Analysis of long-stay patients in the Hospice Palliative Ward of a Medical Center.

Abstract

The Pilot Project on Per-diem Payment for Inpatient Hospice Services of Taiwan's National Health Insurance Program was begun in July 2000. The project monitors hospices to control for a median length of stay (LOS) of not longer than 16 days to prevent inappropriate stasis in hospices. To determine the best utilization of palliative care, patients remaining in the hospice for more than 28 days were analyzed to discover their characteristics and reasons for not being discharged.
The study sample included 1,670 hospice patients who were admitted to the Hospice Palliative Unit in Taipei Veterans General Hospital between July 16, 1997 and December 31, 2002. Two hundred and sixty admissions (21.5%) with LOS > 28 days were identified. Further instrument survey of selected items was performed by 2 trained staff via chart review independently. The basic data were analyzed and comparison between long-stay patients and non-long-stay patients was made.
The mean LOS of 1,670 hospice patients was 16.0 +/- 14.9 days. Two hundred and sixty-eight patients (16.1%) admitted for longer than 28 days were surveyed. Those who had longer mean survival time, a diagnosis of prostate cancer, a metastatic site in the bone, and readmitted patients were associated with long stay. The study also revealed a significant difference in LOS between fee-for-service (FFS) patients and per-diem payment (PDP) patients (mean LOS, 17.5 +/- 16.4 vs. 14.3 +/- 13.4, p < 0.001). Conditions of major physical distress on Day 29 were delirium (41.9%), depression and/or anxiety (20.4%), and severe dyspnea (21.2%). The main reasons for being unable to be discharged on Day 29 after admission included "prolonged terminal phase" (34.2%), "difficult symptom control" (25.8%), "placement problem" (16.9%), and "need of parenteral medication" (15.0%).
Better understanding of the factors related to LOS can help staff in the palliative ward of medical centers to identify patients who are apt to have long stay, and shorten their LOS by successfully dealing with their problems.

Full text

Page 1

Introduction
One of the aims of palliative care services is to provide
emergency beds for patients with urgent symptom con-
trol needs. Due to the constant need of inpatient care,
few palliative care units can afford long-stay care. Long
stay in the hospice palliative unit of medical centers re-
sults in a low turnover rate of beds and thus decreases
the chances of admission for patients with more acute
symptom problems.1Factors associated with long stay
include the cancer diagnosis, referral source, gender,
physician specialty, type of insurance, living status,
and discharge status.2,3In Taiwan, the first medical
center hospice ward was established in 1990. However,
the growth rate of these centers has been very low.
This is partly because the insurance payment was too
small to cover the fee-for-service (FFS) base cost and
partly because of the misunderstandings of the med-
ical staff and public. The increasing incentive to pro-
vide more services for hospice patients generally comes
from higher payment for supportive care and symptom
alleviation services; thus, the Pilot Project on Per-diem
J Chin Med Assoc • June 2008 • Vol 71 • No 6
© 2008 Elsevier. All rights reserved.
294
ORIGINALARTICLE
Analysis of Long-stay Patients in the Hospice
Palliative Ward of a Medical Center
Ming-Hwai Lin, Pin-Yuan Wu, Tzeng-Ji Chen*, Shinn-Jang Hwang
Department of Family Medicine, Taipei Veterans General Hospital and National Yang-Ming
University School of Medicine, Taipei, Taiwan, R.O.C.
Background: The Pilot Project on Per-diem Payment for Inpatient Hospice Services of Taiwan’s National Health Insurance
Program was begun in July 2000. The project monitors hospices to control for a median length of stay (LOS) of not longer
than 16 days to prevent inappropriate stasis in hospices. To determine the best utilization of palliative care, patients
remaining in the hospice for more than 28 days were analyzed to discover their characteristics and reasons for not being
discharged.
Methods: The study sample included 1,670 hospice patients who were admitted to the Hospice Palliative Unit in Taipei
Veterans General Hospital between July 16, 1997 and December 31, 2002. Two hundred and sixty admissions (21.5%)
with LOS >28 days were identified. Further instrument survey of selected items was performed by 2 trained staff via
chart review independently. The basic data were analyzed and comparison between long-stay patients and non-long-stay
patients was made.
Results: The mean LOS of 1,670 hospice patients was 16.0±14.9 days. Two hundred and sixty-eight patients (16.1%)
admitted for longer than 28 days were surveyed. Those who had longer mean survival time, a diagnosis of prostate
cancer, a metastatic site in the bone, and readmitted patients were associated with long stay. The study also revealed a
significant difference in LOS between fee-for-service (FFS) patients and per-diem payment (PDP) patients (mean LOS,
17.5±16.4 vs. 14.3±13.4, p<0.001). Conditions of major physical distress on Day 29 were delirium (41.9%), depression
and/or anxiety (20.4%), and severe dyspnea (21.2%). The main reasons for being unable to be discharged on Day 29
after admission included “prolonged terminal phase” (34.2%), “difficult symptom control” (25.8%), “placement problem”
(16.9%), and “need of parenteral medication” (15.0%).
Conclusion: Better understanding of the factors related to LOS can help staff in the palliative ward of medical centers to
identify patients who are apt to have long stay, and shorten their LOS by successfully dealing with their problems. [J Chin
Med Assoc 2008;71(6):294–299]
Key Words: cancer, discharge planning, hospice palliative care, length of stay, per-diem payment
*Correspondence to: Dr Tzeng-Ji Chen, Department of Family Medicine, Taipei Veterans General
Hospital, 201, Section 2, Shih-Pai Road, Taipei 112, Taiwan, R.O.C.
E-mail: tjchen@vghtpe.gov.tw
● Received: June 4, 2007
● Accepted: April 2, 2008

Page 2

J Chin Med Assoc • June 2008 • Vol 71 • No 6
295
Long-stay patients in the hospice ward
Payment (PDP) for Inpatient Hospice Services of
Taiwan’s National Health Insurance Program was begun
in July 2000. To prevent hospice palliative wards from
earning more profit by increasing the length of hospice
stay (LOS), the project monitors hospices to control
for a median LOS of not longer than 16 days.
Ideally, patients who no longer fit the admission
criteria for acute inpatient palliative care should be
taken care of at home by family members. The transfer
of patients from acute palliative care services to chronic
institutions is necessary when the patient is relatively
stable, but does not have adequate support at home.
However, there are times when care at home or trans-
fer to other chronic institutions is not feasible either,
because of inadequate family support or because of the
complexities of symptom control in terminal cancer
patients. This study examined long stay in the hospice
palliative unit of a medical center in Taiwan and identi-
fied the possible factors and predictors related to pro-
longed stay. This study also compared the LOS between
FFS patients and PDP patients and found a number
of issues requiring further research.
Methods
Taipei Veterans General Hospital, a tertiary teaching
hospital with 2,865 inpatients beds, contains 15 beds
for hospice palliative care and has provided hospice
home care services since July 16, 1997. We carried out
a retrospective study in this hospital-based hospice pal-
liative care unit from July 16, 1997 to December 31,
2002. The basic data including demographic charac-
teristics, age, gender, diagnosis, discharge condition,
and LOS, which were generated from the hospice
database, were analyzed. For comparison with other
studies,2–4we chose 28 days as the cutoff point between
long-stay and non-long-stay cases.
Of the 1,670 terminally ill cancer patients (1,097
men, 573 women) admitted to the hospice palliative
care unit during this period, 268 (16.0%) with LOS
>28 days were identified; 260 long-stay admissions
(8 missing) were further surveyed using an instrument
that we developed by expert panel to determine the
factors associated with long-term hospice stay. The
instrument’s survey items included the prevalence of
major physical distress and the main reason for being
unable to be discharged on Day 29. The chart review
was conducted by 2 trained staff. The initial mean agree-
ment on 5 charts between chart reviewers was 61%. A
critical review of each discrepancy led to modification
of the instrument and clarification of the rules for cod-
ing. To test interrater reliability using the finalized survey
instrument, another 8 charts of the study sample were
examined independently by both chart reviewers. The
mean agreement on the survey items reached 88%.
Since the Bureau of National Health Insurance
changed the payment system from FFS to PDP for in-
patient hospice services on July 1, 2000, 826 (49.5%)
patients admitted after July 1, 2000 were classified in
the PDP group and compared with the conventional
FFS group.5
Data in the text and tables are expressed as mean±
standard deviation, median, and/or range. Descriptive
statistics were summarized as frequencies and per-
centages for categorical variables; median, mean and
standard deviation for continuous variables. We used
2-sample t tests to compare the age distributions and
LOS. We used Pearson’s χ2test to compare the sex,
diagnosis distributions, and other categorical variables
between different groups. For all tests, the difference
was considered statistically significant when p<0.05
(2-tailed). Statistical analyses were performed using
SPSS version 13.0 (SPSS Inc., Chicago, IL, USA) for
Windows.
Results
The mean age of the 1,670 terminally ill cancer patients
was 67.6±13.8 years (range, 13–99.6 years). The most
common diagnoses were lung cancer, colorectal can-
cer, hepatocellular carcinoma, and stomach cancer. The
most common sites of metastasis were bone (30.8%),
liver (27.8%), lung (22.2%), and brain (12.5%). The
mean duration from the diagnosis of malignancy to
hospice ward referral was 22.6±29.7 months (range,
0–325 months). Most patients (57.5%) were referred
from the ordinary ward in the same hospital, and 10.2%
of patients were from emergency room referral. Nearly
70% of cancer patients (69.4%) died in the hospice ward,
and 103 (6.2%) patients were discharged from the hos-
pice in a terminal state because they preferred to die
at home.6Three hundred and twenty-seven (19.5%)
patients were discharged to a home care program in
stable condition, and 81 (4.9%) patients were trans-
ferred to other institutions for chronic care. Readmission
to the hospice was noted in 263 (15.7%) admissions.
Mean survival after the first hospice enrolment was
32.6±68.1 days (range, 0–1,346 days).
Mean LOS in the hospice ward was 16.0±14.9 days
(range, 1–142 days); 268 (16.1%) patients admitted for
longer than 28 days were defined as long-stay patients
(165 men, 103 women; mean age, 68.3±13.4 years).
When compared with non-long-stay patients, long-
stay patients had a longer survival between the time of

Page 3

J Chin Med Assoc • June 2008 • Vol 71 • No 6
296
M.H. Lin, et al
cancer diagnosis and hospice referral, longer mean
survival time, were more frequently among the read-
mitted patients, less in the PDP group, and were more
likely to be discharged to a home care program. They
were significantly more likely to have bone metastasis,
less likely to have liver metastasis, and were more com-
monly prostate cancer patients (Table 1). However,
age, gender, rate of lung and brain metastases, and
other cancer diagnoses were not significantly different
between these 2 groups.
The comparisons of patient profiles between FFS
patients and PDP patients are shown in Table 2. Age,
gender, cancer diagnosis, time of cancer diagnosis and
hospice referral were not significantly different between
the 2 groups. Around 83.9% of the hospice patients
died within 28 days of admission. Because there were
many more short admissions (≤3 days’ admission) in
the PDP group than in the FFS group, LOS in the PDP
group was shorter than in the FFS group (mean LOS,
17.5±16.4 vs. 14.3±13.4, p<0.001). PDP group pa-
tients also had a shorter mean survival time, and were
more likely to be discharged to another institution.
Excluding the 8 sets of missing data, 260 admissions
of longer than 28 days were found and surveyed. The
main reasons for not being discharged on Day 29 after
admission included “prolonged terminal phase” in 89
(34.2%) patients, “difficult symptom control” in 67
(25.8%), “placement problem” in 44 (16.9%), “need
of parenteral medication” in 39 (15.0%) and “incom-
plete course of radiotherapy” in 21 (8.1%) (Table 3).
Among the 44 patients who had placement problems
and could not be discharged on Day 29, 16 chronic
institution transfers were made, 7 chose to go home
for home care program support, 3 moved to another
Table 1. Comparisons of patient profiles between long-stay patients and non-long-stay patients
Long-stay patients
(n=268)
Non-long-stay patients
(n=1,402)
p
Age (yr)68.3±13.467.4±13.9 0.325*
Gender (% male)61.666.50.121†
Time from diagnosis to hospice referral (mo) 27.6±35.821.5±28.40.023*
Mean survival (d) 63.7±51.626.7±69.3
<0.001*
Patient source, n (%)
Readmission after hospice discharge
0.037†
54 (20.1)209 (14.9)
Insurance payment, n (%)
Fee-for-service
Per-diem payment
<0.001†
163 (60.8)
105 (39.2)
681 (48.6)
721 (51.4)
Discharge status, n (%)
Death in hospice
Discharged with terminal status
Stably discharged with home care
Transferred to another institution
<0.001†
164 (61.2)
10 (3.7)
81 (30.2)
13 (4.8)
995 (71.0)
93 (6.6)
246 (17.6)
68 (4.8)
Diagnosis, n (%)
Lung cancer
Colorectal cancer
Hepatocellular carcinoma
Stomach cancer
Pancreatic cancer
Prostate cancer
Other
<0.001†
75 (28.0)
37 (13.8)
24 (9.0)
21 (7.8)
8 (3.0)
16 (6.0)
87 (32.4)
331 (23.6)
208 (14.8)
160 (11.4)
161 (11.5)
79 (5.6)
38 (2.7)
425 (30.4)
Metastasis site, n (%)
Bone
Liver
Lung
Brain
<0.001†
103 (38.4)
50 (18.7)
69 (25.7)
42 (15.7)
412 (29.4)
415 (29.6)
301 (21.5)
167 (11.9)
*Two-sample t test, 2-tailed; †Pearson’s c2test, 2-sided.

Page 4

J Chin Med Assoc • June 2008 • Vol 71 • No 6
297
Long-stay patients in the hospice ward
hospice, and 15 patients died eventually at our hos-
pice, with a range of additional LOS of 1 to 114 days
(median, 17.0 days; mean, 29±37.1 days).
Conditions of major physical distress on Day 29
of the 260 long-stay terminal cancer patients were
delirium (41.9%), depression and/or anxiety (20.4%),
severe dyspnea (21.2%), unstable body temperature
and/or infection (15.8%), malignant bowel obstruction
(11.2%), obstructive uropathy and/or hematuria (8.8%),
difficult pain control (8.1%), active tumor bleeding
(7.7%), intractable ascites (7.3%), and malignant ulcer
(6.5%) (Table 4).
Table 2. Comparisons of patient profiles between fee-for-service (FFS) and per-diem payment (PDP) patients
FFS patients (n=844) PDP patients (n=826)
p
Age (yr)67.3±14.368.4±13.0 0.475*
Gender (% male)67.2 64.2 0.198†
Time from diagnosis to hospice referral (mo)23.6±33.3 22.1±27.70.249*
Mean survival (d)37.3±83.627.9±46.9 0.001*
Mean LOS (d)17.5±16.4 14.3±13.4
<0.001*
Distribution of LOS, n (%)
≤3d
4–7d
8–14d
15–21d
22–28d
29–42d
43–56d
>56d
133 (15.8)
134 (15.9)
194 (23.0)
127 (15.0)
93 (11.0)
99 (11.7)
41 (4.9)
23 (2.7)
174 (21.1)
137 (16.6)
197 (23.8)
122 (14.8)
91 (11.0)
70 (8.5)
21 (2.6)
14 (1.7)
Discharge status, n (%)
Death in hospice
Discharged with terminal status
Stably discharged with home care
Transferred to another institution
<0.001†
587 (69.5)
56 (6.6)
180 (21.4)
21 (2.4)
572 (69.2)
47 (5.7)
147 (17.8)
60 (7.3)
Diagnosis, n (%)
Lung cancer
Colorectal cancer
Hepatocellular carcinoma
Stomach cancer
Pancreatic cancer
Other
0.056†
189 (22.4)
130 (15.4)
91 (10.8)
99 (11.7)
39 (4.6)
296 (35.1)
217 (26.2)
115 (13.9)
93 (11.3)
83 (10.0)
48 (5.8)
270 (32.7)
*Two-sample t test, 2-tailed; †Pearson’s c2test, 2-sided. LOS = length of hospice stay.
Table 3. Main reasons for inability to be discharged on Day 29
after admission
RankClinical problem
n (%)
1
2
3
4
5
Prolonged terminal phase
Difficult symptom control
Placement problem
Need of parenteral medication
Uncompleted radiotherapy
89 (34.2)
67 (25.8)
44 (16.9)
39 (15.0)
21 (8.1)
Table 4. Prevalence of major physical distress on Day 29
RankClinical problem
n (%)
1
2
3
4
Delirium
Depression and/or anxiety
Severe dyspnea
Unstable body temperature and/
or infection
Malignant bowel obstruction
Obstructive uropathy and/or hematuria
Difficult pain control
Active tumor bleeding
Intractable ascites
Malignant ulcer
109 (41.9)
53 (20.4)
55 (21.2)
41 (15.8)
5
6
7
8
9
29 (11.2)
23 (8.8)
21 (8.1)
20 (7.7)
19 (7.3)
17 (6.5)10

Page 5

J Chin Med Assoc • June 2008 • Vol 71 • No 6
298
M.H. Lin, et al
Discussion
In our study, long-stay patients had a longer survival
from cancer diagnosis to hospice referral, longer mean
survival time, more readmitted patients, fewer in the
PDP group, and were more frequently discharged for
home care programs in a stable condition. Patients with
prostate cancer and with sites of metastasis in the bone
seemed to have more long stays. This finding was com-
patible with similar LOS studies.2–4This implicates
the slow-growing nature of some cancers and the greater
hospital dependency due to disability and deprivation
of activities of daily living functions resulting from
bone metastasis.
Though insurance will reimburse acute hospice care
for cancer patients with expected survival time less
than 6 months, the majority of patients are not referred
to hospice until they are very close to death.4This is
why 83.9% of our hospice patients died within 28 days
of admission, and around 18.4% died within 3 days of
admission (Table 2). Types of insurance payment also
affect LOS. When we examined the distribution of
LOS, patients in the conventional FFS group were
found to have more long stays. This might be because
the frequency of short admissions (≤3 days’ admission)
increased from 15.8% in the FFS group to 21.1% in
the PDP group.
In the past, if some terminal patients were admit-
ted for short pre-dying care only, they would lose pri-
ority in acute inpatient hospice admission to terminal
cancer patients who were suffering from acute physi-
cal or psychospiritual distress. However, because the
outcome of clearly moribund patients was easy to pre-
dict, the strategy was changed to fulfill the require-
ment that a median LOS should be less than 16 days
using the PDP insurance system. Some studies7,8con-
firmed the value of short terminal admissions (death
within 48 hours of admission) to patients and fami-
lies, though hospice staff felt that a number of patients
transferred from the hospital should not have been
moved. Whether such transfers benefit patients or just
increase the stress on patients and their families and
the burdens on staff is worthy of further study. Early
transfer from conventional care to hospice care should
be initiated to overcome barriers to hospice care, and
education of physicians and patients about palliative
care is needed so that greater numbers of patients can
benefit more fully from hospice care at the end of life.
Compared with FFS, PDP offers potential benefits
during end-of-life care, including integrated care and
appropriate accountability for quality, and reduced finan-
cial incentives for nonessential tests and procedures.9,10
However, the traits of PDP that may be detrimental to
palliative care include the cost constraints, the conflicts
regarding placement, and the devaluation of persons
at the end of life.11,12These issues include precipitous
readmissions to the hospital for terminal care, and long-
stay terminal admissions.13
Because of the uncertainty and unpredictability of
the course of terminal malignant disease, some lengthy
admissions due to a prolonged terminal phase and
complexity in symptom control are usually difficult to
avoid.14In our study, patients and their families were
usually compelled by persistent major physical distress,
such as unstable body temperature and/or infection,
fluctuating delirium, severe dyspnea and recurrent
tumor bleeding, to give up their preference for going
back home. The prolonged terminal phase observed in
patients with malignant bowel obstruction and organ
failure usually resulted in lengthy admission. It is obvi-
ously immoral to push discharge planning for such
moribund patients.15van den Eynden et al found that
emotional problems considered to play an important
role in the decision to transfer and/or regarding the
place of death in palliative cancer included anxiety,
psychological distress, problem with acceptance, anger
and depression.16Placement problems and family diffi-
culties were prevalent in our long-stay patients. Though
nearly all the patients and families had been told that
acute care admission was a short-stay unit, the possibil-
ity of impending death appeared to make the reality
of this difficult to handle. Patients’ concerns about their
future care and placement may increase levels of anxi-
ety, and make pain and symptom control more diffi-
cult. Some family members expected their relatives
to die in the hospital instead of at home, and some
patients even said that their final wish was just to die
in our hospice unit because of the sense of security
and confidence.
In Fainsinger et al’s study in 2000,17only 24% of
patients going to a hospice and 7% of their families
preferred home discharge. Functionally dependent and
cognitively impaired patients were generally unable to
return home. To support the patients and their families
in an environment of their choice, access to increased
physical support in the home must be addressed. There
is also a need to find places of care for terminally ill
patients who may require longer-term inpatient stays.
Lack of good nursing homes with high palliative care
support and placement problems were major issues
while the family was deciding on discharge planning.18
Transfer of a patient from a palliative care unit to a
nursing home or other institution may lead to a short-
fall in meeting the needs of patients and relatives due
to the lower level of specialization than in the pallia-
tive care unit. During bereavement follow-up, many