Invisible partners in care: Snapshot of well‐being among caregivers receiving comprehensive support from Veterans Affairs

Abstract Background and aims Since May 2011, over 23 000 caregivers of Veterans seriously injured on or after September 11, 2001 have enrolled in the Program of Comprehensive Assistance for Family Caregivers (PCAFC). PCAFC provides caregivers training, a stipend, and access to health care. The aim of this study is to describe the characteristics of caregivers in PCAFC and examine associations between caregiver characteristics and caregiver well‐being outcomes. Methods We sent a web survey invitation to 10 000 PCAFC caregivers enrolled as of September 2015. Using linear and logistic regressions, we examine associations between PCAFC caregiver characteristics and caregiver outcomes: perceived financial strain, depressive symptoms (Center for Epidemiologic Studies Depression Scale [CESD‐10]), perceived quality of Veteran's Veterans Health Administration (VHA) care, and self‐reported caregiver health. Results We had complete survey data for 899 respondents. Since becoming a caregiver, approximately 50% of respondents reported reducing or stopping work. Mean time spent providing care was 3.8 years (median 3, IQR 1‐5) with an average of 4.9 weekdays (median 5, IQR 5‐5) and 1.9 weekend days (median 2, IQR 2‐2). The mean CESD‐10 score was 8.2 (median 7, 4‐12), at the cutoff for screening positive for depressive symptoms. A longer duration of caregiving was associated with having 0.08 increase in rating of financial strain (95% CI, 0.02‐0.14). Caregiver rating of the Veteran's health status as “fair” or better was a strong predictor of better caregiver outcomes, ie, self‐reported caregiver health. However, higher levels of education were associated with worse caregiver outcomes, ie, lower global satisfaction with VHA care, higher CESD‐10 score, and higher rating of financial strain. Conclusions Higher depressive symptoms among longer duration caregivers, coupled with high rates of reductions in hours worked, suggest interventions are needed to address the long‐term emotional and financial needs of these caregivers of post‐9/11 Veterans and identify subpopulations at risk for worse outcomes.


| INTRODUCTION
In the United States, uncompensated family members and friends deliver the majority of long-term services and support as informal caregivers. (Reaves & Musumeci 1 ) An estimated 5.5 million informal caregivers, 2.3% of the adult population in the United States, provide medical care, emotional support, and/or physical assistance to Veterans (these caregivers are referred to as military caregivers henceforth). 2 Approximately 1.1 million military caregivers provide care for Veterans who served 2 post-9/11. Post-9/11 Veterans and their caregivers are younger than pre-9/11 Veterans and their caregivers.
Posttraumatic stress disorder (PTSD), traumatic brain injury (TBI), and depression are the signature injuries of the Operation Enduring Freedom/Operation Iraqi Freedom (OEF/OIF) conflicts. 3 As a result of the signature injuries of OEF/OIF and the advances in medicine, the primary caregiving activities of post-9/11 caregivers often differ from traditional caregiving activities. For example, post-9/11 military caregivers provide assistance with fewer activities of daily living (eg, toileting) and instrumental activities of daily living (eg, managing finances) than civilian caregivers; however, they provide greater assistance helping "care recipients cope with stressful situations or avoid triggers of anxiety or antisocial behavior." 2 With differences in caregiving tasks coupled with an increased number of years providing care, the long-term burden of caregiving on post-9/11 military caregivers could include work strain, financial strain, and difficulty planning for the future (eg, caregiver's own retirement). 2 Moreover, these younger caregivers may be balancing multiple demands, including young children and careers. Evidence of high levels of burden and strain for informal caregivers of patients with chronic diseases exists. [4][5][6] However, the impacts of the expected prolonged informal caregiving by post-9/11 military caregivers are currently unknown.
In addition to the differences in caregivers of pre-9/11 Veterans to caregivers of post-9/11 Veterans, caregivers of Veterans may have different informal caregiving experiences compared with caregivers of civilians. Prior work has found that "compared with national statistics on nonmilitary caregivers, military caregivers were found to be younger, serve as caregivers longer, and have greater caregiver burden, stress, and financial strain." 2,7 To further compound these challenges, caregivers of Veterans with PTSD, a signature injury in post-9/11 Veterans, may have higher burden and more negative caregiving experiences compared with caregivers of Veterans without PTSD. 2,8 The RAND Hidden Heroes report identified that caregivers of post-9/11 Veterans had approximately "five times the odds of meeting criteria for probable depression and scored an average of 19 points higher on anxiety symptoms" compared with noncaregivers. For context, caregivers of pre-9/11 Veterans and nonmilitary caregivers had "twice the odds of meeting criteria for probable depression and scored between six and eight points higher on the anxiety scale." 2 Historically, National Caregiver Support Programs in the United States have been fragmented and limited in scope. 9 Currently, the optimal combination of services and supports for long-term caregivers to maintain or improve their health and well-being is unknown. The support provided to caregivers in PCAFC is unmatched in the United States. Gaps in the literature of caregiver health and well-being of post-9/11 caregivers exist. Additionally, the demographics and health and well-being of caregivers enrolled in PCAFC are unknown.
Because PCAFC provides unmatched support in the United States, understanding caregiver health and well-being for those post-9/11 caregivers receiving intensive caregiver supports and services in the United States is currently unaddressed in the caregiving literature.
Examining PCAFC caregivers' health and well-being outcomes, such as caregiver depressive symptoms and financial strain, contribute to the gaps in the existing literature on post-9/11 caregivers and on caregivers receiving intensive supports and services in the United States.
The primary objective of this study is to describe the characteristics of caregivers in PCAFC using survey data collected in September 2015. The secondary objective is to examine associations between caregiver characteristics and caregiver well-being outcomes, including caregiver perceived financial strain, depressive symptoms, Veteran's VHA quality of care, and self-reported health status. The findings from this study highlight those caregiver characteristics associated with more positive caregiver outcomes of well-being in the study sample.
While these findings are not generalizable to the entire population, they provide some insight regarding caregiver characteristics associated with increased risk for poor outcomes, which may help inform future research to develop appropriate interventions for these individuals.

| Study sample
The analysis uses cross-sectional data collected through a national survey of caregivers as part of a larger coordinated effort to survey caregivers of Veterans. 10 For the analysis of the PCAFC participants, we determined a target sampling pool of 10 000 caregivers based on projected response rate of 50%. To ensure wide geographic representation of the target sample, we invited a 10-percent stratified sample of caregivers applying for support from each Veterans Affairs Medical Center (VAMC) (n = 8118), and an additional nation-wide group of caregivers who were surveyed 9 months earlier (n = 1882) to participate. We required all caregivers in the analysis to have been enrolled in PCAFC for at least 90 consecutive days as of September 1, 2015.
We conducted the survey through a secure website hosted by a third-party vendor, Intellica Corporation, from September to October 2015. Paper surveys were also available upon request. The final analytic cohort was n = 899 and included survey responses, which had no missing data in the outcomes modeled (described below). Because of the low response rate, the sample does not claim to be representative of the total population.

| Measures
Caregiver health and well-being outcomes hypothesized to be impacted by PCAFC include perceived financial strain, depressive symptoms, global rating of satisfaction with VHA care for the Veteran, self-reported health, level of community engagement, positive aspects of caregiving experienced, and employment status since becoming a caregiver. These outcomes encompass metrics of well-being potentially impacted by caregiver supports and services and outcomes that are commonly measured in caregiver populations, ie, geriatric care recipients to younger, disabled care recipients. [4][5][6] We also measured caregiver and Veteran demographic characteristics and Veteran health status indicators. We examined all variables descriptively. Given the small sample size and generalizability limitations, we prioritized metrics of caregiver health and well-being and covariates for regression modeling a priori. We identified four measures of greatest interest for regression modeling, as informed by the program office administering PCAFC: perceived financial strain, depressive symptoms, global rating of satisfaction with VHA care for the Veteran, and self-reported health of the caregiver. The significance levels for analyses were P < 0.05.

| Outcomes examined descriptively only
We examined three additional outcomes of interest: level of caregivers' community engagement, positive aspects of caregiving, and caregivers' current work status.

| Level of caregivers' community engagement
We measured the caregivers' level of community engagement as defined by the number of times the caregiver went out into the community in the prior month to shop, see a movie, attend a sporting event, volunteer, attend religious services, or do something else he/she enjoys. 11 Response categories were "never," "one time," "two times," "three times," "four times," or "five times or more."

| Positive aspects of caregiving
We assessed positive aspects of caregiving as captured by Tarlow and colleagues' nine item, validated measure among caregivers to patients with dementia. 12 Scores range from 9 to 45, where a higher score indicates more positive aspects of caregiving experienced. Respondents were asked to rate how much he/she agreed with nine statements referring to positive feelings due to caregiving as "disagree a lot," "disagree a little," "neither agree or disagree," "agree a little," "agree a lot." For example, respondents are asked how much they agree with the statements such as "Providing help to the Veteran has made me feel strong & confident" and "Providing help to the Veteran has made me feel appreciated." The positive aspects of caregiving has a Cronbach's α of 0.94 in our sample of caregivers.

| Caregiver's current work status
Caregivers were asked to select the option that best described their work status since becoming a caregiver ("I am working my usual hours for pay," "I am working reduced hours for pay," "I started working for pay," "I started working more hours for pay," "I stopped working for pay completely," or "I was not working before and am not now.") 11,13 2.4 | Outcomes examined descriptively and with regression modeling

| Caregiver perceived financial strain
We measured perceived financial strain through the three-item Impact on Finances subscale from the Caregiver Reaction Assessment.
Responses included "strongly disagree," "disagree," "neither agree nor disagree," "agree," or "strongly agree" for statements regarding the degree of financial strain experienced. For example, caregivers were asked how much he/she agreed with the following statements: "It is difficult to pay for the things the Veteran needs"; "Caring for the Veteran puts a financial strain on me"; and, "My financial resources are adequate to pay for things that are required for caregiving". Scores range from 3 to 15, where a higher score indicates higher strain. The caregiver perceived financial strain scale has a Cronbach's α of 0.73 in our sample of caregivers. 14

| Caregiver depressive symptoms
We measured caregiver depressive symptoms through the Center for Epidemiologic Studies Depression 10-item Scale (CESD-10). 15 Responses include "never," "rarely," "sometimes," or "often" regarding statements of frequency of depressive symptoms experienced. For example, caregivers were asked how often "were you bothered by things that don't usually bother you" and "did you feel depressed." The reference period is within the last week of survey administration.
Scores range from 0 to 30, where higher scores indicate more depressive symptoms. Depending on the use of the CESD-10, a score of more than or equal to 8 or more than or equal to 10 is often used to indicate screening positive for depressive symptoms and probable depression, respectively. 15 The CESD-10 has a Cronbach's α of 0.84 in our sample of caregivers.

| Caregivers' self-reported health status
We measured caregivers' self-reported health status through a single item used 13,17 in the SF-36. Caregivers were asked "how would you say your health is now" with possible responses including "poor," "fair," "good," "very good," or "excellent." This was dichotomized into (1) responses of "good," "very good," or "excellent" compared with (0) responses of "poor" or "fair" because of the distribution of responses across categories. (1-lowest, 2 or 3-highest), and Veteran health status ("poor," "fair,"

| Key explanatory variables and covariates
"good," "very good," or "excellent"). We also controlled for the follow- prefer not to answer. If a respondent selected "other," then he/she had the option of writing in his/her race. For regression modeling, race categories were collapsed to "white" and "all other" because of the distribution. Regarding Hispanic ethnicity, possible responses included yes, no, or prefer not to answer.

| Statistical analyses
We calculated descriptive statistics for the sociodemographic characteristics of caregivers, caregiver well-being outcomes, and

| Associations between caregiver and veteran characteristics with caregiver well-being
Regression model results are presented in Table 3. Having an education level of some college, vocational/trade school, or associate's degree compared with those with less than some college educational level was significantly associated with a 0.74 unit lower score of global     The driving force behind these recommendations was to improve the quality of care provided for the patient/care recipient. The report highlights that understanding the caregiver experience is a critical piece to providing appropriate support that can enhance outcomes for both the caregiver and care recipient.
In an effort to understand further the caregiver experience of PCAFC caregivers, we examined established outcomes commonly reported in the caregiving literature. We found that PCAFC caregivers substantially decreased their labor force participation after becoming a caregiver. Overall, PCAFC caregivers also reported high levels of depressive symptoms. Our findings are consistent overall with the RAND Hidden Heroes post-9/11 caregiver profiles. Increased depressive symptoms and perceived financial strain were also associated with longer caregiving duration. Caregivers with higher levels of education generally fared more poorly than those with lower levels of education, which bears further inquiry to understand why. It may be that higher educated caregivers expected to have more control over how they would spend their lives than lower educated caregivers. Thus, it may be possible that these higher educated caregivers fared more poorly and struggling more with the caregiving role, but this is merely speculation that bears more rigorous inquiry. Unsurprisingly, the receipt of a higher stipend was associated with decreased perceived financial strain. Finally, we found that the caregiver's report of the Veteran's health status was strongly associated with caregiver outcomes, with better Veteran health associated with better caregiver well-being.
PCAFC caregivers experienced pronounced self-reported changes in employment status since becoming a caregiver, with approximately half of caregivers reporting they were no longer working for pay or have reduced hours at work due to the demands of caregiving. Additionally, caregivers reported a very high average number of hours of work taken off due to caregiving. The reduction in labor force participation by younger caregivers is of concern, as reduced earnings in the The reference group included caregivers whose self-reported health was "poor" or "fair".
i P values were estimated using t tests.
j P values were estimated using Wald Chi-Squared test.
present, and the potential for reduced earnings in the future will affect the amount of social security received in retirement. Third, this analysis also does not consider other caregiving demands, such as childcare or caring for aging parents. Fourth, while we control for length of time a caregiver has been a caregiver, we are unable to control for length of time a caregiver has been in PCAFC because of data quality limitations. Fifth, the goodness of fit tests suggest the linear models explain a small percent of the variation around the mean and the logistic regression has mediocre predictive accuracy. Finally, we rely on self-reported outcomes, including caregiver report of Veteran's health status. Future research should take into consideration how Veteran's health status is categorized, as caregiver-reported Veteran health status may not correspond with the Veteran's health status documented in the medical record. A Veteran with many comorbidities documented in the medical record may not necessarily be categorized as being in "Poor" health by his/her caregiver.
Despite these limitations, these survey results are the first caregiver-reported data on the participants in a new national program supporting caregivers comprehensively, PCAFC. These data complement, but are distinct from, the RAND "Hidden Heroes" report, because these caregivers are identified as receiving support and being caregivers of Veterans engaged in the VHA health care system. 2 These findings contribute to the body of literature and operational understanding of the caregivers of Veterans who served post-9/11. Griffin et al found slightly higher, but similar, results for employment changes for caregivers of Veterans with Traumatic Brain Injury (TBI) and polytrauma. 22 These findings suggest that even with comprehensive caregiver support, there are individual factors that render the caregiver at risk for worse outcomes. Thus, these findings highlight the need for longitudinal studies to determine the impact of long-term informal caregiving on caregiver health and well-being. Moreover, these findings provide a unique description of caregivers of post-9/11 Veterans participating in PCAFC, which is consistent with the overall RAND report profile of post-9/11 military caregivers 2 and of selfreported data on PCAFC participants. As the type of caregiving in this population differs from the traditional expectations of caregiving, eg, not Activities of Daily Living (ADL) focused, it is critical for researchers to develop appropriate interventions to better serve these often younger caregivers of post-9/11 Veterans, such as considering support services with options for pension support to help offset leaving the workforce. These findings of caregiver health and well-being in a caregiver population receiving unprecedented level of systematic support in the United States illustrate the continued need to develop novel combinations of trainings and supports for caregivers to offset negative outcomes of caregiving challenging this population.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.