Article Text

Download PDFPDF

How common are palliative care needs among older people who die in the emergency department?
  1. Teresa Beynon1,2,
  2. Barbara Gomes2,
  3. Fliss E M Murtagh2,
  4. Ed Glucksman3,
  5. Andy Parfitt2,
  6. Rachel Burman3,
  7. Polly Edmonds3,
  8. Irene Carey1,
  9. Jeff Keep3,
  10. Irene J Higginson2
  1. 1Guy's and St. Thomas' NHS Foundation Trust, St. Thomas' Hospital, London, UK
  2. 2King's College London, Department of Palliative Care, Policy & Rehabilitation, Weston Education Centre, London, UK
  3. 3King's College Hospital NHS Foundation Trust, London, UK
  1. Correspondence to Dr Teresa Beynon, Department of Palliative Care, St. Thomas' Hospital, Westminster Bridge Road, London SE1 7EH, UK; teresa.beynon{at}


Objective To determine the prevalence and nature of palliative care needs in people aged 65 years or more (65+) who die in emergency departments (EDs).

Methods This was a retrospective analysis of routine hospital data from two EDs in South London. Patients aged 65+ living in the hospitals' catchment area who died in the ED during a 1 year period (2006–2007) were included. Palliative care needs identified by diagnosis and symptoms, and problems likely to benefit from palliative care documented in clinical records 3 months prior to the final ED attendance were extracted.

Results Over 1 year, 102 people aged 65+ died in the ED, frequently following an acute event (n=90). 63.7% presented out of hours. 98/102 were admitted by ambulance, over half (n=59) from home. Half (n=50) had attended the same ED or been admitted to the same hospital in the previous 12 months. Over half (58/102) presented with diagnoses that signalled palliative care need. Of these, 29 had recorded symptoms a week before death and 28 had complex social issues 3 months prior to death. Only eight were known to palliative care services.

Conclusions There is considerable palliative care need among older people who ultimately died in the ED, of whom only a minority were known to palliative care services in this study. Previous ED and hospital admission suggest opportunities for referral and forward planning. More older people in need of palliative care must be identified and managed earlier to avoid future undesired admissions and deaths in hospital.

  • Analgesia/pain control
  • emergency ambulance systems
  • emergency care systems
  • emergency care systems
  • admission aviodance
  • prehospital care

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.


Attendances at emergency departments (EDs) throughout England and other countries have been increasing, particularly among older people.1–4 More older people are likely to be admitted to hospital as the population ages; often this is via the ED as the gateway to hospital care. Older peoples' health problems are often complex clinically and managerially, thus time consuming and challenging for medical staff.5 The number of people aged ≥65 years is escalating, with the fastest growing group being those aged ≥80 years. In England and Wales, the numbers in this age group increased by >1.2 million between 1981 and 2007 (from 1.5 to 2.7 million), from 2.8% to 4.5% of the population.6 Increases in attendance or admissions of any age are important as the National Health Service (NHS) faces the challenge of possible reductions in funding.7

The British Geriatric Society and the Royal College of Physicians have both recently published documents on advance care planning, the first recommending that advance care plans are respected when an older person is admitted to an ED.8 This resonates with the UK government End of Life Care Strategy emphasis on improved recognition of an approaching death and advance care planning, with a view to increasing choice, particularly around the place of death.9 Palliative care assessment and interventions targeting ED users may facilitate achievement of this goal, as the ED will rarely be a preferred place of death.10 11 One palliative care service in Nova Scotia, Canada, which implemented an integrated ‘shared care’ model of palliative care, crossing all places of care (ie, hospital, community and hospice) reported a 7% reduction in the number of patients with cancer presenting to EDs in the year after the service started.12 Assessing and managing people with a non-malignant diagnosis and an older population pose tougher challenges.13 14

Little is known about the presence and the nature of the need for palliative care in older people presenting and dying in EDs, and whether their admission may have been preventable. Therefore, this study aimed to determine the prevalence of palliative care needs experienced in the 3 months prior to death, among people aged ≥65 years who died in the ED.



The study was carried out in two EDs in South East London covering an area of 27.8 square miles (44.9 km2) with an estimated population of 537 400 for the two hospitals involved. A total of 9.1% of the local population is >65 years of age; 27.3% of this population describe themselves as non-white. Currently in the UK all patients who require unscheduled care must be seen, assessed and 98% be admitted or discharged within 4 h, with an expected 2% having clinical reasons to remain longer in the ED.15

Data sources

ED, hospital and palliative care databases from King's College Hospital and Guy's and St. Thomas' Hospital were reviewed to identify those ≥65 years of age who presented to the ED during the 1 year period of the study (November 2006–November 2007).

Inclusion/exclusion criteria

We included all patients ≥65 years of age who died in the ED. Patients who lived outside the catchment area of the two hospitals were excluded from the notes review.

Data extraction

We reviewed and extracted data on to a standard template from ED, hospital and palliative care records. A random sample (12/102 records) were independently extracted by two researchers and cross-checked to standardise data extraction at the outset. Demographic information on age, gender and ethnicity was recorded. Borough of residence was determined from the postcode. Data on presentation to the ED were recorded, including time of arrival, means of transport to hospital, place of the incident and presenting complaint. For the purposes of this study, out-of-hours attendance was defined as between 18:00 and 09:00 on week nights and from 18:00 on a Friday until 09:00 the following Monday.

Information was sought on previous admissions to the same hospital in the last 3 months. Palliative care need was determined by identifying those likely to benefit from specialist palliative care involvement, the definition and criteria based on a literature review and expert consensus, derived from the steering group.16–19 The group comprised the authors of this paper (four full-time palliative care consultants, two clinical academic palliative care consultants, three emergency medicine consultants and a research associate/psychologist). Two definitions were agreed—one designed to be very inclusive or ‘broad’ and one more narrowly defined. The ‘broad’ definition consisted of all disease groups commonly referred to palliative care (as people are often referred to palliative care services on the basis of diagnosis alone). It included people with ≥2 co-morbidities measured using the Charlson Index (which describes co-morbidities which might have prognostic significance).20 People meeting the ‘broad’ criteria were viewed as ‘potentially’ having palliative care needs. The second ‘narrow’ definition of need reflects patients more likely to require specialist palliative care, not only the basis of their disease, but also because they had a significant level of physical symptoms or complex social needs, requiring additional help for the patient and/or their family. This ‘narrow’ definition was used to describe a group of people who were thought very likely to have palliative care needs (box 1).

Box 1

Definitions of palliative care need

  • Broad: cancer or chronic obstructive pulmonary disease (COPD) or heart failure or renal failure or liver failure or neurological disease (multiple sclerosis, Parkinson's, dementia, motor neuron disease) or >2 co-morbidities on Charlson Index

  • Narrow: cancer or COPD or heart failure or renal failure or liver failure or neurological disease (multiple sclerosis, Parkinson's, dementia, motor neuron disease) or >2 co-morbidities on Charlson Index20 AND symptoms (pain, breathlessness, nausea, weight loss, confusion, anxiety) and/or complex social needs

Other data

Performance status on admission was estimated from the paper records using the Karnofsky score (ranging from 100, normal function; to 0, dead). This approach to performance status has been successfully used in other studies.21 Any reference to use of end-of-life care tools (eg, the Liverpool Care Pathway for the Dying (the LCP)22 or an advance care plan) was recorded, although the LCP was in use in only one of the two EDs at the time of the study. Medical expectation of treatment as recorded in the notes was included. Data on involvement of community services where recorded in hospital notes were also collected.


The data were analysed using SPSS version 16.0 and are reported here using summary statistics.


The study received ethical approval from the Royal Marsden Ethics Committee (REC number 07/H0801/151) and the Patient Information and Advisory Group (PIAG PI 4-05(o)/2007).


There were 27 543 attendances of people ≥65 years of age across the two EDs over the 1 year period, of which 14 296 (51.9%) ended in admission (table 1). The ≥65 years age group accounted for 11.7% of all attendances but 28.6% of all admissions. However, 46.3% of the 281 deaths in the ED were of people aged ≥65 years. Further detail on this population is reported elsewhere.23 A total of 102 of the 130 (78.5%) deaths of those ≥65 years of age in the ED were recorded as living in the local catchment area (table 2).

Table 1

Annual attendances, admissions and deaths across two emergency departments

Table 2

Data on deaths in the emergency department from the local catchment area

Most of the local older residents who died in the ED were female (n=58), and 24 (23.5%) were from black or ethnic minorities. The ages were evenly spread over three decades, with a median age of 81.5 years (IQR 73–86).

Many of the patients included in this study had full cardiopulmonary resuscitation prior to arrival in the ED and continued to be resuscitated after arrival in ED, presumably while adequate information was obtained. Medical expectation of treatment was recorded in 81 cases, usually by the admitting doctor or consultant. This often noted that the person was imminently dying (n=67) or ‘palliative’ as defined by the person recording or extracting from the notes (n=12) (table 3). Do not attempt resuscitation (DNAR) orders were completed in ED for 32 cases. Preferred place of care (as home) was only noted in one case. No end-of-life care tools were recorded as being implemented and no advance care plans were identified.

Table 3

Medical expectation of treatment in the emergency department and palliative care definition

Using our ‘broad’ definition of palliative care, 58 (56.8%) of included cases had palliative care needs in the 3 months prior to admission to the ED (figure 1). Of these, 38 had one diagnosis, 13 had two diagnoses, 1 had three diagnoses and 1 had four diagnoses. This contrasts with 38/102 (37.2%) of included cases if our narrow definition is used. For this group there was evidence in the notes that they experienced symptoms and/or had complex social needs in the 3 months prior to death. Complex psychosocial needs usually referred to the increasing inability of the patient to care for themselves (n=23) but also included family anxieties (n=3) or difficulty for the carer in coping (n=2). It also included housing issues in two cases and financial concerns in one. Breathlessness was the most commonly recorded symptom (n=23).

Figure 1

Retrospective identification of patients likely to have palliative care needs.

Eight patients were identified as being known to palliative care services; seven of these had cancer as a recorded diagnosis. All fell into the ‘broad’ definition for palliative care and six into the ‘narrow’ definition. The two people who did not fall into the ‘narrow’ criteria were because one had no symptoms recorded (which could have been because these had been managed) and in the other because there were insufficient data. Two people had been seen by both hospital and community palliative care teams; both had cancer.

The case studies (box 2) are illustrative of the differential circumstances, access to services and management approaches underpinning these data. Case 1 shows why an older patient with multiple conditions known to a palliative care service might still be admitted to an ED, in this case due to an unexpected acute event. Case 2 describes a person living alone with little social support in older age, with a problematic symptom (breathlessness) where referral to either palliative care or other community services might have enabled the patient to have more explicit choice over her place of death and would probably have facilitated a better quality of life.

Box 2

Case studies

Case 1

  • A 71-year-old female ex-smoker who lived with her husband was diagnosed with non-small cell lung cancer. Her previous medical history included ischaemic heart disease with a previous myocardial infarction.

  • Her cancer progressed rapidly following diagnosis and she was referred to a community palliative care team. Brain metastases were diagnosed. She was treated with palliative radiotherapy.

  • At 06:00 one morning she collapsed. Her husband called an ambulance. Resuscitation was started but she died in the ED.

Case 2

  • A 69-year-old female with chronic obstructive airways disease and sleep apnoea lived on her own but had relatives living nearby.

  • Two months prior to death she was seen in the chest clinic. It was noted she should have home oxygen and respiratory function tests. No record was found in her hospital notes of referral to a respiratory clinical nurse specialist, district nurse or palliative care.

  • She had a presumed respiratory arrest at home, was found by a close relative and admitted to the ED where she died.


Our findings show that older people, particularly those with the most serious outcomes—that is, those who are admitted or die, are an important group of users of EDs.

The percentage of ED attendances of older people was similar to the percentage of older people in the local population (11.7% attending compared with 9.1% locally). However, both admissions and deaths were proportionally higher compared with the whole population. The admission rate for all age groups attending the ED was 21.3%, but of those attending the ED who were ≥65 years of age, >51.5% were admitted, similar to the data reported by the Department of Health nationally.1 Deaths were also proportionally higher; 0.12% of all age groups died in the ED whereas 0.47% of those ≥65 years of age died, nearly half of ED deaths (46.3%) involving older people. Many of the deaths in the older population (65/102) took place outside of normal working hours. There appeared to be considerable palliative care need in the older population who died; our estimation of prevalence in EDs was 37–57% which is comparable with other studies in France (35% of deaths in ED) and the USA (66.7% of all admissions), although these studies were not conducted exclusively with an older population.

The data are from two EDs in a large multicultural capital so may not be applicable to EDs in other contexts. The study is also limited by the data sources used, which were indirect (from the staff point of view) and reviewed retrospectively. Importantly, we did not review community records which was beyond the scope of this study but would have given a more comprehensive picture, particularly with respect to symptoms pre-dating the terminal admission. It is therefore likely that we did not detect the involvement of all community services, diseases, symptoms and social needs, so our prevalence estimates of palliative care need are likely to be an underestimate. In addition, we did not review cases of older ED attendees who died close to or shortly after admission to the hospital, many of whom were also likely to have had palliative care needs. The UK emergency care performance waiting target, ensuring patients are seen and discharged or admitted within 4 h, may have had an impact on activity, although it is unclear how this would impact on death rates in the ED.

Of those in need, only a minority (all with cancer) were recorded as having been seen by palliative care services. Specialist palliative care services continue to focus on cancer, as reflected in national figures (84.6% of referrals to hospital palliative care services in England, Wales and Northern Ireland in 2006–2007).24 Despite efforts to extend palliative care to all those in need regardless of diagnosis, as proposed in the End of Life Care Strategy, patients with non-malignant diseases continue to be much less likely to receive specialist palliative care.25 Identifying more reliable prognostic markers for non-malignant disease may encourage referral to palliative care, and other services at an earlier stage, which could prevent emergency admissions at the end of life.12 However, equally, identification of patients with complex and multiple needs, using criteria similar to those in this study, may be a more fruitful approach, given the unreliability of prognostic estimates.26 Both the broad and narrow criteria developed for the purposes of this study appeared to be helpful in identifying those that might have had palliative care need, although those who fell into the narrow category were more likely to be the group who would benefit from palliative care.

In some cases it appeared that admission and subsequent death in the ED could not have been prevented; in others, it appeared that pre-existing palliative care need had not been recognised or acted upon. Our analysis and the case studies demonstrate greater need for community palliative care support than was available at the time of this study, for all disease groups, which may be important for advance care planning.25 Ambulance services in particular are clearly critical to appropriate care for patients who are imminently dying and may have palliative care needs, as they are often the first health professionals to attend the patient in an emergency. Exploration of the training needs for ED staff in end-of-life care may be needed since these are the health professionals who support the patients and their families at a critical, sometimes dying, stage. Systems for more rapid exchange of information are also likely to help. Rapid discharge teams and pathways should also be considered, but are currently not widespread.

In the context of an ageing population, death will take place increasingly at an older age and the numbers of older people in need of palliative care will rise dramatically.6 The evidence from this study suggests that there is a cohort of older patients who have emergency health and palliative care needs, who are not known to specialist palliative care services. Some (those fulfilling the ‘narrow’ definition of palliative care needs) due to their disease or co-morbidities and the presence of symptoms or complex social needs are very likely to benefit from specialist palliative care services at an earlier stage. There may be others (some in the ‘broad’ definition of palliative care need) for whom such intensive intervention may not be required, but for whom other community services (eg, GP support, District Nurse visits, paid home carers) would be appropriate. The patients in this study presented to the ED with an acute deterioration but had usually had contact with medical or nursing services within the last year. Intervention from palliative or other community services at an earlier stage may modify the acute crisis for some, if not all, of those older people who currently die in EDs, through provision of more appropriate care and anticipation of crisis through advance planning. A history of previous hospital admissions or ED attendance suggests an opportunity for service redesign, offering a possible trigger for referral to palliative care services and the potential for easing pressure on acute medical services.

More older people in need of palliative care could be identified and managed proactively to avoid future undesired admissions and deaths in hospital. Triggers for emergency admission need to be explored and service developments considered that would effectively enable people with predictable palliative care needs to be managed at home. Further research is needed into the best ways to prospectively identify patients with complex needs who would benefit from palliative care and to establish whether provision of palliative care reduces ED deaths. Prevention of inappropriate emergency admissions and deaths in this way would improve quality of life and choice over place of death, and reduce pressure on busy EDs.


We thank the project advisory group who oversaw the development of methods and study conduct. We acknowledge staff involved in data collection and entry, in particular Laura Skingle, Sam Hart, Dorothee Bechinger-English and Deidre Howley, secretaries and data analysts from the two ED, medical records staff, EPR, PIMS and IT teams within both hospitals and Dr Gao Wei (statistician) of the COMPASS collaborative. The hospitals represented and the Department of Palliative Care, Policy and Rehabilitation, King's College London are all part of King's Health Partners Academic Health Sciences Centre (AHSC), a pioneering collaboration between King's College London, and Guy's and St Thomas', King's College Hospital and South London and Maudsley NHS Foundation Trusts.



  • Funding Guy's and St. Thomas' Charity and Kings College Hospital Charity.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Royal Marsden Ethics Committee (REC number 07/H0801/151).

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

Linked Articles

  • Primary survey
    Malcolm Woollard