“When I Said I Wanted to Die at Home I Didn’t Mean a Nursing Home”: Care Trajectories at the End of Life

Background

Concerns about the implications of population aging are accompanied by concerns regarding the increased medicalization of death and dying evident over the course of the 20th century and which saw death increasingly institutionalized in acute hospital care settings (Menec, Lix, Nowicki, & Ekuma, 2007; Wilson et al., 2001). When asked, most people say they would prefer to die at home in familiar surroundings and in the presence of loved ones (CIHI, 2007). Yet, whereas death (and dying) at home was the norm until the early years of the 20th century, from the 1950s through to the 1990s, death became increasingly concentrated in hospital settings, reaching a peak in 1994 when 80% of all Canadian deaths took place in acute care hospitals (Wilson et al., 2001). Since then, however, with evidence regarding people’s preferences for home deaths coinciding with governmental interests in reducing hospital costs (CIHI, 2007), increasing attention has been directed towards relocating death (and dying) back to home or home-like (e.g., hospice) settings. As a result, by 2004, approximately 60% all deaths in Canada took place in hospital settings (CIHI, 2011). Similar reductions in hospital deaths have been reported in other countries as well (e.g., Houttekier, Cohen, Surkyn, & Deliens, 2011; Teno et al., 2013).

Reductions in hospital deaths are frequently attributed to the success of efforts to strengthen EOL care provided in community settings and thereby increase home-based deaths (CIHI, 2011; Gruneir et al., 2007). They also are viewed as evidence of enhanced continuity of care and the success of attempts to bring about reductions in preventable transitions in care, particularly those that involve hospitalization (Naylor et al., 2009). However, in the push to relocate deaths outside of hospital settings, what is less frequently acknowledged is that as hospital deaths have been reduced and home deaths have increased, the number of deaths taking place in nursing home settings has also increased. According to Wilson et al. (2009), from 1994 to 2004, the number of deaths taking place in Canadian nursing homes and similar institutions tripled. After hospitals, LTC facilities now represent the second most common place of death (CIHI, 2007). In the province of British Columbia specifically, in 2003–2004, 53% of deaths took place in hospitals whereas 27% took place in nursing home care facilities and 17% took place in people’s own homes (CIHI, 2008).

Yet, limited research has focused on why it is that some people die at home in the community while others experience one or more transitions that lead to death in a hospital or a nursing home setting. Concerns around the negative impact of transitions, particularly repeated transitions, from home to hospital on the health and mortality of older adults as well as on related concerns with preserving continuity of care have led to research on health care use at the EOL (Menec et al., 2007) as well as the number and type of transitions experienced (e.g., Abarshi et al., 2010; Teno et al., 2013; Van den Block, Deschepper, Bilsen, Van Casteren, & Deliens, 2007) and where people die (Broad et al., 2013; Gruneir et al., 2007; Klinkenberg et al., 2005; Menec, Nowicki, Blandford, & Veselyuk, 2009; Motiwala, Croxford, Guerriere, & Coyte, 2006; Temkin-Greener, Zheng, & Mukamel, 2013; Teno et al., 2013). Findings suggest that health care utilization tends to increase near the EOL (especially the last 3–6 months) and transitions between care settings are common, in Canada (CIHI, 2008) and elsewhere (Abarshi et al., 2010; Klinkenberg et al., 2005; Van den Block et al., 2007).

Studies have also focused on the characteristics of those likely to experience multiple transitions in care as well as of those whose lives end in various settings. Research evidence suggests that older women are less likely to experience one or more care transitions at the EOL (Aaltonen, Forma, Rissanen, Raitanen, & Jylhä, 2010; Aaltonen, Rissanen, Forma, Raitanen, & Jylhä, 2012; Abarshi et al., 2010). In addition, socioeconomic factors (education, income), marital status, informal support, health status (e.g., dementia vs. other causes of death), the availability of palliative care or other health services, and several other factors also appear to influence where people die (Forma, Rissanen, Noro, Raitanen, & Jylhä, 2007; Fowler & Hammer, 2013; Martikainen, Moustgaard, Einiö, & Murphy, 2014; Menec et al., 2007, 2009; Wilson et al., 2009), including the likelihood of transition (e.g., to hospital) at the EOL (Aaltonen et al., 2010; Menec et al., 2007; Motiwala et al., 2006). For example, older (age 85+) women, those with more limited economic (education, income) and social resources (e.g., not married, no children, lack a primary caregiver) appear less likely to die at home in the community than in institutional care settings (Aaltonen et al., 2010; Klinkenberg et al., 2005; Menec et al., 2007; Motiwala et al., 2006; Weitzen, Teno, Fennell, & Mor, 2003).

The Current Study

The preceding review points to a need for research on care trajectories at the EOL and the factors that influence them. From a life course perspective, it can be argued that it is not only the number of transitions experienced at the very end of life or where the final transition (death) takes place that is important to consider. Instead, insofar as prior transitions are likely to influence subsequent ones, the longer-term trajectories that end in death are also important. Accordingly, we focus on individuals’ transitions through the formal LTC system, including home and community-based care as well as nursing home/residential care settings and other service providers. We draw on a structural/institutional life course perspective (Dannefer, 2012; Moen, 2013) and conceptualize LTC trajectories as being among the multiple social pathways (work, marriage, family life) through which older adults as a social group are likely to pass. These pathways are embedded in and will therefore reflect macro- and meso-level social structural and contextual factors, including the policy contexts within which they are situated, the socially structured inequalities that attend location within particular social groups (e.g., gender, age, geography), the resources and barriers that emanate from these structural forces (e.g., social and economic resources), and the health-related risks that they impose. Accordingly, our objectives are both descriptive and analytical: to examine what the EOL trajectories of older adults who are recipients of publicly provided LTC services look like and to assess the impact of social structural location, associated social and economic resources, and health status factors on these trajectories.

Methods

Results

Discussion

This study drew on a structural/institutional life course perspective and longitudinal data to: (1) identify the main EOL trajectories experienced by older adults transitioning through the publicly supported LTC system and to (2) assess the roles of social location, social and economic resources, and health-related factors in influencing them.

With regard to the first objective, our findings revealed that, in general, LTC clients’ EOL trajectories were generally linear and straight forward: very few (2.5%) experienced trajectories that involved transitions from RC to HCC or more than a single transition between HCC and RC (e.g., HCC to RC to HCC to death). Instead, the vast majority of older clients began their trajectories with the receipt of HCC services. Just over one-third (33.4%) then transitioned directly from HCC to death, most often in a hospital setting. In addition, almost one-half (48.2%) transitioned from HCC to RC to death, typically in a non-hospital setting. Although a minority began their trajectories in RC, most (66.6%) older clients’ trajectories included RC and the vast majority (81.7%) of those whose trajectories included RC subsequently experienced a non-hospital death. Findings indicating that just over 18% of RC residents died in hospital are comparable to those reported in another Canadian study (Menec et al., 2009) and slightly lower than figures reported for the United States (20.4% from 2003 to 2007—see Temkin-Greener et al., 2013).

Overall, the fact that almost two-thirds (65.4%) of those in our sample died outside of hospital settings and that those most likely to do so were those receiving RC rather than HCC services suggest that concerns regarding a highly medicalized and technological versus a comfort-focused home-based death may be more relevant to recipients of home-based LTC services than to older residents of nursing home care settings. This is further supported by findings indicating that over 70% of those whose LTC trajectories ended with a nursing home death had no hospitalizations in the 3 months prior to death compared to only 44% of those who died at home.

As well, findings indicating that most RC residents who died did not die in hospital and did not experience any hospitalizations in the months prior to their deaths could be considered consistent with recent economic, policy, and quality of care objectives (emphasizing the need to minimize unnecessary and often problematic transitions for older adults, especially those with dementia or multiple chronic conditions and complex care regimens for whom such transitions often pose challenges to continuity and quality of care—e.g., Naylor et al., 2004). On the other hand, not all transitions should be considered unnecessary and specifying an appropriate rate appears problematic (Sivananthan & McGrail, 2016). Also, concerns remain regarding the quality of nursing home EOL care (CIHI, 2013; Mitchell, Teno, Miller, & Mor, 2005). When asked, few people identify nursing homes as places they wish to receive care or where they wish to end their lives (Wilson et al., 2009). Further, it has been noted that nursing homes in Canada provide nursing and personal care but do not provide palliative care services and “do not emphasize that they provide care to dying people” (Wilson et al., 2009, p. 1756). For example, it has been reported that only 18.3% of those who died in RC facilities in British Columbia in 2003/2004 received some form of palliative care (CIHI, 2008).

Our second objective was to assess the roles of social location, social and economic resources, and health-related factors in influencing the care trajectories experienced by older clients at the EOL. As noted, a structural/institutional LCP considers age as well as gender and geography as indicators of location within a stratified (unequal) social structure. Our findings revealed considerable empirical support for the importance of such factors in influencing EOL care trajectories. For example, older clients were more likely to experience trajectories that included RC transitions and also, to experience EOL trajectories that ended outside of hospital. This was evident regardless of the inclusion of social and economic resources, health status/needs, or recent hospitalizations in the model, suggesting that the impact of age was not attributable to such factors. Similar findings have been reported by others (e.g., Menec et al., 2007) and support the view that old age itself is a key determinant of EOL care (Forma et al., 2007). There are several possible explanations. First, there is the possibility that within the public LTC system, those in advanced old age are funneled into institutional (nursing home) rather than community-based LTC settings (Penning, Cloutier, Nuernberger, MacDonald, & Taylor, 2016) and, in the process, into a trajectory that will see their lives come to an end within the nursing home itself. In addition, as noted by Forma et al. (2007, p. 152), “regardless of need, older age groups may have fewer admissions to inpatient hospital care than younger ones because they are less likely admitted and less intensive care is given for them” (see Levinsky et al., 2001). This also suggests a rationing of hospital EOL care away from older adults in RC settings.

EOL care trajectories also differed depending on gender and geography. Consistent with previously reported findings (e.g., Forma et al., 2007), older men were generally more likely than older women to experience trajectories that concluded with death in a hospital setting. Although women are often reported to use more health services than men, this appears to vary with the type of service involved, with gender inequities in the direction of greater use by men reported in access to acute hospital care (e.g., Dunlop, Manheim, Song, & Chang, 2002; Song, Chang, Manheim, & Dunlop, 2007). Forma et al. (2007) speculate that this may reflect the fact that older women spend more time in RC institutions than men due to their greater likelihood of living alone, reduced access to informal care, and more severe disabilities. However, we were able to control for the impact of living arrangements, marital status, as well as several health status factors (and economic resources), albeit not informal care, with the result that the impact of gender remained. Similar findings are reported by Song et al. (2007). With regard to geography, compared to urban-dwelling respondents, those who lived in rural areas were less likely to experience EOL trajectories that involved RC transitions compared to HCC transitions (including hospital and non-hospital deaths).

Findings indicating that as was the case with regard to age, the impact of gender and geography on EOL care trajectories remained significant despite the inclusion of social and economic resources or health status indicators in our models suggests that there is in fact something else about these indicators of social structural location that leads to different EOL care trajectories. One possibility, of course, is that these findings reflect our failure to include the most relevant indicators of such factors (e.g., availability of informal caregivers, specific cause of death) in our models. Another possibility, however, is that they reflect the impact of macro- and/or meso-level contextual factors that lead to the differential social structuring of age, gender, and rural versus urban-related LTC and EOL trajectories. This includes health care policy and service delivery contexts that may consider older women to be less capable of functioning independently in the community setting, that may regard older men as being more appropriate candidates for acute hospital and medical care, and that provide for differential access to hospital, nursing home, and community-based resources in rural and urban areas.

Importantly, social and economic resources also influenced LTC trajectories, including their end-points. The finding that those who were not married were more likely to experience trajectories involving RC than HCC, regardless of whether these ended in death in or outside hospital, appears to support prior research attesting to the importance of close family relationships for the ability to live in the community when health declines (e.g., Gaugler, Duval, Anderson, & Kane, 2007). Controlling for marital status, it was those who were living with others who were more likely to experience EOL care trajectories involving RC, particularly those involving non-hospital death. On the one hand, this points to the possibility that those with greater disability were more likely to live with others and subsequently, to enter RC settings when their care needs increased. However, it may also speak to the role of family members in providing guidance to RC facilities with regard to decision-making around EOL care (e.g., providing support for decisions to pursue EOL care that does not involve hospitalization).

Findings pointing to the importance of economic resources in influencing EOL care trajectories must be interpreted within the context of governmental LTC funding policies. In this study, low income levels reduced the likelihood of older adults experiencing EOL care trajectories that included transition from RC to death (either in or outside hospital) compared to HCC to death (in or out of hospital). In British Columbia, publicly subsidized HCC is primarily targeted to those with low incomes. RC services, in contrast, are accessible to individuals at all income levels with subsidy levels dependent on income and assets. Our findings also revealed an association between low incomes and an increased likelihood of experiencing a HCC to hospital death trajectory rather than a HCC to non-hospital death trajectory. We are not sure why this was the case but speculate that it may reflect the greater involvement of HCC staff in the care trajectories of those with the lowest income levels.

Finally, in terms of health status, we found that greater cognitive and functional impairment increased the likelihood of experiencing EOL trajectories that included RC care (and a hospital or non-hospital death) and reduced the likelihood of experiencing a HCC to hospital death as opposed to a HCC non-hospital death trajectory. Similarly, higher levels of depression and exhibiting behaviors labeled as problematic were associated with the likelihood of experiencing an EOL care trajectory that included RC (and a hospital or non-hospital death) whereas greater medical instability increased the likelihood of RC accompanied by non-hospital death. Conversely, having more chronic conditions increased the likelihood of experiencing a HCC to hospital death rather than a HCC to non-hospital death, but decreased the likelihood of experiencing EOL trajectories that involved RC (including those involving HCC as well as RC and hospital and non-hospital deaths). These findings not only attest to the importance of health factors in influencing EOL care trajectories but also link functional, mental health (cognitive impairment, depression), and behavioral concerns to EOL trajectories centered around RC.

Overall, our findings indicate several relatively distinct patterns with regard to various EOL care trajectories. For example, recipients of HCC EOL trajectories were likely to be rural residents, slightly younger than those who died in RC settings, to have low income levels, to be married, and to have more chronic conditions but lower levels of ADL and cognitive impairment, depression, and problematic behaviors. Among older HCC recipients, those most likely to experience a hospital death were men, those with lower incomes and poorer physical health status (based on chronic conditions and CHESS scores) but better ADL and cognitive functioning, and no hospitalizations within the 90 days prior to death. Conversely, those more likely to experience EOL trajectories involving RC (whether or not these also involved HCC and whether or not they ended in a hospital or non-hospital death) were more likely to be urban residents, unmarried, not poor, and to have fewer chronic conditions but greater ADL and cognitive impairment as well as higher levels of depression and problematic behavior. There were relatively few differences evident when comparing the determinants of EOL care trajectories among RC residents who entered RC care directly to those who moved from HCC to RC. As well, relatively few differences emerged when comparing RC residents who experienced a hospital or non-hospital death. RC residents who experienced a non-hospital death—most likely a death inside the RC facility—tended to be older, female, to be less likely to have lived alone prior to institutionalization, to have unknown income levels, and to have higher medical instability.

Several methodological issues should be considered when interpreting these findings. First, we focused only on the EOL trajectories evident among older adults transitioning through the publicly subsidized Canadian long-term care system in one specific health region. Thus, our results do not speak to the EOL trajectories experienced by the majority of older adults who do not make use of such services. Experiences will necessarily differ among those relying exclusively on privately provided (paid, unpaid) services or living in locales with different policies and service options. In addition, we focused on LTC trajectories that ended in death over a recent 4-year period for which there were complete assessment and client data. These trajectories might look different if assessed over a different period. As well, hospitalizations were included as a predictor and as an outcome (place of death) but not as a component of the trajectories themselves. Finally, although we included changes in functional impairment and cognitive status in our analyses, some factors could be assessed at baseline only (e.g., marital status, income) or were unavailable for all or part of the sample (e.g., informal caregivers). Further, we did not focus on the role of specific diseases or causes of death within our analyses.

These limitations point to a need for further research. Yet, our findings have implications for theory and research as well as for policy and the delivery of EOL and palliative care services. They point to the presence of several distinct but overlapping LTC trajectories at the EOL and also point to the utility of a structural/institutional LCP for understanding these trajectories and the factors that influence them at the individual level. However, they also suggest a need to address the impact of intersecting inequalities (e.g., age, gender, geography) as well as to focus attention further upstream on the role of macro- and meso-level factors. These include whether and how health care policy and service delivery contexts differentially structure EOL and LTC trajectories on the basis of social location (e.g., through policies and procedures that direct older women, those with dementia, and so forth to RC facilities and non-hospital deaths while directing rural residents to HCC and older men to hospital deaths) as well as access to social, economic, and health resources. Findings indicating that advanced age and other factors also structure opportunities for ensuring quality of care while dying suggest avenues for enhancing equitable access to EOL care. Finally, evidence pointing to the fact that most of the older adults we studied died outside of hospital settings, typically in RC, also points to the need to ensure access to quality EOL and palliative care services within such settings. As the population ages, the number of deaths taking place within these places can be expected to increase substantially, making this an important and time-sensitive goal.

Funding

This work was supported by grants from the Canadian Institutes for Health Research (CIHR): Partnerships in Health System Improvement (PHSI) Grant Program; the Michael Smith Foundation for Health Research (MSFHR) (to M. Penning, D. Cloutier, K. Nuernberger, and D. Taylor, 2012–2016, grant number 122184); and a University of Victoria Internal Research/Creative Project Grant (IRCPG) (to M. Penning, 2016–2017).

Conflict of Interest

None reported.