Comparison of the quality of death between primary malignant brain tumor patients and other cancer patients: results from a nationwide bereavement survey in Japan

No studies have investigated the association between malignant brain tumor and the quality of dying, which is an important outcome in end-of-life care. This study aimed to clarify whether the quality of dying and related factors in patients with malignant brain tumor differ from those in patients with other malignant diseases. This was a secondary analysis of data collected by two nationwide, multicenter, bereavement surveys of palliative care units in Japan. This analysis included 14,171 bereaved family members (160 patients with malignant brain tumors). The quality of dying was examined using the good death inventory (GDI), a validated tool widely used in palliative care settings. Patients with malignant brain tumors were younger (p < 0.0001) and had a longer palliative care unit stay during their end-of-life (p < 0.0001) than others. The total GDI score was significantly lower in patients with malignant brain tumors than others (p < 0.0001). Five GDI items were significantly lower in the malignant brain tumor group than other cancer group: “Being able to stay in one’s favorite place” (p = 0.03); “Trusting the physician” (p = 0.003); “Not being a burden to others” (p = 0.01); “Being independent in daily activities” (p = 0.01); and “Feeling that one’s life is worth living” (p = 0.001). This study showed that the quality of dying of patients with malignant brain tumors was lower compared to other cancers, suggesting the need to review care for patients with malignant brain tumors based on their characteristics.


Introduction
Malignant brain tumors, compared with other malignant diseases, are known to cause significant morbidity and mortality that decrease patients' quality-of-life (QOL) [1,2]. Due to the high symptom burden and the fatal illness, the supportive care and/or end-of-life (EOL) care needs of patients with malignant brain tumors may differ from those of patients with other cancers [3][4][5][6]. Previous studies have reported that patients with primary malignant brain tumors require more support for decision-making and physical symptoms than other patients with cancer during EOL care due to cognitive/functional impairment [7][8][9][10]. However, the QOL at EOL, more specifically, the quality of dying, of patients with malignant brain tumors has not been sufficiently investigated.
It is important to evaluate the outcomes of care to improve them [11,12], and QOL at EOL or the quality of dying may be one of the core outcomes in palliative care.
However, studies that have evaluated the outcomes of EOL care for malignant brain tumor patients are limited. Sizoo and colleagues measured the health-related QOL (HRQOL) of high-grade glioma patients at EOL retrospectively by proxy report [13]. They found that HRQOL was poor and deteriorated as death approached, but their study was limited by its heterogeneous sample, in more specific, they only targeted high-grade glioma patients and had no comparison. Therefore it remains unclear whether there are any specific issues in patients with malignant brain tumors considering the characteristics of the disease [13]. The concept of QOL at EOL and that of quality of dying are different, and measuring each independently is known to be important for good understanding of the dying experience [14]. Additionally, a systematic review of the EOL phase of patients with malignant brain tumors suggested that there is a need for more studies focusing on the prevalence of problems and the needs of patients with malignant brain tumors at EOL [15].
The aim of this study was to clarify whether the quality of dying and factors related to it of patients with malignant brain tumors differ from those of patients with other malignant diseases using a nationwide, multicenter, bereavement survey of palliative care units (PCUs) in Japan.

Methods
This study was a secondary analysis of combined data from The Japan HOspice and Palliative Care Evaluation study 3 (J-HOPE3) and J-HOPE4. These J-HOPE studies were nationwide, cross-sectional, anonymous, self-reported questionnaire surveys of bereaved family members after the loss of a close relative in Japan. J-HOPE3 and J-HOPE4 were carried out in May 2014 and July 2018, respectively. The details of J-HOPE3 and J-HOPE4 have been described in the protocol papers of the studies [16,17]. Briefly, in the present study, data of 133 and 187 PCUs, which were members of Hospice Palliative Care Japan (HPCJ) who participated in J-HOPE3 and J-HOPE4, respectively, were used. The patients died of cancer who and whose family member were aged ≥ 20 years were included in the present study. The exclusion criteria were: (1) the patient received palliative care for < 3 days; (2) the bereaved family member was unavailable or could not be identified; (3) the patient of treatment-associated death (i.e. death during a chemotherapy cycle or medical error) or occurrence of death in the intensive care unit; (4) the bereaved family member were assumed to suffer extraordinary emotional distress as a result of their participation (i.e. history of mental diseases), as determined by the primary physician and a nurse; and 5) the bereaved family member was incapable of completing the self-reported questionnaire because of health issues, such as cognitive impairment or a visual disability. Questionnaires were sent to the bereaved family members identified by each participating institution. Returning a completed questionnaire was considered to indicate individual consent to study participation. This study was approved by the institutional review boards of Tohoku University (J-HOPE3: No. 2013-1-334; J-HOPE4: No. 2018-2-290) and all participating institutions.

Measurements
Patient background data were collected from the participating institutions of this study. The data included the deceased patients' age, sex, primary tumor site, date of last admission to a PCU, and date of death. In this study, the patients were divided into two groups based on the answers regarding the primary tumor site: the brain tumor group for all primary malignant brain tumors and the other cancer group.
The questionnaire sent included items regarding sociodemographic background, such as the bereaved family member's sex, age, time since loss, and relation to the deceased. In addition, the item/scale mentioned below was included.

Quality of dying
The good death inventory (GDI) short version was used to evaluate the quality of dying. This scale was developed based on the constructs of QOL at EOL extracted from qualitative and quantitative studies in Japanese subjects as a scale to evaluate the quality of dying of cancer patients through bereaved family assessment and have been translated to several language [18,19], and used in previous studies to evaluate the quality of dying [20][21][22]. The GDI consists of 18 items, namely; "Being free from physical distress", "Not being a burden to others", "Feeling that one's life is worth living", and" Feeling that one's life was completed". All 18 items are listed in Table 2. For each item, the respondents were asked to rate their life in a hospice/palliative care unit on a 7-point scale from "1. totally disagree" to "7. very much agree", with a higher score indicating a higher rating. The reliability and validity of this scale have been verified [23].

Analysis
First, to clarify the characteristics of the brain tumor group, the t-test, Chi-squared test, or Fisher's exact probability test was conducted comparing the brain tumor group and the other cancer group for socio-demographic items. Second, the t-test was used to compare the scores of each of the 18 items of the GDI between the brain tumor group and the other cancer group. Effect sizes (ESs) were calculated based on the definition of Cohen et al. [24]. Next, considering potential bias, multivariable regression analysis was used to adjust all demographic factors obtained with regard to the deceased patients and important bereaved family member factors which were found to be significantly different between the brain tumor group and the other cancer group. A p-value < 0.05 was considered significant, and all tests were two-tailed. SAS software (Japanese ver. 9.4; Cary, NC; BMDP, Los Angeles, CA) was used for the analyses.

Results
In total, 30,179 bereaved family members met the inclusion criteria and 4506 were excluded according to the exclusion criteria. The questionnaire were sent to 25,673 bereaved families in total, and the data collected from 14,275 bereaved family members (valid response rate 56%) from 133 and 187 PCUs from J-HOPE3 and J-HOPE4, respectively, were used for the analysis. The characteristics of the deceased patients and their bereaved family members are summarized in Table 1. In total, 160 of the patients had malignant brain tumors, and 14,011 had other cancers. With regard to the primary tumor site in the other cancer group, 22.7% had lung cancer, and 14.5% had tumors of stomach or esophagus. There were significant differences between the malignant brain tumor group and the other cancer group in patients' age {brain tumor group: 67.1 ± 14.0 years [mean ± standard deviation (SD)] vs. the others: 75.0 ± 11.4 years}, duration of last admission to a PCU (brain tumor group: 55.1 ± 46.4 days vs. the others: 36.5 ± 46.0 days), and relationship to the deceased (ES = 0.04, p < 0.0001).
Comparison of the GDI scores between the patients with brain tumors and other cancers are shown in Table 2.

Discussion
This study examined the differences in sociodemographic characteristics, the quality of dying, and factors related to the quality of dying between patients with malignant brain tumors and those with other malignant diseases. The major findings were: (1) patients with malignant brain tumors were younger, the last stays in PCUs were longer, and primary caregivers were more often parents than the other cancer group; and (2) regarding quality of dying, the total GDI score and 5 items from the GDI scale, namely, "Being able to stay in one's favorite place", "Trusting physician", "Not being a burden to others", "Being independent in daily activities", and "Feeling that one's life is worth living," were significantly lower in the malignant brain tumor group than in the other cancer group. To the best of our knowledge, this is the first large-sample study that focused on the quality of dying of primary malignant brain tumor patients and compared them with patients with other cancers.
First, in the present study, patients in the brain tumor group were younger and had a longer PCU stay during their EOL period than the other cancer group. The results of this study may reflect the characteristics of patients with primary malignant brain tumors. Malignant brain tumors generally affect younger patients and have a lower 5-year survival rate [2]. Regarding the longer stay in PCUs, the poor activities of daily living of the patients and the difficulty choosing home care due to the caregiver burden on the family may have affected the PCU length of stay. Several studies have reported that the decreased functional independence of malignant brain tumor patients leads to their hospitalization more often than other patients with cancer [25][26][27]. In addition, a higher percentage of patients' parents and spouses were their primary caregivers. This might be due to the younger age of patients in the brain tumor group. Spouses and parents of patients have been reported to suffer more in their grieving process over bereavement than other relatives (e.g. child of deceased, siblings) [28]. Moreover, some studies have reported that the deceased patient's poor quality of dying and higher care burden may lead to pathological grief in their caregivers [9,21,[29][30][31]. Future research should clarify the impact of the relationship between the brain tumor patient and the primary caregiver on grief and depression of the bereaved. Second, there were significantly lower scores overall and for a few items of the GDI in patients with primary malignant brain tumors than patients with other cancers. These results are consistent with previous findings that the QOL of patients with primary malignant brain tumors tends to be lower than in other cancer patients [1,4,13,15], which may lead to a lower evaluation of the quality of dying.
Most patients with primary malignant brain tumors have a decreased level of consciousness more than one month before death and a loss of independence [25]. These characteristics may have been reflected in the present results, with the lower scores for "Being independent in daily activities" "Not being a burden to others" and "Feeling that one's life is worth living". These three items are items that are also  [32][33][34][35]. Personal dignity is considered to be important for improving quality of dying; therefore, we believe that care that values personal dignity is necessary to improve the quality of dying of brain tumor patients. Sizoo et al. suggested the importance of satisfaction with the physician providing EOL care and communication ability during EOL care for a dignified death in high-grade glioma patients [36]. In the present results, the score for "trusting physician" was lower in the malignant brain tumor group, which suggests that there may be room for improvement. In addition, the present results suggest that patients with primary malignant brain tumors tend to have difficulties regarding staying or dying in their preferred place than the other cancer group. This might reflect the difficulty in EOL discussions in primary malignant brain tumor patients, which previous studies have pointed out [15,37]. However, the target of the present study was patients who died in PCUs. In most facilities in Japan, one of the criteria for admission to a PCU is that the patient and his/her family have been informed of the disease and the severity of the condition, and that the patient desires and agrees to be admitted to the ward where basically no aggressive cancer treatments are provided [38]. Therefore, the difficulty of primary malignant brain tumor patients of dying in their preferred place may be underestimated, and in other general settings, more patients might not achieve death in their preferred place. Studies have reported that EOL decision-making is often a challenge for primary malignant brain tumor patients [10,15,37]. This is due to the typical symptoms of malignant brain tumor that affect communication from an early stage of the disease. Therefore, early EOL discussion and decision-making are needed. However, studies have shown that primary physicians tend to avoid disclosing the exact diagnosis, especially their severe prognosis to patients with malignant brain tumor [39]; in addition, they frequently did not discuss EOL preferences with their patients [37]. Previous studies of cancer patients have shown that patients who engage in EOL discussions tend to receive care consistent with their wishes, and that the quality of EOL care and the QOL of patients and caregivers are improved [40][41][42].
Healthcare providers need to consider timely interventions such as providing appropriate information to patients and their families so that they can have adequate discussions and make decisions regarding their EOL care preferences.

Limitations
Several limitations of this study warrant consideration. First, this study involved patients with primary malignant brain tumors who died in a hospice/palliative care ward. Patients admitted to a hospice/palliative care ward were likely to have been able to discuss their EOL wishes, including expressing their intention to enter the ward. This may have positively affected the quality of care received by the patients and their assessment of achieving a desired death. Second, background factors of patients with brain tumors, such as the grade of the brain tumor, the time when the level of consciousness decreased, and the time when the physical function (performance status) decreased were not examined. Third, bereaved families who were in a significantly worse physical or mental condition may not have participated in the study. Fourth, because this was a cross-sectional study based on a survey at a single point in time, causal inferences between variables cannot be made. Fifth, recall bias may have occurred in the evaluation of patients, such as the quality of care and achievement of desired death, because the bereaved family members evaluated them in retrospect as proxies. However, the bereaved family survey is a valid method for evaluating end-of-life care because terminally patients with cancer are too ill to participate in the surveys [43]. The sixth issue is the representativeness of the subjects in this study. Among the member facilities of the Japan Hospice and Palliative Care Association, it can be assumed that those facilities that are particularly active in evaluating care and providing feedback on the results participated in this study. This may have affected their involvement with patients and bereaved families before and after bereavement, and their outcomes. In addition, regarding those patients with other malignancies, the data did not include whether they had brain metastases.

Conclusion
This study showed that the quality of dying of patients with malignant brain tumors was lower than that of patients with other cancers. In particular, the following four items were significantly lower: "Being able to stay in one's favorite place"; "Trusting physician"; "Being independent in daily activities"; and "Feeling that one's life is worth living." These results suggest the necessity of reviewing care practices for patients with malignant brain tumors according to their characteristics. Care practices that value personal dignity and early EOL decision-making can be more important to improve the quality of dying in patients with malignant brain tumors. Future studies may be needed to establish EOL care suited to patients with malignant brain tumors and evaluate their effects on the quality of dying.
Consent to participate A document explaining the aim and procedure of this study was sent along with questionnaire, and the return of a completed questionnaire was considered as consent to participate in this study.
Consent to publish Not applicable.