Objective The quality of end-of-life (EOL) care in acute hospitals is variable and interventions to improve this care, such as EOL care pathways, are not always used. The underlying reasons for this variability are not fully understood. We explored healthcare professionals’ views on delivering EOL care within an acute hospital trust in the South West of England.
Methods We employed qualitative methods (focus groups, in-depth interviews and questerviews) within a study investigating the impact of a simple EOL tool on the care of dying patients. We invited a range of staff of all grades with experience in caring for dying patients from medicine, surgery and care of the elderly teams to participate.
Results Six focus groups, seven interviews and five questerviews were conducted. Two main themes emerged: (a) delays (difficulties and avoidance) in diagnosing dying and (b) the EOL tool supporting staff in caring for the dying. Staff acknowledged that the diagnosis of dying was often made late; this was partly due to prognostic uncertainty but compounded by a culture that did not acknowledge death as a possible outcome until death was imminent. Both the medical and nursing staff found the EOL tool useful as a means of communicating ceilings of care, ensuring appropriate prescribing for EOL symptoms, and giving nurses permission to approach the bedside of a dying patient.
Conclusions The culture of avoiding death and dying in acute hospitals remains a significant barrier to providing EOL care, even when EOL tools are available and accepted by staff.
- Terminal care
- healthcare professional
- death and dying
Statistics from Altmetric.com
The Department of Health End of Life Care Strategy for England was published with the overall aim of improving the care of dying patients to the level of the best.1 Many patients in the last year of life will require admission to hospital for management of acute exacerbations or complications of either long-term conditions or advanced disease; unfortunately, some will die during these admissions. Also, evidence is emerging that as patients approach death, a significant number change their preferred place of death from home to either hospital or hospice.2 Deaths will therefore continue to occur in acute hospital trusts and efforts should be made to ensure care is at the level of the best and that teams providing end-of-life (EOL) care in hospitals are educated and enabled to provide this.
Studies have highlighted the variability in the quality of EOL care in acute hospital settings across the world. The SUPPORT study in the USA,3 a French survey4 and UK studies5 ,6 have revealed how active interventions are often continued until close to death at the expense of attention to symptoms. The End of Life Care Strategy advocates the use of the Liverpool Care Pathway (LCP) for the dying person7 or equivalent tool to improve care in the last days of life and these have been adopted by many acute trusts throughout the UK. While it has undoubtedly had a significant impact on EOL care in hospitals, the most recent UK National Care of the Dying Audit—Hospitals showed only 29% of patients died with their care directed by the LCP.8 Some aspects of the pathway such as spirituality are consistently neglected. These findings suggest that implementation of EOL pathways is unlikely to be the sole solution for improving EOL care in acute trusts.
The reasons for the variability in the quality of EOL care in hospitals are complex and not fully understood. This study was designed to explore healthcare professionals’ views on delivering EOL care within a study examining the implementation and impact of a simple EOL tool on the dying experience in a large acute teaching hospital.9 The EOL tool consisted of medical and nursing checklists designed to prompt staff to ensure good communication and care, and a symptom observation chart.
We used qualitative research methods—focus groups (FGs), individual interviews and ‘questerviews’10 (interviews held with participants as they completed the EOL tool checklists, using the participants’ responses to the items on the checklist to access attitudes to EOL care) to explore the healthcare professionals’ attitudes toward and perspectives on delivering EOL care. This study took place in medical, surgical and care of the elderly wards. The aim was for the FGs to have a range of staff (junior doctors, registrars, consultants, trained nursing staff and nursing assistants). Sampling was pragmatic; we approached staff highlighted by the ward managers as having experience in EOL care. Once recruitment to the FGs was complete, we conducted in-depth interviews with other nursing staff who were not available to participate in the FGs. The healthcare professionals were given a staff information leaflet and invitation letter detailing the aims of the qualitative component of the study. After at least 24 h, they were contacted by the researcher to see whether they wished to take part.
FGs took place in a quiet room within the workplace. They were facilitated by one member of the research team (KF) using a semistructured interview schedule with other members of the study team (JG or SB) present, taking field notes. They were conducted both before and after the implementation of the EOL tool and lasted from 60 to 120 min. The interviews, lasting approximately 30 min, were conducted by two members of the research team (SB, MB). Interviews and FGs were recorded and transcribed verbatim. Participants were asked about their experiences of providing EOL care in hospital, how different groups of staff interacted during EOL care and about their experience of using the EOL tool once it had been implemented. The questerviews were conducted on the wards as staff completed the EOL tool checklists. Informed consent was obtained from all of the participants.
The initial transcripts from the FGs were scrutinised by the study team to identify emerging themes. Improvements were made to the interview topic guide to allow further exploration of these themes in subsequent data collection. Likewise, the interview transcripts were anonymised and then assessed by the interviewers to inform future interviews. Final analysis of the entire dataset took place once all interviews and FGs were conducted. Three members of the research team (CR, JG, SB) independently read all of the transcripts in order to develop an index of key concepts derived from the data, using the framework approach for qualitative analysis.11 They met to agree and apply this index to the entire dataset which led to its further refinement. These concepts were then assembled within a chart, allowing interpretation of the dataset as a whole. This led to the emergence of two key themes. Two of the FGs were not recorded due to equipment failure but the field notes from these were used in the analysis. Ethical approval was gained from the Frenchay Local Research Ethics Committee (REC), now South West 5 REC.
Six FGs were conducted with a range of staff. These were conducted in orthopaedic, two different medical and healthcare of the elderly wards. In total they comprised two consultants, four specialist registrars, six junior doctors, one staff grade doctor, five ward sisters, eight staff nurses and two healthcare assistants. Thirteen nurses were approached and seven were available for face-to-face interview. Five questerviews were conducted on a medical ward as they completed the EOL care tool.
Two major themes emerged from the dataset: (a) delays in diagnosing dying (due to difficulty, avoidance, and cultural and structural issues) and (b) the EOL tool providing support for staff. Each is presented below with supporting quotations.
Delays in diagnosing dying
Staff reported that the diagnosis of dying was usually made very late, sometimes within hours of death. Delays in moving to comfort care resulted in patients having treatments that staff considered unnecessary.
I'm sure we've all seen patients who are breathless … and the patient is dying, and people will say … ‘Not for escalation to ITU, but they're still for IV antibiotics.’ And … you've got to go and cannulate him because … he's losing a cannula every 20 minutes. And that all comes back to the decision of—we're not making the decision at the appropriate time.
There seemed to be several reasons for this delay: staff found it difficult to be sure that a patient was dying, particularly if they did not know that patient well; the hospital culture meant it was rarely acknowledged openly that a patient might die and so diagnosing dying was avoided; and hospital systems resulted in frequent ward changes for patients, so their medical and nursing teams were not familiar with them. These are discussed in more detail below.
Difficulties in diagnosing dying
Staff consistently described reasons why it was often difficult to diagnose dying. Sometimes this was because they were not familiar with a patient whom they were asked to review out-of-hours, or they lacked experience in diagnosing dying. Staff found it particularly difficult to acknowledge that a patient was dying when prior to hospital admission he or she had been independent and well.
It's very difficult actually on-call yeah. Because you haven't—particularly if you haven't met the patient, and then you're called to see someone who looks as though they're dying. But it's a difficult decision to make to say, ‘OK that's it, we need to withdraw.’
Medical and nursing staff had different opinions in terms of who should make the diagnosis of dying. Medical staff thought the nursing staff were first to raise the possibility that a patient might be dying, perhaps because they spent more time closely observing patients, but nurses looked to the medical team for confirmation.
Um it's still difficult, because we know that patients are coming in, you know, are admitted, and er there's a good chance sometimes that they might die on the admission, but it is difficult for us to know when to pick up this document [the EOL tool] and start. Because you take the lead really from the doctors.
Staff explained that the hospital culture was rarely to discuss or acknowledge death or dying and to treat all patients actively, assuming that active treatments and investigations are always in a patient's best interest. Some spoke of the ‘bravery’ required to state that a patient was dying.
I think we try and say when people are dying. We don't always get it right, but I think we do try and not ignore it. And we do try and be brave enough to say it. Sometimes we should pull out earlier, I'm sure, but we do try and talk about it. Well I don't know, I feel we do—do we?
Furthermore, not actively treating a patient was seen by some as ‘letting them die’.
It's alright when your team is on and you've had that discussion, and … everybody is agreeing to that same plan of action. But if that's not documented word for word, an on-call will see that and think, ‘Well I can't just let that patient die.’
Staff explained how there were different ward cultures, depending on the nature of the specialty and that the same patient might be treated differently on another ward. Different attitudes to EOL care between consultants also affected decision-making by junior doctors and therefore had an impact on patient care.
And ... different consultants have different opinions ... Um and that's very difficult to work with as a junior doctor. You may agree with one more than the other, but you have to go with their opinion.
The nursing staff interviewed described how the junior doctors were also anxious about prescribing medications used at the end of life, and this often led to delays in appropriate drugs for symptom control being prescribed. Once a decision had been made to change from active to comfort care, it seemed that nursing staff then led the patient's care, with even junior doctors having little further engagement, other than the writing of prescriptions.
It's a daily review, and then most of the times it is a, you know, writing the notes ‘patient comfortable’, if they're on a syringe driver you'd rewrite it, ‘24-hour syringe driver rewritten’, any adjustments that need to be made etc. And then it's just continue for TLC. So it tends to be just like the clerical side of it.
Hospital shift systems causing difficulties in care of the dying and out-of-hours assessments by junior doctors were mentioned frequently. When doctors were asked to assess patients whose care was led by other teams who had not documented clearly the plan of care or ceiling of treatment, they were uncertain about changing to comfort care, even if they felt it would be appropriate.
Another hospital system that did not facilitate good EOL care was the frequency of senior reviews of patients. Junior doctors and nursing staff described needing validation of the diagnosis of dying by the senior medical staff (usually the consultant), and thus asked for this confirmation on the ward round. However, senior staff felt that they did not see patients frequently enough to be confident about diagnosing dying.
Staff also complained about the movement of patients to different wards around the hospital which had an impact on both their ability to diagnose dying and communication with the dying patient.
I suppose the immediate problem in this setting is the fact that you encounter somebody for the first time and they die. And typically we don't even get continuity in death, so patients are shuffled between wards because of the bed pressures and because of the side room issues and the hospital acquired infections. And so frequently even dying patients will meet multiple consultants.
A practical issue was the availability of single rooms. In the past these rooms would have afforded privacy for dying patients, but staff described how usually they are prioritised for patients who pose an infection risk to others. The lack of side rooms for dying patients distressed staff, who then did their best to move patients to quieter areas of the ward.
So if we get a poorly patient, if they're on an open ward, we then try to like put them on um Bed 16 or Bed 1, so there will be a little bit of—there will be privacy.
EOL tool providing support for staff
Staff were positive about the EOL tool describing how it provided a structured approach to EOL care. Previously when decisions were made to stop active treatments and routine observations, the nurses felt as if they ‘weren't doing anything’. The symptom assessment chart in the EOL tool encouraged and enabled nurses to approach the patient and check on symptoms, which in turn meant families could see staff ‘doing something’.
The nurses also mentioned that the EOL tool empowered them to challenge some of the medical interventions that were being continued even when a patient was thought to be dying and whose care was being directed by the tool.
The anticipatory prescribing guidelines incorporated within the tool were considered useful by medical staff who said they had not been taught prescribing for EOL symptoms at medical school. It was felt the tool improved documentation of care but it also functioned as a means of communicating appropriate interventions for out-of-hours staff.
It's almost like putting a sign over their head—this person is dying please keep them comfortable, do not take blood, do not do obs, do not do anything that may cause discomfort unnecessarily, treat symptoms and don't worry about, you know, anything else ...
However, during one of the questerviews, a staff nurse completing the tool said he had not been able to tell the relatives that the patient was dying. Even at this stage, it would seem that the culture around death and dying prevented those relatives from being fully informed about the seriousness of the patient's condition.
Well I … discussed with the family so far about the fact that he's deteriorating … but I have said that the weekend … um … we will just have to see how he—um—goes over the weekend but it is possible that he will continue to deteriorate … but I didn't mention the word death, yet.
This exploratory study has highlighted the difficulties surrounding providing good EOL care in an acute hospital trust. Nursing staff were often the first to recognise the patient was dying, but looked to the medical team for confirmation. Nursing staff felt they could not change the focus of care without a doctor's decision and would therefore deliver interventions to patients even when they felt uncomfortable doing so. However, they continued to request a review of the focus of care for the patient. It was striking how medical staff described the difficulty in deciding when a patient was dying and withdrawing active care. They indicated that their fear was that by not instituting a further course of treatment they would be culpable in the patient's death. This situation was worsened by the hospital culture of not considering or discussing death, or acknowledging it as a possible outcome. This meant diagnosing dying required ‘bravery’ and might be viewed by colleagues as being nihilistic, since decisions to move to comfort care were seen as 'giving up’, or worse, ‘allowing’ a patient to die.
Junior doctors conducting out-of-hours reviews looked to the medical notes to guide them, but the culture of treating actively as default and failure to acknowledge dying as a possible outcome meant that no hospital systems had been implemented to clarify appropriate treatment goals within the medical notes. They might recognise a patient was dying but were unlikely to suggest a move to comfort care. Generally it was felt that senior medical staff should make the diagnosis of dying and lead the change to EOL care, despite them reviewing the patient with the least frequency. This meant that either the diagnosis was further delayed while a consultant opinion was awaited, or perhaps did not happen at all since the consultant might feel they did not know the patient well enough. As a result, a patient's deterioration might be recognised by ward staff, but a shift to more appropriate goals of care did not happen and a patient received active treatment until death.
The EOL tool was welcomed by staff as a means of standardising care, communicating goals of care, prioritising the assessment of symptoms and facilitating appropriate and timely prescribing of EOL medication. However, hospital culture remained such that even when the tool was being used, open discussions about death did not always take place, contrary to guidance within the tool.
This study has strengths and limitations. Its strengths are that it used qualitative research methods with multidisciplinary staff of all levels to explore issues around the diagnosis of dying and the delivery of EOL care in an acute hospital. This has highlighted the cultural and structural issues around the difficulties of diagnosing dying. The main limitation of the study was that it was conducted in one teaching hospital which might limit the generalisability of the findings. The participants might have been influenced by the research team who were mainly specialist palliative care colleagues and who were seeking to pilot the introduction of a simple EOL tool that could change clinical practice. However, the staff did not seem inhibited; instead, they were honest about the difficulties they faced which resonated with clinical experience. While the findings do need to be confirmed in other hospitals, the strength and consistency of the major themes and their plausibility suggest they might be generalisable.
These findings resonate with other studies. The difficulty in diagnosing dying is possibly not a surprise since it has been shown previously that doctors often only really learn about death and dying after they qualify; as medical students they are taught to manage long-term conditions but with little emphasis on understanding either acute or long-term prognosis.12 Gott and colleagues interviewed primary and secondary care professionals about transitions to palliative care in acute hospitals.13 In their interviews, junior doctors also spoke of nursing staff questioning care and the consultant leading decision-making but our interviews explain how these factors subsequently affect patient care. Sheward and colleagues sent a questionnaire survey to staff in a regional hospital, collecting qualitative data in the form of free text comments.14 The five themes generated were similar to those in our interviews, but by virtue of the methods used lacked depth and have been strengthened by our data. A large study from two medical centres in the USA asked physicians to discuss the care of their most recent patient who had died.15 Over a third questioned the care received by patients at the end of life, and described difficulties for physicians in knowing when to stop active treatment; however, the culture in which these physicians were working was not described. A recent UK Nursing Times survey discovered that nearly one in four nurses involved in EOL care do not feel competent to broach the subject of death with patients.16 This might explain why, although nursing staff were able to recognise dying in our study, they needed confirmation from senior medical staff. Many other studies have described the positive attitudes of nursing staff towards EOL care pathways, in particular the LCP,17 but this is the first time an EOL tool has been described as having a role in empowering nursing staff to approach the bedside of a dying patient.
This research confirms the urgent need to tackle the culture of hospital medicine that ignores death as a potential outcome for patients with life-threatening illnesses until the last hours of life. Death should be discussed openly and we need to encourage teams to strive for ‘good deaths’: deaths that are not ‘failures’ of medicine, but are an inevitable consequence of advanced disease or long-term conditions. If death was acknowledged as a possibility from earlier in a patient's admission, then the recognition of the dying patient might be easier. More importantly, communicating therapeutic uncertainty to patients themselves might allow them to make decisions about both the intensity and location of their ongoing care. Interventions such as the AMBER care bundle,18 which promote early recognition of patients with an uncertain prognosis, are designed to promote this culture change. However, we are not aware of any published evidence thus far that this is achieved. The practice of some specialist palliative care teams has been modified in order to effect some of this change, with teams now attending board rounds in acute assessment units and prompting discussions about prognosis.19
These interviews highlight the challenges that hospital teams face while managing dying patients. Appropriate EOL care can only be delivered if hospital culture accepts death and dying as a possible outcome for patients. Changing culture and improving communication have greater potential to influence hospital EOL care than simply implementing EOL care pathways, since the latter can only change care for those patients recognised as dying and, even then, usually only for the last hours or days of life. Future efforts should be directed at addressing and changing hospital culture so that EOL care can be improved for the majority.
We would especially like to thank all of the ward staff who participated in this study. Thanks also to Anne Currie for extensive administrative support and the UHBristol Specialist Palliative Care Team for clinical support. We are grateful to Professor Jenny Donovan for helpful comments on earlier versions of the manuscript.
Contributors KF proposed the idea for the study. KF and JG developed the study. KF facilitated the focus groups and SB and MB conducted the indepth interviews and questerviews. RMcC and KF supervised the research team. JG, SB and CR conducted data analysis. CR wrote the first draft of the manuscript and all authors contributed to the final paper. CR is the guarantor.
Funding The research nurses and JG were funded by the European Palliative Care Research Collaborative (http://www.epcrc.org/); however, this funding was not specifically for the purpose of this study. We would also like to thank the UHBristol ‘Above and Beyond’ charity for a donation to cover stationary and transcription costs.
Ethics approval South West 5 REC.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement We do not currently have any plans for data sharing but would consider appropriate applications for use of the data.
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.