Improved Symptom Change Enhances Quality of Dying in Patients With Advanced Cancer: An East Asian Cross-Cultural Study

Abstract

Background

Symptom burdens tend to increase for patients with cancer and their families over the disease trajectory. There is still a lack of evidence on the associations between symptom changes and the quality of dying and death. In this context, this research investigated how symptom changes influence the quality of dying and death.

Methods

This international prospective cohort study (the East Asian Collaborative Cross-Cultural Study to Elucidate the Dying Process (EASED), 2017-2019) included 22, 11, and 4 palliative care units across Japan, South Korea, and Taiwan. Eligible participants were adults (Japan and Korea, ≥18 years; Taiwan, ≥20 years) with locally advanced or metastatic cancer. Physical and psychological symptoms were assessed by physicians upon admission and within 3 days before death. Death quality was assessed using the Good Death Scale (GDS), developed in Taiwan. Univariate and multivariate regression analyses were used to identify correlations between symptom severity changes and GDS scores.

Results

Among 998 patients (542 [54.3%] men and 456 [45.7%] women; mean [SD] age = 70.1 [± 12.5] years), persistent dyspnea was associated with lower GDS scores when compared to stable dyspnea (β = −0.427, 95% CI = −0.783 to −0.071). Worsened (−1.381, −1.932 to −0.831) and persistent (−1.680, −2.701 to −0.659) delirium were also significantly associated with lower GDS scores.

Conclusions

Better quality of dying and death was associated with improved symptom control, especially for dyspnea and delirium. Integrating an outcome measurement for the quality of dying and death is important in the management of symptoms across the disease trajectory in a goal-concordant manner.

There is a lack of evidence on the associations between symptom changes and the quality of dying and death for patients with advanced cancer. This research addresses the gap.

Implications for Practice.

A multicenter prospective observational study demonstrated that symptom improvements were associated with better Good Death Scale scores in patients with advanced cancer. This study provided robust evidence that improved or mild changes in dyspnea and delirium were associated with better quality of dying and death. There is a particular need for studies aimed at determining factors that are relevant to the quality of dying and death. This research addresses the gap. The integration of an outcome measurement for the quality of dying and death is valuable for physicians aiming to manage symptoms across the disease trajectory in a goal-concordant manner.

Introduction

The quality of dying and death is a grave issue for patients, families, and clinicians. While previous studies have conceptualized the construct of the quality of dying and death, it is essential to note that such assessments are influenced by the cultural background in which the research is situated.^1-4 Moreover, the most appropriate medical care should be concordant with the patient-and family-centered willingness. For example, goal-concordant care is essential for oncology patients, as relevant provisions should prolong the survival period through tumor-directed treatments and incorporate patient values and preferences.⁵ However, limited resources may influence the medical community, stakeholders, and policymakers to allocate funds toward the evaluation of cost-effective medical treatments aimed at patient-directed outcomes such as the quality of life without adequately considering the quality of dying and death. Even in the literature, there are more studies of the quality of life compared to those focusing on the quality of dying and death; in this regard, several domains are in critical need of continuous assessment, including freedom from physical and psychological symptoms at the end of life, preparing for death in the last days of life, finding meaning in life, gaining a sense of peace with personal growth, and matters of spirituality.^1-3,6 In addition to this, many factors associated with improved quality of dying and death may help patients with advanced cancer and their families understand the overall benefits and harms of cancer treatments while also ensuring that they receive goal-concordant care.

Various physical symptom evaluations have been integrated into the treatment outcome assessment model to make prognoses, including cancer treatment.^7,8 In this context, severe symptoms substantially influence a variety of issues, including daily function, emotional stress, difficulty engaging in pleasurable activities accompanied by social withdrawal, caregiver burdens, and decreased dignity during medical treatment. For example, dyspnea is a common symptom associated with lower quality of life, mood distress, and decreased mastery.⁹ Oncology patients also frequently experience fatigue, which significantly impacts the quality of life, daily activities, social relationships, and caregiver burdens.¹⁰ Moreover, other severe conditions include delirium, agitation, cognitive impairment, and hallucination, which are distressing to patients, caregivers, and physicians, as they result in notable medical and psychological comorbidities, especially in the end-of-life care context.^11,12 As reflected by their role in making prognoses, these symptoms drastically increase suffering for patients, thus emphasizing the need for all healthcare professionals to be educated on the importance of symptom control.^13,14 As symptom burdens tend to increase for patients with cancer and their families over the disease trajectory, interprofessional teams also need to address their physical, psychosocial, and spiritual needs, thus ensuring that personalized symptom control goals and relevant management procedures are met.¹⁵

Despite the evident negative impacts of symptom progression, there is still a lack of evidence on the associations between symptom changes and the quality of dying and death. To address the gap in the literature, this research investigated how symptom changes influenced the quality of dying and death in 3 East Asian countries based on data from an extensive international multicenter cohort study. We believe that the findings constitute valuable information for clinicians aiming to manage symptoms across the disease trajectory in a goal-concordant manner.

Methods

Study Design

This international prospective cohort study was conducted in Japan, Korea, and Taiwan, specifically known as the East Asian Collaborative Cross-Cultural Study to Elucidate the Dying Process (EASED).

Setting

As a multicenter study, the research setting included 22, 11, and 4 palliative care units (PCUs) across Japan, South Korea, and Taiwan, respectively. Eligible participants were recruited from among patients newly admitted to these PCUs during the study period from 2017 to 2019. All evaluations were performed during routine clinical practice.

Participants

The specific inclusion criteria were as follows: (1) adults (≥18 years of age in Japan and Korea; ≥20 years of age in Taiwan) with (2) locally advanced or metastatic cancer. The exclusion criteria were as follows: (1) planned discharge within 1 week, and (2) refusal to participate (by patients or their families). Patients who survived and were discharged from any PCUs during the study period were also excluded from the analysis, as the Good Death Scale (GDS) was inapplicable in this case.

Variables

Demographic data were collected from patients upon initial admission, with items including age, sex, country, educational level (illiterate, elementary or junior school, high school, college or graduate school, unknown), and marital status (unmarried, married, widowed, separated, unknown).

Data Measurement

Physical and psychological symptoms were assessed by physicians upon admission and within 3 days before death. The symptom ratings within 3 days of death were retrospectively assessed after a patient died. In this context, dyspnea was rated using the following nominal scale: 0, no; 1, yes (exertional only); 2, yes (at rest). The validated Integrated Palliative Care Outcome Scale was used to evaluate fatigue objectively as follows: 0, not at all; 1, slightly; 2, moderately; 3, severely; 4, overwhelmingly; 5, cannot be assessed because of unconsciousness.¹⁶ Delirium was assessed using Item 9 of the Memorial Delirium Assessment Scale as follows: 0, none; 1, mild; 2, moderate; 3, severe.¹⁷ Hallucination was assessed using Item 2 of the Delirium Rating Scale-revised 98 as follows: 0, not present; 1, mild perceptual disturbances; 2, illusions present; 3, hallucinations present.¹⁸ Notably, we evaluated these highly prevalent symptoms through a literature review, expert group meetings, and general team clinical experience.^9-13,15,19

Good Death Scale (GDS)

Death quality was assessed using the GDS, a tool developed in Taiwan that has since been used in a variety of studies, including the EASED.^3,6,20 The scale consists of 5 items, including the awareness that one is dying (0 = complete ignorance, 3 = complete awareness), accepting death peacefully (0 = complete unacceptance, 3 = complete acceptance), arranging one’s will (0 = no reference to the patient’s will, 1 = following the family’s will alone, 2 = following the patient’s will alone, 3 = following the will of both the patient and their family), death timing (0 = no preparation, 1 = the family alone had prepared, 2 = the patient alone had prepared, 3 = both the patient and their family had prepared), and the degree of physical comfort 3 days before death (0 = considerable suffering, 1 = suffering, 2 = a little suffering, 3 = no suffering). Following the death of each patient, the palliative care team recorded a score ranging from 0 to 15 to summarize the extent to which the death was considered good. For locations in Taiwan, the opinion of each member of the research team was considered, with final scores for each patient jointly decided at a team meeting. In Japan and Korea, GDS scores were determined by attending physicians. In all cases, higher total scores indicate better death quality.

Data Analysis and Statistics

To facilitate analysis, symptom severity and symptom changes during the study period were categorized with reference to our previous studies.^21,22 More specifically, symptom severity was classified according to 3 levels, including no (or mild), moderate, and severe. The raw scores for each symptom were transformed to reflect the corresponding level of severity as follows: no or mild (dyspnea = 0, fatigue = 0-1, delirium = 0-1, hallucination = 0-1); moderate (dyspnea = 1, fatigue = 2, delirium = 2, hallucination = 2), and severe (dyspnea = 2, fatigue = 3-4, delirium = 3, hallucination = 3). Changes in symptom severity between admission and 3 days before death were categorized based on 4 groups as follows: stable, improved, worsened, and persistent. Groupings were defined according to the following rules: stable (no or mild-no or mild), improved (moderate-no or mild, severe-no or mild), worsened (no or mild-moderate, no or mild-severe), and persistent (moderate-moderate, moderate-severe, severe-moderate, severe-severe).

Statistical Methods

We conducted univariate and multivariate regression analyses to identify the correlations between severity changes for the 4 symptoms and GDS scores, with additional consideration of the collected demographic data. All analyses were conducted using IBM SPSS for Windows, version 21.0 (IBM Corp., Armonk, NY, USA).

Ethics

This study was approved by the institutional review board of National Taiwan University Hospital (IRB Number: 201811091RIND) and by all participating institutions. In Korea and Taiwan, informed consent was obtained from the patients or families (if the patient lacked the capacity to decide). Under the ethical guidelines for human research of the Ministry of Health, Labor, and Welfare in Japan, informed consent from the patients was waived in Japan.

Results

Participants

Figure 1 shows a flowchart of the patient selection process. A total of 2638 eligible patients were admitted across all investigated PCUs during the study period. However, 500 were excluded due to survival at discharge; of the remaining 2138 patients, 1140 were excluded for missing data or being scored unknown (cannot access because of unconsciousness) on the symptom scales or GDS. As such, 998 patients were included in the final analysis.

Figure 1.

Flowchart of the patient selection process.

Descriptive Data

Table 1 shows demographic data for the 998 analyzed patients. As shown, the mean (SD) age was 70.1 (± 12.5) years, with 542 (54.3%) men, and 456 (45.7%) women. The average GDS score for all patients was 11.9 ± 2.5.

Table 1.

Demographic data.

Characteristics
Age (mean ± SD)	70.1 ± 12.5
Sex
Male	542 (54.3)
Female	456 (45.7)
Country
Japan	779 (78.1)
South Korea	92 (9.2)
Taiwan	127 (12.7)
Achieved education level
Illiterate/elementary/junior high	124 (12.4)
High school	122 (12.2)
College/graduate school	133 (13.3)
Unknown	619 (62.0)
Marital status
Unmarried	114 (11.4)
Married	629 (63.0)
Widowed/separated	254 (25.5)
Unknown	1 (0.1)
Good Death Scale score	11.9 ± 2.5
Cancer diagnosis
Lung	165 (16.5)
Gastrointestinal	289 (28.9)
Hepato-biliary-pancreatic	219 (21.9)
Breast	71 (7.1)
Prostate/testis	13 (1.3)
Gynecology	60 (6.0)
Urologic	47 (4.7)
Hematologic	20 (2.0)
Head/neck (excluding thyroid)	46 (4.6)
Brain	1 (0.1)
Unknown	13 (1.3)
Others	54 (5.4)

Data are presented as mean values ± standard deviations or numbers (row percentages).

Outcome Data

Figure 2 shows the distributions of symptom severity at initial admission and within 3 days before death; approximately half of the analyzed patients showed symptoms of no or mild dyspnea (52.2%) and fatigue (47.5%) upon admission, while a minority showed severe delirium (0.7%) and hallucinations (3.3%). The most frequent severe symptom within 3 days before death was fatigue (39.4%), followed by dyspnea (39.2%).

Figure 2.

The distributions of symptom severity at initial admission and within 3 days before death.

Figure 3 summarizes the distributions of the 4 categories of symptom changes during the study period. As shown, fatigue was “most commonly persistent” (44.1%) and “worsened” (27.5%). Dyspnea showed the highest percentage for “improvement” (8.5%), “frequently persisted” (39.3%), and “worsened” (16.1%). Compared to the other symptoms, delirium and hallucinations showed much higher percentages for “remaining stable,” at 84.9% and 82.9%, respectively.

Figure 3.

Summary of the distributions of the 4 categories of symptom changes during the study period.

Regression Analyses

Table 2 shows the results of the univariate and multiple regression analyses on the factors that were correlated with GDS scores. The univariate analysis showed that the severity of dyspnea, fatigue, and delirium were independent correlates of GDS scores. Persistent dyspnea showed a significant association with lower GDS scores, as compared to stable dyspnea (β = -0.581, 95% CI = −0.936 to −0.225). When compared to stable fatigue, persistent fatigue was also significantly associated with lower GDS scores (−0.439, −0.854 to −0.023). Finally, both worsened (−1.395, −1.897 to −0.893) and persistent (−1.365, −2.258 to −0.472) delirium showed significant associations compared to stable delirium. In contrast, none of the demographic variables were significantly associated with GDS scores.

Table 2.

Results of the univariate and multivariate regression model to identify the predictors of Good Death Scores during the observation period (upon admission and within 3 days of death; N = 998).

	Univariate analysis				Multivariate analysis
	β	95%CI		P-value*	β	95%CI		P-value**
Age at survey	0.007	−0.005	0.020	.247
Sex
Male	–	–	–	–
Female	0.289	−0.021	0.600	.068
Achieved education level
Illiterate/elementary/junior high	–	–	–	–
High school	0.075	−0.547	0.697	.813
College/graduate school	0.244	−0.365	0.852	.432
Unknown	−0.354	−0.833	0.126	.148
Marital status
Unmarried	–	–	–	–
Married	−0.016	−0.515	0.482	.949
Widowed/separated	−0.022	−0.574	0.530	.939
Unknown	−0.904	−5.822	4.015	.719
Dyspnea
Stable	–	–	–	–	–	–	–	–
Improved	0.150	−0.437	0.736	.617	0.069	−0.513	0.650	.817
Worsened	−0.262	−0.724	0.199	.265	−0.248	−0.711	0.215	.293
Persistent	−0.581	−0.936	−0.225	.001	−0.427	−0.783	−0.071	.019
Fatigue
Stable	–	–	–	–	–	–	–	–
Improved	0.449	−0.184	1.082	.164	0.360	−0.267	0.988	.260
Worsened	−0.119	−0.572	0.334	−.119	−0.017	−0.469	0.434	.940
Persistent	−0.439	−0.854	−0.023	.038	–0.291	−0.703	0.122	.167
Delirium
Stable	–	–	–	–
Improved	0.380	−0.765	1.524	.516	0.188	−0.994	1.369	.755
Worsened	−1.395	−1.897	−0.893	.000	−1.381	−1.932	−0.831	.000
Persistent	−1.365	−2.258	−0.472	.003	−1.680	−2.701	−0.659	.001
Hallucination
Stable	–	–	–	–
Improved	0.063	−0.994	1.119	.907	0.073	−0.999	1.144	.894
Worsened	−0.342	−0.818	0.134	.159	0.253	−0.255	0.761	.329
Persistent	−0.366	−1.306	0.574	.445	0.841	−0.218	1.901	.119

^* P-values were obtained via univariate general linear regression analysis.

^** P-values were obtained via multivariate general linear regression analysis.

The multivariate regression analysis showed that persistent dyspnea and worsened and persistent delirium were significantly correlated with lower GDS scores. Moreover, persistent dyspnea was associated with lower GDS scores when compared to stable dyspnea (−0.427, −0.783 to −0.071). Worsened (−1.381, −1.932 to −0.831) and persistent (−1.680, −2.701 to −0.659) delirium were also significantly associated with lower GDS scores.

Discussion

A multicenter prospective observational study conducted in 3 East Asian countries demonstrated improved symptom changes associated with better GDS scores in patients with advanced cancer. This study provided robust evidence that improved or mild changes in dyspnea and delirium were associated with better quality of dying and death. When viewed as an outcome, the quality of dying and death has important implications for symptom management and goal-concordant care communication in the oncology context.

Previous studies have reported that dyspnea is consistently one of the most prevalent symptoms before death in patients with various diagnoses; the condition often results in suffering and discomfort that contribute to functional decline, poor quality of life, increased caregiver burdens, and psychological stress.^9,23 It is not surprising that primary breathing disturbances are associated with depression, decreased patient well-being, and increased staff anxiety, all of which evidently impact the quality of dying and death.²³

Notably, consciousness levels are typically the primary focus of care for patients, families, and medical teams. In this context, delirium drastically hinders meaningful interactions with the outside world, thus precluding patients from meeting the criteria for appropriate death.^24,25 Given that delirium is well-controlled, the sense of mental and spiritual peace may be enhanced among patients, families, and medical teams. This reflects the sentiments of previous arguments concerning the ethics of appropriate death, focusing on the 3 central dimensions of patient autonomy with individual beliefs, family attitude with bereavement, and professional medical ethics.^4,26

In practice, end-of-life care decisions are most often family-centered. However, this concept may differ between cultures. For example, patients in Asia tend to value family harmony more than absolute autonomy.²⁷ In this context, interventions that require family involvement to improve the quality of dying and death are likely to be disrupted by symptoms of dyspnea and delirium, especially as patients may have extreme difficulty communicating about related issues in a calm manner; this may influence families and medical teams to think patients are not accepting death and thus experiencing lower spiritual well-being.²⁸ Under persistent delirium conditions, the components of “accepting death peacefully,” “honoring patients’ wishes,” and “degree of physical comfort three days before death” were low, especially in the Asia-Pacific cultural context.

In recent years, subjects such as assisted suicide, euthanasia, terminal sedation, and general end-of-life anesthesia have become the focus of research.^25,29 The core of relevant disputes may be rooted in various conceptions of how to arrive at a good death.^1,26 In this study, symptom improvement was associated with good quality of dying and death, thus strengthening the evidence that the alleviation of physical symptoms should be points of reminder in cases where a hastened death is requested. The results also emphasize the importance of continued research aimed at the implementation of more reflective outcome measurements, such as the quality of dying and death rather than the quality of life.

This study also had several limitations. First, there was no standard evaluation or definition for a good death. While several measurements of appropriate death have been suggested in previous studies,^1,4 it is challenging to apply individual study results and tools concerning the quality of dying and death in the real-world setting, primarily due to the many variations in cultural elements and laws on assisted dying between countries. Despite this, the current results on using GDS to measure the quality of dying and death provide robust evidence linking good symptom control with a good death. Second, only 4 symptoms were included in the analysis. As such, additional research is needed to cover a greater variety of symptoms that may influence appropriate death. Nevertheless, the 4 investigated symptoms were carefully selected through a literature review, cross-country expert discussions, and the clinical experiences of the authors. As such, the management of each symptom has significant implications for clinical use. Third, interviewer bias in completing the GDS might be an issue concerning the extensive sample size among the 3 countries.

Conclusions

In an era of person-centered oncology care, continued research is needed to clarify what constitutes an appropriate death. Due to the large concentration of academic focus on the quality of life, there is a particular need for studies aimed at determining measurements that are relevant to the quality of dying and death. This study helped address this gap by showing associations between a better quality of dying and death with improved symptom control, especially for dyspnea and delirium. The integration of outcome measurement for the quality of dying and death is valuable for physicians aiming to manage symptoms across the disease trajectory in a goal-concordant manner. To continually address relevant issues, future research and innovative interventions should take collaborative and cross-cultural approaches, such as those adopted in the EASED.

Funding

This work was supported by a Grant-in-Aid from the Japan Hospice Palliative Care Foundation (2015 grant); MEXT KAKENHI (Grant-in-Aid for Scientific Research) Grant Numbers 16H05212 and 16KT0007; Ministry of Science and Technology, Taiwan, 105-2314-B-002-172. The funding source had no role in the study design, data collection, analysis, interpretation of data, writing, and the decision to publish the results.

Conflict of Interest

Masanori Mori reported honoraria from Daiichi-Sankyo, Takeda Pharmaceutical, Shionogi, Hisamitsu, Kyowa-Kirin, and Ono. The other authors indicated no financial relationships.

Author Contributions

Conception/design: H.-L.H., P.-J.C., M.M., S.-Y.S., S.T., S.-Y.C. Provision of study material or patients: P.-J.C., M.M., S.-Y.S., S.-Y.C. Collection and/or assembly of data: P.-J.C., M.M., S.-Y.S., S.-Y.C. Data analysis and interpretation: H.-L.H., P.-J.C., M.M., S.-Y.S., C.-Y.W., J.-K.P., C.-A.Y., J.-S.T., T.-Y.C., Y.H., S.-H.K., T.M., T.Y., D.H., S.-Y.C. Manuscript writing: H.-L.H., S.-Y.C. Final approval of manuscript: All authors.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.