Objectives This study aims to identify factors among British community-based adults associated with advance care planning engagement. Factors are then compared among six domains of wishes: medical care, spiritual and religious needs, privacy and peace, dignified care, place of death and pain relief.
Methods Cross-sectional data were analysed from a stratified random sample of adults across Great Britain (England, Scotland and Wales) who were interviewed on their attitudes towards death and dying. Weighted multivariable logistic regression tested for associations with expressing any end-of-life wishes and then for each separate domain.
Results Analysis of 2042 respondents (response rate: 53.5%) revealed those less likely to have discussed their wishes were: male, younger, born in the UK, owned their residence, had no experience working in health or social care, had no chronic conditions or disabilities, had not experienced the death of a close person in the last 5 years and feel neither comfortable nor uncomfortable or uncomfortable talking about death. Additional factors among the six domains associated with having not discussed wishes include: having less and more formal education, no religious beliefs, lower household income and living with at least one other person.
Conclusions This study is the first to be conducted among a sample of community-dwelling British adults and the first of its kind to compare domains of end-of-life wishes. Our findings provide an understanding of social determinants which can inform a public health approach to end-of-life care that promotes advance care planning among compassionate communities.
- terminal care
Data availability statement
Data are available in a public, open access repository. Data used in this study are available under an End User License from the ReShare Data Repository operated by the UK Data Service. Accessed from: http://doi.org/10.5255/UKDA-SN-7476-1.
Statistics from Altmetric.com
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.
Dying is set to become more frequent and more complex. The annual number of deaths in England and Wales is set to increase 25% by 2040 with the need for palliative care expected to rise by 42%.1 Behind these deaths lies an increasing number of older people living with frailty2 and multiple comorbidities which undermine traditional trajectories of dying,3 making the task of recognising when a person is approaching the ‘end of life’ more complex.
Advance care planning (ACP) is a tool designed to mitigate such uncertainty by discussing and defining preferences and goals for care with an individual, their healthcare providers and family.4 ACP is associated with receiving care in line with previously defined preferences,5 6 improved quality of life,7 decreased medical expenditure8 and an increased likelihood of dying in a preferred place.9 Despite these benefits, research also shows that individual preferences regarding care, like place of death,10 can fluctuate over time. Recent models have instead reframed ACP as a professional-led process of discussion, rather than definition.11
Discussing about dying is no easy task. Healthcare professionals can be hesitant to initiate conversations with patients,12 especially during prognostic uncertainty.13 Hesitancy is also felt by the British public with around 67% comfortable talking about end-of-life care with their general practitioner,14 and their family members.15 In the UK, efforts have been made by organisations like Dying Matters,16 Good Life, Good Death, Good Grief17 and volunteer-run ‘Death Cafes’18 to encourage community-led conversations and planning for death. In doing so, these initiatives upstream ACP from its origins with professionals in healthcare settings, to community environments among friends, families and neighbours.19
Involving communities in end-of-life care is a priority for the UK.20 Taking a public health approach, models such as ‘compassionate communities’ seek to harness existing, and foster new, networks and community resources to support those facing illness, dying, death and bereavement. They posit that much of what is currently done by professionals can also be done by communities,21 with added scope to alleviate pressures on health and social care staffing and financing.22 Preliminary research suggests an overall benefit of community involvement for end-of-life care,23 though not much is currently understood about ACP in communities.
Previous research on ACP largely takes place in healthcare settings. Studies of community-dwelling adults24–26 also exhibit a clinical focus by examining discussions on preferences for healthcare choices. In doing so, these studies echo the current constructed nature of choice at the end of life27: whereby end-of-life discussions are reduced to choices between preferred care options offered by healthcare professionals. For communities supporting people towards the end of life, an overemphasis on which care options people prefer overlooks the diverse (and often ambitious) things that people may wish to do before they die: which communities may be better equipped at realising.
With a view to supporting communities that can carry out ACP, this study stands as the first attempt to investigate factors associated with discussing end-of-life wishes among community-dwelling adults in Britain. By highlighting associated factors, this study aims to identify inequalities in social determinants of ACP as discussion which can be addressed through a public health approach.
Among a nationally representative sample of community-dwelling British adults, this study aims to identify factors associated with having not discussed any wishes for the end of life with another person.
This study also aims to identify factors associated with discussions on wishes in six domains: medical care, spiritual care, privacy and peace, dying with dignity, place of death and pain relief.
Study design and approvals
This cross-sectional study uses secondary data from the British Social Attitudes (BSA) survey, a national survey conducted by the organisation NatCen, which features questions designed by the Dying Matters coalition on attitudes towards dying.28
Setting and participants
Adults (18+years) from private households across England, Scotland and Wales were invited to a face-to-face computer-assisted interview from June to November 2012.
A multistage, stratified random sampling method was used to identify study participants. Two hundred and twenty-six interviewers distributed three versions of the survey questionnaire (two of which contained questions on dying) and were allocated 10 addresses to visit per version. Interviewers received a briefing session on participant selection procedures and questionnaire administration.
Postcode sectors within Great Britain were selected using the Postcode Address File (PAF): a list of all known UK postal addresses. Sectors were stratified by 37 subregions, population density and percentage of homes that were occupied by their owner. This resulted in 242 postal sectors within each of which 28 addresses were randomly selected leaving 6776 addresses. Institutions, households not registered in the PAF and addresses deemed inaccessible by interviewers were excluded.
Interviewers visited each address from a random starting point selected from the PAF. In cases of multiple residences at an address, only one was selected using computer-generated random selection. The interviewer then selected one respondent from each residence, using the same selection procedure for multiple residents.
The first dependent variable in this study is: whether a person has discussed any of their end-of-life wishes with another person. Interviewers asked: “have you ever actually discussed with someone what your wishes would be in any of these areas, if you did not have long to live?” Participants could then choose one or more domain: (1) ‘medical needs’, (2) ‘spiritual and religious needs’, (3) ‘privacy and peace’, (4) ‘dying with dignity (respectful care and support)’, (5) ‘place of death’, and (6) ‘pain relief’. No formal definitions were provided for domains and hence relied on participants’ interpretation of each domain name. Respondents who had not discussed their wishes were asked: “which of these statements best describes why you have not discussed any of these issues with anyone?” Participants could then answer: ‘I have discussed something (with someone)’, indicating they had discussed wishes outside of the given domains. Both responses were combined to create a binary variable representing participants who had discussed any end-of-life wishes. Six separate dependent, binary variables were constructed to represent each of the six domains.
From previous research, the following variables were included in analyses: sex,24–26 age,24–26 country of birth,24 household income (grouped by government-defined income quartiles),25 presence of a chronic condition or disability (physical or mental)24 26 and religious beliefs.26 Potentially confounding variables were added: country of residence (defined by Government Office Regions), community environment, cohabitation with at least one other person, housing tenure, education (highest qualification), history of employment in health or social care (defined by the Standard Occupational Classification 2010), experience of a close person’s death in the last 5 years, feelings of comfort talking about death and preferred information source on end-of-life care.
Bivariate analyses explored associations between dependent and independent variables using logistic regression. Postanalysis variance inflation factors (VIFs) were estimated and variables demonstrating collinearity (VIF <5) were excluded from models.29
Multivariable logistic regression compared respondents who had discussed any of their end-of-life wishes with those who had not. Separate regression analyses were then carried out for each domain of wishes, resulting in a total of seven multivariable models. Models for each domain were constructed based on bivariate regression analyses and Akaike Information Criteria.30
A weighting factor was used to correct for unequal probability of selection and non-response bias, with selection, non-response and calibration weights.31 Variances in weighted frequencies, proportions and confidence intervals (CIs) were calculated using Taylor series linearisation.32 Missing data (unknown or refused responses) were excluded from analyses. We report crude and weight-adjusted counts, percentages, ORs and 95% CIs.
Prior to visitation, letters detailing the survey procedures and study aims were sent to each selected address with written informed consent taken prior to interviews. A full ethical review of the study’s methodology was carried out in 2009 by NatCen’s Research Ethics Committee with a further review in 2012 regarding the addition of questions relating to death and dying.
Of 6776 addresses, 6070 addresses were accessible to interviewers. Interviews were achieved with 3248 people giving a response rate of 53.5%. Two thousand one hundred and forty-five people were asked questions on dying. Eighty participants refused to be asked any questions on the topic, resulting in 2065 respondents.
Respondents who had discussed any end-of-life wishes with another person occupy a total of 34.1% (n=697/2042) of the sample, weighted: 31.9% (n=657.5/2058). For the six specific domains of wishes (figure 1): 10.8% (n=221/2042), weighted: 10.5% (n=216.6/2058) of respondents had discussed medical needs for the end of life: 12.7% (n=259/2042), weighted: 12.3% (n=252.2/2058) had discussed spiritual and religious needs; 8.2% (n=168/2042), weighted: 7.6% (n=157.4/2058) had discussed privacy and peace; 18.7% (n=382/2042), weighted: 16.6% (n=342/2058) had discussed dying with dignity; 12.2% (n=249/2042), weighted: 11.8% had discussed place of death; and 8.7% (n=178/2042), weighted: 7.9% (n=163.2/2058) had discussed pain relief. Only 0.2% (n=4/2042), weighted: 0.1% (n=3) of respondents had discussed other wishes outside of these domains.
As figure 2 shows, the majority of respondents had only discussed one of the six domains of wishes amounting to 16.2% (n=330/2042), weighted: 15.5% (n=319.6/2058), with 2.4% (n=50/2042), weighted: 2.6% (n=53/2058) reporting discussing all domains with another person.
Any end-of-life wishes
When asked if they had discussed any end-of-life wishes, 12 participants refused to answer and 11 did not know. In the following, we present the results of analyses conducted on 2042 respondents’ data.
Table 2 presents a cross tabulation of respondents’ data by whether they had discussed any wishes for the end of life with someone.
Results of logistic regression analyses
Table 3 presents the results of unadjusted and adjusted logistic regressions for respondents who had or had not discussed any end-of-life wishes.
Domain 1: medical needs
Respondents less likely to discuss their wishes for medical needs were male (adjusted OR (AOR)=0.54, 95% CI 0.34 to 0.85), had no qualifications (AOR=0.43, 95% CI 0.20 to 0.91) or had a secondary school qualification (AOR=0.51, 95% CI 0.28 to 0.92) or had a postsecondary qualification (AOR=0.54, 95% CI 0.31 to 0.92), and no experience working in health or social care (AOR=0.54, 95% CI 0.32 to 0.91).
Domain 2: spiritual and religious needs
Respondents less likely to discuss their wishes regarding spiritual and religious care were male (AOR=0.52, 95% CI 0.34 to 0.77), aged 18–24 (AOR=0.30, 95% CI 0.10 to 0.94) or 25–34 years (AOR=0.42, 95% CI 0.21 to 0.87), owned their residence (AOR=0.46, 95% CI 0.29 to 0.72), had no qualifications (AOR=0.44, 95% CI 0.29 to 0.72) or had a secondary school qualification (AOR=0.51, 95% CI 0.28 to 0.90), no experience working in health or social care (AOR=0.45, 95% CI 0.27 to 0.75), no religious beliefs (AOR=0.53, 95% CI 0.35 to 0.81), no death of a close person in the last 5 years (AOR=0.47, 95% CI 0.31 to 0.73) and were neither comfortable nor uncomfortable (AOR=0.20, 95% CI 0.09 to 0.44) or uncomfortable talking about death (AOR=0.25, 95% CI 0.09 to 0.68).
Domain 3: privacy and peace
Respondents less likely to discuss their wishes for privacy and peace were male (AOR=0.47, 95% CI 0.27 to 0.80), had no experience working in health or social care (AOR=0.53, 95% CI 0.29 to 0.97) and neither comfortable nor uncomfortable talking about death (AOR=0.44, 95% CI 0.21 to 0.91).
Domain 4: dying with dignity
Those less likely to discuss their wishes regarding dying with dignity were male (AOR=0.43, 95% CI 0.30 to 0.61), aged 18–24 (AOR=0.10, 95% CI 0.03 to 0.31), 25–34 (AOR=0.28, 95% CI 0.14 to 0.57), 35–34 (AOR=0.45, 95% CI 0.26 to 0.77) or 45–55 years (AOR=0.46, 95% CI 0.27 to 0.80), had no experience working in health or social care (AOR=0.60, 95% CI 0.39 to 0.93), no chronic condition or disability (AOR=0.65, 95% CI 0.45 to 0.93) and were neither comfortable nor uncomfortable (AOR=0.46, 95% CI 0.27 to 0.79) or uncomfortable talking about death (AOR=0.18, 95% CI 0.07 to 0.45).
Domain 5: place of death
Those less likely to discuss their wishes on place of death were male (AOR=0.45, 95% CI 0.29–0.70), had no experience working in health or social care (AOR=0.46, 95% CI 0.29 to 0.73) and were neither comfortable nor uncomfortable (AOR=0.24, 95% CI 0.11 to 0.49) or uncomfortable (AOR=0.29, 95% CI 0.11 to 0.75) talking about death. Respondents with no qualifications were more likely to have discussed their wishes (AOR=2.18, 95% CI 1.15 to 4.16), compared with those with a university-level qualification (equivalent AOR=0.46, 95% CI 0.24 to 0.87).
Domain 6: pain relief
Respondents were less likely to discuss their wishes regarding pain relief if they were male (AOR=0.38, 95% CI 0.22 to 0.64), living with at least one other person (AOR=0.52, 95% CI 0.31 to 0.85), had a household income equal to or less than £1200 per month (AOR=0.39, 95% CI 0.18 to 0.88), had no experience working in health or social care (AOR=0.46, 95% CI 0.26 to 0.83) and were neither comfortable nor uncomfortable talking about death (AOR=0.32, 95% CI 0.14 to 0.74).
Comparison of independent variables by domain
Table 4 provides a cross tabulation of the results from all seven multivariable regression analyses.
This is the first study to explore factors for discussion about end-of-life wishes among community-dwelling British adults and relies on a multistage, stratified random sampling method with robust population weights for national representativeness.
After exclusion of missing data, a total of 32% of respondents had discussed their end-of-life wishes with someone. Discussions about wishes for dying with dignity were the most frequently reported domain (16.6%) followed by spiritual and religious needs (12.3%), place of death (11.8%), medical needs (10.5%), privacy and peace (8.2%) and pain relief (7.9%).
Factors associated with having not discussed any end-of-life wishes include: male sex, younger age, being born in the UK, owning one’s own residence, no experience working in health or social care, no chronic conditions or disabilities, no death of a close person in the last 5 years and feeling neither comfortable nor uncomfortable or uncomfortable talking about death.
Additional factors for specific domains included: (1) medical needs: not possessing a university qualification; (2) spiritual and religious needs: having a qualification below postsecondary level and having no religious beliefs; (3) place of death: possessing a university qualification; (4) pain relief: a household income of £1200 or below per month and living with at least one other person.
Next, we discuss these findings and their limitations in the context of previous research.
As seen in Belgium,24 Canada25 and the Netherlands,26 males are less likely to discuss their wishes. In the present study, being male was consistently associated with not discussing any wishes or any domain of wishes.
Also in line with previous studies, younger age was associated with not discussing any wishes for the end of life. Additionally, respondents aged 18–24 and 25–34 had a lower odds of discussing spiritual and religious needs. For ‘dying with dignity’, this was seen for respondents aged 18–24, 25–34, 35–44 and 45–54 years.
Those born in the UK were less likely to discuss any wishes. In a Canadian study, those born abroad were more likely to discuss their wishes for healthcare with a healthcare provider.25 In the UK, there are known disparities in place of death for those born abroad, who are more likely to die in hospital.34 However, whether there is a direct link between these discussions and actual location of death is not apparent.
Higher education is associated with ACP engagement in specialist care settings.35 Less education was associated with lower odds of discussing medical needs and spiritual and religious needs, yet curiously, higher odds of discussing place of death. While desiring to express their wishes, people with less formal education may lack appropriate knowledge to discuss potential medical, religious and spiritual care at the end of life. Instead, discussing the topic of where to die may be more accessible to such individuals, and is further encouraged by the current narrative in end-of-life care that a good quality death means death at home.27
Income of £1200 or below per month was associated with not discussing wishes for pain relief. A Canadian study found that having higher than average household income is associated with discussing healthcare preferences with family and friends and lower income with discussing with a healthcare provider.25 In our study, income was not associated with discussing medical needs. We were unable to distinguish who a person had discussed with, hence diverging effects of higher and lower income for family and friends, and healthcare providers, may have been nullified as a result.
Housing and home environment
The role of housing in dying is often overlooked.36 Surprisingly, those who owned their residence had lower odds of having discussed any wishes and specifically spiritual and religious needs. It is unclear why this is. Owning the place one lives may provide a fundamental sense of security which negates uncertainty required to prompt discussion, yet housing tenure was not a factor when discussing place of death. Feelings of security appears to be important for home-based palliative care,37 though further understanding is needed of the relationship between housing tenure and the end of life.
For wishes on pain relief, living with at least one other person was significantly associated with lower odds of discussion. Living with children has been associated with not discussing healthcare at the end of life with a healthcare provider.25 Family members have been highlighted as unwilling to discuss dying and wish to protect their loved ones from upsetting conversations.38 Given that pain is the most widely feared aspect of dying with cancer,39 discussing pain, and its relief, may be particularly upsetting for families. Those living alone may feel more able to discuss about pain with their peers than with family members.
Respondents who had not experienced the death of a close person in the last 5 years were less likely to have discussed any wishes and specifically spiritual and religious needs. This association may be explained by respondents with a recent bereavement having participated in funerals or memorial services for their loved ones. Religious ceremonies may therefore have provided respondents an opportunity to consider how they would prepare their own funeral.
Those not living with a chronic condition or disability had lower odds of discussing any wishes and on dying with dignity. This finding has also been observed in discussions on healthcare at the end of life in Belgium,24 and the Netherlands.26 Unlike the Belgian study, our study was unable to adjust for how often respondents were in contact with their general practitioner. It is unclear whether the lower odds of discussion is explained by a lack of interaction with healthcare professionals. Moreover, it is unclear how participants interpreted the phrase dying with dignity which may have been understood as discussion about assisted dying. The potential for differing interpretations of each domain, resulting from a lack of any formal domain definition, stands as a wider limitation of this study.
Having experience of working in health or social care was significantly associated with discussing any wishes and among all domains. Working in health and social care provides exposure to people approaching the ends of their lives. Caring for those approaching death may allow people to reflect on and share their own thoughts regarding how they wish to die.
Religious belief was only associated with discussing spiritual and religious needs. This is contrary to findings among a Dutch population26 where not having religious beliefs was associated with having discussed preferences for end-of-life care with relatives or healthcare providers. This may be explained by different end-of-life practices in both countries. In the Netherlands, public understanding of end-of-life care may include euthanasia and assisted dying as possible care options. Religious participants who could view these practices negatively may be less likely to discuss their wishes for care in light of this understanding.
In all analyses, being ‘neither comfortable nor uncomfortable’ talking about death was associated with lower odds of discussing any end-of-life wishes and any domain. Interestingly, for spiritual and religious needs and place of death, these odds were even lower than for those uncomfortable talking about death. Individuals who feel neither comfortable nor uncomfortable may have not thought much about their own end of life. When prompted by another person to discuss their wishes, those uncomfortable may have thought sufficiently enough about dying to have some formed wishes, and hence experience discomfort. Location of death and feelings of spirituality and religiosity towards death may specifically be aspects of dying respondents considered less about. Indeed, the final place a person dies and personal feelings of faith could be considered as outside of an individual’s control. Overall, the cross-sectional nature of this study means that the causal direction between comfort talking about death and discussing wishes is unclear: those who have discussed their wishes previously may feel more comfortable having done so.
Practical implications and further research
These findings may be most relevant to a public health approach to end-of-life care when considering how to support communities to discuss their wishes. This approach may feature specific local strategies aimed at the populations highlighted in this study who are less likely to discuss their wishes, in addition to broader systematic responses which modify social determinants to enable discussions on dying. The former would call on actors within communities to lead discussions regarding their specific domains, for instance, religious leaders who can engage their local communities in discussions regarding spirituality and religion at the end of life: thereby increasing their capacity to be involved in care towards the end of life. The latter would require redesigning educational, housing and workplaces environments so as to facilitate discussions about death, though further work to explore such a systematic approach to the end of life is needed.40
Harnessing this study’s findings requires further conceptual clarification of community ACP which discusses wishes for an ideal end of life and not exclusively preferences for care. As such, the completion of formal ACP activities, such as creating advance care plans, which specify care preferences, may arise as a result of upstreaming discussions in the community: activities which are framed by prior community discussion and not the other way around. Promoting this form of ACP could provide communities the opportunity to discuss and then come together so as to realise a person’s last wishes: to enable them to live in the face of death.
The results from this study provide an understanding of factors associated with discussing end-of-life wishes among community-dwelling adults in Britain. These findings can be used when considering how to support community-led discussions on end-of-life wishes as a form of ACP. Further work could explore how the wishes of those in the community can be translated into improving the end-of-life experience together with professionals coordinating their care.
Data availability statement
Data are available in a public, open access repository. Data used in this study are available under an End User License from the ReShare Data Repository operated by the UK Data Service. Accessed from: http://doi.org/10.5255/UKDA-SN-7476-1.
Patient consent for publication
The authors wish to acknowledge the organisations; NatCen and the UK Data Archive of the University of Essex for maintaining and allowing access to these open data.
Contributors JG conceived the project, located the dataset, conducted analysis and drafted the paper. NT oversaw the project and gave feedback on the initial and subsequent drafts of the manuscript. TB gave critical feedback on subsequent drafts of the paper. All authors reviewed and approved the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.