Article Text

Download PDFPDF

Bundle of care to drive improvements in palliative and end-of-life care (PEOLC) in an acute tertiary hospital
  1. Liam Bell1,
  2. Amanda Sebastian2,
  3. Kerrin Palazzi3,
  4. Saxon Farquhar4,
  5. John Attia4,5,
  6. Jeanette Lacey6
  1. 1Department of Intensive Care, John Hunter Hospital, New Lambton Heights, New South Wales, Australia
  2. 2Department of Orthopaedics, John Hunter Hospital, New Lambton Heights, New South Wales, Australia
  3. 3Data Sciences Section, The University of Newcastle Hunter Medical Research Institute, New Lambton, New South Wales, Australia
  4. 4Department of Medicine, John Hunter Hospital, New Lambton Heights, New South Wales, Australia
  5. 5School of Medicine and Public Health, The University of Newcastle, New Lambton Heights, New South Wales, Australia
  6. 6Medicine and Interventional Service, John Hunter Hospital, New Lambton Heights, New South Wales, Australia
  1. Correspondence to Dr John Attia; john.attia{at}


Objectives There is little evidence to suggest the best model of palliative and end-of-life care (PEOLC) in an acute care hospital. We introduced a bundle of care to drive improvements in PEOLC; this bundle included three full-time nursing positions providing a palliative care clinical consult service with physician backup, as well as educating staff, using the NSW Resuscitation Plan and the Last-Days-of-Life Toolkit.

Methods Two audits were performed at John Hunter Hospital, a tertiary hospital in Newcastle, Australia, each sampling from all deaths in a 12-month period, one prior to and one after the bundle of care was introduced. Sampling was stratified into deaths that occurred within 4–48 hours of admission and after 48 hours. Key outcomes/data points were recorded and compared across the two time periods.

Results Statistically significant improvements noted included: lower mortality on the wards after 48 hours of admission, better recognition of the dying patient, increased referral to palliative care nurses and physicians, reduction in the number of medical emergency team calls and increase in the use of comfort care and resuscitation plans. Currently, 73% of patients have their end-of-life wishes observed as per their advance care directive.

Conclusion A bundle of care involving dedicated nurses with physician backup providing a consult service and education is an effective method for driving improvements in PEOLC.

  • Audit and feedback
  • Healthcare quality improvement
  • Hospital medicine
  • Palliative Care

Data availability statement

Data are available on reasonable request. The data referred to in this study can be obtained from LB, who can be contacted at

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • An increasing number of deaths in developed countries occur in hospital and there is increasing scrutiny on quality of such end-of-life care. There is little evidence for what service models best deliver quality palliative care.


  • A unique pre/post audit study examining the effect of implementing a bundle of care in a tertiary public hospital. This bundle was found to increase the standard of palliative and end-of-life care for ward patients.


  • A bundle of care incorporating dedicated nurses, with physician back up, providing both clinical care and education is a viable option which can be considered in other tertiary public hospitals.


Palliative care is defined by the WHO as improving the ‘quality of life of patients and families through prevention and relief of suffering’ and is recognised under the human right to health.1 In developed countries, somewhere between one-third and two-thirds of deaths occur in hospital2 and this number is increasing over time.3 Surveys over a decade indicate that quality of end-of-life care in hospitals has not improved4 and there have been calls within the palliative care community to intentionally focus on measures of quality in order to make a case for value for money.5 Indeed, articles on quality at end of life have increased year-on-year, although most of these are simply descriptive and observational.6 Reviews have been published pulling together the work identifying what elements are important to patients and carers in quality end-of-life care.7 8

The Australian Commission on Safety and Quality in Health Care (ACSQHC) published a National Consensus Statement in 2015, outlining the key elements for quality end-of-life care. Ten key elements, including patient-centred care, responding to concerns, using triggers and evaluation/feedback, were highlighted. Meeting these standards is now a requirement for New South Wales (NSW) hospital accreditation9 although there is little rigorous research evidence regarding the best way to implement them.10

The John Hunter Hospital, a ~750 bed tertiary level teaching hospital in Newcastle, Australia implemented of a bundle of care aimed at improving palliative and end-of-life care (PEOLC) which included three dedicated, on-site nurses providing clinical care and education that incorporated the NSW Resuscitation Plan11 and roll-out of a Last-Days-of-life Toolkit.12 These interventions took place in between two national audits of deaths (including our hospital),13 providing an opportunity to evaluate if the new service model improved quality PEOLC.



Two audits were performed as part of our participation in a national audit of quality end-of-life care in hospitals13; measures were chosen based on the ACSQHC standards and were standardised nationally. The audit randomly sampled cases from all deaths across a public tertiary hospital in two separate 12-month periods; the first was from June 2015 to June 2016 and the second from June 2018 to June 2019.


Prior to the first audit, a palliative care consult service was provided from a neighbouring hospital (Newcastle Calvary Mater Misericordiae), with 5 Full Time Equivalent palliative care staff specialists visiting the John Hunter Hospital when needed but with no on-site presence during regular business hours. After the first audit, three full-time nursing positions were introduced to the hospital. These consisted of an end-of-life care nurse practitioner (for patients who were at expected to die in the current admission) starting in 2016, a nurse practitioner candidate and a clinical nurse consultant/educator (focused on upskilling nursing staff) starting in mid-2018. All nurses shared the role of supporting staff, patients, families and carers when end-of life care was required, educating staff, as well as liaising with palliative care physicians if necessary. During 2016–2017, there was facility wide education and training in the use of the NSW Resuscitation Plan11 and the Last-Days-of-Life Toolkit12 mentioned above. The team took referrals from treating medical and surgical teams every weekday from 08:00 to 18:00 hours, via faxed forms or phone calls to the service. Nurses provided the initial consults on patients; nurses were backed up by specialist physicians either via phone advice or in person assessments. Advice on weeknights and weekends was available from a palliative care on-call service 24/7 by a palliative care medical officer and this was present throughout the study period, that is, both before and after the first audit.

Data collection

Patients were separated into four groups: those who died in non-critical care wards between 4 and 48 hours after admission (‘early’ group), those who died in those wards more than 48 hours after admission (‘late’ group), those who died in the intensive care unit (ICU) between 4 and 48 hours after admission and those who died in the ICU more than 48 hours after admission. Fifty patients were randomly selected from each of the four groups, yielding 200 patients overall for each audit; these were taken from medical and surgical admissions. Given that the nurse-led model of care was not often required by ICU staff, due to the model of care we focused mainly on the ward patients.

Due to incomplete patient information, 174 patients were included in the final analysis from 2015 to 2016, and 194 patients were included from 2018 to 2019. A team of auditors was tasked with retrospectively reviewing admission notes for all patients in the audit to determine which key events occurred.

The number of deaths occurring in hospital across the two time periods, in absolute counts and rates were compared, the latter using the total number of patients admitted that year as the denominator (this was drawn from hospital data rather than the audits). Seven key outcomes deemed necessary for quality palliative and end-of-life care were selected, based on the National Consensus statement and from those frequently raised in the literature. The seven key outcomes were: recognition of the dying patient, referral to palliative care, prescribing of palliative medications, initiating comfort care, use of resuscitation plans and involvement of palliative care nurses and physicians prior to death.

Several key terms to indicate a recognition of ‘dying’ were listed for auditors. Comfort care was defined as ceasing all non-beneficial interventions (eg, intravenous fluids, nasogastric feeds, imaging) and initiating symptom control observations. These definitions were preserved across both audits.

To determine adherence to a patient’s wishes at the end of life, all interventions in the final 48 hours of life were recorded and compared with the wishes of patients in their advance care plan (ascertained and documented in the community) or resuscitation plan (ascertained and documented in hospital); this was only audited in the second time period and hence no pre-post comparisons were available for this outcome.

Auditors also looked at six process measures, recording time from each of the following to death: resuscitation plan documentation, recognition of death, referral to palliative care, first contact with a palliative care nurse and first contact with a palliative care physician.

The central question of this study, that is, the model of care developed to improve end-of-life care in hospital, was born out of clinical experience with patients and carers. There was no formal consultation process with patients and the public, although the audit forms and process were designed in consultation with a consortium of hospitals and clinicians. Dissemination of results has occurred through grand rounds and conference presentations, and will be continued through journal publications and local health district management meetings.

Statistical analyses

The two cross-sectional surveys were used to calculate change over time. Poisson and negative binomial regressions analyses were conducted to compare total deaths preintervention and postintervention. χ2 and Fisher’s exact tests were conducted for each key variable between preintervention and postintervention groups and weighted to reflect total patients. Parametric bootstrap estimation was used when determining significance between timing of events. Statistical analyses were performed using SAS V.9.4 and statistical significance set at p<0.05.


We looked at the total number of people dying in hospital as a percentage of those who were admitted. For ward patients dying or discharged shortly after admission (4–48 hours, ‘early’ group), there was no significant change in the rate of deaths, likely due to the short time frame to influence an already advanced process. However, for those dying or discharged later in their admission (> 48 hours, ‘late’ group), there was a significant reduction of 26%–30% in the percentage of people dying in hospital, (p<0.001; table 1).

Table 1

Number and rate of deaths preintervention and postintervention

Tables 2 and 3 show the change in the seven quality indicators of palliative and end-of-life care between the two time periods for patients on medical or surgical wards.

Table 2

Quality indicators for ‘early’ group

Table 3

Quality indicators for ‘late’ group

For the ‘early’ group, several statistically significant improvements were seen: more patients were recognised as dying (76% improving to 90%), referred to palliative care (4% improving to 38%), had resuscitation plans documented (67% improving to 84%) and were visited by palliative care nurses (0% (no local nursing referral available in the pre-period) improving to 22%) and physicians (4% improving to 12%) prior to death. All of these results were statistically significant. There were improvements in initiating comfort care (73% improving to 78%) and prescribing palliative medications (58% improving to 70%) prior to death, but these did not reach statistical significance.

Among the ‘late’ group, several improvements were also noted, as seen in table 3; referral to palliative care (23% improving to 68%), prescribing of palliative medications (57% improving to 70%) and visitation by palliative care nurses (3%–68%) prior to death, improved to a statistically significant degree.

Overall, in the hospital, rates of advance care directives (ACD) fell from 13% to 7.25% while rates of resuscitation plans increased from 71.8% to 84.4%, which was statistically significant (see table 4). The use of medical emergency team (MET) calls also significantly reduced prior to death, as was the need for MET calls to trigger changes to palliative or comfort care. Both of these were statistically significant results.

Table 4

Advance care plans and MET calls

In the second audit, based on resuscitation plans or ACDs, 72.7% of patients had their wishes followed exactly, with 12.4% of patients not having their wishes documented. Of the 14.9% whose wishes were not followed, the most common instances were MET calls (seven events), administering intravenous fluids (five events), imaging (three events), with two events each for Cardio-Pulmonary Resuscitation (CPR), bag-valve masking, vasoactive drugs and intravenous antibiotics.

Researchers also examined process measures, that is, time from various team contacts to death. Table 5 shows the median time between an event occurring and patient death in all seven measures. In the ‘difference’ column, a positive result indicates a longer time, that is, earlier median time of completion prior to death.

Table 5

Expediency of delivering end-of-life care

Resuscitation plans were written earlier, and patients were recognised as dying earlier in both ward groups. Palliative medications were prescribed earlier in the ‘early’ group patients, but later to a statistically significant degree in ‘late’ group patients. The first evidence of referral to palliative care was later in both groups, significantly so in the ‘early’ group.

Comfort care was administered earlier in both groups, with a significant change in the ‘late’ group. No data from preintervention were present for palliative and end-of-life care nurse visitation for the ‘early’ group, but ‘late’ group patients were seen earlier by palliative care nursing staff. Physicians visited ‘early’ ward patients later, but ‘late’ ward patients were seen significantly earlier.


Standards of palliative care

As noted in the introduction, quality of PEOLC in hospitals has been highlighted as an area of need.5 Qualitative studies have demonstrated that areas of concern include: recognising and communicating that patients are dying, after hours care, and lack of access to senior doctors to make decisions around the goals of care.14

Recognition of dying prior to death in Australian Hospitals is reported to range from 72% to 92% and is vital to ensure quality palliative care occurs in a timely fashion.15 However, the proportion of patients still receiving ‘active treatment’ at the time of death was greater than 50% in a study of over 500 ward patients over the course of 1 year at an Australian metropolitan hospital, indicating that there is much room for improvement in triggering changes in care congruent with recognition of imminent death.15

The literature review of dying in Australian Hospitals by Bloomer et al16 demonstrated that there is little quality evidence on the best way to implement PEOLC, despite many guidelines regarding the key aspects it should contain. A need for more preintervention and postintervention studies was identified as a way to explore the optimum way to improve this.10 This literature review, the ACSQHC’s 2015 Statement and the above referenced articles regarding palliative care delivery in Australian hospitals, were used to determine the seven key areas of care analysed in this paper.

Components of bundle of care

In between the two audits that were reported here, several interventions were made as part of a bundle of care. First, a nurse-led service was introduced between 2016 and 2018 with a total of 3 full-time PEOLC nurses. These nurses had the role of supporting and educating staff, patients and their families and carers when end-of-life care was required, as well as liaising with palliative care physicians when necessary.

Second, the NSW resuscitation plan was embedded in the hospital system, through a comprehensive education and communication strategy in 2016, which provided a clear way to document the goals of care and limitations of care for each patient. Crucially, this required staff to document who had a conversation with the patient/person responsible regarding these decisions. The nurse practitioner played a key role in the dissemination and implementation of this plan.

Finally, the ‘Last Days of Life Toolkit’ devised by the Clinical Excellence Commission (CEC) was piloted in 2016 and subsequently embedded within every ward of the hospital. This toolkit provided guidance for generalist palliative care providers around end-of-life planning, charts for comfort observations and anticipatory prescribing of commonly required palliative medications.12 It is not possible to dissect which of these interventions was primarily responsible for the improvements seen here and this would have to be subject to further studies.

The in-hospital mortality rate for the ‘late’ group significantly reduced after the above intervention. A potential explanation is the improved capacity to allow for palliation in the home or hospice. Anecdotally, there were a number of times when the team was able to facilitate discharge and dying at home or catalyse transfer to hospice (both public and private), where previously a patient would have been kept in hospital. However, we cannot rule out the possibility that dying patients may have been shifted to other acute facilities or that dying patients potentially remined in ICU due to bed block; this will have to be explored in other datasets.

Quality of PeOLC

Maubach et al15 found that recognition of dying varied based on subspecialty from 72%–90% in an acute Australian metropolitan hospital. The ‘early’ group improved significantly in this regard, with dying now being recognised 90% of the time, the top end of that reported in similar hospitals. Those in the ‘late’ group were recognised as dying prior to death at a similar rate (83%), which is in keeping with that reported in Maubach et al.15 Interestingly, this rate was lower than that of the ‘early’ group, with a possible explanation being that more acute patient deteriorations are easier to detect than chronic ones.

A retrospective study of 320 patients at a public Australian hospital found that of all patients who had died, 28.2% were referred to specialist palliative and end-of-life care.17 Referrals to palliative care increased to a statistically significant degree for both the ‘early’ and ‘late’ groups. These rates (38% and 68%, respectively) are now well above that reported in the literature. Visitation by PEOLC nurses increased significantly, as did palliative care physicians in the ‘early’ group. There was increased visitation by palliative care physicians for the ‘late’ group as well, but this was not significant.

Overall, use of resuscitation plans across the hospital increased significantly from 71.8% to 84.4%. Paradoxically, ACDs fell to 7.5% but this largely reflects changes in the community since patients are admitted with their ACDs and this is not driven from a hospital-based team. These rates are now approaching those reported in Maubach et al15 of 87% and 8% respectively. ‘Early’ group patients had significantly more resuscitation plans made at 84%. ‘Late’ group patients improved to 91% which was not significant, but both groups are now at or above completion rates seen in the literature.

Bloomer et al16 found that end-of-life care was provided to 68.6% of patients in an Australian public hospital. Comfort care rates increased to 78% in ‘early’ group patients and remained constant at 76% for ‘late’ group patients, while palliative medication prescribing rose to 70% in the former and significantly to 70% in the latter. These rates are at least at par with those described in Bloomer et al.16

Regarding expediency of palliative care, the hospital now has earlier writing of resuscitation plans, earlier recognition of dying and earlier initiation of comfort care. However, we have later changes to palliative medications and referral to PEOLC. A possible explanation is that the described interventions have made treating teams more confident in managing patients at the end of their life, before involving specialist palliative care staff.

The use of MET calls has shifted between the two audits, with significantly less patients having MET calls prior to their death and significantly less MET calls being used to make changes in treatment goals. We see this as having less inappropriate MET calls and decisions being made in a more controlled environment by patients, families and treating teams.

To the authors’ knowledge, there is no study in Australian patients outlining the extent to which documented patient wishes regarding end-of-life are followed. Of all ward patients, almost 73% had their wishes known and followed prior to death (as judged by ACDs from the community and resuscitation plans from the hospital), with approximately 12% not having their wishes documented and 15% having a discrepancy between documented wishes and actions taken in the last 48 hours of life. Major discrepancies such as CPR, intubation and vasoactive drugs occurred in very few patients. These rates can be used as a benchmark for future studies.

While the palliative care literature frequently focuses on oncological patients, less than 10% of papers focus on delivery of palliative care to the general population and 1% of papers focus on the final days of life.17 We believe this study contributes to filling a gap in the literature. It supports the use of a nurse-led model of care to achieve a standard of palliative care that matches or exceeds that reported in the literature.

Strengths and limitations

Strengths of this paper include strict reporting criteria for auditors preserved across audits to reduce bias and a relatively large sample size compared with studies looking at quality of palliative and end-of-life care, in inpatient hospital settings. The major weakness of this study is its observational nature; it is uncertain how much of the change seen between the two audits was due to the introduction of the nurse-led model of care versus normal secular trends in patient care. We also cannot dissect how much of the change is due to the direct effect of the nurses on patient care and how much is due to their role in helping disseminate the CEC Last Days of Life Toolkit and the NSW Resuscitation plan. In some ways this point is moot, in that both these tools are actively promoted and used by the palliative and end-of-life care team and should really be considered a ‘bundle of care’.

Another limitation is that this does analysis does not capture the complete impact of the bundle of care. While the audits were a convenient way to gauge quality of PEOLC, they do not look at quality of care for those whose symptoms were controlled and who returned home with community follow-up (just over one-third of patients on average over the study period), or those who were transferred to hospice. During the study period, the service received approximately 700 requests for consults annually, representing about 2500 occasions of service each year. In addition, only about 40% of hospital deaths had the involvement of the service, so the impact of the bundle of care could potentially be higher if more terminal patients were referred.


Overall, the implementation of a PEOLC bundle, with use of the NSW resuscitation plan and the CEC Toolkit, seems to have had a positive impact in a tertiary care hospital, when looking at a ward-based population. Most key aspects of palliative care which were analysed are now occurring at rates in-line with, or better than that which is reported in the literature for comparable hospitals. Nevertheless, this observational study needs to be confirmed with more robust, interventional study designs, for example, stepped-wedge trials or cluster-based randomised trials by hospital, ward or service unit.

Data availability statement

Data are available on reasonable request. The data referred to in this study can be obtained from LB, who can be contacted at

Ethics statements

Patient consent for publication

Ethics approval

Ethics approval was obtained from the Hunter New England Local Health District and deemed 'negligible risk'. Authorisation number AU201909-22.



  • Contributors JL is the overall guarantor for the study and senior author. LB, SF and AS extracted the data for the audit. KP performed the statistical analysis. JA provided epidemiological expertise and assisted with data interpretation. LB and JA drafted the manuscript. All authors reviewed the manuscript and provided critical feedback.

  • 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 and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

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