Improving the quality of care delivered to people imminently dying in hospital by implementing a care bundle: an observational before and after feasibility study

Abstract

Introduction

Most Australians die predictably in acute hospital settings. Despite this, hospitals remain ill-equipped to care for dying patients with hospital deaths not uncommonly perceived as distressing by both patients and their families. A care bundle for the dying was developed with this aiming to report the feasibility of implementing this quality improvement strategy.

Methods

A before and after observational approach was used to compare whether there were changes in care delivery to imminently dying patients in two medical wards. Data were extracted retrospectively from their inpatient files. The main outcome of this work was the percentage compliance with the bundle elements and proportional change from baseline where comparative care processes had previously been in place.

Results

Over 6 months, 90 deaths occurred with the bundle in place for 74.5% (n = 70) with significant increases in regular symptom monitoring and monitoring of family distress seen (p < 0.001, respectively). There was compliance with prescribing guideline for pain in 59.2% (29/49) charts and breathlessness in 53.1% (26/49). In the absence of previous prescribing guidelines it was not possible to comment on the significance of this latter observation.

Discussion

This work highlights that it is feasible to implement and assess a pilot project aimed at improving the quality of care delivered to people dying in acute hospitals. This preliminary work suggests that it is possible to integrate evidence-based care processes into the care of the dying by adopting a care paradigm more typically reserved for critical care. However, more work is required to confirm this in larger studies.

Keywords

Quality health care care of the dying hospital care bundle

Introduction

Quality health care is care that is considered patient-centered, evidence-based and safe with one of the intentions of such care delivery being the minimisation of iatrogenic harm.^1–4 However, there is an increasing body of literature that highlights the fact that people dying in hospitals may be exempt from the same rigorous quality standards expected at other points in disease trajectories,^5,6 potentially exposing patients and families to unnecessary risks. For example, data from the UK suggests that up to 60% of people who are in hospital are at risk of distressing physical symptoms especially pain, breathlessness and agitation.⁷ Although people wish to die free of such problems,⁸ this is seemingly often not the case.^9,10 This is problematic not only for the dying person but also for his/her family. This is because previous observations have suggested that when families perceive their relatives’ death as uncomfortable they are more likely to experience more complicated bereavements.^11,12

In the Australian context, the fact that quality care at the time of death cannot be guaranteed to all Australians dying predictably in hospital has been raised as a significant issue. Greater than 50% of Australian deaths annually occur in hospital with 80% of these deaths being the predictable consequence of complex, chronic illness.¹³ This places approximately 120,000 people per annum at risk of not receiving quality-driven care appropriate to their needs. The most common solution proffered is to increase the numbers of home deaths, but there are issues that currently preclude this. Firstly, like other developed countries, the demographics of Australian society are changing with more people living longer with complex, chronic illness, and as part of progressively smaller family units.^14–16 Secondly, community services to support home deaths are limited in their capacity to care for the numbers of Australians dying,¹⁴ and lastly, Australian and international data indicate that for some, hospital is actually the preferred place of death.^17,18 Such observations suggest that the trend towards dying in hospital, either by choice or through circumstances, is unlikely to reverse, highlighting a real imperative to address the issue.

The aim of this project was to develop a care bundle for the dying and then test the feasibility of this approach to improving care delivered to people identified as imminently dying in hospital.

Methods

Setting

The project was conducted in two medical units of a regional teaching hospital in NSW (Australia). This hospital has 200 beds and admits over 16,000 people per year in medicine, oncology, hematology, palliative care and surgery.

Patients

The cohort of interest for this work were people who had been identified as transitioning to the last hours to days of life as an expected part of their disease trajectory. For this project, the documentation that supported that people had entered this phase of life was based on one of two types of written documentation. The first was a “not for resuscitation order” that included a written assessment based on poor prognosis, such an intervention would be futile. The second was documentation that outlined that for this person, death was likely to occur shortly.

Quality improvement process

A structured, sequential quality improvement process was chosen for this work which included: (1) an audit to best understand how usual care of the dying was delivered in order to understand the issues within the larger health care system; (2) establishment of a multidisciplinary team to agree on a solution based on the identified problems; (3) engagement of appropriate stakeholders to assist with the implementation of the solution; (4) piloting the change by engaging staff with the delivery of appropriate education, incorporation of the change into routine clinical use and (5) evaluation of the change.

Understanding usual care

An audit of current practice was conducted with the intent of identifying whether a gap existed between the ideal state as summarised in the literature of people’s expectations and actual care delivered as documented. This audit was undertaken in two regional teaching hospitals in NSW with the approval of the local Human Research Ethics Committee (HREC). This included the hospital where this project was undertaken together with another hospital located close by. In order to standardise the data extracted, two nurses trained in its use utilised a proforma developed specifically for this audit. The proforma’s development was informed by a mixed methods systematic literature review¹⁹ previously undertaken to summarise issues perceived as important by hospitalised patients facing end of life. The files of 100 people who had died in hospital were randomly selected for review. In the time that people had been identified dying imminently, their files were examined to assess whether there was written evidence of monitoring of the dying person’s physical symptoms and, if so, how severe the symptoms were. Other details that were sought included how physical symptoms were palliated and the degree to which families’ needs and distress were considered. Lastly, the numbers of investigations ordered in the final days of life were examined. Of particular interest was whether or not investigations, including blood tests, radiological investigations and screens for infection, were documented as changing patient’s management after the diagnosis of dying had been made.²⁰ The results of the audit are summarised in Table 1.

Table 1.

Summary of results of the initial audit conducted to inform the quality project.

1. Were physical symptoms monitored?

Physical symptoms were not routinely documented. Of the 100 charts audited, only 80 were identified as imminently with only 13% (n = 10) of these having regular pain assessment documented on either on routine observation charts or mentioned in the inpatient notes. Furthermore, breathlessness was highlighted as an issue in only 1% (n = 1) of the files.

2. How symptoms such as pain and dyspnea were addressed?

The poor documentation of people’s experiences precluded understanding of whether evidence-based recommendations were adhered to address problems. Of particular concern was the observation that there was very little difference between the daily doses of opioids prescribed for people who had previously received opioids for longer than 7 days (opioid-tolerant) and those commenced on opioids in the final days of life (opioid-naïve) {oral morphine equivalent 55 mg [SD ± 32.70]³⁰ for opioid-dependent compared with oral morphine equivalent of 68 mg [SD ± 103.57;] for opioid-naïve}.

3. Was families’ distress considered?

Families’ needs were very rarely documented as assessed and as a result, there were very few details to describe what support, if any, had been given to the relatives of the dying person

4. Were unnecessary burdens minimised?

The number of investigations performed including blood work, screening for infections and radiology were examined after a diagnosis of dying had been made. These were considered unnecessary if there was no documentation to advise how these investigations had changed the management of the dying person. The audit revealed 40% (n = 32) of people had such investigations with the most commonly ordered but not acted upon test being a full blood count.

SD: standard deviation; mg: milligram.

Solution development

The audit of care suggested that there were issues that required attention. One solution to try and address these problems was the development of a care bundle. A care bundle is defined as a group of three to five evidence-based care processes systematically grouped together with the intent of improving patient’s access to evidence-based care processes while reducing unwanted care variations^21,22 (Table 2).

Table 2.

Steps in development of a care bundle.²¹

1. Identify the problems to be addressed

2. Collaboratively identify solutions to address the problems

3. Conduct a literature search to identify the best available evidence to implement solutions

4. On the basis of the extracted evidence, develop 3–5 evidence-based interventions to make up the care bundle

When considering the items for inclusion, the literature was reviewed to specifically examine whether evidence already existed to address issues uncovered by the audit. The group involved in this project agreed that such solutions needed to be qualitatively measurable and reflective of practice that was consumer-centered and safe. The final items to be included, the rationale for their inclusion and the level of evidence that supports this are summarised in Table 3.

Table 3.

Bundle items.

Bundle item	Rationale for inclusion	Highest level of evidence to support inclusion
A track and trigger approach to monitoring patients	The baseline audit that documented staff’s observations of patient’s needs were sparse. Few reports highlighted the symptoms that people had, let alone the severity and how such problems were managed. Regular monitoring of people to direct care is routine in health care but the appearance of the chart may impact the quality of the response. This has been highlighted in research undertaken to support the implementation of track and trigger charts which make up the first element of the bundle. Such charts include not only space to document observations but also highlight at which point observations require attention.’	Pseudo-randomised controlled trial (Chatterjee et al., 2005, Christofidis et al., 2013, Kansal and Havill, 2012, McDonnell et al., 2013).
Regularly monitoring people for physical symptoms in particular pain and dyspnoea.	Pain, dyspnoea and other distressing physical symptoms are articulated as issues that those facing imminent death wish to avoid. Populating the track and trigger chart with those symptoms identified as most likely to be bothersome for people aligns care of the dying with best-evidence that highlights that a lack of systematic assessment of symptoms has been identified as the most significant barrier to optimising symptom control.	Systematic review³¹
Regularly screening families for distress	The audits highlighted that very rarely were family’s needs assessed. This is in contrast to families self-identified needs. Studies confirm that regularly screening families for distress objectively reduces the degree of distress that families might experience.	Systematic review (Anon, 2014)
Recommending evidence-based, best-accepted approaches to palliating pain and breathlessness	The audit confirmed that even while symptoms were rarely screened for, actual prescribing practises were poor. Worrying observations included the fact that previously opioid-naïve people were not uncommonly commenced on much higher doses of opioids than would be expected. For this reason, evidence-based approaches to prescribing for pain and dyspnoea as summarised in quality guidelines were included in the bundle.	Systematic review²⁹

The next step was to operationalise this bundle. The initial step was the development of an appropriate track and trigger chart. With permission, the observation chart alluded to in Table 3 was developed with the intent of having a similar appearance to the currently accepted, NSW state-wide routine observation chart.²³ The main difference was that this new chart omitted usual vital signs (blood pressure, temperature and pulse) and replaced them with numerical analogue scores for problems reported as highly likely to be bothersome for dying people. This included pain, breathlessness, agitation and nausea.^24–27 The chart also included the family distress item of the Palliative Care Problem Severity Score (PCPSS),²⁸ a screening tool used in the majority of Australian palliative care services. Like other track and trigger charts, the care of the dying observation chart included explicit calling criteria to help staff gauge when extra support for the dying person or their family was needed.

The second aspect of the bundle required the development of supporting documentation to allow health professionals to understand what is currently considered optimal prescribing to palliate common physical symptoms at the time of death. This included the development of best-practise prescribing algorithms to inform safe symptom control strategies with these algorithms based on current best available evidence.^29–31

Implementing the solution

An implementation plan was devised which included identification of key stakeholders, senior medical and nursing staff, and identification of champions including nurse educators, pharmacists and the medical emergency team. Planned education sessions for as many people as possible were scheduled with the times chosen to be suitable to the working days of these groups.

Overall, 18 sessions were delivered, all of which took the format of face-to-face teaching. The ward nurses required the most sessions, with six sessions targeting a total of 50 nurses. In addition, there were a number of impromptu sessions for nurses who were unable to attend the formal sessions, with this occurring less than 10 times. Other sessions included one for the identified nurse champions for this project (n = 6), one for the medical emergency team (n = 5) combined with intensive care staff (n = 3). Three sessions were conducted with 15 senior medical staff and another with representatives from the hospital’s pharmacy (n = 5). Lastly, four sessions for junior medical officers were held, addressing a total of 21 doctors. Additionally, the project was presented at the hospital’s main education session for doctors, typically attended by approximately 30 people. No assessment of the recall from these sessions was undertaken.

Data collection

Baseline data were collected from a randomly selected cohort of deaths that had occurred from January 2012 to July 2012 inclusively. The bundle was implemented over a 6-month period, commencing July 2013. In order to examine whether the baseline group was similar to the group cared for on the bundle, the following data were collected: demographic details including age; gender; duration of time spent in the terminal phase and cause of death as extracted from the death certificate. The Cumulative Illness Rating Scores were also calculated.³²

For those who died with the bundle in place, and where possible in the audited deaths, the following were collected after the person had been identified as dying:

– Proportions of people who had evidence of regular symptom scores (pain and breathlessness) assessments;

– Proportions of families with evidence that staff regularly assessed their needs as shown by completion of the family distress rating scores;

– Percentage of times that prescribing for pain and breathlessness correlated with the assessments of the severity of the problems documented by staff. This was extrapolated as a measure of compliance with prescribing recommendations.

Dying patients who were prescribed opioids were identified as being either opioid-naïve or opioid-dependant. Allocation to one of these subgroups was based on whether or not they had been prescribed regular opioid analgesia for more than 7 days before they entered the terminal phase of life. For those who met these criteria, the initial dose of opioid prescribed after dying had been diagnosed was extracted. This was converted to the mean oral morphine equivalent (OME) dose in milligrams (mg) using accepted dose conversion tables.³³

Statistical analysis

Extracted data were summarised and analysed using Stata (Version 13.1, StataCorp LP, College Station, Texas, USA, 2014). A p value of <0.05 was considered significant. The data from the audit provided the baseline data against which to compare the effects of the bundle. Descriptive statistics were used to summarise patient’s characteristics. Two sample t-tests of means were used to compare ages and days before death. Two sample tests of proportions were used to compare gender and whether a diagnosis of dying had been made. Chi-square tests for univariate frequency distributions were used to compare the cause of death and cumulative illness rating values. Compliance with the whole bundle was measured as a percentage of the total numbers of deaths over 6 months. Compliance with specific elements of the bundle was extracted from the specific cases where it was in use. In order to provide meaningful comparisons of the change between the baseline audit data and the data collected from those deaths that occurred on the bundle, the proportions of people with symptom scores (pain and dyspnoea scores) and family distress scores were compared using Pearson Chi-square statistics. Lastly, the mean OME in mg of people previously opioid-naïve before death was diagnosed were compared using a t-test with unequal variances with the resulting standard deviations compared using the F-statistic.

Project governance

A detailed description of the proposed project was provided to the local Human Ethics committee allowing this to be approved as a low-risk project. The project was conducted and reported in accordance with the Standards for Quality Improvement Reporting Excellence guidelines.³⁴

Results

Patient characteristics

To validate that the groups of people whose deaths were initially audited on the ward where the project took place (n = 46) were mostly similar to the group who died with the bundle in place (n = 70), the demographic details of both are summarised in Table 4. The major difference between the audit data and the post-intervention data was the cause of death.

Table 4.

Demographic details.

	Characteristics of people whose deaths were audited before the bundle was in place (n = 46)	Characteristics of people who died on the bundle (n = 70)	p Value
Mean age (±SD) in years	Mean 77.0 (11.2)	Mean 72.8 (14.4)	0.10
Number of men/women (missing value)	25/ 20 (1)	30/ F:40 (0)	0.87
Primary cause of death:
% cancer (n)	41 (19)	43 (30)	0.06
% heart failure (n)	30 (14)	10 (7)
% respiratory failure (n)	9 (4)	10 (7)
% sepsis (n)	7 (3)	13 (9)
% neurological failure (n)	7 (3)	7 (5)
% renal failure (n)	2 (1)	3 (2)
% other (n)	4 (2)	9 (6)
% missing value (n)	0 (0)	6 (4)
Cumulative illness rating scores:
% low (n)	45.7 (21)	40.0 (28)	0.03
% medium (n)	47.8 (22)	45.7 (32)
% high (n)	6.5 (3)	14.3 (10)
Is there a diagnosis of dying? Yes/No (n)	45/1	69/1	0.76
Mean number of days before death diagnosed as dying (±SD)	Mean 3.6 (±6.3)	Mean 4.5 (±3.8)	0.34

Use of the whole bundle

Over the 6-month pilot period 94 deaths occurred with 70 of these deaths incorporating the care of the dying bundle reflecting 74.5% compliance (Table 5). With respect to those cared for on the bundle the mean time that people were on the bundle was 1.4 days (SD ± 4.10).

Table 5.

Compliance with bundle elements expressed as proportions.

Bundle element	Baseline audit (%)	After 1 month (%)	After 3 months (%)	After 6 months (%)	p Value
Observation chart	Not routinely in place	16/23 (69.6)	35/46 (76.1)	70/94 (74.5)
Pain scores	11/46 (23.9)	15/16 (93.7)	27/35 (77.1)	49/70 (70.0)	<0.001
Breathlessness scores	2/46 (4.6)	15/16 (93.7)	27/35 (77.1)	49/70 (70.0)	<0.001
Family distress scores	0/46 (0.0)	14/16 (87.5)	27/35 (77.1)	49/70 (70.0)	<0.001
Compliance with prescribing guidelines:
1. Pain	Prescribing guidelines were not in place	7/15 (46.7)	11/27 (40.7)	29/49 (59.2)
2. Breathlessness	Prescribing guidelines were not in place	8/15 (50.3)	14/27 (51.8)	26/49 (53.1)

Compliance with bundle elements

The pre- and post-implementation compliance figures are summarised in Table 5. The cumulative proportions of charts with evidence of monitoring of physical symptoms, family distress and prescribing in accordance with guidelines are summarised in Table 5. The comparison of the proportions of patients with pain scores between the baseline group and post-intervention groups showed a significant difference (0.22 vs 0.70; χ² with 1 df = 27.9, p < 0.001), as did breathlessness (0.04 vs 0.70; χ² with 1 df = 52.0, p < 0.001). Lastly, family distress scoring significantly changed (0.00 vs 0.70; χ² with 1 df = 59.2, p < 0.001). Despite the fact that the frequency of symptom observations had increased, very rarely were observations almost never recorded at the recommended fourth-hourly frequency (Figure 1).

Figure 1.

Run chart to display compliance with documentation of physical symptoms and family distress.

Changes in opioid prescribing

In the baseline group, of the 16 people who fitted this criteria the mean OMED was 90 mg (SD ± 180) compared with 12 people who fitted this criteria where the mean OMED was 55 mg (SD ± 39) after the pilot was commenced. While these figures are insufficient to display a meaningful difference, notably a comparison of the standard deviations of the morphine doses between the two groups using a variance ratio test shows a significant difference (F-statistic with 15 and 11 df = 21.3, p < 0.001). While exploratory, this is suggestive of a reduction in the range of morphine doses being prescribed in the post-implementation group.

Discussion

The aim of this work was to systematically develop, pilot and evaluate a quality improvement approach in order to improving dying hospitalized patients’ access to quality care. This was developed with consideration of consumers’ (i.e. patients and their families) needs, existing evidence and medication safety.

The need for the project was based on a review of the literature and an audit of care with the latter suggesting that the care processes documented in people’s files would not have provided quality care, i.e. care that is patient-centered, safe and evidence-based. The quality improvement solution selected for this site was development and implementation of a care bundle. The results of the work presented here suggest that a care bundle is a feasible approach to improve end-of-life care. This statement is based on the compliance with the initiation of the bundle and the percentage with which the individual elements were completed, particularly tracking of symptoms and family distress.

One of the main intentions of this work was to improve identification and documentation of problems likely to face imminently dying people. Observation charts such as the one developed for this work may play a critical role in supporting staff to document problems and by this action, trigger the most appropriate response. The proportion of people who had documentation of symptoms (especially pain and dyspnoea) and family distress significantly improved from baseline figures, although documentation rarely complied with the recommended fourth-hourly frequency. This failure to monitor physical symptoms at the same frequency of monitoring as usual vital signs (i.e. blood pressure, temperature, pulse rate) has been reported elsewhere in the end-of-life care literature.³⁵ It is not possible to comment here whether this is detrimental to overall symptom control and this question is only answerable with further work. It is also not possible from the preliminary nature of this project to conclude that better documentation will have resulted in better outcomes for the families’ of dying patients. However, good documentation has been identified as a predictive of better patient outcomes.³⁶ The last component of the bundle was the implementation of prescribing guidelines to best palliate pain and breathlessness. As such guidelines had not previously been in place, it is not possible to comment if practice had changed.

One of the most significant concerns from the baseline audit was the observation that medication prescribing was highly variable. Most worrying was the observation that at times, prescribing could be considered unsafe given that some people had been commenced on much higher doses of opioids than would be expected when people were opioid-naïve before they entered the terminal phase of life. Once the bundle had been introduced, it was notable that the average doses of opioids prescribed in those who were opioid-naïve at the time of a diagnosis of dying trended lower. Further, the doses were more aligned with recommended doses.²⁹ However, the small numbers of people to date preclude a more detailed analysis. This work does provide both data to inform sample size calculations for a definitive study and supports the fact that progressing this further is highly feasible.

Strengths of this work

This work, although preliminary, is important for a number of reasons. End-of-life care is not usually seen as the responsibility of health professionals working in acute hospitals but rather the remit of sub-specialty providers such as palliative care services. This belief is paradoxical given that in fact, acute hospitals are the most likely place of death for the majority of people both by choice and circumstance. This pilot work highlights that it is possible to address the care of people dying in hospital and even more importantly, evaluate how this may change the care as documented as delivered. Lastly, this work incorporated accepted approaches to caring for people with complex needs. To date, much of the work that has been undertaken to improve care of the dying has been based on developing separate models not aligned with usual hospital practice. One of the intentions of this quality improvement project was to consider an approach to caring for the dying that reinforces that it is part of usual scope of care provided by acute hospitals. It is imperative to design systems that facilitate care at this time of life delivered with the same attention to quality as at any other time point.

Weaknesses of this work

There are weaknesses with this work. This work is a small non-randomised observational pilot study that compares pre-intervention audit data with post–intervention data. However, comparisons were only possible when at least comparable practices had been in place previously. As such, it does not provide definitive data to support the use of a care bundle to improve end-of-life care. Further, it is not possible to state more than this work has shown an improvement in documentation of people’s symptoms. Definitively recommending this approach will require further research although this work highlights the feasibility of undertaking this necessary research. This work did not allow for clustering despite the fact that it was conducted in a non-randomised fashion in two wards. However, as pilot work, this is acceptable given the baseline data were collected from the same locations prior to introduction of the bundle. The bundle was then introduced simultaneously to the two wards subsequent to which the intervention data was collected. However, the more definitive trial planned to support this approach to care will take this into account.

Future directions

The numbers of people dying in hospital is unlikely to reduce in the imminent future. There is a real need to address the care delivered to people in order to improve the experiences of those imminently dying and their relatives. Further, care at the end of life requires interdisciplinary input to address the issues identified as important to dying people. As with care at any point in disease trajectories, this is best delivered in a coordinated fashion with all involved staff sharing important patient and family information.³⁷ Coordination of care in such a fashion facilitates the provision of the necessary care required to address the often complex needs of dying people and their families. This project was undertaken with the aim of better observing patients and their families, with the expectation that these observations should trigger the most appropriate response.

Based on the work reported here, a larger body of work is planned. This will include both a stepped-wedge, cluster randomised trial together with qualitative family interviews. This work will be conducted across six hospitals in regional NSW with each hospital contributing 35 deaths each. For this trial, an on-line education resource has been developed with the content of this reinforced by a spaced education program. Completion of this program will be accepted as credentialing, with only those credentialed able to participate in care of the dying.

Lastly, there are other areas of care that represent quality of people facing imminent death not covered in this preliminary work. Acknowledging the breadth of the problem it is possible to conclude that perhaps one of the most important aspects of this work is that it may provide a template to inform other investigators aiming also to improve end-of-life care.

Footnotes

Acknowledgments

The authors thanks Research Assistants: Vicki Sproule, Yvonne Harrower; Calvary Mater Palliative Care Staff: Noeline Carson, Erica Cameron-Taylor, Pili Vasquez, Jessica Cain, Lyn Campbell, Milli Sneesby, Mark Mather, Stacey Diana; Calvary Mater Newcastle End of Life Committee: Mary Ringstead, Lyn Herd, Todd Tobin, Linda Liversidge, Tim Stanley, Kim Kolmajer; Hunter New England End of Life Committee: Jane Phillips, Malcolm Green, Susan Newton, Caroline Short, Peter Saul, Jill Lack.

Conflicts of interest

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Anon. Patient-centred care: improving quality and safety through partnerships with patients and consumers Sydney Australian Commission on Quality and Safety in Health Care, 2011.

Anon. Definitions of safety and quality of health care, http://www.aihw.gov.au/sqhc-definitions (2013, accessed 1 August 2014).

Anon. The Healthcare Quality Strategy for NHS of Scotland, http://www.scotland.gov.uk/Resource/Doc/311667/0098354.pdf (2010, accessed 4 November 2014).

Cochrane AL. Effectiveness and efficiency: random reflections on health services. Nuffield Provincial Hospitals Trust 1972: 20–25. (reprinted 1989).

King

Kerr

Walker

. Care for the dying in NSW, Sydney: Clinical Excellence Commission, 2013.

Myers

Lynn

. Patients with eventually fatal chronic illness: their importance within a national research agenda on improving patient safety and reducing medical errors. J Palliat Med 2001; 4: 325–332.

Murtagh

FEM

Bausewein

Verne

. How many people need palliative care? A study developing and comparing methods for population-based samples. Palliat Med 2013; 28: 49–58.

Steinhauser

Christakis

Clipp

. Factors considered important at the end of life by patients, family, physicians, and other care providers. JAMA 2000; 284: 2476–2482.

Hickman

Tilden

Tolle

. Family reports of dying patients' distress: the adaptation of a research tool to assess global symptom distress in the last week of life. J Pain Sympt Manage 2001; 22: 565–574.

10.

Georges

Onwuteaka-Philipsen

van der Heide

. Symptoms, treatment and “dying peacefully” in terminally ill cancer patients: a prospective study. Support Care Cancer 2005; 13: 160–168.

11.

Christakis

. The health impact of health care on families: a matched cohort study of hospice use by decedents and mortality outcomes in surviving, widowed spouses. Social Sci Med 2003; 57: 465–475.

12.

Wright

Zhang

Ray

. Associations between end-of-life discussions, patient mental health, medical care near death, and caregiver bereavement adjustment. JAMA 2008; 300(14): 1665–1673.

13.

Anon. Causes of death in Australia, www.aihw.gov.au/deaths/causes-of-death (2013, accessed 1 August 2014).

14.

Swerissen H and Duckett S. Dying Well Sydney The Gratten Institute; ISBN:978-1-925015-61-4, 2014.

15.

Anon. National End of Life Care Intelligence Network, http://www.endoflifecare-intelligence.org.uk (2010, accessed 1 August 2014).

16.

Beard J and Suzman R. Global Health and Aging, U.S. Department of Health and Human Services: National Institute on Aging, 2011, pp.2–15.

17.

Agar

Currow

Shelby-James

. Preference for place of care and place of death in palliative care: are these different questions? Palliat Med 2008; 22: 787–795.

18.

Gomes

Higginson

. Factors influencing death at home in terminally ill patients with cancer: systematic review. BMJ 2006; 332: 515–521.

19.

Pace

Pluye

Bartlett

. Testing the reliability and efficiency of the pilot mixed methods appraisal tool (MMAT) for systematic mixed studies review. Int J Nurs Stud 2012; 49: 47–53.

20.

Clark

Byfieldt

Green

. Would usual care in of the dying in acute hospitals meet national Australian quality standards? Austr Health Care Rev 2014; 38: 223–229.

21.

Fulbrook

Mooney

. Care bundles in critical care: a practical approach to evidence-based care. Critcal Care Nurs 2003; 8: 249–254.

22.

Barochia

Cui

Vitberg

. Bundled care for septic shock: an analysis of clinical trials. Critical Care Med 2002; 38: 668–678.

23.

Anon. Between the Flags program, http://www.cec.health.nsw.gov.au/programs/between-the-flags (2010, accessed 1 August 2014).

24.

Breitbart

Alici

. Agitation and delirium at the end of life: “We couldn't manage him”. JAMA 2008; 300: 2898–2910.

25.

Fainsinger

Miller

Bruera

. Symptom control during the last week of life on a palliative care unit. J Palliat Care 1991; 7: 5–11.

26.

Klinkenberg

Willems

van der Wal

. Symptom burden in the last week of life. J Pain Sympt Manage 2004; 27: 5–13.

27.

Oi-Ling

Man-Wah

DTSE

Kam-Hung

DNG

. Symptom distress as rated by advanced cancer patients, caregivers and physicians in the last week of life. Palliat Med 2005; 19: 228–233.

28.

Eagar

Green

Gordon

. An Australian casemix classification for palliative care: technical development and results. Palliat Med 2004; 18(3): 217–226.

29.

Anon. Opioids in palliative care: safe and effective prescribing of strong opioids for pain in palliative care of adults. United Kingdom UK Cochrane and NICE, 2012.

30.

Jennings

Davies

Higgins

JPT

. Opioids for the palliation of breathlessness in terminal illness. Cochrane Database Syst Rev 2001; 4: CD002066.

31.

Lorennz

Lynn

. Evidence for improving palliative care at the end of life. Ann Internal Med 2008; 148: 147–159.

32.

Salvi

Miller

Grilli

. A manual of guidelines to score the modified cumulative illness rating scale and its validation in acute hospitalized elderly patients. J Am Geriatr Soc 2008; 56: 1926–931.

33.

Anon. On-line opioid conversion table, https://www.eviq.org.au/ (2009, accessed 2 February 2014).

34.

Ogrinc

Mooney

Estrada

. The SQUIRE (Standards for QUality Improvement Reporting Excellence) guidelines for quality improvement reporting: explanation and elaboration. Qual Saf Health Care 2008; 17: 13–32.

35.

Reid

Gibbins

Bloor

. Can the impact of an acute hospital end-of-life care tool on care and symptom burden be measured contemporaneously? BMJ Supportive Palliat Care 2013; 3: 161–167.

36.

Zegers

de Bruijne

Spreeuwenberg

. Quality of patient reecord keeping: an indicator of the quality of care? BMJ Qual Saf 2011; 20: 314–318.

37.

Schultz

Pineda

Lonhart

. A systematic review of the care coordination measurement landscape. BMC Health Services Res 2011; 13: 119.

38.

Chatterjee MT, Moon JC, Murphy R, et al. The “OBS” chart: an evidence based approach to re-design of the patient observation chart in a district general hospital setting. Postgraduate Medical Journal 2005; 81: 663–666.

39.

Christofidis MJ, Hill A, Horswill MS, et al. A human factors approach to observation chart design can trump health professionals' prior chart experience. Resuscitation 2013; 84: 657–665.

40.

Kansal A and Havill K. The effects of introduction of new observation charts and calling criteria on call characteristics and outcome of hospitalised patients. Critical Care & Resuscitation 2012; 14: 38–43.

41.

McDonnell A, Tod A, Bray K, et al. A before and after study assessing the impact of a new model for recognizing and responding to early signs of deterioration in an acute hospital. J Advanced Nursing 2013; 69: 41–52.

42.

Anon. National Institute for Clinical Excellence. Guidance on cancer services. Improving supportive and palliative care for adults with cancer; the manual. [Internet]. NICE. Available from: http://www.nice.org.uk/nicemedia/live/10893/28816/28816.pdf (2004, accessed 7 October 2014).