Full text of DOI:10.1093/geront/41.2.153

The Gerontologist
Vol. 41, No. 2, 153–160
Copyright 2001 by The Gerontological Society of America
Hospitalization Patterns and Palliation in
the Last Year of Life Among Residents in
Long-Term Care
Shirley S. Travis, PhD, RN, CS,1 Gary Loving, PhD, RN,2
Lue McClanahan, BS, RN,3 and Marie Bernard, MD4
care needs resulted in the conceptualization of a liv-
ing–dying trajectory that included intervals for liv-
ing, living–dying, and dying and death (Pattison,
1977). More recent applications of this continuum to
long-term care (LTC) include the suggestion that
nearly all permanently placed nursing facility residents
enter care in the chronic living–dying interval (Engle,
1998; Lawhorne, 1999; Pattison, 1977). Practitioners
and organizations that care for these residents face
multiple decision points during the residents’ long-
term care. At these times, an assessment of benefits
over harmful effects (beneficence) should guide treat-
ment decisions (Rhymes & McCullough, 1994) that
are based on an understanding of the resident’s prog-
nosis and placement on the living–dying continuum.
In this article we report: (a) hospitalization preva-
lence and incidence rates, (b) differences in the timing
and duration of palliative care modalities for nursing
facility residents who were hospitalized versus those
who were not, and (c) aspects of caregiver coalitions
that appear to distinguish between residents who
were hospitalized and those who were not.
Purpose: This study compared patterns of care, including
hospitalization, during the last year of life for a group of res-
idents in institutional long-term care. These subjects were ei-
ther implicitly or explicitly in palliative care modes versus
those who remained in active treatment or blended care.
Design and Methods: The study used a retrospective chart
review and both quantitative and qualitative methods of
data collection and analysis to examine indepth the end-of-
life experiences of 41 nursing home residents who died in
the nursing care unit of one large continuing care retirement
community during an 18-month period. Results: Most resi-
dents die in palliative care modes, but their movement into
palliation with comfort care and symptom management is
often slowed by indecision or inaction on the part of key de-
cision makers, interrupted by aggressive acute care, or
delayed until the last few days of life. Implications: Transi-
tions from active curative care to palliative care are impor-
tant for residents in permanent long-term care placements.
Improved end-of-life care requires more attention to these
transitions and to the decisions that residents, their families,
and care teams are called upon to make.
Key Words: Advance directives, Caregiver coalitions, End-
of-life care
Review of the Literature
Palliation is the term now used to describe care
that is no longer aimed at cure or active treatment of
medical conditions. Instead, the goals of palliation
are resident comfort and symptom management (Mann
& Welk, 1997; Post & Dubler, 1997). For some
LTC residents with chronic illness, blended care (ac-
tive treatment with palliation) may last for years
with relatively short periods dedicated solely to palli-
ation. For others, such as those with Alzheimer’s dis-
ease, it has been argued that palliative care should be
the dominant mode of care for the long term (So-
lomon & Jennings, 1998). When death is imminent,
generally in the last 6 months of life, the specialized
knowledge of the hospice team, now conceptualized
as a component of palliative care, is used for the final
transition from dying to death (Mann & Welk, 1997).
One of the more critical junctures in the care of an
LTC resident, in terms of both resource utilization
In the mid 1970s, efforts to understand transitions
in care that would be required for growing numbers
of individuals surviving to advanced ages with chronic
The authors thank Leslie Hussey, Gloria Hagopian, Jim McAuley, and
Susan Sherman for their helpful critiques of earlier drafts of the
manuscript.
Address correspondence to Dr. Shirley S. Travis, College of Nursing
and Health Professions, University of North Carolina at Charlotte, 9201
University City Boulevard, Charlotte, NC 28223. E-mail: sstravis@
email.uncc.edu
1College of Nursing and Health Professions, University of North Caro-
lina at Charlotte.
2University of Oklahoma College of Nursing, Oklahoma City.
3Susanna Wesley Center at Epworth Villa Lifecare Community, Okla-
homa City, OK.
4Donald W. Reynolds Department of Geriatrics, University of Okla-
homa College of Medicine, Oklahoma City.
Vol. 41, No. 2, 2001
153

and quality of life, is when the need arises to make a
decision about whether or not to hospitalize the indi-
vidual. From the increasing amount of information
available about these decision points, it is frequently
argued that treatment options should be based on
where the key decision makers (resident, family, phy-
sician, and staff) either explicitly or implicitly place
the resident on a living-dying trajectory, such as the
one in Figure 1. In other words, more than just the
immediate clinical condition of the resident guides
interpretations of what comprises “good” or “harm-
ful,” appropriate or inappropriate treatment options.
Figure 1 is a schematic representation of this liv-
ing–dying trajectory with an additional refinement
that adds treatment modalities, including palliative
and hospice care, to the righthand side of the figure
across from the corresponding intervals in the trajec-
tory. This updated schematic was used to guide our
conceptualization of data collection procedures, se-
lection of data analysis techniques, and the interpre-
tation of the results of a project that studied the last
year of life for a group of residents in institutional
long-term care.
As depicted in Figure 1, acute care is compatible
with a need for blended care early in the living–dying
interval and may include hospitalization or other ag-
gressive care. As the individual’s condition progresses
toward the dying interval, transitional care is used to
move from curative care treatment to a plan of palli-
ation for comfort and symptom management. The
living–dying interval ends with a commitment to and
a plan for palliation with symptom management.
Therefore, high rates of hospitalization among resi-
dents who are established in palliation and/or hos-
pice treatment modalities would represent potential
inconsistencies between actions and prevailing rec-
ommendations about palliation and LTC. It should
be noted that at no time is “no care” an acceptable
standard of care in this schematic. Rather, the mode
of care changes as the resident’s condition changes
from living through the process of actively dying, in
which body functions are beginning to decline and
ultimately shut down as death approaches.
When frameworks such as the one shown in Fig-
ure 1 are well understood by treatment teams and
family members, dramatic reductions in hospital
deaths and enhanced end-of-life care in the nursing
facility have been reported (Ackermann & Kemle,
1999). Unfortunately, neither the term palliative care
nor the associated approaches to care have been well
integrated into most LTC facilities. Indeed, from our
review of more than 225 journal articles and book
chapters, the barriers to fully implementing high-
quality, comprehensive palliation in LTC are better
documented in the literature than are reports of suc-
cessful end-of-life programs.
Among other things, the current regulatory cli-
mate in LTC settings causes some providers to fear
that they will not be seen by nursing home surveyors
as sufficiently aggressive in their care if they allow a
resident to decline without intervention (Meier &
Morrison, 1999). Prognostications and end-of-life
staging continue to be difficult and inexact, espe-
cially for noncancerous, LTC residents whose pro-
gression from curative care to palliation requires a
blended type of care for some period of time (Blatt,
Figure 1. Living–dying care in long-term care.
154
The Gerontologist

1999; Casarett & Karlawish, 1999; Pendley & White,
1999; Smith, 1998; Zoloth-Dorfman, 1998). Further
confounding the problem is failure of care teams, es-
pecially physician members, to know and to docu-
ment the wishes of residents and/or responsible deci-
sion makers, and to make the mutually agreed upon
goals of treatment clear in the written plans of care
(Lawhorne, 1999; Levin et al., 1999). In addition,
some practitioners worry that offering palliative care
to a resident and the family may be misinterpreted as
“no care” and are hesitant to discuss transitions
from curative care to palliative care with a focus on
symptom management (Mann & Welk, 1997). Fi-
nally, families and their residents are often reluctant
to have frank and open discussions of treatment pref-
erences in advance of the need for action. Even when
discussions do occur, an unwillingness by one or
more parties to accept palliation as the treatment
mode can abort even the most well intended efforts
to provide high-quality end-of-life experiences (Dubler,
1995; Roberto, 1999; vonGuten & Twaddle, 1996).
Of course, physicians, other members of the care
planning teams, families, and the residents themselves
all make judgments about the degree of futility associ-
ated with curative care regardless of whether a pallia-
tive care philosophy can be clearly articulated or the
term “palliative care” is even well understood. On
any given day these judgments may be of little conse-
quence to routine care plan decisions. But when criti-
cal events occur, the ways in which key decision mak-
ers have come to think about the futility of curative
care, assessed beneficence, and placed the resident on
the living–dying trajectory (as depicted in Figure 1)
influence the decisions that are made (McCue, 1995;
Zuckerman, 1999). If there is inconsistency or disso-
nance, these situations need to be better understood.
life in the context of the entire institutional histories,
the complete medical record was used to review each
resident’s entire length of stay. Institutional Review
Board approval for the study was obtained from the
University of Oklahoma Health Sciences Center.
Setting and Sample
The cooperating organization was a private, non-
profit continuing care retirement community (CCRC)
in the southwestern United States. The community
consisted of 20 duplexes, 209 apartments for indepen-
dent living, a 45-bed assisted living center, and a 60-
bed nursing facility, which was the focus of this inves-
tigation. Most of the residents in the nursing facility
were life-care residents of the retirement community.
Occasionally, residents were admitted from outside
the CCRC on a private-pay basis. Medicare and Med-
icaid reimbursement were not accepted by the facility.
The sample consisted of 41 nursing home residents
who died during the 18-month period January 1,
1998, to June 30, 1999. Deceased residents ranged in
age from 76 to 102 (mean ϭ 87). Consistent with
other institutionalized populations, they were predom-
inately female (85.3%) and widowed (85.4%). Fully
61% had a diagnosis of dementia on their medical
records. The residents in the sample were in care on the
nursing unit from as little as 2 days to as many as
3,114 days until death occurred either in the facility or
in a hospital (mean ϭ 704, median ϭ 390). Most of
the residents (90.2%) had advance directives on the
chart. The person to be called in case of emergency
was usually an adult daughter or son (75.6%). In all
but four cases, these same individuals were listed as the
responsible party for the residents’ care and also held
Durable Power of Attorney for decisions regarding the
residents’ health care. Eighty-five percent of these re-
sponsible parties lived in state. Four residents were
listed as their own responsible party and were consid-
ered capable of making decisions for themselves.
Methods
Design
A retrospective study design that included medical
chart reviews and the creation of case summaries for
a group of recently deceased residents was used. In
order to place the events of the residents’ last years of
Data Collection
Both quantitative (Phase 1) and qualitative (Phase
2) methods of data collection were used. Table 1 dis-
Table 1. Description of the Data Collected Using Quantitative and Qualitative Approaches
Approach
Method
Phase 1 – Quantitative Data
Step 1 – Pretest of instrument using current residents’ records
Step 2 – Medical record reviews of deceased residents to collect information about:
• Demography • Medical conditions • Hospitalization
• Functional status • Medication histories • Palliative care designation
Phase 2 – Qualitative Data
Step 1 – First reading of all notes in the medical record to create a picture of the:
• General course of care and critical events in the last year of life
• Timing of advance directives and individuals involved in decision making
• Trends in the resident’s decline over time
Step 2 – Second reading of notes around critical events to determine:
• Awareness of the event by team members
• Conflicting perspectives of the event
• Care during the final critical event, the death experience
Step 3 – Data organized into case summary narratives for presentation to the Care Planning Team
Vol. 41, No. 2, 2001
155

plays the phases and steps included in the data col-
lection procedures. The initial data collection process
took between 45 minutes and 3.5 hours for each
case, which was a reflection of the range in lengths of
stay and number of hospitalizations.
team members read each summary one at a time,
commented on any relevant memories they had of
the resident’s care, and verified the accuracy of the
summaries.
Phase 1 consisted of two steps. First, the data re-
trieval tool was pretested using the medical records
of three residents who were currently in care. Next,
quantitative data were collected from the records of
the deceased residents for the last year of life or the
entire length of time in the nursing home, if less than
one year. The information included the resident’s de-
mographic characteristics, functional status, medical
conditions, medication histories, and hospitalization
data for each hospitalization.
Phase 2 consisted of three steps. First, the full set
of “notes” was reviewed to develop an overview of
the resident’s general course of care and to identify
critical events, defined as acute changes in the resi-
dent’s condition or circumstances that required changes
in the plans of care. Elements of the course of care
that were noted included: (1) documented exchanges
among residents, their family members, the resident’s
physician, and the staff regarding care issues (person-
alized care, quality of care, agreement about specific
treatment decisions); (2) the timing of advance direc-
tives and individuals cited in the medical records dur-
ing any decision-making times; and (3) trends in the
resident’s decline over time (e.g., spontaneous, rapid
or progressive, linear or stair-step). Critical events in-
cluded, but were not limited to, changes in the resi-
dent’s medical condition. For example, the death of a
spouse who was the resident’s responsible party and
the need for an adult child to become the primary de-
cision maker, as noted in the social service notes, was
recorded as a critical event.
In the second step of Phase 2, the date of entry for
each critical event was used to guide a second read-
ing of notes that were written before and after the
critical event in order to gain the perspectives of all
members of the care team. Questions guiding this
reading included the following: Did other team mem-
bers seem to be aware of the event? Was it an event
that affected the care provided by other team mem-
bers? Were there any new or conflicting perspectives
on the event from other team members? The resi-
dent’s final critical event, the death experience, was
reviewed in detail. Specific attention was given to ev-
idence of symptom management, inclusion of end-of-
life issues in the interdisciplinary plan of care, the
level of involvement of family members and profes-
sional caregivers at the end of life, and conflicted de-
cision making (parties who did not agree with or ex-
pressed concern about decisions) that may have taken
place around the time of death.
Operationalizing Palliative Care
Because “palliative care” is a term that is not in
common use and specific criteria for operationalizing
the variable for LTC populations do not exist, a
method for determining whether or not a resident
was either explicitly or implicitly in a palliative care
mode had to be created from existing chart informa-
tion. We selected criteria that could justify limiting
aggressive curative care, such as evidence of func-
tional decline with no rehabilitation potential, and
would support decisions for comfort care and symp-
tom management. In addition, a “do not resuscitate”
(DNR) order was deemed an essential criterion be-
cause without such an order, aggressive treatment
(cardiopulmonary resuscitation, activation of 911,
and transfer to the hospital) would have been re-
quired regardless of any other implicit or explicit evi-
dence of the futility of such intervention that we
might have found in the medical record.
Briefly, a resident was classified in palliative care
if he/she had a DNR order on the chart and at least
one of the following entered in the medical record in
the 12 months preceding death: (a) a medical order
for hospice (explicit palliative care), (b) a physician’s
note stating that the resident and/or responsible party
understands that cure is not possible and requests
“comfort care” (explicit palliative care), (c) evidence
in the interdisciplinary plan of care that the resident’s
prognosis was poor and recovery was not an ex-
pected outcome (implicit palliative care), or (d) a
progress note by the physician and/or an occupa-
tional or physical therapist stating that the resident’s
rehabilitation potential was “poor” (implicit pallia-
tive care).
Because the exact wording of entries in the medi-
cal records tended to deviate slightly from the lan-
guage in the operational definition, the investigators
performed a coding reliability check for each case.
For this step, the first two authors reviewed the ver-
batim documentation retrieved from the medical
records and independently determined if the case met
the criteria for palliative care. They agreed with each
other’s coding decisions 100% of the time.
Analysis
Quantitative data analysis initially consisted of de-
scriptive statistics to summarize the data. Once the
reliability of the case summaries was established,
content analysis of the case summaries enabled us to
explore possible inconsistencies between treatment
modalities and care decisions. Because we were inter-
ested in who was involved with care decisions, espe-
cially decisions to hospitalize, part of the analyses of
these data looked at the concept of caregiver coali-
tions in long-term care. The term “caregiver coali-
Phase 2 concluded with the development of a writ-
ten case summary from the Phase 2 information that
represented the context of the resident’s end-of-life
story. These written summaries were presented to the
nursing facility’s interdisciplinary team for review.
Time was set aside during the regular weekly meet-
ings of the team for this purpose. As a group, the
156
The Gerontologist

tions” has been used to describe the shift from a tra-
ditional dyadic relationship of client–provider to a
new triad of client–provider–family that is needed as
individuals become increasingly dependent (Caplow,
1968; Coe & Pendergast, 1985). This metaparadigm,
or global framework (Fawcett, 1984), was used to
help the team focus on who was communicating with
whom about care decisions in the last year of life.
Access to both quantitative and qualitative data about
the residents was important because it allowed the
team to move back and forth from inductive to de-
ductive reasoning and to test hunches as we worked
with the data. A strategy that we labeled “concept
tables” was used to display the concepts of interest
(conflicted vs nonconflicted hospital admissions, types
of hospital admissions, and caregiver coalitions) with
selected resident characteristics. Chi-square tests of
independence and t tests were used to test “hunches”
about possible associations among the variables, as
suggested by the existing literature and our work
with the qualitative data.
tional definition for palliative care within the last 12
months of life. These residents died in palliative care
modes ranging in length of time from 2 days to 284
days. About one third of the hospitalized residents
(37%, n ϭ 7) were in palliative care modes at the
time a decision to hospitalize was made. Forty-two
percent of the hospitalized residents entered pallia-
tive care modes following their hospital discharges.
The hospitalization experience was an important
trigger for hospice care. About two thirds (63%, n ϭ
12) of the hospitalized residents were discharged
with orders for hospice care or were approved for
hospice care within 6 months of their last hospital-
izations. For one half of this group of residents, the
hospice order was the first evidence of a shift to pal-
liative care in the year preceding death. Clearly, a
large percentage of residents from LTC were being
provided aggressive hospital care. Returning to our
initial conceptualization of a living–dying continuum
for permanently placed residents in long-term care, it
seemed possible that at least some of these cases
might represent the cognitive dissonance currently
surrounding palliative care and end-of-life care for
LTC residents (Zuckerman, 1999). In an effort to
better understand the hospitalization patterns identi-
fied in the sample, we turned to the case summaries.
Results
Hospitalization Prevalence and Incidence Rates
Slightly less than half of the deceased residents
(46.3%, n ϭ 19) were admitted to the hospital, for a
total of 27 separate admissions, during their last 12
months of life. Two thirds (68%, n ϭ 13) of these
residents experienced one admission, 21% (n ϭ 4)
were admitted twice, and 11% (n ϭ 2) had three ad-
missions during the 12 months preceding death. In
addition to this inpatient care, 26.8% of the total
sample (n ϭ 11) had between one and three visits to
the emergency room in the last 12 months of life that
did not result in hospitalization. Falls with minor in-
jury (lacerations, rib fractures) were the primary rea-
son for these outpatient contacts.
The reasons for the hospital admissions included
three broad categories: accidental injury (e.g., frac-
tured hip; 15%, 4 admissions), acute conditions (52%,
14 admissions), and conditions related to end-stage
chronic disease (33%, 9 admissions). Over half of
the admissions for acute conditions were pneumonia-
related. Cardiopulmonary failure represented the
vast majority (66%) of the chronic care admissions.
There were no statistically significant differences in
age or weight loss, a classic clinical indicator of ter-
minal decline, between residents who were hospital-
ized and those who were not. In order to see if hav-
ing a diagnosis of dementia, implying reliance on
others for care decisions, made any difference in hos-
pitalization patterns, we compared the number of
hospital stays for the two groups of residents. The
demented residents had somewhat higher rates of
hospitalization in the last year of life than their non-
demented counterparts (t test ϭ Ϫ1.53, p ϭ .067).
The Decision To Hospitalize
A defining feature of effective caregiver triads is
open communication among all members. Therefore,
our analysis focused on the decisions, actions, and in-
teractions of the members of the triad in order to un-
derstand decisions to hospitalize. After reviewing all
summaries, it became clear that the hospitalized cases
fell into two distinct groups. One group was labeled
“nonconflicted” admissions and the other group as
“conflicted” admissions. Among the nonconflicted
admissions, we found no evidence of disagreement
about the admission decision among any of the coali-
tion members. The physician ordered the hospital ad-
mission for either accidental injury or acute illness sit-
uations (usually respiratory infection), and there was
no evidence of contradictory wishes by the residents,
family, or other responsible parties. These admissions
represented about 47% of the hospital cases (n ϭ 9).
We were surprised to find that 55% (n ϭ 5) of the
residents experiencing nonconflicted admissions fit
the palliative care definition at the time of hospital
admission. Their reasons for admissions represented
a mix of accidental and acute conditions, both of which
could justify hospitalization. In addition, based on
the dates at which each resident met the palliative
care definition, and was hospitalized, we determined
that most of these residents had recently segued to
palliative care (6 months or less). These residents
may still have been in blended care modes, a finding
that offers preliminary validation for the inclusion of
blended care in the schematic. The extent to which
the long-term care environment may have had limits
on the availability of certain types of care that might
have been used to provide for adequate symptom
Palliative Care Modalities
For the total sample, the overwhelming majority
of the residents (87%, n ϭ 35) satisfied the opera-
Vol. 41, No. 2, 2001
157

management in lieu of a hospital stay could not be
determined from the chart reviews.
record from family members that stipulated “no
transfer” orders. The care planning team documented
the presence of family members at care planning con-
ferences and their explicit wishes for no hospital care.
Physicians’ notes tended to detail conversations with
family members and the resident, when possible,
about the course of care that would be followed, even
for those residents who appeared to be in blended or
transitional care. As might be expected, 81.8% of the
nonhospitalized residents eventually died in hospice
care compared to 58% of the hospitalized residents.
In summary, the hallmarks of care among the non-
hospitalized cases were: open communication between
all decision makers (family, physician, staff, and resi-
dent when possible); knowledge of the residents’
wishes by a responsible party; physician and staff
commitment to following the wishes of the residents
and families; explicit notes in the medical records
about care decisions; and a commitment to stay the
course of care. These findings were combined with the
caregiving metaparadigm and prior research, espe-
cially the insights of Ackermann & Kemle (1999), to
create the preliminary conceptual model for providing
end-of-life care in LTC settings (see Figure 2).
The solid arrows in the middle of the Figure 2 rep-
resent the open communication that typifies the basic
caregiver coalition in LTC. Ideally, all members of
the coalition are active participants in the decision-
making process. However, as our results demon-
strated, there are certain tasks that are largely the re-
sponsibility of one member of the coalition and that
most directly affect the behavior of another member
of the coalition as the need for managing end-of-life
care approaches. Therefore, while the center arrows
represent simultaneous interaction among the triad
members, the outside arrows and the associated call-
out boxes are intended to represent directional pat-
terns of responsibility for effective end-of-life care.
Recently, it was reported that nearly one half of
nursing home transfers to hospitals may be inappro-
priate, based on interpretations by expert physician
panels regarding the chief complaint on admission
and the clinical care that was required in combina-
tion with the presence of an advance directive (Saliba
et al., 2000). We believe that our present study and
this earlier work may have also identified a subset of
LTC residents whose hospitalizations may have been
avoided. However, in our study we base that deter-
mination on a belief that one or more of the critical
elements in Figure 2 may have been lacking. Further
research on the appropriateness of nursing home
transfers that combine clinical indicators (e.g., Saliba
et al., 2000), location on the living–dying trajectory,
and the responsibility assigned to each member of
the triad in Figure 2 may improve our understanding
of why some LTC residents are hospitalized and oth-
ers in similar circumstances are not.
The remaining 53% (n ϭ 10) of the hospitalized
cases included three distinct groups of conflicted
cases, all of which stemmed from the actions or inac-
tions of the individual members of the caregiver coa-
lition. Eighty percent (n ϭ 8) of the conflicted cases
were not in palliative care modes at the time of hos-
pitalization, although all but one of the cases were in
palliative care at the time of death. Ninety percent of
the conflicted cases died within 6 months of their last
hospital stay. Three scenarios summarize the con-
flicted cases.
In the first scenario, representing 30% (n ϭ 3) of
the conflicted cases, the resident’s primary care physi-
cian was out of town or not on call when the resi-
dent’s condition became unstable and the decision to
hospitalize had to be made. In the absence of explicit
written instructions on the chart about hospital care
or the general goals of care, the physician or physi-
cian’s assistant on call opted to send the resident to
the hospital. Thus, the physician/LTC staff comprised
the weak link at a critical decision-making juncture.
In the second scenario, consisting of 30% (n ϭ 3)
of the conflicted cases, a nuclear family member
(most often an adult child) was the resident’s respon-
sible party. For reasons that were not apparent in the
medical record, this person was not able to agree to
or stay on a palliative care course, even when it was
clear from the functional status of the resident that
his or her condition was deteriorating. In each case,
when the resident’s condition became unstable, the
responsible party was unable to forgo aggressive cur-
ative care and opted for hospitalization.
The final set of conflicted cases (40%, n ϭ 4) was
represented by a scenario in which the responsible
party who held power of attorney was either not a
family member (friend or attorney) or was a member
of the extended family (niece or nephew). There was
very little, if any, documentation in the medical
record (social services and physician’s progress notes)
about the resident’s wishes for care. Despite dire
medical conditions, none of these residents met the
palliative care operational definition because their re-
sponsible parties consistently chose to “do every-
thing possible.” It was also the case that this group
of residents appeared to be the most vulnerable for
tumultuous end-of-life experiences. Two of these
cases had multiple hospital stays the last year of life
and two of the residents died without benefit of hos-
pice care, family, or friends at their bedside. All of
the residents in this subgroup died in the hospital or
within one month of discharge.
When Hospitalization Is Not the Choice
Our analysis then turned to a review of the non-
hospitalized residents’ case summaries. The most out-
standing contrast between hospitalized and nonhospi-
talized residents was the documentation on the
medical records clearly stating the wishes of the resi-
dent and/or the family for end-of-life care and hospi-
tal decisions. Letters were often found in the medical
Discussion
As LTC residents move through living–dying in-
tervals, the current standards by which their care is
158
The Gerontologist

Figure 2. Conceptual model for end-of-life care in institutional long-term care.
evaluated have often resulted in murky decision making
regarding such issues as aggressive acute care, transi-
tions to palliative care, and the timing of hospice or-
ders. A great deal of prior research has focused on
clinical indicators to explain decisions to hospitalize
LTC residents. As a result, authors frequently discuss
the need for enhanced medical monitoring and su-
pervision in LTC facilities. In large part these studies
have ignored the need to identify residents’ place-
ments on a living–dying trajectory, opting instead to
apply active treatment standards and curative goals
to the residents’ plans of care.
In this study we chose to search for implicit or ex-
plicit evidence that key decision makers and influen-
tial others had placed residents in palliative modes of
care. We then examined their course of care in the
last year of life from this perspective, searched for in-
consistencies between the modes of care and subse-
quent hospitalization decisions, and sought to untan-
gle some of the contextual variables that supported
or interfered with care decisions and the provision of
end-of-life care.
(Happ et al., 1999; Wilson & Daley, 1999). Unfortu-
nately, when hospitalization is the trigger event for
end-of-life care planning, little time is left for appro-
priate end-of-life preparation and palliative care.
From these results it is apparent that many LTC
residents need more clearly defined palliative plans in
advance of the need to decide about hospital care. If,
as we have suggested, hospitalization and subsequent
orders for hospice care fit the concept of medical last
rites, then one needed change is to find better ways
of helping families understand the futility of treat-
ment and to recognize a need for a shift to palliative
goals of care. To this end, the model in Figure 2 is of-
fered as a springboard for additional research on ef-
fective programs of end-of-life care. As research pro-
grams continue to mature, it is likely that other
elements will be added and that cause-and-effect re-
lationships will be clarified. For example, the influ-
ence of prevailing public policy and related reim-
bursement issues pertaining to end-of-life care in
LTC settings (Petrisek & Mor, 1999) were not im-
mediately apparent in the present study of private-
pay, life-care retirement community residents of high
socioeconomic circumstances. Both the generalizabil-
ity of the study results and the applicability of the
model across long-term care populations are affected
by our small sample size and the study setting.
Recently, the issuing of DNR orders 24 hours or
more after hospital admission has been called “medi-
cal last rites” (DeJonge, Sulmasy, & Gold, 1999).
Hospitalization may represent a similar message of
futility and finality to families of LTC residents who
are known to have tendencies toward delaying end-
of-life discussions and agreement on appropriate care
We close with acknowledgment of our rather
crude operational definition of palliative care and a
Vol. 41, No. 2, 2001
159

ing treatment decisions for nursing home residents: Who discusses,
who decides and what is decided? Journal of the American Geriatrics
Society, 47, 82–87.
call for greater attention to ways in which clinicians
and researchers can measure this important concept
of care. Prognosticating an individual’s ability to sur-
vive 6 months or less for purposes of implementing a
hospice plan is widely recognized as a very difficult
task. But until diagnosticians are able and willing to
make these determinations, other members of the
care planning teams, family members, and the care
recipients will invariably struggle to find direction
and make informed decisions about end-of-life care.
A significant area for future research will be better
delineation of inevitable transitions in care among
residents in LTC and the role of palliation treatment
goals and hospice care during the final transition to
death.
Mann, S. M., & Welk, T. A. (1997). Hospice and/or palliative care?
American Journal of Hospice and Palliative Care, 14, 314–315.
McCue, J. D. (1995). The naturalness of dying. Journal of the American
Medical Association, 273, 1039–1043.
Meier, D. E., & Morrison, R. S. (1999). Old age and care near the end-of-
life. Generations, 23, 6–11.
Pattison, E. M. (1977). The experience of dying. Englewood Cliffs, NJ:
Prentice Hall.
Pendley, D., & White, R. (1999). Applying hospice principles in the pallia-
tive care debate. American Journal of Hospice and Palliative Care, 16,
412–415.
Petrisek, A. C., & Mor, V. (1999). Hospice in nursing homes: A facility-
level analysis of the distribution of hospice beneficiaries. The Geron-
tologist, 39, 279–290.
Post, L. F., & Dubler, N. N. (1997). Palliative care: A bioethical defini-
tion, principles, and clinical guidelines. Bioethics Forum, 13(3), 17–24.
Roberto, K. A. (1999). Making critical health care decisions for older
adults: Consensus among family members. Family Relations, 49, 167–
175.
REFERENCES
Rhymes, J. A., & McCullough, L. B. (1994). Nonaggressive management
of the illnesses of severely demented patients: An ethical justification.
Journal of the American Geriatrics Society, 42, 686–687.
Saliba, D., Kington, R., Buchanan, J., Bell, R., Wang, M., Lee, M., Herbst,
M., Lee, D., Sur, D., & Rubenstein, L. (2000). Appropriateness of the
decision to transfer nursing facility residents to the hospital. Journal of
the American Geriatrics Society, 48, 154–163.
Smith, S. A. (1998). Facing the challenge of the hospice and palliative care
marriage. American Journal of Hospice and Palliative Care, 15, 181–
183.
Solomon, M. Z., & Jennings, B. (1998). Palliative care for Alzheimer’s pa-
tients: Implications for institutions, caregivers, and families. In L.
Volicer, & A. Hurley (Eds.), Hospice care for patients with advanced
progressive dementia (pp. 132–154). New York: Springer.
vonGuten, C. F., & Twaddle, M. L. (1996). Terminal care for noncancer
patients. Clinics of Geriatric Medicine, 12, 349–358.
Wilson, S. A., & Daley, B. J. (1999). Family perspectives on dying in long-
term care settings. Journal of Gerontological Nursing, 25(11), 19–25.
Zoloth-Dorfman, L. (1998). Ethical considerations at the end of life: The
physician and the inevitably dying patient. Annals of Long-Term Care,
6, 394–400.
Ackermann, R. J., & Kemle, K. A. (1999). Death in a nursing home with
active medical management. Annals of Long Term Care, 7, 313–319.
Blatt, L. (1999). Working with families in reaching end-of-life decisions.
Clinical Nurse Specialist, 13, 219–223.
Caplow, T. (1968). Two against one: Coalitions in triads. Englewood
Cliffs, NJ: Prentice-Hall.
Casarett, D. A., & Karlawish, J. (1999). Working in the dark: The state of
palliative care for patients with severe dementia. Generations, 23(1),
18–23.
Coe, R., & Pendergast, C. (1985). The formation of coalitions: Interaction
strategies in triads. Sociology of Health and Illness, 7, 236–247.
DeJonge, E., Sulmasy, D. P., & Gold, K. G. (1999). The timing of do-not-
resuscitate orders and hospital costs. Journal of General Internal Med-
icine, 14, 1990–1992.
Dubler, N. N. (1995). The doctor–proxy relationship: The neglected con-
nection. Kennedy Institute of Ethics Journal, 5, 289–306.
Engle, V. F. (1998). Care of the living, care of the dying: Reconceptualiz-
ing nursing home care. Journal of the American Geriatrics Society, 46,
1172–1174.
Fawcett, J. (1984). Analysis and evaluation of conceptual models of nurs-
ing. Philadelphia: F. A. Davis.
Happ, M. B., Capezuti, E., Strumpf, N., Wagner, L., Evans, L., & Maislin,
G. (1999). Advance care planning and end-of-life care for nursing
home residents. The Gerontologist, 39 (Special Issue), 60.
Lawhorne, L. W. (1999). Avoidable and unavoidable decline and the nat-
uralness of dying: The nursing home dilemma. Annals of Long-Term
Care, 7, 309–312.
Zuckerman, C. (1999). Looking beyond the law to improve end-of-life
care. Generations, 23(1), 30–35.
Received May 22, 2000
Levin, J. R., Wenger, N. S., Ouslander, J. G., Zellman, G., Schnelle, J. F.,
Buchanan, J. L., Hirsch, S. H., & Reuben, D. B. (1999). Life-sustain-
Accepted November 30, 2000
Decision Editor: Laurence G. Branch, PhD
Correction
In an article on retirement preparedness that appeared in the December 2000 issue (vol. 40, no. 6), Her-
shey and Mowen reported a series of correlations in Table 1 (page 692). Due to a transcription error by
the authors, incorrect valences were reported for 14 of the values contained in that table. The actual va-
lences for all values in column three and row three of the table (involving the variable “emotional stabil-
ity”) should be reversed; positive values are actually negative, and negative values are actually positive.
This error did not affect either the structural modeling analyses the authors describe, nor the conclusions
they draw in the text.
160
The Gerontologist