Symptom Clusters and Influencing Factors in Family Caregivers of Individuals With Cancer

Abstract

Background:

A symptom cluster is a group of 2 or more symptoms that occur together and are related to each other. Although family caregivers of individuals with cancer experience multiple concurrent symptoms, the majority of symptom research has focused on assessing and managing individual, isolated symptoms.

Objective:

The study purpose was to investigate symptom clusters in cancer caregivers and to explore factors that influence symptom clusters.

Methods:

Cluster analysis was performed using cross-sectional survey data from 129 family caregivers of individuals receiving cancer treatment at the National Institutes of Health Clinical Center. PROMIS (Patient-Reported Outcomes Measurement Information System) measures of 5 common symptoms in caregivers (fatigue, sleep disturbance, depression, anxiety, impaired cognition) were used to identify symptom clusters.

Results:

Two symptom cluster groups were identified: low symptom burden (n = 106, 82.2%) and high symptom burden (n = 23, 17.8%). Individuals who reported higher levels of caregiving burden (impact on health subscale) (β = 1.31, P = .005) and loneliness (β = 0.18, P = .024) were significantly more likely to be in the high symptom burden group.

Conclusions:

This study provides evidence that 5 key symptoms among cancer caregivers appear to cluster into 2 groups, those with low symptom burden and those with high symptom burden. Caregiving burden (impact of health) and loneliness were significant factors differentiating symptom cluster membership.

Implications for Practice:

Identifying symptom clusters may lead to better prevention and treatment strategies that target symptoms in cancer caregivers. Identifying factors that place a group at high risk of symptom burden can be used to guide individualized and tailored interventions.

As a result of the growing population of individuals with cancer and survivors, an estimated 2.8 million family members and friends reported serving as informal cancer caregivers in the United States in late 2015.¹ As medical management of cancer becomes more complex, family caregivers are essential partners and emotional companions throughout the cancer trajectory.^2–4 Family caregivers often face changes in their roles and responsibilities, economic circumstances, and interpersonal relationships that they may be underprepared to handle.⁵ Although many individuals report positive experiences as caregivers,^6,7 large numbers also report unmet needs and substantial emotional and practical burdens.^1,8 This burden may cause psychological and physiological changes in the caregivers that may negatively influence their quality of life as well as their mental and physical health.⁹ Yet too often both the role and needs of caregivers are understudied and undertreated.²

In the face of multiple concurrent stressful events, serving as a family caregiver has been correlated with an increase in physical and psychological symptoms including fatigue, sleep disturbance, depression, anxiety, and cognitive impairment.^2,4,10 The presence or exacerbation of preexisting symptoms, as well as the development of new symptoms during the course of caregiving, might interfere with caregivers’ ability to fulfill their caregiving roles. These symptoms not only impact the caregiver’s functional status and quality of life, but also may have a negative impact on the care recipient’s well-being as well.^2,4

While most studies have focused on individual symptoms, it is not uncommon for research regarding caregivers to report the presence of co-occurring symptoms or symptom clusters. A symptom cluster is defined as “2 or more symptoms that are related to each other and that occur together. Symptom clusters are composed of stable groups of symptoms that are relatively independent of other clusters, and they may reveal specific underlying dimensions of symptoms.”^11(p278) Many of these symptoms appear to co-occur, as significant relationships between individual symptoms have been observed in several studies, including fatigue and with sleep disturbance,^12,13 fatigue with depression,¹² sleep disturbance with depressive symptoms,^14,15 depression with anxiety,^5,16 depression with cognition,¹⁶ and anxiety with cognition.¹⁶

Unfortunately, the majority of previous research on symptoms experienced by family caregivers has focused on bivariate relationships between isolated symptoms, with little attention given to symptom clusters. Three published research studies have focused on symptom clusters in caregivers,^17–19 and these studies suggest that symptoms may be reciprocally associated and that they may have common underlying etiologies. Two groups of researchers have used composite scores of multiple, co-occurring symptoms to follow symptoms over time in caregiving populations.^18,19 In spousal caregivers of individuals with dementia,¹⁸ researchers investigated symptom cluster levels using a composite score and found that caregivers experienced concurrent, multiple symptoms, over time including pain, fatigue, and depression. Another longitudinal study in 85 cancer caregiver dyads¹⁹ explored symptom burden calculated by a composite score of 4 of the most severe symptoms using the MD Anderson Symptom Inventory; researchers identified 2 subgroup trajectories, individuals with low symptom burden and those with high symptom burden. However, other researchers have determined that using a composite symptom score may not be appropriate, particularly when there are inconsistencies across the components of the composite score.²⁰ Only 1 study¹⁷ used the person-centered approach, latent class analysis, focusing on similarities or relationships among individuals and symptoms.²¹ They identified 3 symptom cluster groups characterized by 3 distinct patterns of symptom experience, which included pain, fatigue, sleep disturbance, and depression in a combined sample of both cancer patients and caregivers. These clusters included individuals with low depression and low pain, high depression and low pain, and high levels of all symptoms.

While a few researchers have examined the presence of symptom clusters in caregiving populations, fewer have examined factors that may contribute to their development. Jaremka and colleagues^18,22 explored the relationship between loneliness and changes in symptom clusters (pain, depression, fatigue) in diverse populations including caregivers of individuals with Alzheimer’s disease, noncaregiver controls, cancer survivors, and noncancer controls. These researchers found that loneliness was a common risk factor for the 3 co-occurring symptoms across these populations. Research is also limited regarding what factors associated with caregiving specifically contribute to symptom clusters in cancer caregivers, although several researchers have examined factors that contribute to individual symptoms such as fatigue, sleep disturbance, depression, and anxiety. In a systematic review and meta-analysis of 13 research articles, factors found to be associated with an increased risk of depression in cancer caregivers included being younger, female, a spousal caregiver, less educated, an unemployed, as well as having poorer sleep quality, a chronic disease, a longer duration of caregiving, and higher caregiving burden.²³ Govina and colleagues²⁴ found that caregivers who were female, younger, caring for a male patient, and living with the recipient and who had higher levels of perceived burden reported higher levels of anxiety and depression. According to Stenberg et al,²⁵ being female, older, less educated, unemployed, and a child/sibling/parent (versus a spouse) and having higher caregiving burden were factors related to an increased risk of fatigue, sleep disturbance, and depression among cancer caregivers. Understanding factors that influence the development of symptom clusters in caregivers is critical for designing suitable interventions and support services to alleviate caregivers’ symptom burden. To date, no known study has focused exclusively on cancer caregivers or has evaluated how different sets of characteristics of both cancer patients and caregivers are associated with symptom clusters. Therefore, the purpose in this study was to determine (1) if clusters of 5 highly prevalent symptoms (ie, fatigue, sleep disturbance, depression, anxiety, impaired cognition) in cancer caregivers could be identified and (2) what factors influence membership in symptom cluster groups.

Methods

Study Design and Participants

A cross-sectional internet survey design was used to evaluate symptoms and burden in family caregivers of pediatric and adult cancer patients who were undergoing a new cancer treatment at the National Institutes of Health (NIH) Clinical Center. This analysis was part of a longitudinal study examining the feasibility of an electronic patient-reported outcome measurement system to collect and follow symptoms in cancer caregivers.²⁶ Participants were recruited between March 2014 and July 2016. Caregivers were eligible to participate if they (1) were at least 18 years old; (2) were an active caregiver of an individual beginning cancer treatment (±14 days); (3) were able to read and speak English or Spanish; and (4) had internet access and the ability to complete online surveys. A member of the research team first approached the individuals with cancer to introduce the study and to ascertain if they would be supported by informal caregiver during study participation. If supported by a caregiver, permission was sought for a member of the research team to contact the caregiver regarding participation in this study. If the cancer caregiver agreed to participate after providing an overview of the study, they consented and were assigned login information and given instructions for completing the online survey. Care recipients were not enrolled in this study but provided authorization for the study team to access their medical records in order to collect accurate disease and treatment information. Further details regarding study procedures and participant descriptions are published elsewhere.²⁶ This study was approved by the institutional review board of the National Heart, Lung, and Blood Institute at the NIH (NCT01981538).

Conceptual Framework

The analyses were guided by the Theory of Unpleasant Symptoms (TOUS).²⁷ In this theory, physiological, psychological, and situational factors contribute to the development of unpleasant symptoms such as fatigue, sleep disturbances, depression, anxiety, and impaired cognition. According to this theory, problematic symptoms can occur alone, or, as is the case in symptom clusters, the symptoms can occur simultaneously, interacting and influencing each other. This can then lead to negative long-term performance outcomes such as comorbid chronic illnesses or decreased health-related quality of life. Based on previous research and the TOUS, physiological factors considered in this study as possibly contributing to symptom clusters included age, sex, race/ethnicity, and presence of chronic health problems. Psychological factors included self-efficacy and loneliness. Possible situational factors included marital status, socioeconomic status (eg, education, household income), frequency of participation in health-promoting behaviors, caregiving characteristics such as caregiver burden, the type of relationship between the caregiver and care recipient, and patient characteristics (eg, patient age, patient sex, primary disease). These influencing factors that may trigger or shape the development of symptom clusters are reflected in the conceptual model (Figure 1).

Figure 1.

The theory of unpleasant symptoms* adapted for the analysis of symptom clusters in family caregivers of cancer. ^aSocioeconomic status = education, annual household income, work status. ^bCaregiver/patient relationship = relationships with patient, caregiver role, double-duty caregiver. *Adapted from the Theory of Unpleasant Symptoms (Lenz, Pugh, Milligan, Gift, and Suppe, 1997).

Measures

GENERAL CHARACTERISTICS

Demographic characteristics included age, sex, race/ethnicity, marital status, educational level, annual household income, work status, and presence of chronic health problems. Caregiving characteristics included the relationship with patient (spouse, parent, others), caregiver role (sole caregiver vs part of a caregiving team), and whether a caregiver was a double-duty caregiver, which for the purposes of this study was defined as providing care for an individual other than the individual with cancer. Patient characteristics included patient age, sex, pediatric versus adult, primary disease, cancer treatment type, and hospital status.

CAREGIVING BURDEN

Caregiving burden was measured using the Caregiver Reaction Assessment (CRA),²⁷ using a 5-point Likert-type response option from 1 (strongly disagree) to 5 (strongly agree). The measure’s subscales included the extent to which caregiving enhanced their self-esteem (7 items), caregivers perceived lack of family support (5 items), caregiving imposed financial strain (3 items), caregiving disrupted daily schedule (5 items), and caregiving impacted their health (5 items) was assessed. Each subscale score was computed by taking the mean of the scale items, with higher scores suggesting more burden, except caregiver esteem, where a high score denotes a low burden. This measure has been shown to have validity and reliability among cancer caregivers.²⁵ The subscales demonstrated acceptable internal consistency (Cronbach’s α for caregiver esteem = .75, lack of family support = .80, impact on finance = .87, impact on schedule = .80, and impact on health = .70).

HEALTH BEHAVIORS

Health behaviors were assessed using the 52-item Health-Promoting Lifestyle Profile II that measures the frequency with which individuals participate in 5 health-promoting behaviors: physical activity, nutrition, stress reduction, interpersonal relationships, spirituality, and health responsibility.²⁸ Individual items were scored using a 4-point Likert scale from 1 (never) to 4 (routinely), with higher scores indicating more frequent participation in health-promoting behaviors. This measure has been shown to have good reliability in cancer caregivers.²⁹ In this study, the internal consistency of the instrument was acceptable (Cronbach’s α for instrument = .94).

PROMIS, QUALITY OF LIFE IN NEUROLOGICAL DISORDERS, AND NIH TOOLBOX MEASURES

The Patient-Reported Outcomes Measurement Information System (PROMIS), Quality of Life in Neurological Disorders (Neuro-QOL), and NIH toolbox measures provide access health concepts including self-report symptoms, with strong evidence of validity and reliability.³⁰ PROMIS measures of 4 symptoms (fatigue, sleep disturbance, depression, and anxiety), Neuro-QOL measure of cognition, and NIH toolbox measures of self-efficacy and loneliness were used in this study. Items were scored on a 5-point Likert-type scale from 1 (never) to 5 (always). Higher scores indicate higher levels of the concept being measured, except cognition, where a high score denotes lower level of cognitive impairment. The measures were administered by computerized adaptive testing, designed to vary the number of items asked based on the responses provided. Typically, 4 to 7 items per concept are administered by computerized adaptive testing. The measures are standardized to a T score metric, with a mean of 50 and SD of 10 that is centered on the general US population. The measures offer validated cutoff scores for diagnosing clinical symptoms (ie, fatigue, sleep disturbance, depression, anxiety: within normal limits <55, mild = 55–60, moderate = 60–70, severe >70, cognitive impairment: within normal limits >45, mild=45–50, moderate=30–40, severe <30).^30,31

Statistical Analysis

Cluster analysis was performed to identify subgroups of caregivers based on their responses to the 5 symptoms (fatigue, sleep disturbance, depression, anxiety, cognition impairment), using Stata version 13.0.³² An agglomerative, hierarchical cluster analysis was performed with squared Euclidean distance used in the proximities matrix and Ward’s method used as the clustering method. To determine the number of clusters, the Calinski and Harabasz pseudo-F stopping rule index and the Duda and Hart Je(2)/Je(1) index were used jointly³²; a large Calinski and Harabaz pseudo-F statistic, combined with 2 measures from Duda and Hart [ie, a large Je(2)/Je(1) index and its associated small pseudo-T squared value], resulted in 2 as the most appropriate number of clusters for the data.

Once clusters were identified, logistic regression was used to determine what factors predicted symptom cluster membership. Congruent with the TOUS and the literature, physiological factors (age, sex, race/ethnicity, chronic health problems), psychological factors (self-efficacy, loneliness), and situational factors including marital status, education, annual household income, work status, relationship with patient, caregiving role, double-duty caregiver, caregiving burden, and health behaviors; and patient factors (patient age, patient sex, patient type, patient primary disease, cancer treatment type, hospital status) were included in the bivariate analyses (Figure 1). Factors related to class membership at P < .10 were included in the multiple logistic regression model using backward stepwise methods. A priori known factors caregiver’s age and sex were controlled in the model. All the statistical analyses were performed with IBM SPSS software package version 21.0.³³

Results

Three-hundred fifty-one subjects were screened for participation, and 139 enrolled. The baseline survey was completed by 129 caregivers of cancer. Demographic and descriptive characteristics of the study participants and mean symptom scores for fatigue, sleep disturbance, depression, anxiety, and cognition are shown in Table 1. Based on previously established thresholds,^30,31 approximately 30% to 80% of the caregivers in this study were considered “symptomatic,” indicating that they reported at least mild levels of 1 or more of the following symptoms: fatigue (n = 40, 31.0%), sleep disturbance (n = 55, 42.6%), depression (n = 42, 32.6%), anxiety (n = 83, 64.3%), and cognitive impairment (n = 101, 78.3%).

Table 1.

General Characteristics (N = 129)

Variables	Category	n (%) Mean (SD), Range
Caregiver Characteristics
Age y		48.58 (11.76), 20–76
Sex	Female	87 (67.4)
Race/ethnicity	White/non-Hispanic	91 (71.1)
	Non-White/non-Hispanic	18 (14.1)
	Hispanic	19 (14.8)
Marital status	Married/cohabiting	107 (83.6)
	Not marrieda	21 (16.4)
Education	≤Bachelor	85 (65.9)
	≥Postgraduate	44 (34.1)
Annual household income	<$50 000	35 (29.2)
	$50 000-$89 000	28 (23.3)
	>$89 000	57 (47.5)
Work status	Full-time	74 (57.4)
	Part-time	21 (16.3)
	Not workingb	34 (26.3)
Chronic health problems	Yes	65 (50.8)
	No	63 (49.2)
Relationships with patient	Spouse	64 (49.6)
	Parent	45 (34.9)
	Others	20 (15.5)
Caregiver role	Team member	59 (45.7)
	Sole	70 (54.3)
Double-duty caregiverc	Yes	43 (33.6)
Caregiving burdend	Caregiver esteem	4.37 (0.51), 2.57–5.00
	Lack of family support	1.86 (0.75), 1.00–4.40
	Impact on finance	2.81 (1.15), 1.00–5.00
	Impact on schedule	3.45 (0.86), 1.00–5.00
	Impact on health	2.17 (0.75), 1.00–5.00
Self-efficacye		31.30 (5.80), 10–40
Lonelinesse		10.18 (4.28), 5–25
Health behaviorsf		2.63 (0.46), 1.36–3.75
Symptomse	Fatigue	51.04 (7.92), 26.11–81.59
	Sleep disturbance	54.03 (7.92), 26.35–83.79
	Depression	52.47 (7.70), 34.17–81.83
	Anxiety	57.38 (6.71), 26.11–81.59
	Cognition	40.92 (6.72), 21.11–62.32
Patient Characteristics (n = 111)
Patient age, y		41.6 (18.6), 4–76
Patient sex	Male	61 (55.0)
Patient type	Adult	84 (75.7)
	Pediatric	27 (24.3)
Patient primary diseaseg	Carcinoma	52 (46.9)
	Leukemia	25 (22.5)
	Sarcoma	23 (20.7)
	Lymphoma	10 (9.0)
	Myeloma	1 (0.9)
Cancer treatment type	Biotherapy/immunotherapy	83 (64.3)
	Allogeneic HSCT	12 (9.3)
	Chemotherapy	11 (8.5)
	Surgery	11 (8.5)
	Otherh	12 (9.3)
Hospital status	Inpatient	107 (83.6)

Abbreviations: HSCT, hematopoietic stem cell transplantation; Neuro-QOL, Quality of Life in Neurological Disorders; NIH, National Institutes of Health; PROMIS, Patient-Reported Outcomes Measurement Information System. Numbers may not sum to total because of missing data.

^aNot married = never married, divorced, separated, widowed.

^bNot working = student, retired, disability, unemployed.

^cDouble-duty caregivers are those who provide care to one or more individuals (eg, child, parent, spouse, friend) other than the cancer patient.

^dCaregiver burden was measured using the Caregiver Reaction Assessment.

^eSelf-efficacy, loneliness, and symptoms were measured using PROMIS, Neuro-QOL, and NIH toolbox.

^fHealth behaviors was measured using Health-Promoting Lifestyle Profile II.

^gCarcinoma = prostate, melanoma, anal, breast, lung, colon, liver, cervical, ovarian, adrenal cortical, pancreatic, kidney, thyroid, and peritoneal cancer; leukemia = chronic myelogenous leukemia, acute lymphocytic leukemia, acute myelogenous leukemia, and chronic lymphocytic leukemia; sarcoma = brain, gastrointestinal stromal tumor and desmoid tumors; lymphoma = Hodgkin and non-Hodgkin lymphoma; myeloma = multiple myeloma.

^hOther = radiation therapy (n = 5) and combination therapy (n = 8).

Identification of Symptom Clusters

As shown in Figure 2, 2 distinct symptom cluster groups were identified, cluster 1 and cluster 2, which divided the caregivers into subgroups based on the severity of their symptoms. Cluster 1 (n = 106, 82.2%), labeled “low symptom burden,” was characterized by mild levels of anxiety and cognitive impairment, whereas the other 3 symptoms (fatigue, sleep disturbance, depression) were within normal limits. Cluster 2 (n = 23, 17.8%), labeled “high symptom burden,” was characterized by at least moderate levels of all 5 symptoms (fatigue, sleep disturbance, depression, anxiety, cognition impairment) (Table 2, Figure 2).

Figure 2.

Difference in symptom distress scores among the symptom cluster groups. Cluster 1 (low symptom burden), Cluster 2 (high symptom burden). The dashed line indicates the fatigue, sleep disturbance, depression, and anxiety thresholds. The solid line indicates the impaired cognition threshold.

Table 2.

Differences in Symptom t Scores Among the Symptom Cluster Groups (N = 129)

	Mean (SD)
Variables	Cluster 1 (n = 106, 82.2%) Low Symptom Burden	Cluster 2 (n = 23, 17.8%) High Symptom Burden	P
Fatiguea	48.47 (5.56)	62.85 (6.30)	<.001
Sleep disturbancea	52.30 (6.84)	61.99 (7.79)	<.001
Depressiona	50.07 (5.49)	63.53 (6.78)	<.001
Anxietya	55.49 (4.82)	66.08 (7.43)	<.001
Impaired cognitionb	42.69 (5.36)	32.76 (6.39)	<.001

Abbreviations: Neuro-QOL, Quality of Life in Neurological Disorders; PROMIS, Patient-Reported Outcomes Measurement Information System. Numbers may not sum to total because of missing data.

^aFatigue, sleep disturbance, depression, and anxiety were measured using PROMIS.

^bCognition was measured using Neuro-QOL.

Factors Associated With Symptom Cluster Groups

Six factors (CRA impact on finance, CRA impact on schedule, CRA impact on health, health behaviors, self-efficacy, and loneliness) met the bivariate analyses criteria and were included in the multiple logistic regression model predicting symptom cluster membership. Table 3 displays final multiple logistic regression results with predictors using cluster 1 (low symptom burden) as the reference. Controlling for caregiver’s age and sex, caregivers who reported higher CRA subscale scores of impact on health were 3.7 times more likely to be in the high symptom burden group (95% confidence interval, 1.49–9.18). Caregivers who reported higher levels of loneliness also were more likely to be in the high symptom burden group (odds ratio, 1.20; 95% confidence interval, 1.02–1.40).

Table 3.

Final Multiple Logistic Regression Model

Variables	B (SE)	OR (95% CI)	P
Age (years)	−0.02 (0.02)	0.98 (0.93, 1.02)	.334
Femalea	0.78 (0.72)	2.17 (0.54, 8.81)	.278
Caregiving burden (Impact on health)b	1.31 (0.46)	3.70 (1.49, 9.18)	.005c
Lonelinessd	0.18 (0.08)	1.20 (1.02, 1.40)	.024c

Abbreviations: CI, confidence interval; NIH, National Institutes of Health; OR, odds ratio.

Controlling for caregiver’s age and sex.

^aMale was used as a reference group.

^bCaregiver burden was measured using the Caregiver Reaction Assessment.

^cP < .05.

^dLoneliness was measured using NIH toolbox.

Discussion

This is the first known study that identified symptom clusters and the factors that influenced symptom clusters in cancer caregivers. Our findings provide evidence that supports the premise of the TOUS that multiple symptoms are experienced simultaneously and that these symptoms are influenced by physiological, psychological, and situational factors.²⁷ In this study, 2 symptom cluster groups were identified among family caregivers of individuals with cancer: cluster 1 (low symptom burden) and cluster 2 (high symptom burden). Fortunately, most of the caregivers in this study (82.2%) were in cluster 1, exhibiting low symptom burden. This is consistent with one previous study, which found 2 subgroup trajectories in the low symptom burden group (60%) and high symptom burden group (40%) in cancer patient–caregiver dyad.¹⁹ Our results are different from that of Illi et al,¹⁷ who found 3 symptom cluster groups in a combined sample of cancer patients and their caregivers, rather than the 2 groups found in these analyses. Interestingly, caregivers in the low symptom burden group reported mild levels of cognitive impairment as well as anxiety. Most of the existing studies have focused on pain, fatigue, sleep disturbance, depression, and anxiety, and little attention has been given to cognitive impairment in cancer caregivers. In this population, cognitive impairment was one of the most problematic symptoms reported by the caregivers, with the large majority reporting levels of cognitive impairment that could be considered symptomatic. Given that caregivers are often involved in important therapeutic decisions made for the care recipient, these results suggest a need for oncology nurses to be alert for potential cognitive impairment in cancer caregivers.

Our analysis provides further evidence regarding risk factors associated with high levels of caregiver burden in cancer caregivers. In this study, caregivers experiencing greater caregiving burden, particularly those caregivers who reported that the burden of caregiving had a greater impact on their health, were also more likely to fall into the high symptom burden group. The CRA impact on health subscale measures the caregivers’ perception that his/her health has suffered as a result of the obligations of caregiving. This finding is consistent with previous studies.^16,25 According to Karabekiroğlu et al,¹⁶ caregivers’ burden level measured by Zarit Caregiver Burden Inventory was found to be a predictor of depression and anxiety among caregivers of hospitalized advanced cancer patients. Stenberg et al.²⁵ also reported significant relationships between CRA impact on health and individual symptoms, including sleep disturbance and depression, respectively. This may be explained by the fact that caregivers are likely to prioritize the needs of the patient over their own and may modify their lifestyle to accommodate the patient’s needs, including limiting time for exercising, maintaining good nutrition, and engaging in needed health evaluations.⁶ Indeed, caregivers are least likely to engage in health-promoting activities that require time away from the patient, such as exercising and participating in stress reduction activities.^34,35 It is understandable that caregivers place the needs of the patient over their own, but the unfortunate result may be that caregiving responsibilities can increase symptom burden and ultimately contribute to the declining health of the caregiver.

The most notable finding of this study was that loneliness, a negative and unpleasant feeling of disconnectedness or social isolation,³⁶ was reported as a significant risk factor for symptom clusters among cancer caregivers. In this study, lonelier cancer caregivers experienced more concurrent fatigue, sleep disturbance, depression, anxiety, and cognitive impairment than their less lonely counterparts. This is consistent with the previous studies in caregivers of individuals with Alzheimer’s disease,¹⁸ who found that loneliness was a common risk factor for pain, depression, and fatigue symptom clusters. This is possibly because of the fear of leaving the care recipient alone or the time demands associated with cancer caregiving, both of which can hinder caregivers from engaging in their regular social interactions.^6,37 In instances where the care recipient is a spouse or close friend, the caregiver may be reluctant to burden the care recipient with their own problems and concerns, and this might further exacerbate feelings of isolation.⁶ While the psychosocial challenges of cancer caregiving have been extensively researched, and loneliness in family caregivers of the elderly and/or individuals with dementia has received some research attention,³⁷ surprisingly little research has examined loneliness and its sequela in cancer caregivers.

The strengths of this study included psychometrically sound and standardized measurements, PROMIS,³⁰ and a person-centered analysis approach, cluster analysis. However, future research is needed using a mixed-methods approach that incorporates qualitative as well as quantitative methods. Adding a qualitative approach to understanding the symptom experience in cancer caregivers might yield novel, more in-depth information that could help researchers and clinicians better understand the symptom experience and guide future research and clinical practice. Future studies of symptom clusters would benefit from integrating biological markers (eg, cytokines, genomic DNA) into caregiving research. Such research would contribute to continued understanding of the mechanisms underlying the effects of caregiving on the symptom experience in caregivers of individuals with cancer. For example, the mechanisms explaining why and how loneliness puts individuals at higher risk of developing symptoms are not well understood. Expanding the scientific study of biological markers will contribute to our understanding of the physiological pathways through which loneliness may influence caregivers’ symptoms.

Limitations

Some limitations of this study should be acknowledged. Because this analysis is cross-sectional, we did not explore symptom cluster membership over time. A longitudinal design could further be conducted to identify the onset and rate of change in symptoms over time and how symptom clusters change at different phases of the caregiving trajectory. The cross-sectional nature of the current study also did not disentangle causality between symptom clusters and risk factors. For example, loneliness may worsen the symptom cluster, which then exacerbates levels of loneliness. Further research is needed to examine whether the process is unidirectional or cyclical. Second, the sample size for this study was relatively small, and that may have contributed to our inability to find a relationship between certain situational factors such as patient treatment type or double-duty caregiving with symptom cluster membership. In addition, the current study was limited by lack of a comparison group and heterogeneous sample in terms of patient primary disease or cancer treatment type. Further studies using large, homogeneous samples and those using comparison groups are needed to confirm the subgroup findings of the present study. Finally, the caregivers in this study mostly were White, relatively young, and well-educated. Furthermore, this study recruited only caregivers of individuals receiving cancer treatment at the NIH Clinical Center, a unique research hospital that provides care only to individuals enrolled in research protocols. Thus, the findings may not be generalizable to caregivers who are old, poorly educated, or part of a minority population, as well as of patients receiving more traditional cancer care in general hospitals or clinics.

Clinical Implications

The results of this study have implications for clinical practice. The approach focusing on the existence of symptom clusters provides important insights that may lead to the development and testing of symptom management interventions that target multiple troublesome symptoms within a cluster simultaneously. It is possible that the direct treatment of one symptom may indirectly impact another symptom in the cluster. This has implications for oncology nurses. For example, strategies to improve sleep by improving pain management may contribute to decreasing fatigue and depression. Additionally, oncology nurses can explain to caregivers that the symptoms that they experience may not occur in isolation. For instance, not getting adequate sleep not only will contribute to increased fatigue, but also may lead to cognitive changes such as short-term memory loss or exacerbate pain-related conditions. Such education could facilitate better self-management and, as a result, ultimately could enable them to better meet the caregiving needs of the care recipient.

Clinicians might advocate for the incorporation of caregiver health management and support programs in cancer care services to decrease the negative health impact on the caregivers. At the very least, clinicians should give caregivers permission to step away from the bedside in order to participate in health-promoting self-care activities. Clinicians also might emphasize that healthy caregivers are critical to the health and well-being of the individual with cancer.³ In addition, this study found that caregivers with high levels of loneliness appeared to be more likely to be in the high symptom burden group. Thus, healthcare providers might develop and implement interventions aiming at increasing opportunities for social contact, enhancing social support, and addressing maladaptive interpersonal interactions.^36,37 These strategies could help to relieve levels of loneliness and, as a result, might reduce symptoms such as depression, anxiety, cognitive impairment, and sleep disturbances. Additionally, this study suggested that health policymakers might elevate the discussion of caregiver loneliness to the policy level and could use such information to facilitate the best support for cancer caregivers.

Conclusion

This study used a cluster analysis approach to identify subgroups of cancer caregivers based on 5 key symptoms, including fatigue, sleep disturbance, depression, anxiety, and impaired cognition. Two symptom cluster groups were identified: cluster 1 (low symptom burden) and cluster 2 (high symptom burden). This study adds to the knowledge regarding the relationships between concurrent multiple symptoms among family caregivers of individuals with cancer. Most importantly, factors that contribute to symptom cluster group membership were identified, including CRA impact on health and loneliness. The findings in this study may guide the development of more effective and tailored interventions that target multiple co-occurring symptoms.