth, affection, and resolve.
46–48 Total scores range from 5 to 15, with higher scores representing a higher level of family functioning.
46
Stress Indicators
Depression was measured using the validated 20-item Center for Epidemiological Studies-Depression (CESD) scale.
44,49 Total scores range from 0–60, with scores suggesting the severity of depression as follows: < 16 (non-depressed), 16–20 (mild), 21–25 (moderate), and ≥ 26 (severe).
49,50
The perceived burden scale of the Dutch Objective Burden Inventory (DOBI) was used to measure caregiver burden.
16 This validated 38-item instrument measures four domains of caregiving (personal and practical care, motivational and emotional support), with each item representing a particular task.
16,51 Caregivers rate the perceived burden associated with each task on a Likert-type scale. If a specific task was not performed, no perceived burden is reported. Total scores range from 1 to 3, with higher scores indicating higher burden.
16
Perceived life changes were measured using the validated 15-item BAKAS Caregiving Outcomes Scale, which measures caregivers’ perceptions of how caregiving has influenced their social functioning, subjective well-being, and physical health using a 7-point Likert-type scale.
52 Total scores range from 15 to 105, with higher scores indicating more positive caregiving-related life changes.
52
Data Analysis
Sample characteristics were analyzed using SPSS Version 23.
53 Using Mplus Version 7, a manual three-step latent variable mixture model was used to identify latent classes of HF caregivers and examine conditional effects on depression, caregiver burden, and life changes.
54–57 In step one, unconditional latent class models were tested using 11 indicators (age, gender, race, education, number of caregiver comorbidities, amount of time in caregiving role, geographic location, relationship to patient and standardized scores on the ISEL, SPSIR, and APGAR). Model fit was assessed by comparing values of Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) for 1–5 class models. The Vuong-Lo-Mendel-Rubin Likelihood Ratio Test (VLRT) and the Bootstrapped Likelihood Ratio Test (BLRT) were used to test the probability of a proposed model (
k) against a model with fewer classes (
k −1). Additionally, values of entropy were compared to evaluate class separation. Posterior probabilities for the final, best fitting model were saved, and, in step two, the optimal model was re-run fixing the logits for each class to account for classification error. In step three, four conditional effects models were run to examine mean differences in distal outcomes across classes, controlling for years since the HF patient was diagnosed, number of patient comorbidities, and NYHA class.
Results
Caregivers (
n = 530) were 41.39 (± 10.38) years old, with an approximately equal number of men (50.9%) and women (49.1%). Most were white (78.3%), college or post-graduate educated (61.7%), and urban dwelling (77.6%). HF patients were older (54.28 [ ± 14.76]), white (80%) men (56%), with NYHA Class I or II HF (70.7%), and ≤ 2 comorbidities (76.4%). Overall, based on the study measures, caregivers reported low social support (21.20 ± 6.99), poor problem-solving (11.55 ± 2.73), and some family dysfunction (11.44 ± 2.28). Additionally, most caregivers reported some degree of depression (21.37 ± 12.47), burden (1.68 ± 0.398), and negative life changes (60.36 ± 12.446) (
Table 1).
Latent Class Model
The results for 1-5 unconditional latent class models are summarized in
Table 2. Regardless of the number of classes, the values of AIC and BIC decreased with diminishing gains with each added class. The BLRT continued to be statistically significant regardless of the number of classes (results not shown). The VLRT was statistically significant when comparing a 2 versus 1 class model and 3 versus 2 class model, in both cases supporting that adding an additional class improved model fit. When comparing a 4 versus 3 class model, the results were no longer significant supporting the three-class model. Values of entropy for all models indicated good class separation (>.80). These results, combined with substantive interpretation, provided the strongest support for a three-class solution. Average posterior probabilities for the three-class model, presented in
Table 3, supported minimal classification error. Class descriptions are presented in
Table 4. In accordance with latent class naming conventions, each class was assigned a descriptive label (
Adequately Resourced, At Risk for Decompensation, Inadequately Resourced) highlighting the distinguishing features within the class. Classes were labeled based on relative results of coping resources thought to influence caregiver mental health.
Latent Class Descriptions
The Adequately Resourced class comprised 42.3% of the sample and included mostly healthy, educated, white men with fewer years of caregiving and more coping resources. Particularly, these caregivers were in their early forties with the fewest comorbidities, had a college/post graduate education, lived in an urban area, and were primarily family friends or children who had been in the caregiving role for 1–2 years. The Adequately Resourced class scored highest on measures of social support and problem-solving providing evidence of exogenous and endogenous coping resources. Family support was higher than the Inadequately Resourced class but lower than the At Risk for Decompensation class. Fewer personal health concerns, as well as high scores on social support and problem-solving, may have contributed to a lower risk for negative indicators of stress such as depression and burden in this class.
The At Risk for Decompensation class (25.1%) was the smallest class, including sicker, slightly older white caregivers in an equal proportion of males and females, who had a college/post graduate education, lived in an urban area, and had been providing care for longer periods of time. Specifically, caregivers in this class were in their early- to mid-forties and had been providing care to a spouse for ≥ 5 years. Furthermore, this class averaged just over three comorbidities, that is, three-times more than both other classes; suggesting they were also more likely to manage their own health conditions in addition to the HF patients. Caregivers in this class reported higher levels of social support and problem-solving, than the Inadequately Resourced class, but lower levels than the Adequately Resourced class. Scores on family function were highest in this class which may have provided some protective barrier against indicators of stress such as burden. However, how long that protection will last is unknown, as their perceived burden scores were almost as high as the Inadequately Resourced class, suggesting that they may be at an inflection point for future worsening caregiver outcomes.
Representing 32.6% of the sample, the Inadequately Resourced class was the second largest and included a more racially diverse group of younger, less educated, females who were newer to the caregiving role. Largely, these caregivers reported a high school education or lower, lived in an urban setting, and had been providing care for 1–2 years to an extended family member or friend. The Inadequately Resourced class scored lowest on measures of social support, problem-solving, and family function. Lack of adequate resources appear to place them at higher risk for experiencing indicators of stress, such as depression, burden, and negative life changes.
Identified Classes and Stress Indicators
CESD scores were highest in the
Inadequately Resourced class, with scores suggestive of severe depression. In the
At Risk for Decompensation class, scores were just below the threshold for mild depression, while scores for the
Adequately Resourced class indicated no depression (
Table 5). Depression was significantly higher in the
Inadequately Resourced class compared to the
Adequately Resourced (
p <.000; 95%CI [19.66, 22.66]) and the
At Risk for Decompensation classes (p <.000; 95%CI [13.54, 17.44]). Additionally, depression was significantly higher in the
At Risk for Decompensation class compared to the
Adequately Resourced class (p <.000; 95%CI [4.18, 7.17]) supporting the group naming convention, which indicates that while this group is still below the depression threshold, they are well above the
Adequately Resourced group.
Level of caregiver burden varied slightly, with the
Inadequately Resourced class reporting the highest perceived burden followed by the
At Risk for Decompensation and
Adequately Resourced classes, respectively (
Table 5). Caregiver burden was significantly higher in the
Inadequately Resourced class compared to the
Adequately Resourced (
p <.000; 95%CI [0.46, 0.56]) and
At Risk for Decompensation classes (
p <.02; 95%CI [ 0.04, 0.22]). Compared to the
Adequately Resourced, perceived burden was significantly higher in the
At Risk for Decompensation class (
p <.000; 95%CI [0.29, 0.46]).
Across classes, scores on the BAKAS were ≥ 60, with the
Inadequately Resourced class reporting the least positive life changes. The
Adequately Resourced class reported slightly more positive life changes, followed by the
At Risk for Decompensation class, which reported the most positive life changes (
Table 5). BAKAS scores were significantly higher in the
At Risk for Decompensation class compared to the
Inadequately Resourced (
p <.000; 95%CI [7.80, 17.90]) and
Adequately Resourced classes (
p <.000; 95%CI [8.12, 16.96]). There were no significant differences in BAKAS scores between the
Inadequately and
Adequately Resourced classes.
Discussion
Using a robust dataset, we identified a well-fitted 3-group model which categorized HF caregivers as Adequately Resourced, At Risk for Decompensation, and Inadequately Resourced using information that can be easily collected at health care visits. Further, our findings suggest that protective resources may provide an important buffering effect against the increased stress common in HF caregiving. Below, we discuss several noteworthy findings from this analysis.
Younger caregivers new to the caregiving role with less formal education and deficient coping resources (
Inadequately Resourced) reported severe levels of depression. Previous research supports that more caregiving experience and education may contribute to less depression in HF caregivers.
21,58 For example, research shows that depression decreases the longer one provides care.
21 Similarly, the
At Risk for Decompensation class had been providing care the longest and scored lowest for depression. Other studies suggest that depression is higher in HF caregivers with a high-school education or less versus those with more education.
58 Supporting this, the
Inadequately Resourced class, who were less educated, had the highest depression scores. Younger caregivers may also be more susceptible to mental distress, with research showing that depressed HF caregivers are an average of 6 years younger than non-depressed caregivers.
9,19 Our findings agree, with caregivers reporting severe depression (
Inadequately Resourced), being, on average, 4.29 years younger than caregivers reporting no depression (
Adequately Resourced). What our findings add to the literature is that the
Inadequately Resourced group also scored lowest on measures of social support, problem-solving, and family function suggesting modifiable mechanisms for intervention to reduce depression.
Again, our findings show that caregivers with inadequate coping resources (
Inadequately Resourced) also are at greater risk for caregiver burden, with these caregivers reporting the lowest levels of social support, problem-solving, and family function. Prior studies agree, with poor social support found to predict burden in HF caregivers.
59 Similarly, family function, as well as relationship quality appears to influence burden. Specifically, poor family function is associated with greater burden, while better relationship quality is associated with less burden.
30,31,60 Data is limited on problem-solving and caregiver burden in HF, but in other caregiver populations, improved problem-solving is associated with reduced burden.
10,35 Our study suggests that interventions which enhance coping resources by strengthening social support, problem-solving, and family function may be useful to combat increased burden, particularly in caregivers at high-risk for burden.
Largely, all caregivers reported positive life changes; however, those who were younger (
Inadequately Resourced) reported the least, while those with less caregiving experience (
Adequately Resourced) reported slightly more. Spousal caregivers with more education and caregiving experience (
At Risk for Decompensation) reported the most positive life changes. Like depression and burden, coping resources were lowest in those reporting the least positive life changes (
Inadequately Resourced), emphasizing the significance of these resources to caregiver mental health. Notably, family function was highest in the
At Risk for Decompensation class and may have provided some protective barrier against negative perceptions of life changes. Previous research shows that family function is important to maintaining caregiver psychological well-being and may be beneficial in helping caregivers cope with daily stressors, as family members usually share similar values and beliefs.
14,32,61,62 Interventions which improve family function may enhance the caregiving experience and influence perceived life changes.
Notably, our findings also suggest a gender imbalance among those who were adequately versus inadequately resourced to care. The
Adequately Resourced class included primarily men, while those in the
Inadequately Resourced class were mostly women. Stress-coping theories may partially explain this disparity. Often, women use emotion-focused coping and ineffectual coping styles such as denial and avoidance. In contrast, men use more effective coping strategies (eg, social support, problem-solving, acceptance, detachment) than women.
63 Prior research also indicates that men who have fewer social networks (eg, social support and family) are more likely to ask for and receive outside support from formal and informal sources compared to women.
64 Furthermore, while women value the problem-solving process and more frequently discuss their feelings in developing solutions, men tend to focus on the goal or outcome of problem-solving in identifying effective solutions.
65 Gender roles may also influence group imbalance, with women often performing primary roles including taking care of children, spouses, and parents, while men often assume the identify of financial provider.
66 Despite the possible reasons, there is inadequate research examining HF caregiving interventions, especially those which target gender.
7
Our study included mostly younger caregivers, with an almost equal number of males and females, adding to the HF palliative care literature. However, our findings may not be representative of the typical HF caregiver population. Further, most caregivers were highly educated, urban dwelling, and white limiting generalizability of findings. The use of a self-report online questionnaire may have contributed, not only to which population of caregivers participated in the study, but also to response bias regarding the study variables. Likewise, caregivers primarily cared for patients with less severe HF (NYHA Class I or II) and few comorbidities (≤ 2), potentially influencing the stress indicators examined. Lastly, although quality control measures were undertaken, participants self-identified, thus, non-caregivers may have been erroneously included.