Palliative care needs and specialist services post stroke: national population-based study

Abstract

Objectives
(1) To compare palliative care needs of patients admitted primarily with stroke and (2) to determine how the care needs of these patients affect their use of different types of specialist palliative care services.

Methods
Observational study based on point-of-care data from the Australian Palliative Care Outcomes Collaboration. Multivariate logistic regression models were used to explore the association between patients’ palliative care needs and use of community versus inpatient specialist palliative care services.

Results
The majority of patients who had a stroke in this study population had mild or no symptom distress, but experienced a high degree of functional impairment and needed substantial help with basic tasks of daily living. A lower Australia-modified Karnofsky Performance Status score (OR=1.82, 95% CI 1.06 to 3.13) and occurrence of an ‘unstable’ palliative care phase (OR=28.34, 95% CI 9.03 to 88.94) were associated with use of inpatient versus community palliative care, but otherwise, no clear association was observed between the majority of symptoms and use of different care services.

Conclusions
Many people with stroke could potentially have been cared for and could have experienced the terminal phases of their condition in a community setting if more community support services were available for their families.

Full text

1
LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Palliative care needs and specialist
services post stroke: national
population- basedstudy
Huiqin Liu ,1,2 Angus Cook,3 Jinfeng Ding,4,5 Hongwei Lu,1,2
Jingjing Jiao,4 Wenhui Bai,6 Claire E Johnson7
To cite: LiuH, CookA,
DingJ, etal. BMJ Supportive
& Palliative Care Epub ahead
of print: [please include Day
Month Year]. doi:10.1136/
spcare-2023-004280
►Additional supplemental
material is published online
only. To view, please visit the
journal online (http:// dx. doi.
org/ 10. 1136/ spcare- 2023-
004280).
1Health Management Center,
Third Xiangya Hospital, Central
South University, Changsha,
Hunan, China
2Department of Cardiology, Third
Xiangya Hospital, Central South
University, Changsha, Hunan,
China
3School of Population and Global
Health, University of Western
Australia, Crawley, Western
Australia, Australia
4Xiangya School of Nursing,
Central South University,
Changsha, Hunan, China
5Yale School of Internal Medicine,
New Haven, Connecticut, USA
6Department of Nursing,
Zhengzhou University People’s
Hospital, Zhengzhou, Henan,
China
7AHSRI, University of Wollongong
Faculty of Business, Wollongong,
New South Wales, Australia
Correspondence to
Dr Jinfeng Ding, Central South
University, Changsha, Hunan,
410017, China;
jinfeng. ding@ csu. edu. cn
Received 20 March 2023
Accepted 7 July 2023
Original research
© Author(s) (or their
employer(s)) 2023. No
commercial re- use. See rights
and permissions. Published by
BMJ.
ABSTRACT
Objectives (1) To compare palliative care needs
of patients admitted primarily with stroke and
(2) to determine how the care needs of these
patients affect their use of different types of
specialist palliative care services.
Methods Observational study based on point-
of- care data from the Australian Palliative Care
Outcomes Collaboration. Multivariate logistic
regression models were used to explore the
association between patients’ palliative care
needs and use of community versus inpatient
specialist palliative care services.
Results The majority of patients who had a
stroke in this study population had mild or
no symptom distress, but experienced a high
degree of functional impairment and needed
substantial help with basic tasks of daily living. A
lower Australia- modified Karnofsky Performance
Status score (OR=1.82, 95% CI 1.06 to 3.13)
and occurrence of an ‘unstable’ palliative care
phase (OR=28.34, 95% CI 9.03 to 88.94)
were associated with use of inpatient versus
community palliative care, but otherwise, no
clear association was observed between the
majority of symptoms and use of different care
services.
Conclusions Many people with stroke could
potentially have been cared for and could
have experienced the terminal phases of their
condition in a community setting if more
community support services were available for
their families.
INTRODUCTION
Stroke is an acute cerebrovascular disease,
the second most frequent cause of death
and a leading cause of permanent disability
worldwide.1 Globally, approximately
12 million people have a diagnosis of
stroke each year, with nearly three million
of them dying.2 Although the majority of
patients undergo inpatient rehabilitation
or skilled healthcare immediately after
experiencing stroke, the wider palliative
care needs of such patients are often not
fully addressed.
A policy statement published in 2016 by
the American Stroke Association recom-
mended integrating palliative care and
stroke care to benefit patients who had
a stroke both physically and psycholog-
ically.3 However, the particular role of
palliative care in patients with stroke is less
clear, and there is limited current litera-
ture on the detection and management of
WHAT IS ALREADY KNOWN ON THIS TOPIC
⇒Stroke for many patients may be a
‘terminal’ condition, and people with
advanced stroke can benefit from
palliative care.
⇒We have little knowledge on the specific
palliative care needs of people with
terminal stroke and how these care needs
affect their use of different specialist
services.
WHAT THIS STUDY ADDS
⇒The majority of people who accessed
palliative care principally for stroke
experienced low levels of symptom- related
distress but high levels of functional
impairment.
⇒Large disparities in symptom- related
distress and function impairment were
not observed between people seen in
community versus inpatient palliative care
settings.
HOW THIS STUDY MIGHT AFFECT
RESEARCH, PRACTICE OR POLICY
⇒Improved access to supportive services in
community settings could facilitate care
within the home environment for people
with terminal stroke.
⇒Development of a stroke- specific palliative
care strategy with consideration of unique
needs is needed to promote appropriate
care models for this group of patients.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

2 LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
palliative care needs specific to stroke survivors.4 The
palliative care strategies for cancer and other chronic
progressive diseases may not be directly translatable to
patients who had a stroke because the clinical trajec-
tory of stroke is often distinct from such conditions.
Ideally, individuals approaching the end of life should
have access to palliative care in a setting that satisfies
their care needs and preferred choice, which is based
on either inpatient or community options (including
within private residences and resident care facilities).
Compared with inpatient options, community- based
palliative care interventions can often substantially
improve symptom management and quality of life
at lower levels of health service use and expense for
terminally ill individuals.5 The community palliative
care model has been regarded as the ‘new frontier’
in supporting patients and families longitudinally
and across a variety of care settings due to its cost-
effectiveness and convenience.6 However, people with
complex and acute care needs may often require and
should have ready access to inpatient palliative care.
To date, there are few detailed analyses of the clinical
characteristics of patients who had a stroke receiving
specialist palliative care, their current patterns of palli-
ative care use, or how well service use is aligned with
patients’ palliative care needs and other clinical char-
acteristics. These knowledge gaps limit our ability to
develop evidence- based policy and specific palliative
care strategies for patients who had a stroke.
Given the increasing needs for palliative care among
patients with stroke, this study aimed to (1) describe
and compare clinical characteristics, symptom distress
and functional impairment of patients who required
palliative care principally for stroke on their first
admission to community and inpatient palliative care
services; and (2) determine whether and how care
needs and other clinical factors are associated with
their use of different types of palliative care services.
This analysis sought to assess whether the principle of
‘right care in the right place’ applies to patients with
stroke in practice and also to inform policy making,
resource allocation and establishment of stroke-
specific care models.
METHODS
Study design and data sources
Deidentified, point- of- care data were collected by
palliative care services registered with the Australian
Palliative Care Outcomes Collaboration (PCOC), a
national quality programme funded by the Australian
Government Department of Health and Aged Care
(Australian Health Services Research Institute,7 2019).
The PCOC dataset captured between 33% and 61% of
deaths from causes commonly seen by specialist pallia-
tive care providers within its jurisdictions in 2018. This
proportion has reportedly increased by around 3%
annually since then. In 2019–2020, 35 319 episodes
of care within inpatient settings were reported to
PCOC, which represents substantial coverage of the
49 200 palliative care- related type of hospitalisations
recorded in the Australia’s National Health Mortality
Database.8 Inpatient palliative care services conduct
detailed assessments on admission and then assess
the care of individual patients at least daily. Commu-
nity palliative care services perform an assessment on
admission and during each subsequent patient contact.
Data—including information regarding admission,
phase changes and discharge—are reported to PCOC
biannually. Prior to data aggregation and analyses, the
PCOC programme conducts data cleaning procedures.
Participating services receive 6 monthly reports on care
needs of their patients which allow comparisons with
national results and benchmarks established against
industry standards. ‘Improvement facilitators’ collabo-
rate with clinical staff to ensure the standardisation of
assessments and data collection (PCOC,9 2018).
Population and setting
This study included people who met the following
criteria: (1) required specialist palliative care from
services across Australia registered in PCOC; (2)
required palliative care where stoke was listed as the
principal life- limiting illness; (3) first episode of care
occurred between 1 January 2013 and 31 December
2020; and (4) died between 1 January 2013 and 31
December 2020. This study included two types of
specialist palliative care episodes: community and
inpatient (including consult liaison services) episodes.
An ‘inpatient episode’ refers to people who have been
seen in designated specialist palliative care units as well
as in non- palliative care designated beds by specialist
palliative care consultants/teams. A ‘community
episode’ refers to people who received specialist palli-
ative care within private residences or residential care
facilities.9
Variables and instruments
Clinical needs were assessed using five validated clin-
ical instruments.9 Levels of distress associated with
common symptoms (ie, difficulty sleeping, appe-
tite problems, nausea, bowel problems, breathing
problems, fatigue and pain) were assessed with the
11- point PCOC Symptom Assessment Scale (SAS)
(0, absent; 10, worst possible)10 11 by patients (or
proxies where patient- reporting was not possible).
Complexity and severity of patients’ palliative care
problems (pain, other symptoms, psychological/
spiritual and family/carer problems) were rated
using the 4- point Palliative Care Problem Severity
Score (PCPSS) (0, absent; 3, severe).12 Patients’
level of functional dependency (eating, toileting,
bed mobility and transfers) were evaluated using
the Resource Utilisation Groups–Activities of Daily
Living (RUG- ADL). The RUG- ADL scores eating
based on 3- point items (1, independent or super-
vision only; 3, total dependence/tube fed) and the
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

3
LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
other three activities with 5- point items (1, inde-
pendent; 5, two or more persons assist) .13
Patients’ performance status were measured with the
11- point Australia- modified Karnofsky Performance
Status (AKPS) (0, dead; 100, complete function).14
The non- sequential PCOC phases (stable, unstable,
deteriorating or terminal) were employed to assess
each patient’s palliative care needs based on a holistic
assessment of patients and their families.15 Assessments
were performed by trained clinicians.
Other clinicalcharacteristics of each patient were
also evaluated, including length of palliative care stay,
referral sources, accommodation at the start of the
episode, palliative care phases and place of death.
This study focused on the first assessment that was
conducted on admission to the first episode of pallia-
tive care for each patient. The term ‘episode of care’ is
defined as a continuous period of care for a patient in
one setting of care.
Data analysis
Characteristics of participants and episodes were
described using frequency counts and percentages.
Standard clinical measures were presented using
means (with SD) and medians (with IQRs). Differences
between the two groups were assessed using Pearson
χ2 tests or Mann- Whitney U tests.
Multivariable logistic regression models were devel-
oped to test whether and how clinical characteristics
affected use of inpatient versus community palliative
care services. Significant associated factors of use of
different specialist services were determined by back-
ward stepwise procedures. Due to the high degree of
multicollinearity within elements of the RUG- ADL
family, we only selected ‘total RUG- ADL’ for the step-
wise procedures. We set inclusion and exclusion criteria
at significance levels of 0.05 and 0.10, respectively. We
used C- index to assess the fitness of the final model.
p Values for trend were estimated. People admitted
community services were used as the reference group.
In the regression analyses, PCOC SAS, PCPSS, total
RUG- ADL and AKPS were recoded as categorical vari-
ables as follows: (1) The percentage of patients with
scores >1 of four PCOC SAS symptom scores were
less than 11% of the total, and therefore PCOC SAS
was classified as follows: 0=absent (corresponding to
PCOC SAS=0), 1=mild to severe (PCOC SAS=1–10);
(2) The percentage of patients with scores ≥2 of two
PCPSS symptom scores was lower than 10%, and
therefore PCPSS was classified as follows: for PCPSS:
0=absent (corresponding to PCPSS=0), 1=mild
(PCPSS=1), 2=moderate to severe (PCPSS=2–3); (3)
for AKPS: 1=not completely bedfast (corresponding to
AKPS=30–100), 2=completely bedfast (AKPS=20),
3=comatose or barely rousable (AKPS=10) and (4) for
RUG- ADL: 1=requires less than two assistants (total
RUG- ADL=4–17) and 2=requires two assistants for
full care (total RUG- ADL=18).
We fitted Kaplan- Meier curves and performed log-
rank tests to compare survival time after admission for
both groups of patients. Survival was defined as the
date of first admission to a palliative care service to the
date of death to enable comparisons between commu-
nity patients and inpatient patients with a censor date
of 31 December 2020.
Descriptive statistics, logistic regressions and collin-
earity analysis were performed using SPSS V.22.0.
Estimate of C- index and Kaplan- Meier curves were
conducted using R statistical software V.4.0. p Values
of <0.05 were considered statistically significant.
RESULTS
Study population
In total, 3930 patients were included in our study, of
which 14.4% (566) accessed palliative care in a commu-
nity setting and 85.6% (3364) in an inpatient setting.
For community patients versus inpatients, there was a
higher percentage of patients who were female (63.1%
vs 57.2%), who were aged over 85 years (63.8% vs
47.7%) and who indicated that they preferred to speak
a language other than English (58.7% vs 30.9%). This
first episode of palliative care ended in death for the
majority of people (3486, 88.7%) in both settings
(table 1).
The time interval between palliative care admis-
sion and death for the community group was more
than two times of that for inpatients (7 days vs 3 days,
p<0.0001) (figure 1).
Episode of care characteristics
Compared with inpatients, patients receiving commu-
nity palliative care had a lower proportion of referrals
from hospital (29.0% vs 88.6%), a higher percentage
of referrals from community services (62.7% vs 3.5%,
p<0.001) and a higher rate of dying either at home
(37.3% vs 0, p<0.001) or in a residential aged care
facility (45.8% vs 0.1%, p<0.001). For the inpatient
group, more than half lived in private residences
(60.6%) and 12.0% in residential aged care facilities
before palliative care admission; corresponding figures
for community patients were 48.9% and 43.1%
(p<0.001). For phase types, inpatient groups had a
far higher percentage of individuals starting with a
terminal phase compared with community groups
(62.1% vs 37.5%). Inpatient groups had a slightly
higher percentage of episodes ending with death than
the community group (89.2% vs 85.5%, p=0.003).
Comparisons of standard clinical measures between
inpatient and community groups
Both groups of patients had high RUG- ADL and low
AKPS scores (table 2). RUG- ADL scores of community
palliative care patients were significantly lower than
those of inpatient palliative care patients, whereas
AKPS scores were significantly higher (p<0.001).
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

4 LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
Table 1 Baseline sociodemographic and clinical characteristics for people with stroke by episode settings
Characteristics of patients All N (%) Community n (%) Inpatient n (%) P value for differences by setting
Sociodemographic characteristics
Total number 3930 566 (14.4) 3364 (85.6) –
Gender
Male 1647 (41.9) 209 (36.9) 1438 (42.7) 0.009*
Female 2281 (58.0) 357 (63.1) 1924 (57.2)
Missing 2 (0.1) 0 (0.0) 2 (0.1)
Age (years)
65–85 1748 (44.5) 194 (34.3) 1554 (46.2) <0.001*
<65 215 (5.5) 11 (1.9) 204 (6.1)
>85 1967 (50.1) 361 (63.8) 1606 (47.7)
Preferred language
English 2558 (65.1) 234 (41.3) 2324 (69.1) <0.001*
Non- English 1372 (34.9) 332 (58.7) 1040 (30.9)
Admission of year
2013 229 (5.8) 16 (2.8) 213 (6.3) 0.007*
2014 321 (8.2) 37 (6.5) 284 (8.4)
2015 542 (13.8) 69 (12.2) 473 (14.1)
2016 489 (12.4) 77 (13.6) 412 (12.2)
2017 504 (12.8) 70 (12.4) 434 (12.9)
2018 511 (13.0) 86 (15.2) 425 (12.6)
2019 612 (15.6) 93 (16.4) 519 (15.4)
2020 722 (18.4) 118 (20.8) 604 (18.0)
Referral source
Hospital 3144 (80.0) 164 (29.0) 2980 (88.6) <0.001*
Community services 474 (12.1) 355 (62.7) 119 (3.5)
Others 122 (3.1) 25 (4.4) 97 (2.9)
Missing 190 (4.8) 22 (3.9) 168 (5.0)
Accommodation at start of episode
Private residence 2316 (58.9) 277 (48.9) 2039 (60.6) <0.001*
Residential aged care 648 (16.5) 244 (43.1) 404 (12.0)
Other 936 (23.8) 34 (6.0) 902 (26.8)
Missing 30 (0.8) 11 (1.9) 19 (0.6)
Episode end mode
Death 3486 (88.7) 484 (85.5) 3002 (89.2) 0.003*
Others 401 (10.2) 78 (13.8) 323 (9.6)
Missing 43 (1.1) 4 (0.7) 39 (1.2)
Phase type on admission
Stable 170 (4.3) 60 (10.6) 110 (3.3) <0.001*
Unstable 243 (6.2) 14 (2.5) 229 (6.8)
Deteriorating 1216 (30.9) 280 (49.5) 936 (27.8)
Terminal 2301 (58.5) 212 (37.5) 2089 (62.1)
Place of death
Home 212 (5.4) 211 (37.3) 1 (0.0) <0.001*
Residential aged care facility 261 (6.6) 259 (45.8) 2 (0.1)
Hospital 1611 (41.0) 14 (2.5) 1597 (47.5)
Unknown 362 (9.2) 63 (11.1) 299 (8.9)
Missing 1484 (37.8) 19 (3.4) 1465 (43.5)
Survival days after admission
0–3 days 2131 (54.2) 189 (33.4) 1942 (57.7) <0.001*
4–7 days 1044 (26.6) 130 (23.0) 914 (27.2)
≥8 days 755 (19.2) 247 (43.6) 508 (15.1)
‘Episode end mode’ refers to ‘the reason that this episode of palliative care ended’.
Comparisons between the two groups were conducted using Pearson’s χ2 tests. Missing data entries were not accounted for in the analyses.
*P≤0.05.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

5
LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
PCPSS and PCOC SAS scores for community palli-
ative care patients were significantly lower than those
for inpatients (p<0.001). Average scores were below
two on all symptoms within PCPSS and PCOC SAS for
both groups, suggesting that the majority of patients in
both groups presented with absent or mild symptom
severity and distress. PCOC SAS scores were highest
for pain- related distress, followed by distress caused
by fatigue in community patients and breathing diffi-
culties in inpatients. For the PCPSS, scores were higher
for ‘family problems’ compared with other domains.
Associated factors of use of different specialist palliative
care services
The final regression model (results shown in table 3)
was obtained through a backward stepwise process.
The model had a high C- statistic (0.90, based on
multiple logistic regression models in which clinical
need measures were specified as categorical vari-
ables). The final model included the following vari-
ables: gender, age, preferred language, admission year,
accommodation at start of clinical episode, phase type,
days until death, AKPS scores, PCPSS–other symp-
toms, PCOC SAS–appetite, and PCOC SAS–pain.
In the adjusted model, lower ORs for entry into
inpatient care were observed for patients who used
a language other than English as their preferred
language (OR=0.14, 95% CI 0.09 to 0.20; p<0.001),
were admitted after 2013 (ORs ranged from 0.04 to
0.20, p<0.01 for all), were accommodated at resi-
dential aged care facilities versus private residences
(OR=0.28, 95% CI 0.19 to 0.42; p<0.001) and
were referred from a community service (OR=0.02,
95% CI 0.01 to 0.03; p<0.001) versus those referred
from hospital. Patients in an inpatient palliative care
setting had a greater likelihood of being in an unstable
phase (OR=28.34, 95% CI 9.03 to 88.94; p<0.001)
and being classified as ‘completely bedfast’ based on
the AKPS score (OR=1.82, 95% CI 1.06 to 3.13;
p<0.029). Conversely, lower ORs for use of inpatient
versus community services were estimated for patients
reporting concerns relating to appetite (OR=0.37,
95% CI 0.23 to 0.61; p<0.001) and pain (OR=0.42,
95% CI 0.29 to 0.61; p<0.001). For PCPSS, those
with ‘mild’ levels of psychological concern (relative to
‘absent’ as the reference category) also had decreased
odds of the use of inpatient services (OR=0.62, 95% CI
0.42 to 0.93; p<0.021). There were no consistent
correlations between total RUG- ADL and odds of use
of different services.
DISCUSSION
This study described the care needs and other clinical
characteristics of patients who required palliative care
principally for stroke and explored how those clinical
factors affected their use of community- based and inpa-
tient specialist palliative care. Overall, we found that
patients had relatively low recorded levels of symptom
severity and distress but moderate to high functional
deterioration. Although the symptom burden was, in
many cases, not associated with the use of different
types of specialist palliative care, patients who had a
stroke with poorer function were more likely to receive
inpatient palliative care. Patterns of use of the different
care services were also highly associated with palliative
care phases and variables such as referral source and
type of accommodation before the episode.
Overall, as noted with previous studies,16 our
analysis found that the majority of patients had
no or only mild levels of symptom severity or
distress (refer to online supplemental figures S1
and S2). Inpatient palliative care services, which
are designed to provide 24- hour, highly special-
ised management and to quickly respond to the
complex and urgent needs of patients,17 18 may not
be required for many of the patients who had a
stroke included in the study. However, it should
be noted that acute symptom management is not
the only reason for accessing inpatient palliative
care. Some patients sought access to inpatient care
for care needs that are not captured by PCOC,
such as urgent dysphagia management and other
clinical issues. People receiving community- based
palliative care were more likely to present with
mild- to- severe levels of distress for problems such
as pain and loss of appetite when compared with
inpatient care. Reasons for this apparent inconsis-
tency could relate to underdetection of symptom
distress among patients receiving inpatient care
because they were closer to death and more likely
to be in a comatose state compared with their
Figure 1 One- month survival curves for patients who had a
stroke in inpatient versus community palliative care.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

6 LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
community- based counterparts. Patients may be
unable to communicate the extent of distress due
to the cognitive impairment or more severe impair-
ment of communication.
People with mild psychological distress are also
more likely to receive community palliative care
versus inpatient palliative care compared with those
with no recorded psychological distress. However,
no increased use of community care was observed
among patients with moderate/severe psychological
distress, although the small number of patients on
this group may have limited the statistical power of
the model. Kendall et al reported that inpatient pallia-
tive care tends to focus more on physical pain and less
Table 2 Clinical need measures for patients with stroke by episode settings
Clinical outcomes N (patients)
All mean (SD),
median (IQR)
Community mean
(SD), median (IQR)
Hospital mean (SD),
median (IQR)
P values for differences
by settings
Days until death 3930 12.42 (73.16)
3 (1–6)
41.79 (159.62)
7 (3–17)
7.48 (42.47)
3 (1–6)
<0.001*
RUG- ADL
Total RUG- ADL 3859 17.63 (1.44)
18 (18–18)
17.28 (1.96)
18 (18–18)
17.69 (1.33)
18 (18–18)
<0.001*
Transfer 3890 4.92 (0.39)
5 (5–5)
4.85 (0.50)
5 (5–5)
4.93 (0.37)
5 (5–5)
<0.001*
Mobility 3897 4.89 (0.49)
5 (5–5)
4.79 (0.67)
5 (5–5)
4.90 (0.46)
5 (5–5)
<0.001*
Toileting 3895 4.92 (0.39)
5 (5, 5)
4.85 (0.51)
5 (5, 5)
4.93 (0.37)
5 (5, 5)
<0.001*
Eating 3863 2.91 (0.36)
3 (3–3)
2.80 (0.50)
3 (3–3)
2.93 (0.33)
3 (3–3)
<0.001*
AKPS, n (%)
Not completely bedfast
(AKPS=30–100)
426 (10.84) 149 (26.33) 277 (8.23) <0.001*
Completely bedfast (AKPS=20) 1609 (40.94) 238 (42.05) 1371 (40.76)
Comatose or barely rousable
(AKPS=10)
1845 (46.95) 177 (31.27) 1688 (49.58)
Missing 50 (1.27) 2 (0.35) 48 (1.43)
PCPSS
Pain 3837 0.53 (0.72)
0 (0–1)
0.76 (0.72)
1 (0–1)
0.49 (0.71)
0 (0–1)
<0.001*
Other symptoms 3790 0.74 (0.79)
1 (0–1)
0.88 (0.69)
1 (0–1)
0.71 (0.81)
1 (0–1)
<0.001*
Psychological 3890 0.42 (0.65)
0 (0–1)
0.60 (0.67)
0 (0–1)
0.39 (0.64)
0 (0–1)
<0.001*
Family 3774 1.05 (0.84)
1 (0–2)
1.14 (0.75)
1 (1–2)
1.03 (0.86)
1 (0–2)
0.001*
PCOC SAS
Difficulty sleeping 3343 0.26 (1.05)
0 (0–0)
0.43 (1.21)
0 (0–0)
0.23 (1.01)
0 (0–0)
<0.001*
Appetite problems 3367 0.43 (1.47)
0 (0–0)
0.94 (1.89)
0 (0–1)
0.33 (1.35)
0 (0–0)
<0.001*
Nausea 3374 0.14 (0.76)
0 (0–0)
0.26 (0.99)
0 (0–0)
0.11 (0.70)
0 (0–0)
<0.001*
Bowel problems 3364 0.40 (1.26)
0 (0–0)
0.75 (1.58)
0 (0–1)
0.33 (1.18)
0 (0–0)
<0.001*
Breathing problems 3378 0.83 (1.75)
0 (0–1)
0.93 (1.70)
0 (0–2)
0.81 (1.76)
0 (0–0)
<0.001*
Fatigue 3373 0.82 (1.98)
0 (0–0)
1.39 (2.37)
0 (0–2)
0.71 (1.88)
0 (0–0)
<0.001*
Pain 3377 1.06 (1.79)
0 (0–2)
1.59 (1.86)
1 (0–3)
0.95 (1.76)
0 (0–1)
<0.001*
Comparisons between the two groups were conducted using Mann- Whitney U test with RUG- ADL, PCPSS, PCOC SAS; comparisons of AKPS between the
two groups were conducted using Pearson’s χ2 tests.
*P≤0.05.
AKPS, Australian- modified Karnofsky Performance Status; PCOC, Palliative Care Outcomes Collaboration; PCPSS, Palliative Care Problem Severity Score;
RUG- ADL, Resource Utilisation Group–Activities for Daily Living; SAS, Symptom Assessment Scale.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

7
LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
on mental distress compared with community pallia-
tive care.19 People with psychological distress in our
study may also have more commonly been cared for in
community settings where familial support, such as the
company of close and trusted family members, is more
likely to be available.20 It is also possible that proxy
assessments in inpatient settings tend to be completed
by healthcare providers who are more likely to under-
rate patients’ psychological distress compared with
family and community care providers.21
Our study identified that most patients in both
groups had very limited functional ability and required
Table 3 Regression analyses for factors associated with inpatient versus community palliative care
Inpatient versus community unadjusted
OR (95% CI), P value*
Inpatient versus community adjusted
OR (95% CI), P value† P value for trend
Characteristics of patients and episodes of care
Gender (ref. male) 0.783 (0.65 to 0.94), 0.009 0.80 (0.56 to 1.15), 0.227
Age groups (ref. 65–85 years)
<65 years 2.32 (1.24 to 4.32), <0.008 3.59 (0.98 to 13.09), 0.053 0.909
>85 years 0.56 (0.46 to 0.67), <0.001 0.89 (0.62 to 1.26), 0.496
Preferred language (ref. English) 0.32 (0.26 to 0.38), <0.001 0.14 (0.09 to 0.20), <0.001
Admission of year (ref. 2013)
2014 0.58 (0.31 to 1.06), <0.078 0.20 (0.06 to 0.66), 0.008 <0.001‡
2015 0.52 (0.29 to 0.91), <0.022 0.12 (0.04 to 0.37), <0.001
2016 0.40 (0.23 to 0.71), <0.002 0.06 (0.02 to 0.20), <0.001
2017 0.47 (0.26 to 0.82), <0.008 0.11 (0.03 to 0.33), <0.001
2018 0.37 (0.21 to 0.65), <0.001 0.04 (0.01 to 0.13), <0.001
2019 0.42 (0.24 to 0.73), <0.002 0.05 (0.02 to 0.14), <0.001
2020 0.38 (0.22 to 0.66), <0.001 0.07 (0.02 to 0.20), <0.001
Referral source (ref. hospital)
Community service 0.02 (0.01 to 0.02), <0.001 0.02 (0.01 to 0.03), <0.001
Others 0.02 (0.13 to 0.34), <0.001 0.57 (0.27 to 1.22), 0.147
Episode start accommodation (ref. private residence)
Residential aged care 0.23 (0.18 to 0.28), <0.001 0.28 (0.19 to 0.42), <0.001
Other 3.60 (2.50 to 5.19), <0.001 10.53 (5.24 to 21.18), <0.001
Clinical measures
Phase type (ref. stable)
Unstable 8.92 (4.78 to 16.66), <0.001 28.34 (9.03 to 88.94), <0.001
Deteriorating 1.82 (1.30 to 2.57), 0.001 1.17 (0.59 to 2.30), 0.656
Terminal 5.38 (3.81 to 7.59), <0.001 1.55 (0.72 to 3.34), 0.260
Days until death (ref. 0~3 days)
4–7 days 0.68 (0.54 to 0.87), 0.002 0.69 (0.45 to 1.06), 0.087 <0.001‡
≥8 days 0.20 (0.16 to 0.25), <0.001 0.19 (0.12 to 0.30), <0.001
AKPS (ref. not completely bedfast)
Bedridden 3.10 (2.43 to 3.95), <0.001 1.82 (1.06 to 3.13), 0.029 0.793
Comatose 5.07 (3.94 to 6.52), <0.001 1.42 (0.73 to 2.77), 0.304
PCPSS–other symptoms (ref. absent)
Mild 0.44 (0.35 to 0.54), <0.001 0.69 (0.47 to 1.03), 0.072 0.387
Moderate/severe 0.58 (0.44 to 0.77), <0.001 1.60 (0.91 to 2.82), 0.103
PCPSS–psychological (ref. absent)
Mild 0.45 (0.37 to 0.55), <0.001 0.62 (0.42 to 0.93), 0.021 0.557
Moderate/severe 0.50 (0.36 to 0.70), <0.001 0.91 (0.44 to 1.89), 0.806
PCOC SAS–appetite (ref. absent)
Mild/moderate/severe 0.24 (0.19 to 0.30), <0.001 0.37 (0.23 to 0.61), <0.001 <0.001‡
PCOC SAS- pain (ref. absent)
Mild/moderate/severe 0.36 (0.30 to 0.44), <0.001 0.42 (0.29 to 0.61), <0.001 <0.001‡
OR values were calculated based on transformed categorical variables; ref. category for RUG- ADL total: requires less than two assistants; for AKPS: not completely bedfast; for PCPSS
and PCOC SAS: absent.
*Unadjusted and adjusted OR values estimated based on univariate and multivariate binary logistic regression models. Patients admitted to community palliative care were used as the
ref. group.
†Adjusted models include all variables selected through the backward stepwise procedures; unadjusted models include each specific variable.
‡P≤0.05.
AKPS, Australian Modified Karnofsky Performances Status; PCOC, Palliative Care Outcomes Collaboration; PCPSS, Palliative Care Problem Severity Score; ref, reference; RUG- ADL,
Resource Utilisation Group–Activities for Daily Living; SAS, Symptom Assessment Scale.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

8 LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
substantial assistance for basic living activities (refer to
online supplemental figures S3 and S4). Compared
with community palliative care services, inpatient
palliative care facilities had a higher proportion of
patients in a comatose state at admission. These find-
ings are consistent with the recent studies that patients
who had a stroke admitted to hospital palliative care
had greater needs for physical assistance than those
admitted to community palliative care.22–24 In addi-
tion, when the inpatient group first accessed pallia-
tive care, they were closer to death than those in the
community group (average of 7 days vs 42 days before
death). Consistent with our previous findings,23 the
survival time of patients who had a stroke in commu-
nity palliative care was longer than that of those from
inpatients, which is unsurprising, given the relatively
less impaired functional status among the community
group. For the majority of people in our study who
did not have severe or complex symptoms, the optimal
management model is likely to be patient- centred,
based at the individual’s usual place of residence,
with highly trained staff to provide support for family
members and caregivers.17 18 25
Those patients referred from hospital settings had a
higher rate of progressing to inpatient palliative care,
possibly for two reasons. On the one hand, the under-
lying conditions or comorbidities of patients referred
from the hospital most likely warranted ongoing care
in a comparable setting. Given that the majority of
hospitals also operate or are closely affiliated with
palliative care units, in- hospital transfer may also have
been viewed as the more convenient option. Home
care may also be very difficult for hospitalised palli-
ative care patients who are comatose or near death,
especially if support from family carers is not available
or sufficient.18 In contrast, patients who had a stroke in
community settings are more likely to access commu-
nity palliative care services. This highlights the need
for strong professional connections and cooperative
relationships among relevant health professionals in
the community (including with general practitioners,
disease- specific specialists and community nursing
teams). Home care is also dependent on the avail-
ability, willingness and ability of family members or
other non- professional carers, all of which are funda-
mental contributors to care and death at home for
patients with any disease.26
Given that more than 60% of inpatient pallia-
tive care patients lived in private residences prior to
admission, this raises the possibility that more patients
may be able to remain in the community if adequate
short- term family support is provided to families/
carers during this period. Community palliative care
allows patients to stay in their familiar environment
with their relatives in the last stage of their lives,
which is an important indicator of ‘good death’ for the
majority of people.27 Many previous studies, including
numerous randomised controlled studies, suggest that
palliative care symptoms can usually be well managed
by community- based multidisciplinary palliative care
teams. However, it is important to ensure that inpa-
tient palliative care is available and accessible for all
patients who had a stroke should they require acute
management.28 29 Suitable support should also be avail-
able for carers in the community as the patient deteri-
orates. The current PCOC data used in this study lack
economic data about the cost of providing additional
help for families to support dying at home. Therefore,
it would be worthwhile to conduct future studies that
specifically address the economic implications and
explore potential strategies to optimise resource allo-
cation and cost- efficiency within the context of palli-
ative care. In addition, this study did not delve deeply
into the influence of language barriers; it does provide
an opportunity for future research to explore this
topic further.
To our knowledge, this is the first large- scale national
study to comprehensively examine the clinical char-
acteristics and care needs of palliative care patients
with stroke in Australia. This study identified asso-
ciations between the clinical characteristics and use
of different types of specialist palliative care services
among patients with stroke. While our findings were
supported by the standardised and validated assess-
ment tools used routinely at point of care, there are
a number of potential limitations in this study. First, a
large proportion of patients dying of stroke may have
limited self- reporting abilities due to cognitive impair-
ment, language dysfunction or being in a comatose
state.30 In these situations, symptom levels were instead
reported by surrogates. Previous studies have reported
inconsistencies in reported outcomes between surro-
gates and patients,21 and it was possible that this study
may have also incorrectly estimated levels of symptom
distress. Second, patients diagnosed with stroke may
also have other comorbidities, such as coronary heart
disease and depression. However, the PCOC system
only records the patient’s primary diagnosis and
usually omits details on other comorbidities that may
require palliative care,3 and the relationship between
comorbidities and the need for palliative care services
is therefore often unclear. Third, a substantial amount
of data are missing with respect to the variable ‘place
of death’. The data were collected from numerous of
palliative care services throughout Australia and subse-
quently submitted to PCOC. As a result, rectifying or
amending these missing data points proves to be chal-
lenging. Therefore, the result regarding place of death
in this study should be interpretated with caution,
recognising the limitations imposed by the missing
data. Finally, the regression models used in this study
were primarily centred on physical symptoms and
functions, whereas variables related to cognitive and
behavioural disorders were not considered. However,
patients with advanced stroke potentially experience
cognitive and behavioural challenges which are likely
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

9
LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
to impact their use of different types of palliative care
services. The impact of such cognitive and behavioural
problems on the use of different types of service
requires further study.
CONCLUSIONS
This study found that the majority of patients who
required palliative care principally for stroke tended
to have relatively low scores for the presence and
intensity of physical symptoms, but many individuals
were affected by mild to severe functional impairment.
Poorer function tended to predict greater use of inpa-
tient versus community palliative care services, but
the nature and severity of physical symptoms typically
did not affect patterns of service use. Patients who
had a stroke experiencing physical and psychological
distress were more likely to receive palliative care in
a community setting compared with those without
distress. These findings suggest that palliative care
models specifically for patients who had a stroke are
required to improve the allocation and use of limited
palliative resources, thereby achieving ‘right care at
the right place for the right person’ with regard to this
population. Targeted strategies should include support
and promotion of community- based palliative care
(including trained staff and wider social networks, and
material and financial resources), as well as the provi-
sion of expanded opportunities for care in residential
settings.
Acknowledgements We thank the Australian Palliative Care
Outcomes Collaboration for providing data for our study.
Contributors JD conceived the concept, developed the study
design, led the data application from Palliative Care Outcomes
Collaboration and conducted data analysis and interpretation.
JD is also the author acting as the guarantor. H Liu drafted the
manuscript. JD, AC, H Liu, H Lu, JJ, WB and CEJ critically
reviewed and commented on the manuscript. All authors
provided final approval for this submission.
Funding The authors have not declared a specific grant for this
research from any funding agency in the public, commercial or
not- for- profit sectors.
Competing interests None declared.
Patient consent for publication Not applicable.
Ethics approval The Palliative Care Outcomes Collaboration
(PCOC) programme was approved by the human research
ethics committee of the University of Wollongong (HE06/045).
This study was based on a secondary analysis of PCOC data
and an exemption from ethics review was approved by the
human ethics office at the University of Western Australia
(RA/4/20/6280).
Provenance and peer review Not commissioned; externally
peer reviewed.
Data availability statement Data may be obtained from a third
party and are not publicly available.
Supplemental material This content has been supplied by the
author(s). It has not been vetted by BMJ Publishing Group
Limited (BMJ) and may not have been peer- reviewed. Any
opinions or recommendations discussed are solely those of
the author(s) and are not endorsed by BMJ. BMJ disclaims
all liability and responsibility arising from any reliance placed
on the content. Where the content includes any translated
material, BMJ does not warrant the accuracy and reliability of
the translations (including but not limited to local regulations,
clinical guidelines, terminology, drug names and drug dosages),
and is not responsible for any error and/or omissions arising
from translation and adaptation or otherwise.
ORCID iD
Huiqin Liu http://orcid.org/0000-0002-5754-2623
REFERENCES
1 Harshfield EL, Sims MC, Traylor M, et al. The role of
haematological traits in risk of ischaemic stroke and its
subtypes. Brain 2020;143:210–21.
2 Mattson MP, Arumugam TV. Hallmarks of brain aging:
adaptive and pathological modification by metabolic States.
Cell Metab 2018;27:1176–99.
3 Braun LT, Grady KL, Kutner JS, et al. Palliative care and
cardiovascular disease and stroke: a policy statement from
the American heart association/American stroke association.
Circulation 2016;134:e198–225.
4 Stevens T, Payne SA, Burton C, et al. Palliative care in stroke: a
critical review of the literature. Palliat Med 2007;21:323–31.
5 Whitney RL, Bell JF, Tancredi DJ, et al. Hospitalization rates
and predictors of rehospitalization among individuals with
advanced cancer in the year after diagnosis. J Clin Oncol
2017;35:3610–7.
6 Kavalieratos D, Gelfman LP, Tycon LE, et al. Palliative care in
heart failure: rationale, evidence, and future priorities. J Am
Coll Cardiol 2017;70:1919–30.
7 Australian Health Services Research Institute. Palliative care
outcomes collaboration. 2019. Available: https://ahsri.uow.edu.
au/pcoc/index.html [Accessed 1 Dec 2019].
8 Daveson BA BM, Clapham S, Banfield M, et al. The palliative
outcomes collaboration: data quality statement. Australian
Health Services Research Institute University of Wollongong;
2022.
9 Palliative care outcomes collaboration. Palliative care outcomes
collaboration clinical manual. 2018. Available: https://ahsri.
uow.edu.au/content/groups/public/@web/@chsd/@pcoc/
documents/doc/uow129133.pdf [Accessed 01 Dec 2019].
10 Aoun SM, Monterosso L, Kristjanson LJ, et al. Measuring
symptom distress in palliative care: psychometric properties
of the symptom assessment scale (SAS). J Palliat Med
2011;14:315–21.
11 Daveson BA, Allingham SF, Clapham S, et al. The PCOC
symptom assessment scale (SAS): a valid measure for daily use
at point of care and in palliative care programs. PLoS One
2021;16:e0247250.
12 Masso M, Allingham SF, Johnson CE, et al. Palliative care
problem severity score: reliability and acceptability in a
national study. Palliat Med 2016;30:479–85.
13 Fries BE, Schneider DP, Foley WJ, et al. Refining a case- mix
measure for nursing homes: resource utilization groups (RUG-
III). Med Care 1994;32:668–85.
14 Abernethy AP, Shelby- James T, Fazekas BS, et al. The Australia-
modified karnofsky performance status (AKPS) scale: a
revised scale for contemporary palliative care clinical practice
[Isrctn81117481]. BMC Palliat Care 2005;4:7.
15 Masso M, Allingham SF, Banfield M, et al. Palliative care
phase: inter- rater reliability and acceptability in a national
study. Palliat Med 2015;29:22–30.
16 Mazzocato C, Michel- Nemitz J, Anwar D, et al. The last days
of dying stroke patients referred to a palliative care consult
team in an acute hospital. Eur J Neurol 2010;17:73–7.
17 Kamal AH, Currow DC, Ritchie CS, et al. Community- based
palliative care: the natural evolution for palliative care delivery
in the U.S. J Pain Symptom Manage 2013;46:254–64.
18 Wiencek C, Coyne P. Palliative care delivery models. Semin
Oncol Nurs 2014;30:227–33.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from

10 LiuH, etal. BMJ Supportive & Palliative Care 2023;0:1–10. doi:10.1136/spcare-2023-004280
Original research
19 Kendall M, Cowey E, Mead G, et al. Outcomes, experiences
and palliative care in major stroke: a multicentre, mixed-
method, longitudinal study. CMAJ 2018;190:E238–46.
20 Fuller- Iglesias HR, Webster NJ, Antonucci TC. The complex
nature of family support across the life span: implications for
psychological well- being. Dev Psychol 2015;51:277–88.
21 de Graaf E, Zweers D, de Graeff A, et al. Symptom intensity
of hospice patients: a longitudinal analysis of concordance
between patients' and nurses' outcomes. J Pain Symptom
Manage 2018;55:272–81.
22 Ding J, Cook A, Qin X, et al. Palliative care needs and
utilization of specialist services for people Imminently dying
with dementia: a national population- based study. Int J Nurs
Stud 2020;109:103655.
23 Ding J, Johnson CE, Lee YCO, et al. Characteristics of people
with dementia vs other conditions on admission to inpatient
palliative care. J Am Geriatr Soc 2020;68:1825–33.
24 Ding J, Johnson CE, Qin X, et al. Palliative care needs and
utilisation of different specialist services in the last days of
life for people with lung cancer. Eur J Cancer Care (Engl)
2021;30:e13331.
25 Mitchell SL, Teno JM, Kiely DK, et al. The clinical course of
advanced dementia. N Engl J Med 2009;361:1529–38.
26 Bakitas M, Allen Watts K, Malone E, et al. Forging a new
frontier: providing palliative care to people with cancer in
rural and remote areas. J Clin Oncol 2020;38:963–73.
27 Gomes B, Higginson IJ, Calanzani N, et al. Preferences for
place of death if faced with advanced cancer: a population
survey in England, Flanders, Germany, Italy, the Netherlands,
Portugal and Spain. Ann Oncol 2012;23:2006–15.
28 Lutz BJ, Young ME, Creasy KR, et al. Improving stroke
caregiver readiness for transition from inpatient rehabilitation
to home. Gerontologist 2017;57:880–9.
29 Sit JWH, Wong TKS, Clinton M, et al. Stroke care in the
home: the impact of social support on the general health of
family caregivers. J Clin Nurs 2004;13:816–24.
30 Adhikari MH, Hacker CD, Siegel JS, et al. Decreased
integration and information capacity in stroke measured
by whole brain models of resting state activity. Brain
2017;140:1068–85.
Protected by copyright. on July 31, 2023 at Central South University CALIS.http://spcare.bmj.com/BMJ Support Palliat Care: first published as 10.1136/spcare-2023-004280 on 27 July 2023. Downloaded from