BMJ Support Palliat Care 3:77-83 doi:10.1136/bmjspcare-2012-000331
  • Research

Who, why and when: stroke care unit patients seen by a palliative care service within a large metropolitan teaching hospital

  1. Brian Le1
  1. 1Department of Palliative Care, Melbourne Health, Royal Melbourne Hospital, Parkville, Victoria, Australia
  2. 2Melbourne Epicentre (The Centre for Clinical Epidemiology, Biostatistics and Health Services Research), Melbourne Health, Royal Melbourne Hospital, Parkville, Victoria, Australia
  3. 3Department of Neurology, Melbourne Health, Royal Melbourne Hospital, Parkville, Victoria, Australia
  1. Correspondence to Dr Peter Eastman, Department of Palliative Care, Melbourne Health, Royal Melbourne Hospital, Grattan Street, Parkville, VIC 3050, Australia; eastman{at}


Objectives To investigate factors associated with referral of patients from an Australian stroke care unit (SCU) to an inpatient palliative care service (PCS).

Methods This retrospective observational cohort study included patients who were referred to the PCS after SCU admission between 1 January and 31 December 2008. Variables measured included patient demographics, premorbid functional status, premorbid living situation, stroke type, history of previous stroke and discharge outcomes. Group differences between all SCU patients seen and not seen by the PCS were compared using univariate analyses. Multivariate logistic regression analysis was undertaken to identify factors associated with PCS involvement. Group differences were also compared between deceased stroke patients seen and not seen by the PCS.

Results 544 patients were admitted to the SCU during the study period with 62 (11.4%) referred to the PCS. Assistance with end-of-life care was the commonest reason for referral. From univariate analyses, factors significantly associated with PCS involvement included age, gender, premorbid modified Rankin score, living situation prior to stroke and stroke type. Factors predicting PCS involvement for SCU patients from logistic regression were: increasing age, higher premorbid modified Rankin score and haemorrhagic stroke. 87 (16.0%) SCU patients died during their admission, with 49 (56.3%) seen by PCS. Deceased patients seen were significantly older, more disabled premorbidly and lived significantly longer.

Conclusions This study indicates there are patient and condition-level factors associated with referral of stroke patients to PCS. It highlights factors that might better stratify hospitalised stroke patients to timely palliative care involvement, and adds an Australian perspective to limited data addressing this patient population.


After coronary heart disease,1 stroke is the second commonest single cause of deaths in Australia, and is a leading cause of morbidity and mortality worldwide.2 ,3 It is estimated that 60 000 strokes occur in Australia each year, and that one in six people will suffer a stroke during their lifetime. Following a stroke, 20% of sufferers die in the acute phase,2–6 and a third will be dead within a year.4 About half of all stroke deaths take place in a hospital setting,3 and the majority occur in those aged 75 years or older.7 High rates of symptom burden have been reported in dying stroke patients,8 as well as a range of complex end-of-life and psychosocial issues5 ,6 for patients and their families.

The benefits of palliative care are well established in malignant disease and increasingly recognised within non-malignant settings in accordance with a needs-based approach to care. Recent figures on Australian palliative care service provision found around 16% of palliative care patients had a non-malignant primary diagnosis.9 Stroke, with its high mortality and morbidity and complex end-of-life issues, is a non-malignant condition in which palliative care involvement can be appropriate. This approach is supported by funding agencies, such as the Victorian Department of Human Services7 and key international craft and consumer groups,10–12 including the Intercollegiate Stroke Working Party of the Royal College of Physicians London,10 and the National Stroke Foundation in Australia, who recommend ‘stroke patients and their families/carers should have access to specialist palliative care teams as needed and receive care consistent with the principles and philosophies of palliative care’.12

While considerable research on stroke has focussed on important factors, such as acute treatment and rehabilitation, much less attention has been paid to the interface between stroke and palliative care. While data are limited, both Australian13 and international3 ,14 evidence suggest the overall number of stroke referrals to inpatient PCSs is increasing. In keeping with the heterogeneity of stroke presentations and outcomes, considerable variation has been reported in the proportion of admitted stroke patients who receive a palliative care referral. Holloway et al3 examined all stroke admissions to a large US hospital, and found that palliative care consultation was sought in 6.5% of cases. Other authors,8 ,15 however, looking more specifically at dying stroke patients, have reported palliative care referral rates closer to 25%.

The lack of available evidence guiding management for palliative stroke patients prompted the stroke and palliative care services at the Royal Melbourne Hospital (RMH), to collaborate on a project addressing this patient group. The title of this comprehensive project was ‘Improving access to palliative care for people who have had a stroke and their carers’.16 Its outcomes included the development and implementation of a stroke palliative care pathway, and the consolidation of connections between stroke and PCSs. As part of the preimplementation phase of the project, a retrospective database review was undertaken to better characterise stroke patients referred by the Stroke Care Unit (SCU) to the PCS. The current study presents data collected during this phase. It aims to explore the proportion of patients referred, the characteristics of this group, and factors influencing referral. It is hoped that this information may provide guidance in the stratification of inpatient palliative stroke patients, and aid clinical decision making with this group.



The RMH is a 350 bed, academic affiliated, tertiary referral public hospital servicing almost 1 000 000 people in metropolitan and regional areas.17 All patients presenting to the RMH with stroke are admitted under the care of the SCU. Subarachnoid haemorrhages are not included in this group as they are admitted under the neurosurgical team. The SCU is an acute focused comprehensive stroke centre18 staffed by a multidisciplinary team of stroke neurologists, neurology advanced trainee registrars, junior medical officers, stroke nurse practitioners, stroke nurse specialists, ward nurses and allied health professionals. Additional support is provided by a range of departments, including neurosurgery, geriatrics and radiology. There are no specific palliative care staff embedded in the SCU. For patients assessed by the multidisciplinary SCU team as requiring palliative care input, a referral was made to the PCS consult team. At the time of study, this decision was based upon consensus expectation of poor prognostic outcome among the multidisciplinary SCU staff, rather than on established referral criteria. Accordingly, the decision to refer was a clinical one predicated on each individual patient's clinical course. The PCS consult team consists of a specialist palliative care consultant, registrar and clinical nurse consultant. It works closely with intrahospital allied health professionals, and has strong ties with community-based service providers.

Subjects/data collection

All patients admitted under the RMH SCU between 1 January 2008 and 31 December 2008 with an acute ischaemic stroke (AIS) or a primary intracerebral haemorrhage (PICH), excluding subarachnoid haemorrhages, were included in this retrospective observational cohort study. Data were collected on patient demographics, premorbid living situation, stroke type, history of previous stroke, length of stay, PCS involvement and premorbid and discharge modified Rankin scores (mRS). This functional measure scored between zero and six, in which increasing debility is represented by higher scores, is widely used in stroke research with a recent systematic review reporting moderate reliability.19 Principal reason for PCS referral was chosen by the referrer from options on a drop-down menu embedded in the PCS referral form. Data were accessed from the RMH SCU and PCS databases. There were no missing data. The project was approved by the Melbourne Health Human Research Ethics Committee.


Descriptive statistics were used to summarise demographic data, as well as the reasons for palliative care referral. Differences between SCU patients seen and not seen by the PCS were assessed using independent t tests for normally distributed data; Wilcoxon Rank-sum tests were used if the data were not normally distributed. χ2 tests/Fisher's exact test were used for categorical data. Multivariate logistic regression analysis was performed to identify factors associated with PCS involvement.

For patients who died following their stroke, Kaplan–Meier survival curves were generated for patients seen and not seen by the PCS. Univariate analyses were also undertaken to compare differences between these two groups, as this has been highlighted as an area of practice requiring further research.3 A χ2 test for goodness-of-fit was used to compare PICH rates in the general stroke population with those in the sample seen by the PCS. Level of significance was set at <0.05 for all tests. Analysis was performed using SPSS V.19 (SPSS Inc, Chicago, Illinois, USA) and Stata12 (StataCorp, Texas, USA).


During the audit period, 544 patients were admitted to the SCU, with men making up 54.0% of admissions. Sixty-two (11.4%) of these patients were referred to the PCS. As illustrated in table 1, request for assistance with end-of-life care was the commonest reason for referral. Of the 62 referred patients, five were assessed by the PCS as either more appropriate for rehabilitation services or inappropriate for palliative care. This decision was made by the specialised palliative care team based on individual patient needs assessment, clinical judgement and consultation with other health professionals. This group was not included in further analysis. For patients seen by the PCS, the average time from admission to PCS referral was 5.5 days, with 91.2% of patients seen on the day of referral.

Table 1

Principal reasons for referral from SCU to PCS

Table 2 shows the characteristics of all patients admitted to the SCU and compares those who were seen by the PCS with those who were not. Univariate analyses indicated that patients seen by the PCS were older and more commonly women. They were more likely to have been living alone or in a residential aged care facility premorbidly, and were more disabled prior to their stroke. While AIS were more common than PICH in both groups, haemorrhages were significantly over-represented in patients seen by palliative care. Patient age, premorbid mRS and a PICH predicted PCS involvement after adjustment for all other factors (table 3). Sixty per cent of all SCU admissions had a separation mRS of four or greater, indicating either discharge with moderate to severe ongoing disability, or death following the stroke. The majority of patients seen by the PCS died in hospital (discharge mRS=6), and those who survived were all discharged with an mRS of 5 indicating severe disability requiring extensive nursing care.

Table 2

Patient characteristics of stroke care unit admissions, and comparison of those seen and not seen by palliative care service

Table 3

Factors from multivariate logistic regression predicting palliative care service involvement for stroke care unit patients

Eighty-seven (16.0%) patients admitted under the SCU died during their admission, with 56.3% of these patients referred to the PCS. Over half of all deceased SCU patients, and 42.8% of patients seen by PCS, had a premorbid mRS of 0 indicating complete independence prior to their stroke (table 4). While there was little difference in the overall number of men and women who died following their stroke (46.0% vs 54.0%, respectively), men were less likely to be referred to the PCS. Non-significant differences were seen in stroke types, with the proportion of haemorrhages greater in the non-referred group. Time to death was significantly longer for patients referred to the PCS, with those not referred significantly more likely to die during the first 3 days of their admission (table 4/figure 1).

Table 4

Group differences between deceased SCU patients (n=87) seen and not seen by the PCS

Figure 1

Survival curves for decreased stroke care unit patients seen and not seen by palliative care service. PCS, palliative care service.


This study identified a number of factors that influenced referrals from the SCU to palliative care, including age, gender and stroke type. Although men were more commonly represented in admissions to the SCU, women were significantly more likely to be seen by the PCS. Increasing age was significantly associated with PCS involvement and, again, women seen by palliative care tended to be older than their men counterparts. This is in keeping with previous data,20 ,21 including the Framingham Heart Study which found that women suffered first-ever strokes at an older age than men, and that in the acute phase experienced significantly more functional disability.21 The greater life expectancy of women may explain some of this difference, as it increases the likelihood of a stroke occurring at an older age when the risk of compromised health is greatest.

Stroke types associated with expected poor outcomes were over-represented in referrals to the PCS. PICH, for instance, which have high 30-day mortality rates,22 ,23 occurred in 35% of patients seen by the PCS. This was a significantly greater proportion than occurred in those not seen by the PCS, and is also a significantly greater proportion (p<0.001) than would be expected to occur in the general stroke population, where PICH have been reported to make up 10–15% of strokes.22

A considerable proportion of overall SCU presentations, PCS referrals and patients who died following their strokes, were fully independent premorbidly. The dramatic change in condition associated with their stroke presents unique challenges for patients, families and for staff. While the trajectory of many oncological and non-malignant diseases afford the opportunity for end-of-life planning to take place, the suddenness of a devastating stroke often precludes this. Effective and timely communication has been highlighted as a crucial component of care in the acute stroke setting, with evidence suggesting patients and their families/carers desire this to be honest and clear even when prognosis appears poor.5 Providing assistance with communication between health professionals, patients and families/carers is consistent with the skillset of a palliative care team, and has been reported as a reason for requesting referral.6 ,13

Assistance with end-of-life care was the commonest reason for referral to the PCS in the current study. This suggests a more traditional view of palliative care as synonymous with terminal management, and may reflect a lack of awareness among non-palliative care health professionals regarding other services palliative care can offer. While ‘end-of-life care’ is a nebulous term, it incorporates many of the other referral reasons reported in the literature, including symptom management, support for families and staff with complex, ethical, decision making and discharge planning.8 ,13 As a descriptor, the generality of the term ‘end-of-life care’ makes comparison to more specific referral data from other authors difficult. Importantly, however, this non-specificity reflects the broad complexities encountered in many of these situations, and is in keeping with the clinical reality.

More than 50% of patients who died following their strokes were seen by the PCS. This is a higher proportion than has been previously reported,8 ,15 and likely reflects growing acknowledgement of the benefits PCSs can offer. By contrast with the comparison results for the total SCU population, the proportion of haemorrhages was greater in deceased stroke patients who were not seen by the PCS, and this group was significantly more likely to die early in their admission. The extensive nature of their disease accounts for much of this difference, with data suggesting that SCU staff generally feel comfortable managing patients in a clearly terminal phase.6 For longer-term survivors, however, the situation can be more complicated, as the needs and concerns of patients and families/carers after a severe stroke change in accordance with the evolving clinical picture. Rogers et al6 highlighted differences in the perceived quality of care provided to patients who died within 5 days of their stroke, and those who survived a longer period. The authors concluded that the longer a patient survived after a severe stroke, the more complex ongoing decision making became with particular emphasis on factors, such as prognostication, hydration and artificial feeding.

Only 12.0% of patients seen by the PCS were discharged alive from hospital, limiting the opportunity to analyse this group in any detail. All these patients had a discharge mRS of 5, and were generally discharged to residential aged care facilities or palliative care units. By contrast, those discharged with an mRS of 5, but not seen by the PCS, tended to go to Geriatric Evaluation and Management units, rehabilitation or other acute settings. Only three (4.5%) of the total SCU patients discharged with an mRS of 5 were able to go home. Despite this low figure, the possibility of offering severely disabled patients and their carers more choice in relation to preferred venues of care warrants further exploration.

There are limitations associated with this study. Its retrospective nature meant that there was no preplanned definition driving the nature of PCS provision, and this potentially limits conceptual clarity. Retrospective data collection also meant other factors potentially contributing to decision making about PCS involvement might not have been identified. These factors might include patient or family preference and variations in the decision making of individual referrers. A prospective observational study could provide additional insight into these issues. The research was undertaken in a single tertiary centre with an established SCU and inpatient PCS, and thus, the results may not be generalisable to smaller or more regional centres. Importantly, however, this study adds an Australian perspective to the limited data addressing referrals to PCS for hospitalised stroke patients. This is pertinent information, given that stroke rates are anticipated to increase with the ageing population, meaning, more events will occur in an already frail and debilitated group.


The authors acknowledge the contribution of Dr Peter Hand for his expert opinion and comments on the final manuscript.


  • Contributors Per the ICJME-recommended criteria for authorship, all the listed authors met the three criteria: (1) substantial contributions to conception and design, acquisition of data or analysis and interpretation of data; (2) drafting the article or revising it critically for important intellectual content and (3) final approval of the version to be published.

  • Funding The project ‘Improving access to palliative care for people who have had a stroke and their carers’ was supported and funded by the Australian Government, Department of Health and Ageing under the National Palliative Care Program, with specific thanks to Ellen Sheridan and Jackie Kearney of the Victorian Department of Health.

  • Competing interests None.

  • Ethics approval Melbourne Health Human Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Received 17 July 2012.
  • Revision received 20 November 2012.
  • Accepted 20 November 2012.
  • Published Online First 3 January 2013