Symptom relief during last week of life in neurological diseases

Abstract Objectives The aim of this study was to investigate symptom prevalence, symptom relief, and palliative care indicators during the last week of life, comparing them for patients with motor neuron disease (MND), central nervous system tumors (CNS tumor), and other neurological diseases (OND). Material & Methods Data were obtained from the Swedish Register for Palliative Care, which documents care during the last week of life. Logistic regression was used to compare patients with MND (n = 419), CNS tumor (n = 799), and OND (n = 1,407) as the cause of death. Results The most prevalent symptoms for all neurological disease groups were pain (52.7% to 72.2%) and rattles (58.1% to 65.6%). Compared to MND and OND, patients with CNS tumors were more likely to have totally relieved pain, shortness of breath, rattles, and anxiety. They were also more likely to have their pain assessed with a validated tool; to receive symptom treatment for anxiety, nausea, rattles, and pain; to have had family members receive end‐of‐life discussions; to have someone present at death; and to have had their family members offered bereavement support. Both patients with CNS tumor and MND were more likely than patients with OND to receive consultation with a pain unit and to have had end‐of‐life discussions. Conclusions The study reveals high symptom burden and differences in palliative care between the groups during the last week of life. There is a need for person‐centered care planning based on a palliative approach, focused on improving symptom assessments, relief, and end‐of‐life conversations.

Treatment of distressing symptoms in neurological diseases is an important focus throughout the illness trajectories and during endof-life care. Two studies show that around 20%-35% of patients with MND undergo noninvasive ventilation (NIV) and around 35% receive enteral feeding at the end of life (Gil et al., 2008;Oliver et al., 2010).
Even though it is known that many neurological diseases experience increasing symptom burden as the illness progresses, little is known about symptom prevalence and the extent of symptom relief during the last week of life. To symptom relief in end-of-life care for people with neurological diseases, it is increasingly recommended that a palliative approach be centrally integrated into the care provided by all healthcare professionals' relief (Randall & Downie, 2006). Regardless of the diagnosis, a number of factors are important when integrating a palliative approach to care, including comprehensive assessment, patient and family engagement, communication, care coordination, attention to symptom relief, and preparation for an imminent death through end-of-life conversations with the dying person and family members (Ferrell, Twaddle, Melnick, & Meier, 2018). It is therefore important to study symptom relief in palliative care at the end of life (De Roo et al., 2013) and also to obtain such knowledge from a population perspective (Cohen & Deliens, 2012). Although disease progression is often more aggressive for MND and CNS tumors than for other neurological diseases, there is limited knowledge about symptom prevalence and relief among these disease groups at end of life. In addition, since palliative care knowledge is historically primarily focused on care of patients with cancer, there is a gap in knowledge about differences in a palliative approaches to care across the different disease groups. Better knowledge about possible differences might give rise to strategies to improve the care regardless of the disease in question.
Thus, the aim of this study was to investigate symptom prevalence, symptom relief, and palliative care during the last week of life, by comparing the experiences of patients with neurological diseases grouped as MND, CNS tumor, and other neurological diseases (OND).

| Study population and data collection
This was a retrospective register study. Data were obtained between 1 January 2011 and 31 December 2012 from the Swedish patients with CNS tumor and MND were more likely than patients with OND to receive consultation with a pain unit and to have had end-of-life discussions.

Conclusions:
The study reveals high symptom burden and differences in palliative care between the groups during the last week of life. There is a need for person-centered care planning based on a palliative approach, focused on improving symptom assessments, relief, and end-of-life conversations.

K E Y W O R D S
amyotrophic lateral sclerosis, brain neoplasms, end of life, motor neuron disease, neurological disease, palliative care Register for Palliative Care (SRCP) (http://palli ativ.se), which is a national quality register documenting care during the last week of life, irrespective of diagnosis (Lundström, Axelsson, Heedman, Fransson, & Fürst, 2012). During this period, the register covered 53% and 62% of all deaths in Sweden for 2011 and 2012, respectively (Öhlén et al., 2017). Information for the register was obtained from healthcare professionals (physician, nurse, or team) based on their responses to a questionnaire completed after the patient's death.

| Study variables
Selected variables from the SRCP were organized into patient characteristics, symptom presence and symptom relief, and key indicators of palliative care. Variables for patient characteristics were age, sex, place of death, and type of care setting, as well as the number of days enrolled in the service. Place of death and type of care setting were categorized into general home care, shortterm care facility for elderly people, hospital ward/department, specialized palliative home care, hospice or palliative care unit, long-term care for older people, and other. Presence and degree of relief were assessed for the following symptoms: pain, nausea, shortness of breath, rattles, anxiety, and confusion. Each symptom was reported as "not present," "totally relieved," "partly relieved," "not relieved at all," or "unknown." Assessments of pain were based on use of validated tools in clinical practice, such as visual analog scale, numeric rating scale, and Abbey pain scale.
Information about individual prescription of pro re nata injections (PRN injections) was documented for the each of the above symptoms, except for shortness of breath and confusion. Also included were time since last examination by a physician, consultation of external expertise for symptom relief (e.g., pain or palliative care team), presence of someone at the time of death, patient's death occurring in preferred place, and whether bereavement support for family members had been offered. Variables pertaining to endof-life conversations with patients and family members were not limited to the last week of life. These variables were entered as "yes," "no," or "unknown."

| Statistical analysis
Descriptive statistics were used to present patient characteristics, symptom presence, symptom relief, and palliative care during the last week of life from the SCRP, as well as data from the Swedish Causes of Death Certificate Register.
Following well-established procedures for categorical data analysis (Agresti, 2010), logistic regression models were used to compare symptoms, symptom relief, and palliative care indicators among the three patient groups, while adjusting for covariates (age, sex, place of death, days enrolled in the service). Separate multinomial logistic regression analyses were completed for each symptom as a dependent variable with categories "no," "yes, but totally relieved," and "yes, partly relieved or not relieved at all" (the last two categories were collapsed due to sparse data). Similarly, multinomial logistic regression analyses were completed for dummy-coded palliative care indicators that had more than two categories, and logistic regression models were used for indicators that had only two categories. Comparisons were made by including neurological diseases (MND, CNS tumor, OND) as dummy-coded independent variables, together with the covariates.
The adjusted odds ratios (ORs) are reported as forest plots to facilitate interpretation and comparison across the three diagnostic groups.
Multiple imputation was used to accommodate missing data, and results were compared to these based on complete case analysis (Little & Rubin, 2019). We imputed 15 data files using expectation-maximum likelihood imputation with all variables included as covariates, as implemented in SPSS. Statistical significance was defined as p < 0.05. The statistical analyses were performed using SPSS Statistics version 23 (IBM Corp).

| Symptoms, symptom relief, and palliative care indicators
Pain (52.7%-72.2%) and rattles (58.1%-65.6%) were the most prevalent symptoms, and nausea (9.0%-12.3%) was the least prevalent symptom in all three study groups ( Figure 1). Anxiety (62.7%) and shortness of breath (58.3%) were also among the most prevalent symptoms in patients with MND, as was confusion (30.0%) in patients with CNS tumor. None of these were reported as being totally relieved. While pain (36%-57.6%) was the symptom with the highest percentage of being totally relieved (in all three groups), shortness of breath (39.9%) had the highest percentage of not being totally relieved in MND. Rattles (29.7%-37.8%) was also a symptom with a relatively high percentage of not being totally relieved in all three study groups, as was confusion in CNS tumor (25.2%).
Palliative care indicators are described in Table 2. In general, patients with CNS tumor were more likely to receive PRN prescriptions than patients with MND or OND.

| Comparison between patients with MND, CNS tumor, and OND
Compared to MND and OND, patients with CNS tumor were more likely to have their pain, shortness of breath, rattles, and anxiety totally relieved, rather than partly or not at all relieved, with odds ratios ranging from 2.3 to 1.7 (Figure 2). There were no statistically significant differences in the odds ratios comparing patients with MND and OND.
With regard to the palliative care indicators, compared to patients with MND and OND, patients with CNS tumor were more likely to have their pain assessed with a validated tool; to receive symptom treatment for anxiety, nausea, rattles, and pain; to have family members receive end-of-life discussions; to have someone present at death; and to have bereavement support offered to family members. The most substantial difference for patients with CNS tumor was for prescribed pain medications, with odds ratios of 3.9 (MND) and 4.0 (OND). Both patients with CNS tumor and MND were more likely than patients with OND to receive consultation with a palliative care team (OR = 7.5 and 8.8, respectively), to receive end-of-life discussions from health professionals (OR = 4.8 and 4.3, respectively), and to have their wishes met regarding desired place of death (OR = 2.4 and 1.7, respectively). Conversely, patients with OND were substantially more likely than patients with CNS tumor to receive intravenous fluid or nutritional support on the last day of life (OR = 8.6). In addition, compared to OND, patients with MND were more likely to have PRN prescriptions for nausea (OR = 2.1) and anxiety (OR = 1.8); to have family members present at death (OR = 1.6); and to have had family members offered bereavement support after the patient's death (OR = 1.9; Figure 3).

| D ISCUSS I ON
The prevalence of severe symptoms and lack of symptom relief that was found in the present study suggests that patients with % symptom presence and relief % imputed data Depending on the diagnoses, the progression of symptoms and its rapidity differ between the three studied groups. Health service professionals' knowledge about OND and its trajectory might be insufficient (especially since 55.5% of these patients lived at long-term care facilities for older people) and the symptoms might not be noted as in more aggressive diseases like MND or CNS tumor. It might also suggest that the patients with OND did not receive the same degree of palliative care as the other groups studied. Another possible explanation for the differences might be that many neurological diseases have a chronic course with longer disease trajectories than for MND (Wijesekera & Leigh, 2009) and many cancer diagnoses . The differences in the number of days enrolled to the service between the groups of MND and CNS tumor versus OND probably depend on differences in disease trajectories and progression in MND and CNS as compared to OND. Consequently, it might be difficult to assess symptoms and decide when patients are in need of palliative care and when to give appropriate treatment. As recommended based on previous research (Oliver et al., 2016), a focus on a palliative approach to care, also at an earlier stage, might lead to higher attention being paid to what type of care is appropriate.  Person-centeredness combined with a palliative approach to care is necessary to focus on individual differences in symptoms and symptom relief in striving for equal care.
Only around 32% of the patients with OND were offered end-oflife discussions, while the figure was around 69% in the case of their family members. It may be that these numbers are related to cognitive or consciousness decline and similar results are found for other diagnoses Axelsson et al., 2018). It would be interesting to explore why patients do not receive this discussion. It is important to offer participation in end-of-life discussions as such interaction represents an essential part of palliative care (Ferrell et al., 2018). Having end-of-life discussions with patients and family members is a significant prerequisite for professionals to learn more about the patient's personal preferences regarding both place and planning of care at the end of life.
To our knowledge, there is a lack of studies focusing on patients with neurological diseases grouped into MND, CNS tumor, and OND, examining and comparing symptom prevalence and relief, as well as palliative care during the last week of life. The goal of the SRCP is to improve the quality of care at the end of life (Lundström et al., 2012), and this register study provides knowledge that could be used for such improvement. This is the first study on neurological diseases based on SRCP data, focusing specifically on diseases from the central and pe-

ACK N OWLED G M ENTS
The authors thank the Swedish Register of Palliative Care for providing the registry data. The study was supported by the Ulla-

Carin Lindquist Foundation and the Foundation for Neurological
Research. In addition, this research was undertaken, in part, thanks to funding from the Canada Research Chairs program; Sawatzky holds a Canada Research Chair (Tier 2) in personcentered outcomes.

CO N FLI C T S O F I NTE R E S T S
None of the authors have any conflicts of interest to disclose.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author.

O RCI D
Anneli Ozanne https://orcid.org/0000-0003-1737-3359 F I G U R E 3 Odds ratios for palliative care indicators. Note: Results based on multinomial logistic regression and multiple imputation (n = 2,627). Odds ratios are adjusted for age, sex, place of death, days enrolled in the service. See Table 2