Abstract
Background
Little is known about how unpaid family caregivers may already be engaged in caregiving activities prior to their care recipient’s cancer diagnosis. We examined pre-cancer diagnosis caregiving patterns and their association with caregiving strain.
Methods
We conducted a population-based analysis of 2011–2017 National Health and Aging Trends Study (NHATS) linked with National Study of Caregiving (NSOC) and Medicare claims data. Latent class analysis was used to examine patterns of 16 health-focused caregiving tasks (e.g., tracking medications, making appointments) of family caregivers assisting adults ≥65 years prior to an incident cancer diagnosis. High caregiving strain was defined as a total score ≥85th percentile of 6 caregiving strain items (e.g., financial difficulty, no time for self). Association between caregiving patterns and strain were examined using multivariable logistic regression, adjusting for care recipient and caregiver characteristics.
Results
An estimated 4.2 million caregivers cared for older adults prior to care recipients’ new cancer diagnoses during 2011–2017. They engaged in a median of four health-focused caregiving activities. Nearly 1-in-5 (18.7%) pre-cancer caregivers had high caregiving strain. Caregivers were classified into 3 health-focused caregiving activity classes: Low-level (41.2%), Moderate-coordination (29.3%), and High-intensity (29.4%). Higher caregiving activity was associated with higher caregiving strain (adjusted odds ratio (aOR)=3.85, 95% CI: 2.34–6.33). Caregivers in the High-intensity class had the highest caregiving strain (39.9%), and include more spouses (28.1% vs <18%).
Conclusion
One-third of U.S. caregivers who help older adults prior to their cancer diagnoses are already highly strained and engaged in high-level health-focused caregiving tasks. Oncology clinicians should assess the capacity and strain of family caregivers who may already be supporting patients with new cancer diagnoses and refer caregivers to additional supportive care services.
Keywords: cancer, unpaid caregivers, caregiving strain, older adults, supportive care, latent class analysis
1. Introduction
By 2030, 22.1 million Americans are projected to be living with cancer, and most will be ages 65 years and older.[1] Family and close friends often play an indispensable frontline role in providing unpaid support to older individuals with cancer, up to an average of over 8 hours per day.[2] These family caregivers perform a wide range of medical and health-related tasks, such as transportation to appointments, coordinating care, managing medications and symptoms, and communicating and advocating on behalf of patients.[2–4] A growing national spotlight has been focused on the caregiving role over the past decade as concerns grow that caregivers are overly strained in providing this support.[5–7] National studies on informal unpaid caregivers find that nearly 20% (range: 11.7%−34.4%) of caregivers are in fair or poor health.[4] One in five caregivers caring for adults 50 years and older reported that caregiving worsened their own health.[3] In addition, caregivers are increasingly experiencing difficulties as they navigate complex and fragmented healthcare systems (31% in 2020).[8] Mounting literature has also highlighted caregiving strain, [2,5,9–17] including a recent meta-analysis of over 21,000 cancer caregivers with estimated prevalence of depression and anxiety to be 42% and 47%, respectively.[18]
Despite this growing evidence-base documenting caregiving strain after their care recipient’s cancer diagnosis, little is known about the strain among caregivers for older adults before their cancer diagnoses. Using a nationally representative sample of U.S. older adults, we previously found that 84% of older adults received help from family and other unpaid caregivers prior to being diagnosed with lung, breast, prostate, or colorectal cancer.[19] In the current study, we aimed to examine the pre-cancer diagnosis caregiving patterns of health-focused caregiving tasks, and to explore the association between caregiving activity patterns and caregiving strain. Given the shift of cancer care delivery to the home and the reliance on family caregivers to provide high quality complex care,[3,5,20,21] it is imperative to understand both strain and situational complexities to provide a comprehensive supportive care plan that complements the patient’s cancer treatment plan.
2. Methods
2.1. Data sources and study population
We used the 2011–2017 National Health and Aging Trends Study (NHATS) and the linked National Study of Caregiving (NSOC) and Medicare claims database (2010–2018) to identify family and other unpaid caregivers who helped older adults prior to their cancer diagnoses. Since 2011, NHATS has been collecting longitudinal data through annual in-person interviews to monitor patterns of physical and cognitive capacity in later life based on a nationally representative sample of Medicare recipients age 65 and over.[22] In conjunction with NHATS, caregivers of NHATS respondents were interviewed using the NSOC protocol by telephone at periodic intervals (2011, 2015, and 2017) to provide a nationally representative study of family and other unpaid caregivers.[23] Multiple caregivers (up to 5) may be interviewed per NHATS respondent. The response rates ranged from 59.7% to 61.9% in NSOC and from 71.3% to 94.0% in NHATS.[22,23] This secondary data analysis was approved by the Institutional Review Board of the Icahn School of Medicine at Mount Sinai.
We used the caregiver’s NSOC interview date as the index date to identify their care recipients who had a cancer diagnosis within one year after the index date but no cancer diagnoses one-year prior to the index date. Care recipients’ new cancer diagnoses were identified using the International Classification of Diseases codes indicative of neoplasms (ICD-9: 140–239 and ICD-10: C00-D19, D30-D49) from all available diagnoses records in Medicare claims. We further limited our sample of care recipients to those who had a continuous Medicare Part A and B coverage but without Medicare Advantage one year prior to and one year post the caregiver’s index date, to ensure complete Medicare claims during the observation window. Out of 6572 caregivers from NSOC, we identified 444 caregivers and their corresponding 302 care recipients from NHATS that met our selection criteria. After excluding seven participants who lacked information on caregiving strain measures, the final analytical sample included 437 caregivers and their corresponding 300 care recipients.
2.2. Measures
Health-focused caregiving activities:
Because of our interest in the compounding of caregiver health-focused caregiving tasks associated with their care recipient undergoing a new cancer treatment plan, we focused on sixteen health-focused caregiving tasks. Informed by prior studies,[24–27] we grouped these tasks into four activity domains: health management (i.e., care for a special diet, exercises, teeth, feet, and skin care wounds included), healthcare tasks (i.e., keep track of medications, take shots or injections, manage medical tasks, or help order prescribed medicines), healthcare system logistics (e.g., help make appointments for care recipients with a medical provider, change or add a health insurance or prescription drug plan), and patient advocacy (i.e., speak to medical providers about care recipient care, and sit in with care recipients during their visits to the doctors).
Caregiving strain:
Consistent with prior approaches,[28] we computed caregiving strain by summing responses to three items regarding caregivers’ emotional, physical, and financial difficulty of helping and the following three items: “exhausted when go to bed at night”, “have more things to do than can handle”, and “don’t have time for self” (score range 0–9; higher scores=higher strain). “High” caregiving strain was defined as scoring ≥85th percentile (≥5).[28]
Other covariates:
We included care recipients dementia status, as it is well established that caring for persons with dementia contribute to caregiving strain.[25,28–30] Dementia status in NHATS is classified as probable, possible, and no dementia based on a combination of self-reported diagnosis, proxy responses to the AD8 Dementia Screening interview, and cognitive testing.[31] In the current study, we defined those with probable dementia as dementia while possible/no dementia as no dementia. We also included sociodemographic (e.g., age, sex, race/ethnicity, education, income) and health status variables (e.g., number of common chronic conditions treated as a continuous variable, probable anxiety and depression, self-rated genera health) related to caregiving strain.[19,24,27] To understand the profiles of caregivers, we also included variables about caregiving experience, such as caregiving schedule and time, and the relationship between care recipient and caregivers.
2.3. Statistical analysis
The overall descriptive pattern of caregivers and their care recipients were compared by caregiving strain status using Rao-Scott Chi-square tests for categorical variables and F tests in linear regression for continuous variables.
We used latent class analysis (LCA), a finite mixture model-based clustering approach, to identify distinct latent (“unobserved”) patterns of caregiving activities based on caregiver’s responses (“observed”) to the 16 specific health-focused caregiving tasks, without survey weight. From the distributions of item responses, LCA estimated the probabilities of latent class membership in the study population and the item-response probabilities, which are class-specific likelihoods of individuals’ item endorsement.[32] We tested models with 2–10 latent classes. The number of optimal classes were selected based on commonly used measures (i.e., the Akaike’s Information Criterion, the Bayesian Information Criteria, the adjusted Bayesian Information Criteria, entropy, as well as percentage of seeds associated with best-fitted model and class interpretability).[32,33] Once the optimal latent classes were determined, we compared sample characteristics by the identified latent classes.
We used logistic regression to examine the association between caregiving activity pattern and high caregiving strain, adjusting for care recipient’s dementia status and race/ethnicity, as well as relevant caregiver-level characteristics. The caregiving activity pattern was treated as a continuous variable in the main model, as exploratory analyses indicated a dose-response relationship (data not shown). We also conducted several sensitivity analyses to test the robustness of the main model, including treating the caregiving activity pattern as a categorical variable, and using alternative indicators of caregiving activity pattern (i.e., using the summed numbers of health-focused care tasks) and caregiving strain (i.e., using summed scores of caregiving strain), treated as continuous variables.
Unless otherwise specified, all analyses took into account the NSOC survey design by incorporating strata, cluster, and sampling weights. The analyses were conducted using SAS (v9.4) and R (v4.0.2) with RStudio (v1.3.1073).
3. Results
3.1. Characteristics of pre-cancer care-recipients and caregivers by caregiver’s caregiving strain status
After accounting for the NSOC survey design, we estimated that there are approximately 4.2 million pre-cancer caregivers helping older adults during 2011–2017. About 56% of the care recipients had more than one caregiver (Table 1). The median age of caregivers was 58.4 (interquartile range, IQR 48.1–67.8) years and for care recipients was 79.9 (IQR 73.0–86.4) years. The median number of comorbidities was 1.8 (IQR 0.4–3.6) among caregivers and 3.5 (IQR 2–5.1) for care recipients. The median time between the caregiver’s interview and care recipient’s cancer diagnosis was 3.4 (IQR 1.1–6.3) months.
Table 1.
Characteristics of caregivers and care recipients overall and by caregiving strain status.
| Overall | Non-high caregiving strain | High caregiving straina | ||
|---|---|---|---|---|
| Weighted sample size (weighted %) | 4,177,313 | 3396972 (81.3) | 780341 (18.7) | |
| Unweighted sample size | 437 | 356 | 81 | |
| Variable | Weighted % | Weighted % | Weighted % | p-valueb |
| Care Recipient Characteristics | ||||
| Care recipient sociodemographic | ||||
| Age in yearsc | 79.9 (73–86.4) | 80 (72.7–86.6) | 79.5 (74.6–85.4) | 0.94 |
| Female | 68.8 | 70.3 | 62.4 | 0.13 |
| Non-Hispanic white | 79.9 | 80.8 | 75.9 | 0.06 |
| Non-Hispanic black | 9.8 | * | * | |
| Hispanic | * | * | * | |
| Other race/ethnicity | * | * | * | |
| Above high school Education | 73.8 | 74.3 | 71.8 | 0.62 |
| Married | 35.1 | 32.9 | 44.3 | 0.14 |
| Medicaid | 21.8 | 20.3 | 28.7 | 0.11 |
| Size of social networkc | 1.3 (0.4–2.4) | 1.4 (0.5–2.5) | 0.8 (0–2.1) | 0.04 |
| Care recipient health status | ||||
| Fair/Poor general health | 48.7 | 48.5 | 49.5 | 0.89 |
| Anxiety symptoms | 27.5 | 25.9 | 34.5 | 0.18 |
| Depressive symptoms | 25.7 | 22.2 | 41.1 | 0.007 |
| Need help with ≥1 ADL | 55.3 | 53.8 | 61.9 | 0.32 |
| Need help with ≥ IADL | 68.4 | 68.5 | 68.1 | 0.96 |
| Probable dementia | 26.8 | 22.4 | 45.7 | <.0001 |
| Number of common chronic health conditionsc | 3.5 (2–5.1) | 3.4 (1.9–5) | 4.1 (2.3–5.9) | 0.25 |
| Months to care recipient’s cancer diagnosisc | 3.4 (1.1–6.3) | 3.4 (1.2–6.3) | 3.5 (0.8–6.2) | 0.87 |
| Having more than one caregiver | 55.9 | 57.1 | 55.6 | 0.38 |
| Caregiver Characteristics | ||||
| Caregiver sociodemographic | ||||
| Age in years (Median (IQR))c | 58.4 (48.1–67.8) | 58.5 (47.6–67.7) | 58.3 (50.3–67.7) | 0.83 |
| Female | 60.7 | 58.1 | 72.2 | 0.05 |
| Above high school education | 93.7 | * | * | 0.24 |
| Married/Living with Partner | 69.6 | 69.0 | 71.9 | 0.70 |
| On Medicaid | 14.3 | 13.0 | 20.2 | 0.25 |
| Living with children | 76.8 | 75.4 | 82.3 | 0.32 |
| Currently work for pay | 37.4 | 39.4 | 29.0 | 0.17 |
| Caregiver health status | ||||
| Fair/Poor general health | 19.7 | 15.3 | 38.8 | 0.001 |
| Anxiety symptoms | 11.7 | 9.9 | 19.5 | 0.07 |
| Depressive symptoms | 13.6 | 10.6 | 27.2 | 0.002 |
| Number of common chronic health conditionsc | 1.8 (0.4–3.6) | 1.5 (0.2–3) | 3.6 (1.7–5.4) | <.0001 |
| Caregiving experience | ||||
| Primary caregiver | 62.5 | 59.9 | 73.8 | 0.10 |
| Caregiver also the spouse | 19.9 | 17.0 | 32.7 | 0.02 |
| Co-resides with the care recipient | 43.6 | 40.6 | 57.4 | 0.16 |
| Have friends or family that help with care for you, in last month | 71.3 | 74.2 | 58.8 | 0.013 |
| Help on a regular schedule | 37.5 | 35.3 | 46.8 | 0.123 |
| Duration of caregiving > 1 year | 91.6 | * | * | 0.34 |
| Caregiving hours per weekc | 7(2.5–18.3) | 6.8 (2.3–14.9) | 13.4 (6.9–23.8) | 0.059 |
| Total caregiving strain scores c | 1.5 (0.1–3.2) | 1.1 (0–2.1) | 5.1 (5–6.4) | <.0001 |
| Health-focused caregiving activities | ||||
| Total number of health-focused caregiving tasks c | 3.8 (1–6.7) | 2.8 (0.7–5.6) | 6.9 (4.8–9) | <.0001 |
| Health management | ||||
| Special diet | 25.0 | 21.0 | 42.4 | 0.002 |
| Exercise | 21.5 | 19.8 | 29.1 | 0.164 |
| Foot care | 24.0 | 20.1 | 40.8 | <.0001 |
| Skin care | 18.1 | 13.4 | 38.4 | <.0001 |
| Dental care | 13.9 | 13.4 | 16.2 | 0.56 |
| Healthcare tasks | ||||
| Take shots or injections | 8.1 | * | * | 0.037 |
| Track medication | 41.2 | 34.0 | 72.6 | <.0001 |
| Order medicine | 39.1 | 32.4 | 68.4 | <.0001 |
| Manage medical tasks | 7.7 | 6.3 | 13.6 | 0.080 |
| Health system logistics | ||||
| Handle insurance | 31.3 | 25.1 | 58.1 | <.0001 |
| Make appointments | 48.4 | 43.1 | 71.1 | 0.0003 |
| Get mobility devices | 47.5 | 42.9 | 67.2 | 0.0008 |
| Make health related home improvement | 39.5 | 36.0 | 54.7 | 0.015 |
| Find paid helper | 18.2 | 17.8 | 20.1 | 0.72 |
| Patient advocacy | ||||
| Speak to or email CR’s medical provider | 50.5 | 42.8 | 83.9 | <.0001 |
| Sit in with CR at doctor’s appointment | 38.4 | 34.5 | 55.5 | 0.007 |
Note: The analyses took into account the National Study of Caregiving (NSOC) survey design.
Caregiving strain was based on six items: emotionally, physically, or financially difficult to help, exhaust at night, care more than one can handle, and no time for self. High caregiving strains was defined using the cutpoint of 5, corresponding to ≥85th percentile in this study.
Comparisons of variables by high caregiving strain status were conducted using Rao-Scott Chi-square tests for categorical variables, and F-test in linear regression for continuous variables.
Median (interquartile range, IQR) are shown for continuous variables. ADL=Activities of Daily Living, IADL=Instrumental Activities of Daily Living, CR=care recipient.
not reportable due to cell size suppression policy to protect the confidentiality of patients.
Approximately 18.7% of the caregivers had high caregiving strain (Table 1). Those with high strain were more than twice as likely as those without high strain to care for older adults with dementia (45.7% vs 22.7, p<−.0001), and had depressive symptoms (41.1% vs 22.2%, p=0.007) (Table 1). Compared to those who did not experience high strain, caregivers with high strain were more likely to have fair or poor general health (38.8% vs 15.3%, p=0.001), depressive symptoms (27.2% vs 10.6%, p=0.002), and more chronic health conditions (median 3.6 vs 1.5, p<0.0001). In addition, they were more likely to be the spouse of the care recipients (32.7% vs 17.0%, p=0.02) and receive less help with care from friends or family (58.8% vs 74.2%, p=0.013).
3.2. Health-focused caregiving activities by caregiving strain status
Caregivers with high caregiving strain had a median caregiving strain score of 5.1 (IQR 5–6.4; Table 1), while those without high caregiving strain had a median score of 1.1 (IQR 0–2.1; Table 1) out of a possible score range of 0 to 9. Caregivers with (vs without) high caregiving strain were more involved in healthcare-related caregiving tasks, particularly in activities such as tracking (72.6% vs 34.0%, p<0.0001) and ordering (68.4% vs 32.4%, p<0.0001) medication; handling insurance (58.1% vs 25.1%, p<0.0001); communicating with medical provider (83.9% vs 42.8%, p<0.0001); as well as assisting with foot (40.8% vs 20.1%, p<0.0001) and skin care (38.4% vs 13.4%, p<0.0001). They were also involved in a higher number of health-focused caregiving tasks (median 6.9 vs 2.8 tasks, p<0.0001, Table 1).
3.3. Latent classes of health-focused caregiving activities
Caregivers were classified into 3 health-focused caregiving activity classes: Low-level (41.2%), Moderate-coordination (29.3%), and High-intensity (29.4%), based on their class-specific probability of partaking in health-focused caregiving activities (Table 2). For example, the probability of caregivers assisting with any of the 16 health-focused caregiving activities was below 33% for those in the Low-level activity class, while those in the High-intensity activity class had greater than 50% probability of performing 11 out of 16 activities. Those in the Moderate-coordination class tended to have high propensity of engaging in health system logistics activities, such as making appointments (probability 99%); advocating for patients such as sit in with care recipient at doctor’s appointment (80%) and communicating with care recipient’s provider (62%); as well as engaging in healthcare tasks, such as tracking and ordering medication (53–54%, Table 2).
Table 2.
Health-focused caregiving activities were classified into Low-level, Moderate-coordination, and High-intensity activity latent classes.
| Low-level | Moderate-coordination | High-intensity | |
|---|---|---|---|
| % of caregivers in each of the latent class | 41.2% | 29.3% | 29.4% |
| Health management | Class-specific probability | ||
| Special diet | 0.09 | 0.17 | 0.52 |
| Exercise | 0.09 | 0.07 | 0.59 |
| Foot care | 0.12 | 0.18 | 0.58 |
| Skin care | 0.09 | 0.16 | 0.46 |
| Dental care | 0.07 | 0.02 | 0.42 |
| Healthcare tasks | |||
| Take shots or injections | 0.02 | 0.03 | 0.15 |
| Track medication | 0.15 | 0.53 | 0.95 |
| Order medicine | 0.12 | 0.54 | 0.89 |
| Manage medical tasks | 0.02 | 0.03 | 0.24 |
| Health system logistics | |||
| Handle insurance | 0.06 | 0.38 | 0.64 |
| Make appointments | 0.07 | 0.99 | 0.88 |
| Get mobility devices | 0.33 | 0.41 | 0.68 |
| Make health related home improvement | 0.28 | 0.37 | 0.74 |
| Find paid helper | 0.13 | 0.14 | 0.37 |
| Patient advocacy | |||
| Sit in with care recipient at doctor’s appointment | 0.15 | 0.80 | 0.87 |
| Speak to or email care recipient’s medical provider | 0.19 | 0.62 | 0.64 |
3.4. Characteristics of pre-cancer care-recipients and caregivers by latent classes of health-focused caregiving activities
Characteristics of caregivers and their care recipients differed across the three health-focused caregiving activity latent classes (Table 3). Care recipients with probable dementia were 12.6%, 40.9%, and 41.1% in the Low-level, Moderate-coordination, and High-intensity caregiving activity classes, respectively. Compared to those in the Low-level activity class, caregivers in the High-intensity and Moderate-coordination activity classes were more likely to be female (>70% vs 49.6%, p=0.001), and the primary caregiver (>74% vs 49.4%, p<0.0001). Highly engaged caregivers, compared to those in the Moderate and Low classes, also worked longer hours (median 15.9 vs <7 hours per week). Approximately 50.8% of caregivers in the Moderately-coordination activity class were daughters, stepdaughters, or daughters-in-law of the care recipients, compared to 20.4% and 37.6% found in the Low-level and High-intensity activity groups (Figure 1).
Table 3.
Characteristics of caregivers and care recipients by health-focused caregiving activity latent classes.
| Low-level | Moderate-coordination | High-intensity | ||
|---|---|---|---|---|
| Weighted sample size (weighted %) | 2,093,553 (50.1) | 1,054,541 (25.2) | 1,029,218 (24.6) | |
| Weighted% | Weighted% | Weighted% | p-valuea | |
| Care Recipient Characteristics | ||||
| Care recipient sociodemographic | ||||
| Age in yearsb | 79.2 (71.9–84.6) | 83.2 (76–89.6) | 79.6 (71.5–86.8) | 0.0003 |
| Female | 68.5 | 74.8 | 63.3 | 0.27 |
| Non-Hispanic white | 85.2 | 80.2 | 68.5 | |
| Non-Hispanic black | * | * | * | |
| Hispanic | * | * | * | |
| Other race/ethnicity | * | * | * | |
| Above high school education | 78.4 | 75.5 | 62.3 | 0.03 |
| Married | 40.5 | 23.0 | 36.3 | 0.02 |
| Medicaid | 19.0 | 23.0 | 26.1 | 0.55 |
| Size of social networkb | 1.4 (0.7–2.4) | 1 (0–2.1) | 1.1 (0.2–3) | 0.18 |
| Care recipient health status | ||||
| Fair/Poor general health | 49.6 | 50.1 | 45.3 | 0.80 |
| Anxiety symptoms | 26.9 | 30.2 | 25.9 | 0.81 |
| Depressive symptoms | 22.0 | 25.1 | 33.8 | 0.167 |
| Need help with ≥1 ADL | 49.2 | 50.6 | 72.3 | 0.01 |
| Need help with ≥ IADL | 62.7 | 69.7 | 78.7 | 0.087 |
| Probable dementia | 12.6 | 40.9 | 41.1 | <.0001 |
| Number of common chronic health conditionsb | 3.2 (1.8–5) | 3.8 (2.2–5) | 3.6 (2.2–5.7) | 0.27 |
| Months to care recipient’s cancer diagnosisb | 3.4 (1–7) | 3.3 (1.3–5.5) | 3.4 (0.9–5.7) | 0.74 |
| Caregiver Characteristics | ||||
| Caregiver sociodemographic | ||||
| Age in years (Median (IQR))b | 59.2 (47.8–69.2) | 58.5 (51.1–67.7) | 55.4 (46–63.8) | 0.15 |
| Female | 49.6 | 70.6 | 73.2 | 0.001 |
| Above high school Education | 93.4 | * | * | 0.96 |
| Married/Living with Partner | 66.2 | 77.6 | 68.1 | 0.09 |
| On Medicaid | 13.5 | 9.3 | 21.2 | 0.12 |
| Living with children | 71.7 | 84.3 | 78.8 | 0.08 |
| Currently work for pay | 40.0 | 37.8 | 31.7 | 0.51 |
| Caregiver health status | ||||
| Fair/Poor general health | 18.7 | 18.9 | 22.4 | 0.80 |
| Anxiety symptoms | 10.7 | 11.3 | 13.9 | 0.82 |
| Depressive symptoms | 9.7 | 11.7 | 23.6 | 0.06 |
| Number of common chronic health conditions | 1.7 (0.3–3.3) | 1.7 (0.4–4) | 2.3 (0.7–3.9) | 0.3346 |
| Caregiving experience | ||||
| Primary caregiver | 49.4 | 77.3 | 74.0 | <.0001 |
| Caregiver also the spouse | 17.1 | 17.7 | 28.1 | 0.15 |
| Co-resides with the care recipient | 40.2 | 36.8 | 57.9 | 0.0004 |
| Have friends or family that help with care for you, in last month | 68.9 | 75.5 | 72.0 | 0.60 |
| Help on a regular schedule | 26.8 | 30.8 | 65.8 | <.0001 |
| Duration of care giving > 1 year | 90.5 | * | * | 0.78 |
| Caregiving hours per weekb | 4.6 (2.1–10.4) | 6.9 (2.2–13.1) | 15.9 (9.2–35.1) | <.0001 |
| Caregiving Strain and Activities | ||||
| High caregiving strain c | * | * | 39.9 | <.0001 |
| Total caregiving strain scores b | 1 (0–2.1) | 1.6 (0.5–3.8) | 3.1 (1.6–4.9) | <.0001 |
| Total number of health-focused caregiving activities b | 1 (0.1–2.4) | 4.8 (3.9–5.7) | 9.1 (7.9–10.2) | <.0001 |
Note: The analyses took into account the National Study of Caregiving (NSOC) survey design.
Comparisons of variables by caregiving activity latent classes were conducted using Rao-Scott Chi-square test for categorical variables, and F-test in linear regression for continuous variables.
Median (interquartile range, IQR) are shown for continuous variables.
Caregiving strain were based on the following items emotionally, physically, or financially difficult to help, exhaust at night, care more than one can handle, and no time for self. High caregiving strains was defined using the cutpoint of 5, corresponding to ≥85th percentile in this study. ADL=Activities of Daily Living, IADL=Instrumental. Activities of Daily Living.
not reportable due to cell size suppression policy to protect the confidentiality of patients.
Figure 1.
Distributions of caregivers’ relationship to care recipients by health-focused caregiving activity latent classes. (Note: Other included other family and non-family members).
3.5. Caregiving strain and latent classes of health-focused caregiving activities
The proportion of caregivers with high caregiving strain in the Moderate-coordination and High-intensity caregiving activity classes were 23.2%, and 39.9%, respectively, which were higher than those in the Low-level activity class (p<0.0001) (Table 3). Similar patterns were seen in the individual caregiving strain items (Figure 2), as well as in the caregiving strain scores and the total number of health-focused caregiving tasks performed (Table 3).
Figure 2.
Proportions of selected caregiving strain items by health-focused caregiving activity latent classes. (Note: Weighted% for financial and physical difficulty strain items were not reportable due to cell size suppression policy to protect the confidentiality of patients.)
After adjusting for care recipient’s dementia status and race/ethnicity, as well as other relevant caregiver-level characteristics, increasing level of caregiving activity remained positively associated with caregiving strain (OR=3.85, 95%CI 2.34–6.33, Table 4). Similar results were found when alternative indicators of caregiving activity (e.g., treated as a categorical variable) and caregiving strain were used (Appendix Table A1).
Table 4.
Higher degree of health-focused caregiving activity is positively associated with high caregiving strain, results from the multivariable logistic regression model.
| Adjusted Odds Ratio (95%CI) | ||
|---|---|---|
| Caregiving activity pattern | Increasing level of health-focused caregiving activities | 3.85 (2.34 – 6.33) |
| Care recipient characteristics | ||
| Dementia status | Possible/No dementia | 1 (Reference) |
| Probable dementia | 2.36 (1.24 – 4.48) | |
| Sociodemographic | non-Hispanic White | 1 (Reference) |
| non-Hispanic Black | 0.56 (0.21 – 1.55) | |
| Hispanic | 0.61 (0.15 – 2.47) | |
| Other race/ethnicity | 5.6 (1.85 – 16.94) | |
| Caregiver characteristics | ||
| Sociodemographic | Age in years | 0.98 (0.96 – 1.01) |
| Male | 1 (Reference) | |
| Female | 1.44 (0.8 – 2.58) | |
| High school education or less | 1 (Reference) | |
| Above high school education | 3.7 (0.55 – 24.85) | |
| Not Married | 1 (Reference) | |
| Married/Living with partner | 0.72 (0.24 – 2.15) | |
| Caregiver health | Good/excellent health | 1 (Reference) |
| Poor/Fair general health | 3.71 (1.71 – 8.06) | |
| Relationship to care recipient | Other | 1 (Reference) |
| Spouse | 2.67 (0.88 – 8.12) | |
| Caregiving situation | Without friend help | 1 (Reference) |
| With friend help | 0.28 (0.14 – 0.55) | |
| Non-primary caregiver | 1 (Reference) | |
| Primary caregiver | 0.83 (0.34 – 2.04) |
Note: The analyses took into account the National Study of Caregiving (NSOC) survey design. n=414 for the adjusted model. Caregiving strain were based on the following items emotionally, physically, or financially difficult to help, exhaust at night, care more than one can handle, and no time for self. High caregiving strains was defined using the cutpoint of 5, corresponding to ≥ 85th percentile in this study. The 3-level caregiving activity pattern was treated as a continuous variable in the main model.
4. Discussion
Using a nationally representative sample, we found that approximately 4.2 million pre-cancer caregivers cared for older adults six months prior to a cancer diagnosis during 2011–2017. Half of them were involved with four or more health-focused caregiving tasks. In addition, the majority of these caregivers (~60%) were already engaged in moderate- to high-intensity healthcare activities, and experienced associated high levels of caregiving strain. Given the indispensable role caregivers play in helping older adults with cancer, these findings suggest that a substantial proportion are already highly burdened with tasks and strain, which may be further exacerbated with the compounding of additional responsibilities associated with the patient’s new cancer diagnosis and subsequent treatment. In response, strategies are needed to assess caregiver strain and have supportive care services available.
As older adults increasingly make up the population of newly-diagnosed cancers, navigating pre-existing illnesses and supporting a rising number of caregivers caring for patients with multimorbid conditions will be critical to future models of comprehensive oncological care. Between 2013 and 2017, approximately 4.6 million older adults aged 65 and above were diagnosed with a new cancer, representing 55% of all new cancer diagnosed in the United States.[34,35] Advancement in prevention and treatment strategies, combined with the aging population trend,[36] will result in an increase in the number of older adults living with cancer. An expected parallel increase is the demand of caregiving support and care delivery support for older adults across the cancer care continuum.[37]
Our study demonstrated that increasing levels of health-focused caregiving activities were positively associated with increasing levels of caregiving strain, even after adjusting for care recipient’s non-cancer related illness severity, including dementia status. Moreover, the association was robust to different measures of caregiving strain and caregiving activities. With anticipated increases in health-focused caregiving activities following a cancer diagnosis,[38,39] caregiving strain is likely to increase as well, as suggested by high rates of distress in samples of cancer caregivers.[18] For clinicians, older patient’s new cancer diagnosis will also increase the complexity of the treatment decision-making process and the subsequent oncological care delivery for older adults with dementia.[40,41] As caregivers play a major role in how patients manage cancer treatment, understanding caregivers’ experiences and recognizing caregivers with high caregiving strain prior to older adults’ cancer diagnoses can help guide forthcoming cancer care treatment decision making and oncological care delivery.[42] Incorporating caregiving distress screening into cancer treatment planning may help assess and strengthen caregiving capacity to support multimorbid older adults.[43] The care plan and decision-making must involve caregivers who are already highly involved in assisting patients.[43] The health and well-being of patient’s caregivers should also be part of the treatment planning and decision-making.
Critically, we found that pre-cancer caregivers were heterogeneous relative to their healthcare-related activities. Close to one third of pre-cancer caregivers were highly involved in nearly every aspects of care recipient’s health-focused activities. Not surprisingly, 40% of these high-intensity caregivers also experienced high strain. Caregivers in the Moderate activity class were primarily involved in healthcare tasks, health system logistics, and patient advocacy, and about a quarter of them experienced high caregiving strain. In addition, about 40% of caregivers with Moderate-coordination and High-intensity activity classes assisted older adults with dementia. While the health and economic costs for caring for persons with dementia is well established and recognized,[25,28–30] our findings highlight the burden of caring for persons with multiple age-related morbidities. Future treatment options for dementia care will likely result in prolonged caregiving tenure that may in turn lead to increased caregiving strain. For healthcare practitioners, older patient’s new cancer diagnoses may increase the complexity of the treatment decision-making process and subsequent care delivery, especially in the context of dementia.
The identified caregiving activity patterns move beyond existing approaches that rely on simple total counts of caregiving activities to define caregivers.[25–27] The total number of caregiving activities has the advantage of being easy to use and serves as a sufficient indicator of caregiving intensity. However, this method gives all the caregiving activities the same weight, which may not be sufficiently nuanced. For example, Moderate-activity caregivers had high propensity of helping with five selective tasks related to care coordination, and those caregivers are likely to play an essential part of impending oncology treatment. Analyses limited to activity counts may miss this distinction thereby reducing opportunities for targeted caregiver intervention.
Our study has several limitations. First, we only examined the period prior to care recipient’s cancer diagnosis and do not have sufficient data (n=75) for post-diagnosis analysis of change in caregiving activity. However, by examining the caregiving activity profile and caregiving strain prior to care recipient’s cancer diagnosis, the study provides potential actionable insights early in cancer care and treatment planning that may benefit both soon-to-be cancer patients and their caregivers along the cancer care continuum. Second, care recipients in NHATS were sampled from the Medicare fee-for-service population. Their caregivers, thus, may not be representative of all caregivers assisting older adults (e.g., such as those with Medicare Advantage). However, this study drew on a nationally representative sample and the results could be generalizable to caregivers helping Medicare fee-for-service (~38 million), which is about 70% of the older adults aged 65 and above (~54 million) in the United States.[44,45] In addition, the demographic characteristics of caregivers in our sample, e.g., majority female, non-Hispanic White, and caring for parent/parent-in-law, were consistent with those found in national surveys.[3,4,8,27] Finally, the identified caregiver profiles were based on a comprehensive list of health-related care activities and was not designed specifically to capture cancer-care activities.
In conclusion, an analysis of a national dataset of older adults with soon-to-be-diagnosed cancer found that a large proportion of them were cared for by family caregivers who were already providing intense support and who were already highly strained. Given the likely increase in cancer diagnoses among older adults with pre-existing caregiving needs, it is imperative that models of oncology care evolve to include structured screening, assessment, and support of family caregivers who are critical partners in cancer care delivery.
Acknowledgement
The National Health and Aging Trends Study (NHATS) is NHATS is being supported by the National Institute on Aging under a cooperative agreement with the Johns Hopkins University Bloomberg School of Public Health (U01AG032947), with data collection by Westat. The National Study of caregiving (NSOC) I (2011) and II (2015) were conducted with funding from the Assistant Secretary of Planning and Evaluation, DHHS. NSOC III (2017) was funded by the National Institute on Aging R01AG054004. NHATS is funded by the National Institute on Aging (U01AG032947).
Funding
Supported in part by a grant from the National Institute on Aging (NIA) to the Claude D. Pepper Older American Independence Center Grant 5P30AG028741-07.
Appendix
Table A1.
Sensitivity analysis results: Higher degree of health-focused caregiving activities is positively associated with high caregiving strain, and the results are consistent across different models.
| Outcome variable | Exposure variable | Adjusted Odds Ratio (95%CI) | Models |
|---|---|---|---|
| High caregiving strain (Yes vs No) | Caregiving activity pattern (categorical variable) | Sensitivity analysis, logistic regression model | |
| Low-level activity class | 1 (Reference) | ||
| Moderate-coordination activity class | 7.52 (2.3 – 24.54) | ||
| High-intensity activity class | 17.58 (5.49 – 56.35) | ||
| Outcome variable | Exposure variable | Adjusted Odds Ratio (95%CI) | |
| High caregiving strain (Yes vs No) | Summed numbers of health-focused caregiving activities (continuous variable) | 1.39 (1.26 – 1.53) | Sensitivity analysis, logistic regression model |
| Outcome variable | Exposure variable | Marginal effect (95%CI) | |
| Summed score of caregiving strain (continuous variable) | Caregiving activity pattern (categorical variable) | least square mean difference) | Sensitivity analysis, linear regression model |
| High-intensity vs Low-level activity class | 1.78 (1.13 – 2.42) | ||
| High-intensity vs Moderate-coordination activity class | 1.03 (0.56 – 1.51) | ||
| Low-level vs Moderate-coordination activity class | −0.74 (−1.23 – −0.26) |
Note: Caregiving strain were based on the following items emotionally, physically, or financially difficult to help, exhaust at night, care more than one can handle, and no time for self. High caregiving strains was defined using the cutpoint of 5, corresponding to ≥85th percentile in this study. All models adjusted for care recipient’s dementia status and race/ethnicity, and caregiver-level characteristics (age, sex, education, marital status, relationship with the care recipient, general health, having friend/family help with care, and primary caregiver status). Sample size for the adjusted model was 414. All analyses took into account the National Study of Caregiving (NSOC) survey design.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of competing interest
The authors have declared no conflicts of interest.
References
- 1.Miller KD, Nogueira L, Mariotto AB, et al. : Cancer treatment and survivorship statistics, 2019. CA Cancer J Clin 69:363–385, 2019 [DOI] [PubMed] [Google Scholar]
- 2.Yabroff KR, Kim Y: Time costs associated with informal caregiving for cancer survivors. Cancer 115:4362–73, 2009 [DOI] [PubMed] [Google Scholar]
- 3.NAC: Cancer Caregiving in the U.S.: An Intense, Episodic, and Challenging Care Experience. https://www.caregiving.org/wp-content/uploads/2020/05/CancerCaregivingReport_FINAL_June-17-2016.pdf. Accessed April 22, 2021. National Alliance for Caregiving, in partnership with the National Cancer Institute and the Cancer Support Community, 2016 [Google Scholar]
- 4.Edwards VJ, Bouldin ED, Taylor CA, et al. : Characteristics and Health Status of Informal Unpaid Caregivers - 44 States, District of Columbia, and Puerto Rico, 2015–2017. MMWR Morb Mortal Wkly Rep 69:183–188, 2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kent EE, Rowland JH, Northouse L, et al. : Caring for caregivers and patients: Research and clinical priorities for informal cancer caregiving. Cancer 122:1987–95, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Porter LS, Dionne-Odom JN: Supporting cancer family caregivers: How can frontline oncology clinicians help? Cancer 123:3212–3215, 2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.National Academies of Sciences E, and Medicine: Families Caring for an Aging America, in Schulz R, Eden J (eds): Families Caring for an Aging America. Washington (DC), The National Academies Press. 10.17226/23606., 2016, pp 1–15 [DOI] [PubMed] [Google Scholar]
- 8.NAC: Caregiving in the U.S. 2020 report, The National Alliance for Caregiving (NAC), 2020 [Google Scholar]
- 9.Coumoundouros C, Ould Brahim L, Lambert SD, et al. : The direct and indirect financial costs of informal cancer care: A scoping review. Health Soc Care Community 27:e622–e636, 2019 [DOI] [PubMed] [Google Scholar]
- 10.Washington KT, Pike KC, Demiris G, et al. : Unique characteristics of informal hospice cancer caregiving. Support Care Cancer 23:2121–8, 2015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ware OD, Cagle JG: Informal Caregiving Networks for Hospice Patients With Cancer and Their Impact on Outcomes: A Brief Report. Am J Hosp Palliat Care 36:235–240, 2019 [DOI] [PubMed] [Google Scholar]
- 12.Romito F, Goldzweig G, Cormio C, et al. : Informal caregiving for cancer patients. Cancer 119 Suppl 11:2160–9, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hayman JA, Langa KM, Kabeto MU, et al. : Estimating the cost of informal caregiving for elderly patients with cancer. J Clin Oncol 19:3219–25, 2001 [DOI] [PubMed] [Google Scholar]
- 14.Goren A, Gilloteau I, Lees M, et al. : Quantifying the burden of informal caregiving for patients with cancer in Europe. Support Care Cancer 22:1637–46, 2014 [DOI] [PubMed] [Google Scholar]
- 15.Goldzweig G, Merims S, Ganon R, et al. : Informal caregiving to older cancer patients: preliminary research outcomes and implications. Ann Oncol 24:2635–2640, 2013 [DOI] [PubMed] [Google Scholar]
- 16.Given CW: Family Caregiving for Cancer Patients: the State of the Literature and a Direction for Research to Link the Informal and Formal Care Systems to Improve Quality and Outcomes. Semin Oncol Nurs 35:389–394, 2019 [DOI] [PubMed] [Google Scholar]
- 17.Kim Y, Schulz R: Family caregivers’ strains: comparative analysis of cancer caregiving with dementia, diabetes, and frail elderly caregiving. J Aging Health 20:483–503, 2008 [DOI] [PubMed] [Google Scholar]
- 18.Geng HM, Chuang DM, Yang F, et al. : Prevalence and determinants of depression in caregivers of cancer patients: A systematic review and meta-analysis. Medicine (Baltimore) 97:e11863, 2018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ornstein KA, Liu B, Schwartz RM, et al. : Cancer in the context of aging: Health characteristics, function and caregiving needs prior to a new cancer diagnosis in a national sample of older adults. J Geriatr Oncol 11:75–81, 2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ferrell B, Wittenberg E: A review of family caregiving intervention trials in oncology. CA Cancer J Clin 67:318–325, 2017 [DOI] [PubMed] [Google Scholar]
- 21.Northouse LL, Katapodi MC, Song L, et al. : Interventions with family caregivers of cancer patients: meta-analysis of randomized trials. CA Cancer J Clin 60:317–39, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kasper JD, Freedman VA: National Health and Aging Trends Study (NHATS),. Baltimore: Johns Hopkins University School of Public Health. Available at www.NHATS.org, 2021 [Google Scholar]
- 23.Freedman VA, Skehan ME, Hu M, et al. : National Study of Caregiing (NSCO) I-III User Guide, Baltimore: Johns Hopkins Bloomberg School of Public Health. Available at www.nhats.org, 2020 [Google Scholar]
- 24.Ornstein KA, Kelley AS, Bollens-Lund E, et al. : A National Profile Of End-Of-Life Caregiving In The United States. Health Aff (Millwood) 36:1184–1192, 2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Riffin C, Van Ness PH, Wolff JL, et al. : Family and Other Unpaid Caregivers and Older Adults with and without Dementia and Disability. J Am Geriatr Soc 65:1821–1828, 2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Riffin C, Van Ness PH, Wolff JL, et al. : Multifactorial Examination of Caregiver Burden in a National Sample of Family and Unpaid Caregivers. J Am Geriatr Soc 67:277–283, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wolff JL, Spillman BC, Freedman VA, et al. : A National Profile of Family and Unpaid Caregivers Who Assist Older Adults With Health Care Activities. JAMA Intern Med 176:372–9, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Vick JB, Ornstein KA, Szanton SL, et al. : Does Caregiving Strain Increase as Patients With and Without Dementia Approach the End of Life? Journal of Pain and Symptom Management 57:199-+, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Witham G, Haigh C, Mitchell D, et al. : Carer Experience Supporting Someone With Dementia and Cancer: A Narrative Approach. Qual Health Res 28:813–823, 2018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hurd MD, Martorell P, Delavande A, et al. : Monetary Costs of Dementia in the United States. New England Journal of Medicine 368:1326–1334, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kasper JD, Freedman VA, Spillman BC: Classification of Persons by Dementia Status in the National Health and Aging Trends Study. Technical Paper #5. https://www.nhats.org/sites/default/files/inline-files/DementiaTechnicalPaperJuly_2_4_2013_10_23_15.pdf. Accessed April 9, 2020. Baltimore: Johns Hopkins University School of Public Health., 2013 [Google Scholar]
- 32.Collins Linda M., Lanza ST: Latent Class and Latent Transition Analysis: With Applications in the Social, Behavioral, and Health Sciences, John Wiley & Sons, Inc., 2010 [Google Scholar]
- 33.Weller BE, Bowen NK, Faubert SJ: Latent Class Analysis: A Guide to Best Practice. Journal of Black Psychology 46:287–311, 2020 [Google Scholar]
- 34.CDC: Rate of New Cancers by Age Group (years), All Races, Both Sexes. https://gis.cdc.gov/Cancer/USCS/DataViz.html. Accessed April 20, 2021. 2017
- 35.Bluethmann SM, Mariotto AB, Rowland JH: Anticipating the “Silver Tsunami”: Prevalence Trajectories and Comorbidity Burden among Older Cancer Survivors in the United States. Cancer Epidemiol Biomarkers Prev 25:1029–36, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.He W, Goodkind D, Kowal P: An Aging World: 2015. https://www.alz.org/media/documents/alzheimers-facts-and-figures.pdf. Accessed April 22, 2021. Census Bureau, International Population Reports, P95/16–1, U.S. Government Publishing Office, Washington, D.C., 2016 [Google Scholar]
- 37.Takvorian SU, Balogh E, Nass S, et al. : Developing and Sustaining an Effective and Resilient Oncology Careforce: Opportunities for Action. J Natl Cancer Inst 112:663–670, 2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Kehoe LA, Xu H, Duberstein P, et al. : Quality of Life of Caregivers of Older Patients with Advanced Cancer. J Am Geriatr Soc 67:969–977, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Huang LW, Smith AK, Wong ML: Who Will Care for the Caregivers? Increased Needs When Caring for Frail Older Adults With Cancer. J Am Geriatr Soc 67:873–876, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Lamore K, Montalescot L, Untas A: Treatment decision-making in chronic diseases: What are the family members’ roles, needs and attitudes? A systematic review. Patient Education and Counseling 100:2172–2181, 2017 [DOI] [PubMed] [Google Scholar]
- 41.Dionne-Odom JN, Ejem D, Wells R, et al. : How family caregivers of persons with advanced cancer assist with upstream healthcare decision-making: A qualitative study. Plos One 14, 2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Caba Y, Dharmarajan K, Gillezeau C, et al. : The Impact of Dementia on Cancer Treatment Decision-Making, Cancer Treatment, and Mortality: A Mixed Studies Review. JNCI Cancer Spectr 5:pkab002, 2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Dionne-Odom JN, Williams GR, Warren PP, et al. : Implementing a Clinic-Based Telehealth Support Service (FamilyStrong) for Family Caregivers of Individuals with Grade IV Brain Tumors. J Palliat Med 24:347–353, 2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.CMS: Original Medicare Enrollment: Part A and/or Part B Total, Aged, and Disabled Enrollees, Calendar Years 2014–2019. https://www.cms.gov/files/document/2019cpsmdcrenrollab9.pdf-0. Accessed April, 21, 2021. Centers for Medicare & Medicaid Services, Office of Enterprise Data and Analytics, CMS Chronic Conditions Data Warehouse. [Google Scholar]
- 45.Census U: Stats for Stories: National Senior Citizens Day: August 21, 2020. https://www.census.gov/newsroom/stories/senior-citizens-day.html Accessed April, 22, 2021. RELEASE NUMBER CB20-SFS.99, 2020