| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Allied Health Sciences, Division of Speech and Hearing Sciences; Center on Aging and Diversity, University of North Carolina at Chapel Hill.
2 Cecil G. Sheps Center for Health Services Research and the School of Public Health, University of North Carolina at Chapel Hill.
3 Cecil G. Sheps Center for Health Services Research and the School of Social Work, University of North Carolina at Chapel Hill.
4 College of Social Work, Florida State University, Tallahassee.
5 School of Aging Studies, University of South Florida, Tampa.
6 Cecil G. Sheps Center for Health Services Research and the Department of Family Medicine, University of North Carolina at Chapel Hill.
Address correspondence to Sharon W. Williams, Department of Allied Health Sciences, Division of Speech and Hearing Sciences, Bondurant Hall, CB 7190, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599. E-mail: Sharon_Williams{at}unc.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. We analyzed data using linear mixed models.
Results. Among caregivers, younger age, female gender, more education, financial burden, other dependents, poorer perceptions of care, and more care involvement were associated with more emotional strain. More staff support also was associated with more emotional strain and partially mediated the relationship between having a trusted staff member and emotional strain. Characteristics associated with poorer physical health included unemployment, financial burden, poorer physician communication, and trusted staff member. Informal social support was directly related to better physical health and mediated the relationships between physical health and both physician communication and financial burden.
Discussion. Many characteristics related to end-of-life caregiving outcomes in long-term care are consistent with community-based studies, suggesting that end-of-life caregiving outcomes in long-term care are not markedly different than in other settings. However, the role of staff support may either facilitate or complicate emotional strain and merits additional study.
Key Words: End of life • Long-term care facilities • Social support • Emotional and physical health
THE role of families in providing care to older adults with chronic disease and/or disability is well established. Family and other informal caregivers including friends and neighbors often provide care along a trajectory or caregiving career that includes providing care to an older family member in the community, providing care in a long-term-care (LTC) facility, and providing care at the end of life (EOL; Gaugler, 2005
; Montgomery & Kosloski, 2000; Pearlin & Aneshensel, 1994). Although researchers know that care is typically provided across settings and progressive levels of need, most research has focused on community-based caregiving. Thus, little is known about the experiences of informal caregivers who provide care for residents in LTC facilities, and even less is known about the experiences of informal caregivers who provide care to LTC residents at the EOL (Haley et al., 2002; Tornatore & Grant, 2004; Zarit, 2004). Consequently, although it is well documented that providing community-based care is associated with poor caregiver health outcomes (Dilworth-Anderson, Goodwin, & Williams, 2004; Pinquart & Sörensen, 2003; Schulz & Beach, 1999; Vitaliano, Zhang, & Scanlon, 2003), research on informal caregivers' health outcomes when care recipients are residents of LTC and/or at the EOL has received little attention.
Community-based caregivers report poorer emotional health outcomes (depression, anxiety, role strain) during caregiving and more difficult bereavement after the death of the care recipient than noncaregivers (Knight & McCallum, 1998; Schulz et al., 2004). In terms of physical health, researchers have documented lower immunity, poorer self-reported health, and impaired functional health among caregivers (Kiecolt-Glaser et al., 2003; von Kanel et al., 2006). Furthermore, although both poorer caregiver emotional and physical health have been associated with placement of relatives in LTC (Yaffe, Fox, & Newcomer, 2002), evidence is mixed as to whether LTC placement is associated with improvements in caregivers' health. Some researchers have documented sustained improvement in emotional (Mausbach et al., 2007; Zarit & Whitlatch, 1992) and physical (Mausbach et al., 2007) outcomes following placement, but others have suggested that placement is not associated with improved emotional (Schulz et al., 2004) or physical (Lieberman & Fisher, 2001) health among caregivers.
Although it is logical, and some empirical evidence supports the fact, that burdens and strains are at their greatest during the EOL period (Herbert, Prigerson, Schulz, & Arnold, 2006; Phipps, Braitman, True, Harris, & Tester, 2003; Wolff, Dy, Frick, & Kasper, 2007), this is the period in the caregiving trajectory about which researchers know least. Given that approximately 1 in 4 Americans and 70% of adults with dementia die in LTC facilities (Mitchell, Teno, Miller, & Mor, 2005; Teno et al., 2004), and that LTC facilities are increasingly the site of EOL care for older Americans (Mitchell et al., 2005; Munn, Hanson, Zimmerman, Sloane & Mitchell, 2006), knowing more about the experience of EOL care in these settings is especially important. Furthermore, although scholars know that resources such as social support reduce a host of negative outcomes for caregivers in the community, they know less about social support for informal caregivers who provide care in LTC facilities and, again, even less about those providing care for residents at the EOL.
Social support varies along a host of dimensions, one being whether it is informal or formal. Caregivers of community-based older adults typically receive the majority of their support from informal sources such as family and friends and/or other unpaid helpers; they tend to use limited formal or paid support (Cox & Monk, 1993; Norgard & Rodgers, 1997; Wolff & Kasper, 2006). This situation is different in LTC, however, where staff provide formal support that is important in the context of adjusting to the LTC setting, family involvement, and family satisfaction (Friedemann, Montgomery, Maiberger, & Smith, 1997; Whitlatch, Schur, Noelker, Ejaz, & Looman, 2001). Furthermore, it may be that the presence of formal support provided by LTC staff changes the amount and/or type of informal support that family caregivers of residents in LTC facilities receive from others.
This study used a stress and coping model to examine the emotional and physical health of informal caregivers of LTC residents at the EOL (see Figure 1). The stress and coping model posits that combinations of background characteristics (in this case, decedent and caregiver characteristics), stressors (primary and secondary), and mediators influence an individual's reaction to stress (Pearlin, Mullan, Semple, & Skaff, 1990). We examined whether selected decedent and caregiver characteristics and facility-related perceptions were associated with emotional and physical health, and whether three sources of social support (informal, staff, spiritual) helped explain these relationships.
|
). Primary stressors are related to the cognitive and physical needs of care recipients and also to the surveillance and work required to ensure care recipients' safety (Pearlin et al., 1990). In LTC settings, such surveillance involves monitoring the resident's care through visitation and involvement in care. These stressors are objective and indicate the demands of care; other primary stressors are subjective and reflect how caregivers experience the situation (Aneshensel, Pearlin, & Schuler, 1993; Pearlin et al., 1990; Pearlin, Aneshensel, & LeBlanc, 1997; Tornatore & Grant, 2004). Subjective primary stressors can include the overall perceptions of care and perceived quality of the death experience. Secondary stressors are produced as a result of primary stressors and are associated with roles and activities outside the caregiving situation. Such stressors include family and/or job conflict as well as stress related to financial and/or social limitations. The model proposes that in the face of stressors (primary and secondary), potential mediators such as social support help explain the relationship between stressors and emotional and physical health outcomes. For example, this suggests that caregivers who have stress such as financial burden, but also more social support, will experience comparatively better health than if they did not have such support.
Similar to other studies using the stress and coping model, we examined social support as a mediator between stressors and health outcomes. Our primary research questions were the following: (a) What is the relationship between potential stressors (defined as selected decedent and caregiver characteristics and facility-related perceptions) and the caregiver's emotional and physical health in the context of EOL caregiving? (b) Does social support (informal, staff, spiritual) mediate the relationship between potential stressors and emotional and physical health?
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). To identify decedents and contact their family and staff caregivers, researchers contacted participating facilities each month and asked them to identify resident deaths that had occurred during the previous month. A decedent was eligible if he or she had resided in the facility for at least 15 of 30 days prior to death and had died within the facility or within 3 days of leaving the facility by transfer or discharge to another facility. For each eligible death, facility staff identified the family member or responsible party who was most involved in decisions for the deceased resident during the last month of life and who had visited and/or spoken with the resident or staff at least once during the last month of life. A letter of condolence and a consent form introducing the study were mailed 4 weeks after the date of death to the identified family member or responsible party. Data were collected through telephone interviews, with verbal consent obtained prior to conducting interviews. The institutional review board of the University of North Carolina at Chapel Hill approved all procedures. Between July 2002 and January 2005, family interviews were completed for 451 decedents (44% of eligible decedents; 68% of those for whom a family member was identified by the facility and could be contacted within a 6-month window). Slightly less than three quarters (74%) of the interviews were completed within 4 months post death, and almost all (97%) were completed within 6 months post death. The caregiver sample for this study included the 434 decedents whose informal caregivers provided complete information on the health outcomes of interest in this study.
Measures
The dependent variables were emotional and physical health. The emotional health measure was a 7-item indicator of strain that assessed the stress and strain a caregiver felt as a result of having a relative in an NH (Whitlatch et al., 2001). Examples of questions include the following: "I don't have enough time for myself," "I have more things to do than I can handle," "I feel that my social life is limited," and "I have trouble managing all the demands on my life." Caregivers chose between four response options: (1) strongly disagree, (2) disagree, (3) agree, and (4) strongly agree. Possible scores ranged from 7 to 28, with higher scores indicating greater emotional strain (Cronbach's 
=.89 for this sample).
The physical health measure was the 5-item general health subscale of the Medical Outcomes Study (McHorney, Ware, Lu, & Sherbourne, 1994). Caregivers rated their health as either excellent, very good, good, fair, or poor. In addition, caregivers responded to four statements concerning their health in relation to others' and their expectations about their health. We used an established algorithm to convert responses to a percentile score, with possible values ranging from 0 to 100 (Ware & Sherbourne, 1992). For this study, we recoded the values so that higher scores indicated poorer health (Cronbach's =.82 for this sample).
Background characteristics included the decedent's age, gender, race, and length of stay in the LTC facility in which the resident had died or from which he or she had been transferred within 3 days of death. We also included the informal caregiver's age, gender, race, marital status, relationship to the decedent, level of formal education, employment status, and number of other dependents. Dependents were those for whom the respondent had provided primary physical and/or financial support during the last month of the decedent's life.
Potential primary stressors were decedent and caregiver characteristics selected based on theory and existing literature (Mitrani et al., 2006; Pinquart & Sörensen, 2003; Scharlach, Li, & Dalvi, 2006; Schulz & Beach, 1999; Tornatore & Grant, 2002). Decedent stressors were objective (those related directly to the health of the decedent during the last month of life) and included cognitive status (impaired vs intact) and a symptom burden score based on the caregiver's perception of the decedent's severity and frequency of symptom occurrence in four areas during the last month of life: pain, shortness of breath, skin cleanliness, and nutrition. Scores for each area ranged from 0 to 9, and each area was equally weighted. We summed scores to provide a total score ranging from 0 to 36, with higher scores indicating greater symptom burden (Hanson et al., 2008). We also included the course of illness in the last year of life (stable health; steady, slow decline in health; a series of ups and down in health).
Facility-related primary stressors were both objective (those related to surveillance and monitoring) and subjective (reflecting how caregivers experienced the situation). Figure 1 lists the primary objective stressors of family–physician communication, number of days the caregiver visited the resident during the last month of life, and level of involvement. Family–physician communication was assessed using a 7-item scale (Cronbach's =.96 in this sample; Biola et al., 2007). Caregivers also reported their perceived level of involvement (one item, with five response options ranging from very low to very high). Other potential facility-related stressors were conceptualized as primary subjective stressors and included the caregivers' perception of the overall care the decedent had received in the last month of life (poor, fair, good, very good) and the effectiveness of hygiene care (not at all, a little, quite a bit, very much). For analysis purposes, poor hygiene care was any response lower than "very much." Also, caregivers responded whether they would have preferred more involvement, no change in involvement, or less involvement in the resident's care and whether they had a trusted relationship with any facility staff member (yes/no).
Experiences related to the death of the residents were conceived as primary objective stressors and included the informal caregiver's report of hospice enrollment (yes/no), whether the resident had died in the facility (yes/no), whether family was present at the time of death (yes/no), and whether the death was expected by the informal caregiver (yes/no). In addition, caregivers rated the overall quality of death (conceived as a primary subjective stressor) using a 5-point scale (not at all true to completely true) in response to the following question: "How would you rate the overall quality of [his or her] death; in other words, how much was it like he/she or the family wanted it to be?"
Potential secondary stressors, or those stressors that are triggered by primary caregiving stressors but experienced outside the caregiving role, included the number of other helpers, financial burden, and family conflict. Thus, caregivers indicated the number (none, one, two or more) of other helpers who were available to them in the last month of the decedent's life. Financial burden was the extent (not at all, a little, quite a bit, very much) to which caregivers felt a burden associated with facility fees or health care costs during the last year of the decedent's life. The family conflict measure consisted of eight items (Gaugler, Zarit, & Pearlin, 1999). Respondents indicated the amount of disagreement (none, a little, quite a bit, a lot) regarding the decedent's care during the last month of life in response to statements such as "They gave you unwanted advice." We averaged item scores for a total score (range 0–3), with a higher score indicating more disagreement (Cronbach's =.95 in this sample).
Potential mediators in this study included three sources of social support: informal, staff, and spiritual. The Medical Outcomes Study Social Support Survey was used to assess informal social support (Sherbourne & Stewart, 1991). It includes 19 items that assess four types of support (emotional/informational, tangible, positive social interaction, and affectionate). For each item, respondents indicated how often the type of support was available to them during the last month of the decedent's life, ranging from 1 = none of the time to 5 = all of the time. Scores ranged from 19 to 95, with higher scores reflecting more support.
Staff social support was based on a nursing assistant support for families scale (Whitlatch et al., 2001). An example of 1 of its 11 items is "How often did staff members (not including physicians) reassure you that your family member's behavior was not unusual?" Item scores ranged from 0 = never/almost never or no instance to 3 = always/almost always. We summed item scores for a scale score ranging from 0 to 33. A higher score indicated more support. Spiritual support was measured with five items from the Systems of Belief Inventory (Holland, Kash & Passik, 1998), with a sample question being "When I needed suggestions on how to deal with problems, I knew someone in my religious or spiritual community that I could turn to." Respondents indicated agreement ranging from 0 = strongly disagree to 3 = strongly agree, for a total range of 0 to 15. Higher scores indicated more support. Fewer than 10 caregivers indicated they had no spiritual community.
Analysis
We used descriptive statistics to characterize the sample with respect to decedent and informal caregiver characteristics, potential stressors, mediators, and the physical and emotional health measures. We used linear mixed models, implemented using the SAS MIXED procedure, to account for the clustering of respondents within facilities by specifying a random effect for facility (Littell, Milliken, Stroup, Wolfinger, & Schabenberger, 2006; Verbeke & Molenberghs, 2000). We first estimated a model for each independent variable and each health outcome. We selected for inclusion in a multivariable model for each health outcome the informal caregiver and resident characteristics and potential stressors that were associated with each health outcome in these bivariate analyses (p <.10). We eliminated variables no longer at least marginally significant (p <.10) in the multivariable model once we were assured, via examination of correlation matrices, that this was not due to near collinearity among the independent variables.
Tests for mediation of the direct effects identified in the multivariable models followed the regression strategy outlined by Frazier, Tix, and Barron (2004). We tested whether each independently associated informal caregiver and resident characteristic was associated with each potential mediator by regressing the potential mediator separately on each characteristic. For predictor–mediator combinations that met this requirement, we examined potential mediation effects by adding the potential mediators to the multivariable model for each outcome and computing the percent change in the coefficient for each independent variable.
In order to be considered a mediator, the social support measure also had to be associated with the health outcome measure while we controlled for the selected informal caregiver and resident characteristics (Frazier et al., 2004). Finally, for the emotional health outcome, which had two social support measures meeting criteria for mediation, we estimated a model including the significant resident and informal caregiver characteristics and both mediators. Based on previous empirical and theoretical research supporting potential mediation by social support, we used a fairly liberal p value (p <.10) in determining that the predictor–mediator and mediator–outcome relationships were statistically significant.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
As shown in Table 2, the mean emotional strain score was 18.5 (SD = 5.4; range 7–28), and the mean recoded physical health score (higher scores equal poorer health) was 24.4 (SD = 16.1; range 0–100). Caregivers reporting more emotional strain tended to report poorer physical health, although the strength of the association was modest (r =.20, p <.001). In terms of social support, the mean informal support score was 84.7 (SD = 20.0; range 0–100), mean staff support score was 20.3 (SD = 8.1; range 0–33), and mean spiritual support score was 7.9 (SD = 5.5; range 0–15).
|
|
|
Although in the bivariate analyses caregiver emotional strain was associated with shorter decedent length of stay, more symptom burden, and a fluctuating course of illness (see Table 3), these did not remain significant in multivariable analyses predicting emotional strain. Other characteristics that were statistically significant in the bivariate analyses but were not independently associated with emotional strain in multivariable analyses were caregiver relationship and marital status, employed informal caregiver (likely due to the association with age), report of poor hygiene care and a poorer global assessment of quality of death (likely due to their association with the global assessment of quality of care), informal caregivers preferring a different amount of involvement (likely due to this measure's association with perceived level of actual involvement), and hospice care as part of the death experience.
Mediation effects
In order to assess mediation by the social support measures, we first regressed each of the potential mediators on each of the background characteristics and stressors that were independently associated with emotional strain. For those relationships that were at least marginally significant, we computed the percent change in the regression coefficient for the characteristic or stressor upon addition of the potential mediator to the model. Because spiritual support was not associated with emotional strain in the multivariable model (p =.178), we did not include this type of social support in the assessment of mediation.
As shown in Table 4, analyses identified no mediation effects for informal social support, and indeed informal social support was itself only marginally significantly associated with emotional strain when we adjusted for all of the background characteristics and stressors. However, staff support partially mediated the relationship between having a trusted staff member and increased emotional strain (20%–30% decrease in the coefficient). Such a reduction in the regression coefficient with adjustment for staff support indicated that the trusted staff member relationship with emotional strain may have been mediated through staff support: Those who had a trusted staff member had more staff support, which was in turn associated with more emotional strain.
The addition of staff support to the regression model also affected the relationship between perception of quality of care as less than very good and emotional strain (59%–66% increase in the coefficient). There was an inverse relationship between staff support and quality of care ratings; hence, the strength of the relationship between lower care ratings and emotional strain was enhanced by adjustment for staff social support.
Association of Characteristics, Potential Stressors, and Mediators With Poor Physical Health
Direct effects
Similar to the analyses for emotional strain, we entered into the multivariable model for poor physical health background characteristics and potential stressors that were marginally significant (p <.10) at the bivariate level (see Table 3). Findings based on the multivariable model indicated that employed caregivers reported better physical health (i.e., lower levels of poor health) than those who were not employed (see Table 5). In terms of primary stressors, a higher rating of physician communication was also associated with better physical health. However, having a trusted staff member was associated with poorer physical health, and caregivers who reported financial burden (a secondary stressor) also reported poorer physical health. More informal support was related to better physical health. Neither staff support nor spiritual support had a direct relationship with physical health, so we did not further evaluate these for mediation effects.
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Port et al., 2005). Among the stressors experienced by these caregivers were absence of other helpers (45%) and financial burden related to caregiving (40%).
Our findings that older caregivers, men, and those with lower levels of education reported less emotional strain are consistent with findings from other studies (Amirkhanyan & Wolf, 2003; Bowman, Mukherjee, & Fortinsky, 1998; Cameron, Franche, Cheung, & Stewart, 2002; Pinquart & Sörensen, 2007; Thornton & Travis, 2003; Williams, 2005). The positive relationship between family conflict, financial burden, presence of other dependents, and emotional strain is also consistent with other empirical evidence (Scharlach et al., 2006; Strawbridge & Wallhagen, 1991; Tremont, Davis, & Bishop, 2006). In general, caregivers with poorer family functioning (e.g., family conflict, burden) reported poorer emotional health. The fact that these relationships remain evident during the EOL in LTC suggests that they are pervasive throughout the caregiving career.
It is no surprise that employed caregivers reported better physical health (i.e., lower levels of poor health), as it is reasonable that caregivers with the ability to manage the role of employee in addition to their caregiving role would need adequate health to do so. Conversely, financial burden, which may be linked to absence of employment, was associated with poorer health. Findings also indicated that caregivers who were more likely to report having a trusted staff member were in poorer physical health. It may be that their poorer health status required them to develop more trusting relationships with the staff.
Researchers have previously documented the association between better communication with the physician and better physical health for patients, but not for family caregivers (Franks et al., 2005). However, physician communication, or more specifically time spent discussing advance directives, has been associated with greater satisfaction of family caregivers with EOL care for NH residents (Engel, Kiely, & Mitchell, 2006). In this study, the seven items on the communication measure assessed the degree to which the caregiver had been kept informed and understood the resident's condition, received information about what to expect, was able to ask questions, and felt listened to and understood by the physician. Thus, it would seem that they would be as likely to relate to emotional strain as to physical health, and the fact that they related only to the latter suggests that being able to communicate with physicians requires better physical health rather than results in it. This association between communication and caregiver health provides additional support to recent studies calling for improved physician–family communication for persons who are or may be dying in LTC (Biola et al., 2007; Wetle, Shield, Teno, Miller, & Welch, 2005).
In adjusted analyses, none of the decedent primary stressors were related to emotional or physical health of the caregivers in this sample. Researchers have reported similar findings about the relative lack of importance of care recipient factors in community-based samples (Dilworth-Anderson et al., 2004; Harwood, Barker, Ownby, & Duara, 2000) and also in NHs. Tornatore and Grant (2002), for example, found that resident characteristics were not associated with family caregiver burden, stating that there is "growing evidence that care recipient impairment is not as important as other factors in caregiver outcomes" (p. 503). These results suggest that interventions aimed at improving family caregiver health do not need extensive tailoring based on care recipient characteristics but instead should concentrate on characteristics of the caregivers and the caregiving environment.
Although decedent stressors were not related to health outcomes, facility perceptions and secondary stressors were important. Facility-related stressors such as poorer quality of care, days visited, and caregiver involvement were associated with increased emotional strain, which supports other findings that more involved caregivers report higher levels of emotional distress (Whitlatch et al., 2001). Furthermore, our findings support the stress and coping conceptualization that secondary stressors (i.e., those outside the caregiving role) are critical to understanding outcomes for caregivers. Two of the three secondary stressors (financial burden and family conflict) were related to more emotional strain, and financial burden was related to poorer health. Once again, this study clarifies the consistency of these relationships throughout the EOL of the care recipient and the relevance of a stress and coping model to understanding caregiving for residents at the EOL.
The role of the potential mediators was mixed and, in the case of staff support, somewhat surprising. If staff support were indeed a resource, empirical and theoretical evidence suggests that it would be associated with less emotional strain. However, our data indicate a direct link between staff support and increased emotional strain. A more detailed examination of the content of the staff support items showed that the staff support measure assessed staff communication skills, such as listening, keeping the family member informed, encouraging the family to talk about concerns, and providing support in dealing with feelings about death. One possible explanation for this finding is that staff support increases family awareness of the impending death, thus increasing emotional strain. Whereas Vohra, Brazil, Hanna, and Abelson (2004) documented that informal caregivers value when staff are able to inform them that death is near, staff should consider that such knowledge may be associated with increased strain (Goodridge, Bond, Cameron, & McKean, 2005). Alternatively, one cannot infer the temporal relationship between staff support and emotional strain from this cross-sectional study; it is also plausible that those who experienced emotional strain may have been more likely to receive staff support.
Staff support mediated the relationship between having a trusted staff member and emotional strain, as evidenced by the fact that the regression coefficient for having a trusted staff member was reduced when staff support was added to the regression model. This finding suggests that having a trusted staff member is associated with increased staff support, which is in turn associated with more emotional strain. Thus, the association between having a trusted staff member and increased emotional strain is partly explained by the positive relationship of staff support with both having a trusted staff member and emotional strain.
Staff support had the opposite effect on the relationship between poorer quality of care and emotional strain, however. Here, because of the inverse association between poor quality care and staff support and the positive relationship between staff support and emotional strain, the addition of staff support to the regression model magnified the relationship between poorer quality of care and emotional strain. Again, when one considers that the staff support measure was related to talking about death and communication skills, it is not a surprise that caregivers who perceived poorer quality of care would be more strained when they had increased awareness and communication regarding the impending death of their family member. However, because we cannot infer causality, it may also be that caregivers who were more strained perceived that the quality of care was poorer.
Similar to other findings in the social support literature, more informal support was related to better physical health. In addition, the mediating function of informal support on the relationship between secondary stressors such as financial burden and poor physical health is consistent with Pearlin and colleagues' stress and coping model. The potential mediator of spiritual support was not related (either directly or as a mediator) to either health outcome in this study. This may be because members of one's spiritual community may be less likely than family and friends or LTC staff to know about the frail condition of a care recipient in an LTC facility. Furthermore, the items in the spiritual support measure (e.g., "When I needed suggestions on how to deal with problems, I knew someone in my religious or spiritual community that I could turn to," "I sought out people in my religious or spiritual community when I needed help") necessitated that the caregiver seek out support from his or her spiritual community. Neither the informal nor staff support measure required the caregiver to seek support, but rather they reflected others' (family, friends, staff) responses to the caregiver.
Each of these variables and the hypothesized relationships are shown in Figure 1, which is a modification of Pearlin and colleagues' stress and coping model (Aneshensel, Pearlin, & Schuler, 1993; Pearlin et al., 1990; Pearlin & Aneshensel, 1994; Pearlin, Aneshensel, & LeBlanc, 1997) applied to informal caregivers who provide care to dying residents in LTC facilities. This modification is also based on the work of Whitlatch et al. (2001) and Tornatore and Grant (2002) and advances theories of stress and coping by the addition and detailed examination of facility-related perceptions and the inclusion of staff support as a potential mediator.
Findings related to facility perceptions suggest that additional study is needed to more fully understand caregiver involvement. That is, although other evidence indicates that involvement of informal caregivers in LTC facilities is related to improved resident and caregiver outcomes, our finding that more involved caregivers report more emotional strain suggests a need to better understand caregivers' reasons for involvement and also their emotional needs. Thus, future conceptualization of the stress and coping model to EOL caregiving in LTC should include a measure of bereavement. A bereavement measure could help identify caregivers who are at risk for complicated bereavement and contribute to understanding the relationship between bereavement and the health outcomes of informal caregivers.
This study is one of the first large studies to examine stressors, outcomes, and social support for informal caregivers who provide care for residents in NHs and RC/AL settings at the EOL. As such, it provides new insight into this group of caregivers and their experiences at a critical point in the caregiving trajectory. There are, however, some limitations to the study. First, of 1,020 eligible deaths, 667 identified caregivers were successfully contacted. No family caregiver was identified by the facility (n = 123), or the caregiver could not be reached during the study enrollment period (n = 230) for the other eligible deaths. Among the 667 caregivers who were contacted, a total of 216 caregivers refused to enroll in the study, and 17 who participated did not complete the items on caregiver health. Therefore, analyses were restricted to the 434 informal caregivers who consented to be interviewed and responded to the emotional and physical health items (response rate = 65%).
In comparing participants and nonparticipants, we found that caregivers of decedents from larger facilities and NHs, and of younger or Black decedents, were less likely to participate. Although this should not strongly affect the associations between variables, it could mean that our overall estimates of caregiver health are biased if, for example, caregivers we were unable to contact or who refused to participate were in poorer emotional or physical health than those who participated. Such a reduction in variability might have limited our ability to identify statistically significant associations. Second, ours was a cross-sectional study, so we cannot determine temporality and causality. Finally, given that this was a retrospective study, the responses were indicative of the caregiver's feelings after the death and could have been affected by the caregiver's bereavement.
Our modification of the Pearlin et al. stress and coping model (as represented in Figure 1) does, however, identify both potential caregiver stressors and resources associated with the emotional and physical health of informal caregivers of LTC residents at the EOL. These include certain background variables (employment, education), primary stressors (quality of care), secondary stressors (financial burden), and social support (informal, staff). The role of staff support in ameliorating or exacerbating caregiver stress is difficult to interpret and presents an opportunity for additional research regarding the nature of and outcomes associated with staff/family communication and support. We concur with Whitlatch et al. (2001) that there is much to learn about the complex nature of interactions within LTC settings. Given that deaths in LTC facilities are expected to increase and that families of these decedents will be under stress, it is imperative that researchers better understand characteristics that increase and alleviate this stress so as to provide better care not only for facility residents who are terminally ill, but also for the family members who continue to be actively involved in that care.
|
Acknowledgments |
|---|
S. W. Williams was responsible for conceptualization of the manuscript and selection of measures specific to the manuscript and took overall responsibility for the manuscript. C. S. Williams was responsible for conceptualization and conduct of analyses and tables as well as overall completion and review of the manuscript. S. Zimmerman was responsible for conceptualization of the parent study and overall oversight and conduct of the parent study as well as conceptualization and completion of this manuscript. J. Munn participated in the Methods and Discussion portions of this manuscript and overall review of the manuscript. D. Dobbs participated in data acquisition, Methods, Discussion, and overall review of the manuscript. P. D. Sloane was responsible for conceptualization of the parent study and overall oversight and conduct of the parent study and conceptualization and completion of this manuscript.
|
Footnotes |
|---|
Received for publication June 12, 2007. Accepted for publication December 12, 2007.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. T. Lau, J. D. Kasper, J. M. Hauser, C. Berdes, C.-H. Chang, R. L. Berman, J. Masin-Peters, J. Paice, and L. Emanuel Family Caregiver Skills in Medication Management for Hospice Patients: A Qualitative Study to Define a Construct J Gerontol B Psychol Sci Soc Sci, June 19, 2009; (2009) gbp033v2. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
