This article sets out to answer the following question by reviewing relevant literature from 2008 to 2018: ‘what factors precipitate admission to hospital in the last few days of a person's life for those who had expressed a preference to die at home?’
The question was identified by a local hospice as an important question for non-cancer patients who, despite expressing a preference for a home death, frequently end up dying in hospital. Pressure group Dying Matters (www.dyingmatters.org) estimates that around 70% of people would prefer to die at home, yet around 50% die in hospital. Promoting patient choice in treatment, management and location of care, including place of death, is core to several UK and international policy drivers (Higginson et al, 2013; Choice in End of Life Programme Board, 2015; Department of Health (DH), 2017; World Health Organization, 2018). These policies are supported by initiatives designed to reshape primary and community care and to reduce hospital admission at the end of life (DH, 2008).
To realise individual preferences and policy drivers, it is important to understand why patients decide to access hospital services in the last few days of life. Senior staff in the local hospice were keen to learn how to identify, engage with and support this large group of patients who do not always get the appropriate care and support required to ensure their end-of-life preferences are met. Working with the local university, staff from the hospice undertook a review of the literature to help identify the patient and family experience of dying at home and why they sought hospital support.
Background
There is extensive literature about preference for place of death, which overwhelmingly identifies home as the preferred choice and dying in hospital as the least preferred option for the majority, although not all, of the population (Gomes et al, 2012; Office for National Statistics, 2016; Coombs et al, 2017; Dying Matters, 2017; Higginson et al, 2017). Evidence at an aggregate population level highlights consistent failure to achieve individual preferences and identifies the potential to reduce hospital admissions in the last year of life (Gill et al, 2015; Office for National Statistics, 2016; Higginson et al, 2017).
However, as Higginson et al (2017) caution, much of this analysis is retrospective; applying this evidence prospectively to an individual patient experience requires a more detailed knowledge and understanding of personal context and circumstances of both preference for place of death and ability to achieve that preference.
As the Office for National Statistics (2016) reported, almost three out of four (74%) bereaved relatives responding to the VOICES questionnaire felt hospital was the right place for the patient to die, despite only 3% of all respondents stating patients wanted to die in hospital. This suggests that, as the end of life approaches, patient and family preferences alter.
Enabling people to die at home requires early recognition of the need for palliative care (Abarshi et al, 2010) and sufficient community resources to help alleviate distressing symptoms that can result in a hospital admission at the end of life and to improve support for family carers (Hanratty et al, 2012; Collis and Al-Qurainy, 2013; Yang, et al, 2016; Spilsbury et al, 2017). However, even patients who are receiving palliative care services can end up using emergency department services and being admitted to hospital at the end of life (Smith et al, 2012; Henson et al, 2016).
Henson et al (2016), in their study of patients with cancer who use emergency department services in the last month of life, highlight that, although there is a good evidence base for the demographic, environmental and clinical risk factors associated with hospital use at the end of life, there is little understanding of why palliative care patients with cancer decide to access hospital services. Non-cancer patients have a less predictable trajectory in general than patients with cancer (Hunt et al, 2014; Yang et al, 2016), so the understanding of why they might choose to access hospital services at the end of life is even more limited.
If clinicians are to uphold policy objectives to meet patient preferences for place of death, it is important they understand why patients decide to access hospital services at the end of life despite their stated preference to die at home.
Methodology
The literature review was undertaken guided by the University of York guide to systematic reviews (Centre for Reviews and Dissemination, 2008). The review was guided by the question: ‘what factors precipitate hospitalisation of patients in the last few hours or days of life?’
The literature search was undertaken using the keywords ‘End of life, or End-of-Life or EoL, or die, or dying, or death’ AND ‘Hospital or admission or admit’ AND ‘Carer or family or relative.’ The asterisk wildcard symbol was employed to truncate search terms and widen the search. Exclusions such as children and maternal deaths were removed from the search using the Boolean operator NOT. The keywords relating to family were included to narrow the search to articles that focused on the family or carer experience immediately before a hospital admission where the patient subsequently died in hospital. From the search, a number of articles that described primary and community care staff experiences of hospital admission where the patient died in hospital were identified and included where the content of the article addressed the review question. The search terms were reviewed by the team and by a university librarian to ensure they were specific in focus to address the research question but broad enough to capture the full range of literature relevant to the topic. The electronic databases searched were the British Nursing Index, Medline, OVID and PubMed. All databases were searched from 2008 to 2018. Filters for peer-reviewed articles and journal articles were applied in each database used.
Inclusion and exclusion criteria were identified (Table 1) and used to ensure that each article selected addressed the research aim of understanding the family decision-making processes at the end of life that precipitated hospital admission where the adult patient died in hospital.
| Inclusion criteria | Exclusion criteria |
|---|---|
|
|
|
Titles and abstracts of all studies identified in the search in each database were scanned and studies that addressed the research question were extracted and the abstracts further reviewed by a second member of the team. Using this search process, a total of 80 studies were found that addressed the research question after duplicates had been removed.
The titles and the abstracts of the 80 studies were screened to assess whether they met the inclusion and exclusion criteria. Thirty-three studies met the inclusion criteria and were available for full review. The contents of the selected articles were manually analysed by two members of the research team to assess whether the studies met the criteria in full. Only 13 studies fully met the criteria. To meet the inclusion criteria in full, the studies had to focus on transition to and subsequent non-sudden or expected death in hospital for adults. Table 2 gives a synopsis of each article selected.
| Author and country of study | Title | Study description and findings | Methods and limitations |
|---|---|---|---|
|
O'Brien and Jack (2010)
|
Barriers to dying at home: the impact of poor co-ordination of community service provision for patients with cancer | A qualitative study, exploring community nurses' views on the availability of community services in influencing why patients continue to be admitted to hospital in the terminal phase of their illness when they have expressed a desire to die at home. Poor discharge planning and coordination, difficulties accessing additional equipment and services, and inadequate out-of-hours services all contributed to hospitalisation at the end if life | Two focus groups with district nurses and community specialist palliative care nurses (n=19). Semistructured, open-ended interview questions to prompt discussion. Summary of key themes at the end of the focus group. Thematic analysis of findings. Only included community nurses in two health districts so sample is limited. Based on professional opinion rather than empirical evidence of effectiveness |
|
Jack and O'Brien (2010)
|
Dying at home: community nurses' views on the impact of informal carers on cancer patients' place of death | A qualitative study exploring community nurses' views on the role of carers in influencing why patients continue to be admitted to hospital in the terminal phase of their illness when they have expressed a desire to die at home. The burden on Informal carers was the main reason identified by nurses for admission to hospital at end of life | Two focus groups with district nurses and community specialist palliative care nurses (n=19). Semistructured, open-ended interview questions to prompt discussion. Summary of key themes at the end of the focus group. Thematic analysis of findings. Only included community nurses in two health districts so limited sample |
|
Smith et al (2012)
|
Half of older Americans seen in emergency department in last month of life; most admitted to hospital and many die there | Study included 2157 participants aged 65 and over who visited the emergency department in the last month of life; 77% were hospitalised. Of those who were hospitalised, 39% were admitted to an intensive care unit and 68% died in the hospital | Secondary data analysis of a longitudinal, interview-based data set with participation rates of 80–90%. Participant data matched to health records. The study covered people who died within 6 months of visiting the emergency department and for whom next-of-kin interviews were available (n=4158). Multivariate logistic regression was used to identify the factors independently associated with emergency department use in the last 6 months and final month of life. Data analysis was limited by the descriptors in the dataset |
|
Gott et al (2013)
|
The nature of and reasons for ‘inappropriate’ hospitalisations among patients with palliative care needs: a qualitative exploration of the views of generalist palliative care providers | Qualitative study of ‘inappropriateness’ of hospital admission for patients receiving palliative care as understood by palliative care health professionals. For most participants, a good death was described as a home death and the absence of unnecessary and invasive medical technologies. Reasons identified for inappropriate admissions were family carers being unable to cope, the ‘rescue culture’ of modern medicine, financing and availability of community services, and practice in aged residential care | Data were collected from 13 interviews and three focus groups with a range of clinicians (n=41), predominantly from hospitals. The interview guide focused on participants' understanding about appropriateness of hospital admission at the end of life. Grounded theory used to analyse the data. All data were collected from a single organisation, which could influence generalisability of results |
|
Lakasing et al (2014)
|
A practice-based survey of patients dying in hospital: can we do more to support end-of-life care at home? | Subjects were the last 50 consecutive patients to have died in hospital registered at a GP practice. Study aimed to ascertain if they could have received all their end-of-life care at home. Despite good district nurse support, identifying those likely to die and predicting death is more difficult in non-cancer than in cancer patients. There is a high rate of admission from nursing homes, and a reluctance by patients and relatives to discuss likely death is mirrored by uncertainty among clinicians | Case note review of the deceased patients (29 men and 21 women). Descriptive statistics used to identify their characteristics. Professional judgement used to decide whether the hospital death could have been avoided |
|
Reyniers et al (2014)
|
What justifies a hospital admission at the end of life? A focus group study on perspectives of family physicians and nurses | Qualitative focus group study about what justifies the admission of a terminally ill patient to hospital at the end of life. Dying in hospital emerged as a central theme in the analysis. Hospital admission was justified when the patient preferred it, when the caring capacity of the care setting was considered to be inadequate and when an acute medical situation arose | Five single-discipline focus group discussions with 23 nurses and 39 family physicians using vignettes describing three cases to generate discussion about what justifies hospital admission at the end of life. Social health perspective adopted in qualitative, thematic data analysis. Limited to generalised perspectives rather than empirical cases |
|
Hunt et al (2014)
|
End-of-life care and achieving preferences for place of death in England: results of a population-based survey using VOICES-SF questionnaire | A survey of bereaved relatives to explore reported preferences for place of death and experiences of care. Of the respondents, 37.7% said the deceased had said where they wanted to die; 49.3% reported that the preferred place of death was achieved. Of those who reported a preferred place of death, 73.9% had a stated preference to die at home; 13.3% achieved this | Census survey of bereaved relatives of all deaths registered in two health districts between October 2009 and April 2010; 1422 bereaved relatives met the inclusion criteria and were 6–12 months after bereavement, and the response rate was 33%. Data were collected in a post-bereavement survey of end-of-life care using Views of Informal Carers—Evaluation of Services Short Form (VOICES-SF). The response rate was low and a degree of non-response bias in the data was noted |
|
De Korte-Verhoef et al (2014)
|
General practitioners' perspectives on the avoidability of hospitalizations at the end of life: a mixed-method study | GPs (n=319) completed a survey about their most recent patient who died non-suddenly and who was hospitalised in the last 3 months of life. Of the 319 patients, 32% died in hospital. Avoiding hospitalisation at the end of life depends on GPs' professional judgement and needs further testing. Of all avoidable hospitalisations, 46% could have been avoided by proactive communication with the patient, 36% by more communication between professionals around hospitalisation, 28% by additional care and treatment at home, and 10% by providing support to patients and families. GPs identified barriers in daily practice, such as the timing of proactive communication with the patient, incomplete information transfer in acute situations, and the lack of awareness among patients and family that death was near | Mixed-method study of a random sample of 950 GPs who met inclusion criteria from 8896 GPs asked to complete a survey. There was a 34% response rate; 18 respondents were purposively selected and participated in follow-up interviews. The survey was designed by the research team and included predominantly closed questions that were analysed statistically. Qualitative interviews were thematically analysed |
|
De Korte-Verhoef et al (2015)
|
How could hospitalisations at the end of life have been avoided? A qualitative retrospective study of the perspectives of general practitioners, nurses and family carers | To describe how hospitalisation at the end of life can be avoided, from the perspectives of GPs, nurses and family carers. Includes a comparative case study of a patient who died in hospital. In a sample of 30 patients who died non-suddenly, 20 were hospitalised and 10 not hospitalised in the last three months of life. Hospitalisation could have been avoided by: identifying the approach of death, and shifting the mindset of patient and family; being able to provide acute treatment and care at home; anticipatory discussions and interventions to deal with expected severe problems; guiding and monitoring the patient and family in a holistic way through the illness trajectory; and continuity of treatment and care at home | In-depth interviews with 26 GPs, 15 nurses and 18 family members associated with the 30 patients at the time of death. Interviews were thematically analysed. The analysis was augmented by comparison of in-depth individual patient case studies. Interviews 12–18 months after death. Possible recall bias |
|
Seal et al (2015)
|
Family stories of end-of-life cancer care when unable to fulfil a loved one's wish to die at home | Qualitative study of 10 individuals who experienced the death of a loved one in hospital despite the deceased's wish to die at home. Reasons for cessation of home care were the need to consider children's welfare, exhaustion and admission by professionals because of a medical emergency | Qualitative in-depth interviews conducted 1–3 years after bereavement. Narrative analysis used to understand family experience. Length of time between bereavement and interview might have affected data quality |
|
Bone et al (2016)
|
Factors associated with transition from community settings to hospital as place of death for adults aged 75 and older: a population-based mortality follow-back survey | Survey of bereaved individuals that found that 32.3% of respondents were carers of a person who had been admitted to the hospital and died there. Transition was more likely for individuals with respiratory disease than for those with cancer or severe breathlessness. Transition was less likely if end-of-life preferences had been discussed with a practitioner and where a key healthcare professional was available | Random sampling through invitation of 882 bereaved individuals who cared for people aged 75 and older who had died in a care home or hospital. Sample size calculation indicated a required sample of 310 responses to give 80% power and P=0.05 significance. There was a 50% response rate (n=443). Results could be influenced by differences between responders and non-responders |
|
Reyniers et al (2016)
|
Reasons for end-of-life hospital admissions: results of a survey among family physicians | Cross-sectional survey of family physicians of adult patients who died non-suddenly and expectedly in a large university hospital between January and August 2014. Factors such as the patient feeling safer in hospital or family believing care to be better in hospital frequently played a role in end-of-life hospitalisation. When patients were admitted with a limited anticipated life expectancy, doctors were more likely to indicate that an inadequate caring capacity of the care setting had played a role in the admission | Four-page survey designed by research team sent to GPs 2 weeks after patient's death. Of these, 352 eligible cases were identified, 245 surveys were returned (response rate 69%) and 189 returned surveys met the inclusion criteria. Survey responses linked to patient record. Descriptive statistical analysis of results provided. Recall and responder bias possible |
|
Krawczyk and Gallagher (2016)
|
Communicating prognostic uncertainty in potential end-of-life contexts: experiences of family members | The study reports findings from a mixed-methods telephone survey of the relatives of patients 4–6 months after the patient's death. The article focused specifically on a subset of respondents whose relatives died in hospital. Relatives would have liked more prognostic information including that the patient was ‘sick enough to die’. Communication was often cloaked in confusing euphemisms or provided false hope, or the message was incongruent with the aggressive medical care being given. Family members who reported good communication about prognostic uncertainty reported better satisfaction with care | Next-of-kin of all adults who died in one health district and met the inclusion criteria (n=332) were contacted; 90 participated. Grounded theory was used to analyse responses to open-ended questions asked in telephone interviews in relation to satisfaction with hospital death (n=67). The study focused on ‘communicating prognostic uncertainty’. Responses were written down by hand by the interviewer, who transcribed them as closely to verbatim as possible immediately after each interview. The study was not designed specifically for qualitative analysis |
Studies that estimated avoidable hospital admission at the end of life, where it was thought, often retrospectively, that the admission could have been avoided but did not show how decisions had been reached were excluded. Studies that focused on end-of-life discussions and advanced care planning were also excluded, as were intervention studies. International studies were included where similar healthcare systems to that in the UK are available.
A content analysis table was produced of the 13 articles included in the review (Table 2). The Critical Appraisal Skills Programme (CASP), 2018) criteria were used to review the quality of each study, and the reviews are summarised in Table 2.
Findings
Seven of the 13 studies used a qualitative methodology, four used a quantitative statistical methodology, one used mixed methods and one used a case note review. Krawczyk and Gallagher's (2016) study is classified as a qualitative study as, although the data is derived from a mixed-methods study, the article reports the qualitative arm. In total, 163 health professionals and 95 family members took part in the qualitative studies, 3715 family members and 189 practitioners were included in the quantitative studies and 50 deceased patients were included in the case note review.
A number of themes emerged from the analysis of these studies, many of which are well documented in the literature. These include the importance of community-based palliative care support for patients and families and advanced care planning.
To address the review question, this article focuses specifically on findings that relate to decisions taken by patients, families and clinicians that result in patients recognised to be approaching the end of life being admitted to and dying in hospital.
Patient and family decision-making
The survey of family physicians by Reyniers et al (2016) found that 138 cases of end-of-life hospitalisation (76%) were requested by the patient themselves. The study found that 35% of the patients felt safer in hospital, and 39% wanted to be treated in hospital. In their earlier qualitative study, Reyniers et al (2014) found that family doctors often felt they were unable to prevent hospital admission at the end of life, even though the patients had been informed of their prognosis and had an end-of-life care plan. These patients were still hopeful of prolonging their life and, to them, hospital was a ‘safer place’ where they considered they would receive the right treatment and care to sustain life.
Ten of the studies (Jack and O'Brien, 2010; O'Brien and Jack 2010; Smith et al, 2012; Gott et al, 2013; Lakasing et al 2014; De Korte-Verhoef et al, 2015; Seal et al, 2015; Bone et al, 2016; Reyniers et al, 2014; 2016) found that family decisions to hospitalise patients at the end of life can be influenced by the carers' emotional and physical ability to cope and to provide the care needed. The availability of equipment to support families to care for patients at the end of life and the lack of integrated out-of-hours GP services were implicated in decisions to hospitalise patients at the end of life by community nurses in O'Brien and Jack's (2010) study. Seal et al (2015) and Jack and O'Brien (2010) found that the carers' ability to cope with the complex physical care required by the dying person was determined by their knowledge and skill, with some bereaved relatives stating they did not have the practical experience and skills required to look after the dying person. Seal et al's (2015) study found some carers had multiple and conflicting caring roles, and chose to hospitalise the dying person so they could attend to other caring responsibilities.
Bone et al (2016) found that deteriorating symptoms, such as severe to overwhelming breathlessness during the last week of life, made transition to hospital twice as likely.
Smith et al (2012) showed that one factor associated with the increase in the emergency department visits and hospitalisation of the dying patient was an increase in physical care needs related to activities of daily living.
Hunt et al's (2014) retrospective survey of bereaved relatives found that, while 91.2% of patients with cancer knew they were dying, only 57.7% of patients with non-malignant conditions were aware they were dying. Those who died from non-malignant causes were less likely than those who died from cancer to die at home or in their place of preference. Both Bone et al (2016) and Hunt et al (2014) found that family members were less likely to request end-of-life hospital admission if they had explored end-of-life preferences and future care plans with a health professional.
Communicating end-of-life care needs with patients and families
Krawczyk and Gallagher (2016) found that nearly half of all family members they interviewed said they would have liked more information about possible outcomes of care, including knowledge that the patient was ‘sick enough to die’. Their study suggested that, from a family perspective, prognostic uncertainty was often poorly communicated if at all; this was often cloaked in confusing euphemisms, or families were left with unwanted false hope, or they witnessed incongruence between the message and the aggressive level of care being provided.
De Korte-Verhoef et al (2015) found that having very clear conversations with the family and patient about the short life expectancy of the patient facilitated a shift in mindset in the family away from the expectations of cure towards an acceptance of end of life and this reorientated the family away from hospital admission. However, this shift in mindset needed time to evolve and become accepted by the family. Family doctors recognised that, for some families, the shift in mindset did not happen and suggested that, for these families, hospital admission remained the appropriate response to the patients' situation.
Studies by Bone et al (2016), Gott et al (2013) and Smith et al (2012) found that opportunities to discuss end-of-life preferences and plan with a palliative care professional were often not made available to patients and families. In De Korte-Verhoef et al's (2014) study, GPs explained that proactive communication often did not take place because the doctors found it difficult to find the right moment to talk about dying. This was particularly difficult in the case of people with progressive cancer who experienced an unexpectedly fast deterioration, for patients who still seemed active and not very ill at first glance, and for those who were deteriorating slowly, as is often the case in the very elderly. According to the GPs, retrospective analysis showed that nearly half of the avoidable hospitalisations could have been prevented by proactive communication. In their later article, De Korte-Verhoef et al (2015) identified the difficulties GPs often experience when asking families to cope with distressing symptoms that the family member felt could be managed better in hospital.
Both Gott et al (2013) and Seal et al (2015) identified a reticence among some bereaved relatives to discuss the condition and needs of the person they had been caring for, which meant they did not always acquire information on how to deal with difficult issues related to the patient's deteriorating physical condition. Lakasing et al's (2014) case note review of deceased patients found that some family members denied the possibility that their relative would die, even in the presence of a noticeably progressive decline in the patient's health. Krawczyk and Gallagher (2016) found that bereaved relatives who recognised that they resisted knowing that the patient was ‘sick enough to die’ at the time of death stated retrospectively that they would have liked more prognostic information that death was a possible or probable outcome of the patient's admission.
The findings from this review suggest that many end-of-life hospitalisations were precipitated by prognostic clinical uncertainty associated with identifying the approach of the end of life in patients, particularly those dying from non-malignant causes, and from carers' ability to acknowledge the situation, commit to and cope with the care required by the dying person at home.
Conclusion
This review was conducted to support a local hospice to develop an evidenced-based approach to end-of-life care by identifying the decisions taken by patients, families and clinicians precipitating admission to hospital in the last few days of a person's life. The focus on the last few days of life reflected a desire by the local hospice to identify where they could most effectively contribute to enabling patients to achieve their preference for place of death while, at the same time, relieving the demand on local NHS services by reducing the number of inappropriate admissions of end-of-life patients (Gardiner et al, 2014). This reflects the End of Life Care Strategy (DH, 2008), which advocates supporting patients to be cared for and to die, in the place of their choosing.
The findings from this review identified a number of barriers experienced by people with non-cancer conditions nearing death that inhibit the transition to end-of-life care. These include hopeful optimism that repeated hospital admissions can prolong life, prognostic uncertainty, which makes it difficult for GPs to know when to discuss end-of-life care, and the time required for patients and carers to transition between first conversations and acceptance of the situation. Health professionals supporting these patients and their families are aware of the end-of-life care needs of the patient and family but, out of a combination of clinical uncertainty and compassion, support the decision to hospitalise the patient.
The findings highlight the potential for considerable ambiguity in relation to the role of hospices in supporting non-cancer patients to achieve their end-of-life preferences. To access the services of a hospice, non-cancer patients and their carers first need to make the transition to an end-of-life care framework. However, because of the barriers to acknowledging this trajectory, many patients and carers do not make this transition early enough to benefit from hospice care.
For patients and carers to recognise that they might benefit from hospice care, clinicians first need to address the clinical uncertainty and hopeful optimism that sustains families and prevents discussion and acceptance of the terminal nature of the illness.
The findings from this review suggest that, if hospices want to support patients to achieve their preferences for the end of life, an important intervention would be to provide support for patients, their carers and clinicians through the uncertainty of a deteriorating condition trajectory before the patient and family have even acknowledged the life-limiting nature of the illness. An exploration of the role of hospices in supporting the transition of non-cancer patients and their carers into end-of-life care would address some of the key issues identified in the review. As Hunt et al (2014) pointed out, encouraging discussion about end-of-life care should not be restricted to those known to be dying as this reduces the potential impact of this aspect of care.
The literature reviewed in this article highlights a focus on the medical management of symptoms using technology and hospital interventions. The hospice movement complements NHS care by focusing on emotional and spiritual sustenance for patients and carers as they come to terms with their experience of deteriorating health and approach the end of life. The findings highlight a role for hospices in providing education and support for GPs to have early conversations with non-cancer patients on their life-limiting trajectory and their family/carers about the deteriorating nature of the condition and the full range of support available to them locally, including hospice care. Emotional and spiritual support for patients whose health is deteriorating and their families and carers opens up opportunities for further, more specific, conversations about patient end-of-life preferences and carers' needs and support, both technical and emotional.
Further studies are needed to identify the impact on the quality of life and achievement of end-of-life care preferences for patients, and the experience of carers' early access to hospice care.