Article Text

Download PDFPDF
End-of-life discussions in residential care homes improves symptom control: a national register study
  1. Sofia Andersson1,
  2. Lisa Martinsson2,
  3. Carl Johan Fürst3 and
  4. Margareta Brännström4
  1. 1Department of nursing, Umeå University, Skellefteå, Sweden
  2. 2Department of Radiation Sciences, Umeå University, Umeå, Sweden
  3. 3The Institute for Palliative Care, Lund University and Skåne Regional Council, Lund, Sweden
  4. 4Department of Nursing, Umeå University, Skellefteå, Sweden
  1. Correspondence to Dr Sofia Andersson, Department of nursing, Faculty of Medicine, Umeå University, Umea, Skellefteå, Sweden; sofia.andersson{at}umu.se

Abstract

Background In Europe, residential care homes (RCHs) rather than hospitals are the most common care facilities for the older adult and the place where most deaths occur in this age group. There is a lack of knowledge regarding end-of-life (EOL) discussions and how they correlate with symptoms and symptom relief.

Objective The aim was to examine the correlation between EOL discussions and symptom occurrence, symptom relief and prescriptions or PRN drugs against symptoms for care home residents.

Methods All expected deaths at RCHs from 1 October 2015 to 31 December 2016 and registered in the Swedish Register of Palliative Care were included. Univariate and multivariate logistic regression were used to compare the RCH residents, or their family members, who had received documented EOL discussions with a physician (the EOL discussion group) and the non-EOL discussion group.

Results The EOL discussion group (n=17 071) had a higher prevalence of pain, nausea, anxiety, death rattles and shortness of breath reported, compared with the non-EOL discussion group (n=4164). Those with symptoms were more often completely relieved and had more often been prescribed PRN drugs against that symptom in the EOL discussion group. All differences remained significant when adjusting for age, time living in unit and cause of death.

Conclusion The results indicate that EOL discussions are correlated with higher prevalence of symptoms, but also with better symptom relief and prescription of symptom drugs PRN when symptomatic. A possible explanation for this is that the EOL discussion can work as an opportunity to discuss symptoms and treatment for symptom relief.

  • symptoms and symptom management
  • nursing home care

Data availability statement

No data are available. All data relevant to the study are included in the article.

Statistics from Altmetric.com

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.

Key messages

What was already known?

  • End-of-life discussions with physicians occur to a low extent.

What are the new findings?

  • End-of-life discussions are correlated with better symptom relief and prescription of symptom drugs when symptomatic.

What is their significance?

  • Clinical: End-of-life discussions are important for att better quality of care.

  • Research: End-of-life discussion is a forum for detecting symptoms, and that it thus enables symptom relief.

Background

A ‘good death’ is a death that proceeds in accordance with the wishes of dying persons and their family members.1 To make this possible, end-of-life (EOL) discussions could be an important issue.1 This principle has influenced the development of the Swedish Register of Palliative Care (SRPC). The SRPC is a nationwide register that focuses on the last week in life. The register includes questions about quality of death, diagnosis and age. Data are collected from different care settings. About 60% of all deaths in Sweden were reported to the register during 2013–2017.2

The SRPC collects data about EOL discussions with residents using this question: ‘Did the person receive information about the transition to EOL care, that is, an individually tailored and informed conversation with a physician that is documented in the medical records about being in the final stage of life and about care being focused on quality of life and symptom relief?’. The following question is used by the SRPC to collect data about EOL discussions with family members: ‘Did the person’s close friend(s)/relative(s) receive information about transition to EOL care, that is, an individually tailored and informed conversation with a physician that is documented in the medical records about being in the final stage of life and about care being focused on quality of life and symptom relief?’.3 The definition of EOL discussion used in this study was ‘yes’ reported to these respective question to the SRPC.

Residential care homes (RCHs), also called nursing homes or long-term facilities, are common places of death for older adults.4 5 In Sweden, RCHs are the most common places of death for persons older than 65 years.6 Smedbäck et al7 found in SRPC data from 2011 that only a quarter (27.2%) of the older adults in RCHs received EOL discussions and more than half (53.9%) of the family members received EOL discussion.

Martinsson et al report that family members of patients dying from dementia receive EOL discussions less often than those of patients with cancer.8 A register study shows that people with advanced dementia are provided with an overall better quality of EOL care in nursing homes compared with hospitals.9 However, this is not the case regarding EOL discussions with family members. Family members to people with dementia who died in hospitals (81%) more often received EOL discussions compared with family members to people with dementia who died in RCHs (71%).9

Studies from various settings and examining different diagnosis groups have shown that EOL discussions with physicians occur to a low extent.7 10 11 A study reported less than one-third (28%) of the patients with heart failure and more than half of their family members (60%) received EOL discussions with a physician.10 Furthermore, in a study of people with kidney diseases, 41% of the patients and 71% of their family members received EOL discussions with a physician before death.11

Earlier studies have shown that EOL discussions are associated with better symptom relief. A study of patients dying in specialised palliative care, inpatient hospital care and community/other care (including RCHs) reported that not having EOL discussions was associated with increased risk of unrelieved pain.12 Another study reported that EOL discussions between the care home residents and a physician were associated with higher probability of relief of nausea. Further, the study also showed that EOL discussions between the residents’ family members and a physician were associated with higher probability of relief of pain, anxiety and shortness of breath, when present, among the residents.13 On the other hand, another study reported that there were no differences in symptom relief when cancer patients were informed or not about imminent death.14

No studies were found comparing symptom occurrence between RCH residents who receive EOL discussion and residents who do not.

Aim

The aim was to examine the correlation between documented EOL discussions with a physician and symptom occurrence, symptom relief and prescriptions of PRN drugs against symptoms for care home residents.

Methods

Design

The study was a retrospective study design using a national quality register.

Sample

Data about all expected deaths in RCHs in Sweden between 1 October 2015 and 31 December 2016 and reported to the SRPC were collected from the SRPC. Only deaths that were reported as expected based on the disease trajectory were included. An expected death was defined as the answer ‘yes’ to the question ‘Based on the disease trajectory, was the death expected?’ reported to the SRPC.3 The inclusion of only expected deaths is because symptom data in the SRPC are not collected for residents who have been reported to have died unexpectedly based on their disease trajectory. The SRPC focuses on quality of EOL care, and data are gathered retrospectively after the resident’s death. Data are most often reported by a registered nurse.

Variables

The design of the SRPC web questionnaire does not allow items to be left blank at submission. Background characteristics (age, gender, days living at the RCH and cause of death) were collected. Cause of death data in the SRPC database contain individual-level information according to the 10th revision of the International Classification of Diseases. These data are retrieved by the SRPC from the Cause of Death Register at the Swedish National Board of Health and Welfare. Residents with missing cause of death data were excluded.

Two variables describing EOL discussions were collected: documented EOL discussions between the resident and a physician and documented EOL discussions between the family members and a physician. These variables were used to dichotomise the residents into two groups, ‘EOL discussion with the resident and/or family member’ (EOL discussion group) versus ‘no EOL discussion with neither nor the resident or family members’ (non-EOL discussion group). If the resident and/or family member had an EOL discussion reported, the resident was assigned to the EOL discussion group. If neither the resident nor the family received an EOL discussion, the resident was assigned to the non-EOL discussion group. The SRPC does not contain information about how long time before death the EOL discussions were performed. Since RCH residents are older people with multiple diseases, it is probable that a majority of the EOL discussions reported had been performed before the last week in life. If both items were answered ‘Don’t know’, or one of them was answered ‘No’ and the other with ‘Don’t know’, the resident was excluded from further analysis.

Variables describing reported prevalence of occurrence and relief of pain, nausea, anxiety, death rattles and shortness of breath during the last week in life were also collected. Cases where symptom occurrence was unknown (‘don’t know’ answers) were excluded from the analysis. When analysing symptom relief, only the residents who were reported to have suffered from that corresponding symptom were included. The variables covering symptom relief were dichotomised into two groups: totally relieved versus partly relieved or not relieved at all. Additional variables collected were prescription of individual injections as needed for symptom relief of pain, nausea, anxiety and death rattles. The variables covering pro re nata (PRN) drugs were analysed as yes, PRN drug prescribed versus no PRN drug prescribed. ‘Don’t know’ answers were excluded.

Statistical analysis

Statistical findings are described as mean, median, SDs and proportions.

The χ2 test was used to determine the differences regarding reported gender, length of stay in RCHs before death (<100 days, 101–365 days, 366–1825 days and >1826 days) and cause of death (cancer C00–D48, mental and behavioural disorders F00, F01, F02, F03, G09, G30, cardiovascular diseases I00–I99, other A00–B99, D50–D89, E00–E90, G00–G99, J00–J99, K00–K93, L00–L99, M00–M99, N00–N99, Q00–Q99, R00–R99 and V01–Y98). Following the method from a previous study using a similar material, the variable lengths of stay in RCHs before death were categorised into four groups in the analysis.11

First, a univariate logistic regression model was used to test whether prevalence of symptom, symptom relief and prescription of PRN drugs differed between the EOL discussion and non-EOL discussion groups.

Then, a multivariate logistic regression model was developed to test whether prevalence of symptom, symptom relief and prescription of PRN drugs differed between the two groups, adjusted for age (continuous), length of stay in RCHs before death (<100 days, 101–365 days, 366–1825 days and >1826 days) and cause of death (cancer, mental and behavioural disorders, cardiovascular diseases and other).

Statistical significance level was set at p<0.05. SPSS Statistics V.23.0 (IBM) was used for all statistical computations.

Results

Background characteristics

The sample identified 22 855 residents. Of these, 414 were excluded because of missing data regarding underlying cause of death, and 1206 were excluded because of the two EOL discussion items being answered with ‘don’t know’ or one of them being answered with ‘no’ and the other with ‘don’t know’. EOL discussions with the resident and/or family member had been performed for 17 071 residents, while no EOL discussion had been performed for 4164 residents (table 1).

Table 1

Documented EOL discussion with residents and family members

Mean age in the EOL discussion group was 87.7 years, and 88.5 years in the non-EOL discussion group (p<0.001) (table 2).

Table 2

Background characteristics

The majority of residents were female (65% in both groups) and did not differ significantly between the groups. The most common cause of death was cardiovascular diseases (37%), and these were more prevalent in the EOL discussion group (46%) compared with the non-EOL discussion group (21%). The second most common cause of death was mental and behavioural disorders (30%), which were less common in the EOL discussion group (10%) compare to the non-EOL discussion group (64%). More than 1 in 10 (12.5%) residents had lived in RCHs less than 100 days, while almost two of five (17.7%) had lived in RCHs for 101–365 days. It was most common (53.8%) that the resident had lived in RCHs between one and 5 years. The non-EOL discussion group had spent longer time in the RCHs before death compared with the EOL discussion group (table 2).

Symptom prevalence and symptom relief for the total study population

The most commonly reported symptoms were pain (69.2%), death rattles (49.3%) and anxiety (44.4%). Shortness of breath (14.2%) and nausea (10.3%) were less common (table 3).

Table 3

Comparision of symptoms prevalence between the EOL discussion group and the non-EOL discussion group

Of the residents who were reported to have had those symptoms, respectively, pain was the symptom that was most often totally relieved (85.2%), followed by anxiety (77.4%), nausea (58.9%), death rattles (59.1%) and shortness of breath (48.1%) (table 4).

Table 4

Comparision of relief of symptoms for those who were reported to have suffered from that corresponding symptom between the EOL discussion group and the non-EOL discussion group

Most of the residents had PRN drugs prescribed for symptom relief (pain 94%, nausea 80%, anxiety 92% and death rattles 92%). For those who suffered from the corresponding symptom, prescription of PRN drugs was even more common (pain 98%, nausea 91%, anxiety 97% and death rattles 97%) (table 5).

Table 5

Comparision of PRN drugs against symptoms between the EOL discussion group and the non-EOL discussion group

Differences of symptoms and symptom relief between the groups

Pain, nausea, anxiety, death rattles (p<0.001 for all four) and shortness of breath (p=0.028) were reported more often in the EOL discussion group compared with the non-EOL discussion group (table 3). After adjusting for age, time living in unit and cause of death, the correlation between EOL discussions and pain, nausea, anxiety, death rattles (p<0.001 for all four) and shortness of breath (p=0.033) remained (table 3).

The residents who suffered from a symptom were more often relieved from that corresponding symptom in the EOL discussion group compared with the non-EOL discussion group (p<0.001 for all five symptoms). These correlations remained after adjusting for age, time living in unit and cause of death (p<0.001 for all five symptoms) (table 4).

The residents had more often been prescribed PRN drugs against pain, nausea, anxiety and death rattles in the EOL discussion group (p<0.001 for all four drugs). These correlations remained after adjusting for age, time living in unit and cause of death (p<0.001 for all four drugs) (table 5).

Discussion

This nationwide population-based study showed that documented EOL discussions with residents and/or family members were significantly associated not only with a higher reported prevalence of symptoms during the last week of life in RCHs but also with better symptom relief and prescription of PRN drugs against symptoms. This could be partly explained by a selection of symptomatic residents for EOL discussions, but another possible explanation for this is that the EOL discussion can work as a forum for detecting symptoms and thus enable symptom relief. Furthermore, the documentation in itself, about being in the final stage of life, could be a key aspect clarifying for all team members that the goal of care is quality of life and symptom relief.

The findings in this study are in concordance with the result from Klint et al,12 who reported the absence of an EOL discussion as a risk factor for unrelieved pain in different settings (specialised palliative care, hospital and community care). However, another study showed that there were no differences in relief of pain, anxiety, nausea or dyspnoea between the residents and/or family members who received EOL discussions, compared with those who did not.14 Both in this study and in the study by Lundquist et al,14 data were collected from the SRPC, but data in this study were collected after a revision of the questionnaire used for data collection. The revised questionnaire items used for symptom data collection in this study have been shown to be more valid.3 15

The majority (80%) of the residents and/or family members in this study had received EOL discussions. Almost the same result was shown in another study, where 79% had received EOL discussions.12 In a study examining hospital and community settings, it was more common that patients with non-cancer diagnoses (16%–72%) lacked EOL discussions, compared with patients with cancer (6%–43%).16 Other studies showed a difference between diagnoses concerning whether the residents received EOL discussions or not.10 17 In this study, 30% of the residents died from mental and behavioural disorders, and more than one third died from cardiovascular diseases (37%).

The group in this study that did not receive EOL discussions were significantly older than those who did, and had fewer reported symptoms. Earlier studies have shown that old age could be associated with lower reported symptom burden in a geriatric population,18 and age has also been reported as a risk indicator for poor EOL quality.19 Another study reported that older people had more unmet pain, less access to generalist and specialist palliative care and greater information needs than younger people. Older people’s symptoms are blurred, and they may suffer from multiple diseases and dementia diseases.20 In the light of the findings in this study, where the symptom differences between the groups remained after adjusting for age, one can question whether the lower symptom burden reported in the older adults is not a true difference in symptom occurrence but at least partly because of lack of symptom detection.

In this study, 94% of the residents had a PRN prescription of an injectable drug against pain (mostly opioids). The prescription of drugs against pain was much higher in this study compared with the lower prescription rate reported by Tanghe et al.21

In a study by Huang et al,22 education for healthcare staff was a reported correlation with EOL discussion with care home residents. Education to physicians about EOL discussions has previously been associated with a higher proportion of residents and/or patients in Sweden receiving these discussions.23 This suggests that education can be a key factor for improvement.

The results of this study also show that care home residents who had lived for a longer period of time at the RCHs more seldom received EOL discussions compared with those who had lived for a shorter length of time. In the analysis, it was adjusted for this variable.

EOL discussions could be a good forum for screening symptoms and other problems and for learning about important wishes that residents and/or family members might have. A study showed that it is important for healthcare professionals to get to know the patient and their family member’s goals and priorities.24 A study about patients with advanced heart failure reported the same results.25 An earlier report showed that a majority of patients with severe diseases stated that they want to be informed about their disease and prognosis and to discuss issues related to treatment and care in the terminal stage.26 The result of a qualitative study in nursing homes showed that when family members were involved in decisions, nurses felt reassured in providing care consistent with a resident’s preferences, and symptoms were more likely to be well controlled.27 A scoping review showed the EOL discussions should take place early in the patient’s trajectory while the resident is still stable and while the resident and family members are more likely to participate in decision-making. Involvement increases when the resident’s preferences about EOL are respected.28 However, Groebe et al29 reported that participants described how their wish for support in dealing with conversations about death was hindered. Care professionals reported that they wanted more education and tools to provide EOL conversations. A study showed that intervention with an interactive half-day course for physicians and nurses in RCHs and hospitals increased the number of residents who received information about the transition to EOL care, compared with the control group.23

Accurate prognostic information are recognised as being of clinical importance for patients approaching the EOL. Evidence for patients with non-cancer diagnosis such as dementia and heart failure suggest that clinicians’ determinations of prognosis in these patients may be more inaccurate than those in patients with cancer.30 However, three systematic reviews report that clinicians’ predictions about length of survival for patients with advanced cancer are frequently inaccurate and unreliable.31–33 Also, another systematic review, including all diagnoses, found that clinicians’ prediction are frequently inaccurate.34 In this study, the correlation between symptoms and EOL discussions remained after adjusting for cause of death. This strongly indicates that diagnosis is not the sole explanation for these differences.

Previous study from other countries have shown that factors associated with different kinds of RCHs can correlate with symptom management. For example, care size, ownership model and multichain affiliation on resident comfort and symptom management at the EOL.35 It is possible that RCHs factors such as size and ownership model had an impact on the results. This could be examined in further research.

Methodological considerations

A strength of this study is that the SRPC is a national register, and it covers a majority (61% in 2018) of all deaths, in Sweden (SRPC 2018) and around 75% of all RCHs deaths (Personal communication with Staffan Lundström, Research manager for the SRPC). In this study, only expected deaths were included. The data in the SRPC are rated by care professionals rather than residents themselves or their family members, which could carry a risk for bias. It is also a retrospective study, and as such, could contain recall bias. As this is an observational study, the results can show correlation but not causation. It would be interesting to know how long before death the EOL discussions were performed to gain a deeper understanding of the correlation, but no information about the timing of the discussions are available in the SRPC database. Other aspects that could be interesting include the length of the conversation, and whether other healthcare staff members were present.

Conclusion

In this study, the results indicate that EOL discussions are correlated not only with a higher prevalence of symptoms but also with better symptom relief and prescription of symptom drugs PRN when symptomatic. A possible explanation for this is that the EOL discussion can be an opportunity to discuss symptoms and treatment for symptom relief.

Data availability statement

No data are available. All data relevant to the study are included in the article.

Ethics statements

Ethics approval

This study has approval from the Regional Ethics Committee for Human Research (reference number 2017-89-32M) and by the SRPC register management group December 2020. It was not possible to obtain informed consent from residents or relatives, as this is a register-based study based on data collected after death of the individuals whose data is in focus.

References

Footnotes

  • Contributors All authors contributed to the design of the study. SA and MB managed data collection. SA analysed the data with input from MB and LM. SA, MB and LM wrote the paper with input and critical review from all authors.

  • Funding This work was supported by the Foundation of Medical Research, Skellefteå.

  • Competing interests None declared.

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