Using a convenience sampling method, family members of 203 patients who died in the emergency department of a tertiary hospital in Guangzhou between September 1, 2020, and August 31, 2024, were selected as study subjects. Inclusion criteria: 1) Patients: age ≥ 18 years; emergency hospitalization/observation time ≥24 hours; time since death at the time of survey ranged from 1 month to 3 years. The ≥24-hour criterion was used to ensure that family members had sufficient exposure to the patient’s end-of-life care process and could evaluate the emergency care experience. However, this criterion may exclude the most abrupt ED deaths and therefore narrow the generalizability of the findings. 2) Family members: age ≥18 years; education level of primary school or above; informed consent and voluntary participation; primary involvement in the patient’s end-of-life care and medical decision-making. Exclusion criteria: 1) Patients: no family members accompanying them. 2) Family members: incorrect/out-of-service phone number; inability to complete the questionnaire. Based on the multiple linear regression sample size calculation method described in Fang Jiquan’s Statistical Methods for Biomedical Research (Second Edition) [11], the sample size needed to be 5 - 10 times the number of variables. Relevant studies identified 35 variables influencing a good death [12], resulting in a calculated sample size of 175 - 350. Accounting for a 10% rate of invalid responses, 203 cases were ultimately included. This study received formal approval from the hospital’s ethics committee (Ethics Approval No. Lun Shen Lin [2023] 497).

2.2.1. General Information Questionnaire

Developed by the researchers based on literature review and panel discussions, this questionnaire collected data related to deceased emergency department patients and their family members.

PatientData: Gender, age, education level, marital status, religious belief, payment method for healthcare, relationship with family/friends/colleagues, awareness of diagnosis and impending death, communication of cardiopulmonary resuscitation (CPR) preferences, cause of death, length of hospital stay (observation).FamilyMemberData: Gender, age, education level, occupation, marital status, religious belief, average monthly household income per capita, relationship to the patient, health status during the caregiving period.

2.2.2. Chinese Version of the Good Death Inventory (GDI)

This scale, translated and cross-culturally adapted by Zhao Juanjuan etal. [10], was used to measure the level of good death among emergency department patients. It encompasses 18 dimensions (10 core and 8 secondary dimensions), with 3 items per dimension, totaling 54 items. Responses are recorded on a Likert 7-point scale (from 1 = “strongly disagree” to 7 = “strongly agree”). The total GDI score is the sum of all 54 items, and the item mean is the total score divided by 54. In this study, meeting the basic criterion for a good death was defined a priori as a mean GDI item score ≥5, consistent with the scale interpretation used in prior Chinese GDI research.

2.2.3. Chinese Version of the Care Evaluation Scale (CES)

Also translated and cross-culturally adapted by Zhao Juanjuan etal. [13], this scale was used to evaluate the quality of end-of-life care received by patients. It contains 28 items across three structural dimensions of care (environmental comfort, reasonableness of costs, and accessibility of medical resources) and seven process dimensions of care (physical care provided by doctors/nurses, psycho-existential care, explanations provided by doctors to patients/family members, attention to family health, and coordination and consistency of medical care). Items are rated on a Likert 6-point scale (1 = “strongly disagree,” 6 = “strongly agree”), with an additional “not applicable” option. Following the validated Chinese CES scoring procedure, “not applicable” responses were treated as missing and replaced with the corresponding item mean of valid responses before the domain and total scores were calculated. CES domain scores were analyzed as item-sum scores in regression models and converted to a 0 - 100 scale for descriptive reporting.

Prior to formal data collection, all four investigators received standardized training. Investigators contacted eligible family members by phone. Upon obtaining consent, the questionnaire link was sent via SMS, WeChat, or email. Collected questionnaires were jointly organized, checked, and analyzed by two researchers. From the hospital’s statistics department, information for 741 family members meeting the inclusion criteria was obtained. Among these, 86 (11.6%) were unreachable and 132 (17.8%) refused to participate. Consequently, 523 family members agreed to participate and received questionnaires. A total of 205 questionnaires were returned. After excluding 2 invalid responses, 203 valid questionnaires remained. The effective response rate was calculated as 203/523 (38.8%), using the number of questionnaires distributed to consenting family members as the denominator; the valid response proportion among all initially eligible family members was 203/741 (27.4%).

Data were analyzed using SPSS software (version 27.0). Quantitative data are described using mean ± standard deviation; categorical data are described using frequencies and percentages. Independent samples t-tests or analysis of variance (ANOVA) were used to examine differences in GDI scores based on patient and family member socio-demographic characteristics. Pearson correlation analysis was employed to assess the relationship between the level of good death (GDI score) and the quality of end-of-life care (CES score) among deceased emergency department patients. Stepwise linear regression was performed to identify factors influencing the level of good death. To evaluate potential recall bias caused by the wide interval between death and survey completion, a sensitivity model was fitted by additionally entering time from death to survey (months) into the final regression model. A P-value of <0.05 was considered statistically significant.

The interval between the survey and patient death was 14.1 ± 9.7 months. The mean age of deceased patients was 73.93 ± 16.05 years, 53.7% were male, 5.4% had a religious affiliation, and 80.3% had a high school education or below; 59.6% lived with family prior to death, with most reporting good relationships with family (82.8%) and friends/colleagues (71.9%); 62.6% were aware of their disease diagnosis, 47.8% were aware of impending death before passing away; 50.2% had made end-of-life arrangements; only 28.6% had expressed preferences regarding cardiopulmonary resuscitation (6 accepted, 52 refused). Regarding disease distribution, respiratory system diseases were most common (34.5%), followed by digestive system diseases (16.7%) and circulatory system diseases (16.3%). See Table 1.

The mean age of the deceased patients’ family members was 48.45 ± 11.53 years; 54.2% were male, 8.9% had a religious affiliation, 67.0% had a college degree or above, 66.5% were the patients’ children, and 59.6% lived with the patient. During the patient’s final hospitalization/observation period, family members spent a median of 8 (IQR: 5, 12) hours per day with the patient and provided care for a median of 3 (IQR: 1, 5) days per week. Pre-hospital visit frequency was collected only among family members who did not live with the patient before the final hospitalization (n = 82). See Table 2.

The mean total score of the Good Death Inventory (GDI) in this study was 224.46 ± 47.75, and the mean item score was 4.16 ± 0.88. According to the predefined criterion of mean GDI item score ≥5, 39 of 203 patients met the basic criterion for a good death, yielding a proportion of 19.2% (39/203). The dimensions were ranked in descending order based on their mean item scores (scale 1 - 7). The top four dimensions were Dimension 11 “Receiving sufficient treatment”, Dimension 6 “Good relationship with family”, Dimension 13 “Preparation for death”, and Dimension 17 “Feeling life is worthwhile”. The lowest five dimensions were Dimension 7 “Independence”, Dimension 18 “Religious and spiritual comfort”, Dimension 14 “Control over the future”, Dimension 1 “Physical and psychological comfort”, and Dimension 2 “Dying in preferred location”. See Table 3.

Scores were converted to a percentage (0 - 100 scale) for ease of interpretation. The mean item score for the total scale was 54.82 ± 12.89. The dimensions were ranked in descending order based on their mean scores. The top three were “Explanation provided by physicians to family members”, “Psycho-existential care”, and “Reasonableness of expenses”. The bottom three were “Explanation provided by physicians to patients”, “Accessibility of medical resources”, and “Environmental comfort”. See Table 4.

The frequency of CES mean substitution for “not applicable” responses was low. By domain, substitutions were 0/609 cells (0.0%) for physical care by doctors, 0/609 (0.0%) for physical care by nurses, 8/609 (1.31%) for psycho-existential care, 21/609 (3.45%) for doctors’ explanations to patients, 0/609 (0.0%) for doctors’ explanations to family members, 0/609 (0.0%) for environmental comfort, 2/406 (0.49%) for reasonableness of cost, 0/406 (0.0%) for attention to family health, 21/609 (3.45%) for accessibility of medical resources, and 0/609 (0.0%) for coordination and consistency of medical care.

This study revealed that patients who were over 60 years of age, not employed, aware of impending death, had expressed resuscitation preferences and made end-of-life arrangements, and had harmonious relationships with relatives and friends exhibited significantly higher Good Death scores. Furthermore, female family caregivers, those who visited the patient daily or more frequently among non-cohabiting family members, and those who provided an average of 0 - 8 hours of care per day during the patient’s final hospitalization/observation period rated the patient’s Good Death level significantly higher (P < 0.05). See Table 5.

Pearson correlation analysis was employed to examine the relationships between the total Good Death score and the total care quality score, as well as its subdomain scores, among emergency department decedents. The results indicated a significant positive correlation between the total GDI score and the total CES score (r = 0.627, P < 0.01). Furthermore, the total GDI score was significantly positively correlated with the total CES score and all its subdomain scores. See Table 6.

Using the total GDI score of emergency department decedents as the dependent variable, and the subdomains of the CES as independent variables, a stepwise regression analysis was conducted. This analysis controlled for covariates including general demographic characteristics of both patients and their family caregivers, as well as patient disease characteristics. Specific variable assignments are detailed in Table 7. The results identified the following as significant influencing factors for the Good Death level among emergency department decedents (P < 0.05): the patient’s age at death (B = 0.733), awareness of impending death (B = 15.840), having an educational level of college/university or above (B = 12.697), physical care provided by physicians (B = 5.476), environmental comfort (B = 3.797), accessibility of medical resources (B = 4.652), and the duration of the final hospitalization/observation period (B = −0.335). Collectively, these factors accounted for 60.7% of the variance in the Good Death level among emergency department decedents. See Table 8.

In the sensitivity model additionally adjusted for time from death to survey, the recall interval was not significantly associated with GDI total score (B = 0.010, 95% CI: −0.432 to 0.452, P = 0.965). The direction, magnitude, and statistical significance of the main predictors were materially unchanged, suggesting that the primary findings were robust to the recall-window adjustment.

Table 1. General information and disease characteristics of patients (n = 203).

Table 2. General information of family members (n = 203).

Table 3. GDI questionnaire total score and dimension scores (n = 203).

Table 4. CES questionnaire total score and dimension scores (n = 203).

Table 5. Differences in GDI total score by general information of emergency death patients and family members (n = 203).

Table 6. Correlation between GDI total score, GDI dimensions and CES total score, CES dimensions (n = 203).

Note: *P < 0.05; **P < 0.01.

Table 7. Multi-categorical variable assignment table.

Table 8. Multiple linear regression analysis of factors influencing good death in emergency patients (n = 203).

4.1.1. The Overall Level of Good Death among Emergency Department Decedents Is Moderate

In this study, the total GDI score was 224.46 ± 47.75, with an item mean score of 4.16 ± 0.88. According to the predefined GDI criterion used in this study, a mean item score ≥5 indicates the basic achievement of a good death. Based on this rule, 39 of 203 emergency department decedents (19.2%) essentially attained a good death level. This finding is close to the results of a good death survey conducted by Yang etal. from nurses’ perspective regarding cancer decedents (total good death score 245.40 ± 36.91, proportion with item mean score ≥5 was 20.9%) [14]. Furthermore, the survey results of Zhao Juanjuan etal. on 305 family members of cancer patients showed a total good death score of 241.20 ± 39.45, with the proportion of patients having an item mean score ≥5 being 24.3% [12]. These results indicate that the level of good death for both emergency department patients and cancer patients in China is moderate, and the quality of death for emergency patients urgently needs improvement to better meet clinical needs.

4.1.2. Main Characteristics of Good Death in Emergency Department Decedents

Descending order ranking revealed that the dimension “Receiving sufficient treatment” had the highest item mean score (5.22 ± 1.13), indicating that most emergency department decedents received relatively comprehensive medical interventions at the end of life. This result is higher than that found in Zhao Juanjuan etal.’s study on cancer patients (4.18 ± 1.25) [12]. This difference may stem from the characteristics of rapid response and high-intensity resuscitation in the emergency setting, which, compared to the relatively prolonged treatment process for cancer patients, more readily allows family members to intuitively perceive the full efforts of the medical team.

The item mean score for the dimension “Good relationship with family” ranked second (5.18 ± 1.20). Patients with good family relationships had significantly higher GDI scores (230.22 ± 45.14) than those with average relationships (198.58 ± 51.82) (P < 0.001), which aligns with the tradition of family coping with stress in Chinese culture. Hattori [15] noted that harmonious family relationships are an important prerequisite for achieving a good death.

Death preparation refers to the actions a person takes in preparation for the end of life, constituting a comprehensive process encompassing physical, psychological, social, and legal dimensions [16]. In this study, the dimension “Preparation for death” ranked third (5.03 ± 1.25), indirectly reflecting the relatively good fulfillment of the diverse needs of emergency department decedents and their families at the end of life. Adequate death preparation can alleviate death anxiety, assist in medical decision-making, reduce unnecessary interventions, and thereby enhance the quality of life at the end of life for patients [17]; it can also mitigate negative emotions among family members during bereavement [18]. It is important to note that some emergency patients died from sudden illnesses, lacking sufficient preparation for death, leading to lower scores in this dimension for this group, highlighting the urgency and particularity of death preparation in the emergency context.

In this study, the dimension “Independence” had the lowest score (2.30 ± 1.12), indicating that most emergency patients had difficulty maintaining basic physical mobility at the end of life. Reasons for this phenomenon include the impact of the disease itself and the use of various medical devices, such as cardiac monitoring, intravenous infusions, oxygen therapy, enteral nutrition tubes, various other lines, and physical restraints [19], all of which restrict patient mobility. Simultaneously, insufficient attention from emergency healthcare professionals to patients’ needs for maintaining physical activity is also a significant factor.

The item mean score for the dimension “Physical and psychological comfort” was 3.46 ± 1.88, indicating that the physical and psychological suffering of decedents in the emergency department environment was not effectively controlled. As the emergency department primarily focuses on treatment and resuscitation, coupled with insufficient nursing staff and a lack of related palliative or good death training [20][21], comprehensive physical and psychological care is challenging to achieve.

The item mean score for the dimension “Environmental comfort” was 3.78 ± 1.55. During telephone follow-ups, family members also mentioned issues such as “environment needs significant improvement,” “no bathing facilities, foul odor in toilets,” and “noisy like a marketplace at night.” As the frontline for acute and critical care, the environmental quality of the emergency department directly impacts patient outcomes and recovery processes. However, current emergency department environments in most hospitals have notable shortcomings: compact and complex spaces, lack of privacy and a quiet atmosphere, making it difficult for terminally ill patients and their families to attain peace and dignity [22][23]. Noise interference from equipment, staff movement, and patient moans disrupts patient rest and exacerbates their anxiety and distress [24].

General Characteristics of Patients and Their Family Members

The results of this study suggest that patient age is one of the important factors influencing the good death level of emergency patients. For each one-year increase in age, the good death score increased by 0.733 points (P < 0.001). Compared to young and middle-aged patients, elderly patients are generally more accepting of death and can face it with a calmer mindset, making preparations for death (arranging affairs, stating preferences regarding resuscitation) [12].

Patients who were “aware of impending death” had good death scores 15.840 points higher than those who were unaware (P < 0.001). This finding aligns with Western cultural concepts that consider “awareness of death” an important component of a good death, advocating for diverse end-of-life preparations among the public [25]. If patients are aware of impending death, they can more courageously discuss suitable options with doctors and family, actively seek palliative treatment, plan end-of-life matters, thereby enhancing comfort and sense of dignity in the final stage of life. However, a discrepancy exists between this finding and the conclusion of Zhao Juanjuan etal. [12]. This may be influenced, on one hand, by the Confucian cultural tradition of “not knowing life, how can you know death?” and on the other hand, closely related to the guiding role of Chinese medical ethics principles in clinical practice.

Patients with an educational level of college/university or above had higher good death scores (+12.697 points, P = 0.019). Populations with higher education levels tend to have better quality of death, attributed to their better ability to understand medical information and treatment options, and to adopt a proactive attitude towards illness, pursuing a higher quality of life [26].

The duration of the final hospitalization/observation period had a significant negative impact on the patient’s good death level. For each additional day of hospitalization/observation, the good death score decreased by 0.335 points (P = 0.028). Longer hospitalization/observation periods are associated with increased risks of nosocomial infections, immunosuppression, and delirium, alongside rising medical costs and decreased family satisfaction with medical care, consequently affecting the patient’s quality of death. Furthermore, a prolonged emergency stay may indicate more severe symptoms, greater treatment complexity, and longer exposure to the crowded emergency environment, thereby increasing psychological burden for patients and family members [27].

Physical care provided by doctors had a significant positive impact on good death (B = 5.476, P < 0.001). Studies have shown that stronger professional competence and nursing skills of healthcare providers, coupled with a better grasp of palliative care knowledge, enable them to play a more critical role in end-of-life care [27][28]. Physicians prescribing appropriate medication to promptly alleviate physical discomfort and pain is crucial for the patient’s good death. Certainly, nurses also play an indispensable role in the physical care of patients. However, in this study, only 60.1% of family members believed that “nurses possessed sufficient knowledge and skills to alleviate the patient’s discomfort symptoms.” This also indicates that when caring for terminally ill patients, nurses may pay insufficient attention to physical comfort, and their nursing philosophy and skills need enhancement.

CES dimension 6 “Environmental comfort” was significantly positively associated with good death (B = 3.797, P < 0.001). A good environment not only helps healthcare staff provide more meticulous physical and psychological care but also offers better support for family members. However, the noisy, confined space of the emergency department and frequent medical procedures easily cause psychological stress to patients [22]. During the COVID-19 pandemic, strict isolation policies impeded end-of-life care in the emergency department [29]. When the COVID-19 policy was fully lifted at the end of 2022, emergency departments undertook the task of treating a large number of COVID-19 patients, facing severe shortages of manpower, space, and medical supplies [30]. In this study, family members of the 28 patients who died during the COVID-19 period generally rated the emergency environment below 4 points. Clinically, medical staff use bed curtains and screens to provide a relatively independent space for emergency terminally ill patients, and some have reduced disruptions for patients and families spending their final moments together by turning off monitor alarms [31].

“Accessibility of medical resources” refers to the patient’s ability to be admitted promptly and receive professional care when medical services are needed. Terminally ill patients often suffer from symptoms like physical pain that are difficult to control, leading families to ultimately choose the hospital emergency department for urgent management [32][33]. In this study, CES dimension 9 “Accessibility of medical resources” significantly influenced good death (B = 4.652, P < 0.001). However, the item mean score for this dimension was 45.49 ± 17.52, ranking second lowest among all dimensions. This situation is closely related to factors such as severe overcrowding in tertiary general hospital emergency departments [34], difficulties in admission to specialized wards, and issues with health insurance policies [35]. In recent years, the state has introduced a series of policy measures to improve this situation. In 2017, the National Health Commission promulgated the “Palliative Care Practice Guidelines (Trial)” [36] and launched the first batch of pilot programs. With the initiation of the third batch of pilot work [37], the number of palliative care institutions and service teams has increased, allowing more patients to access professional care.

Many countries have focused on research regarding palliative care and quality of death for emergency department patients [38]. However, in China, the traditional view holds that the core function of the emergency department is emergency rescue, thus overlooking its value in palliative care. It is commendable that some scholars have begun active exploration: Ke Juqing etal. [39] introduced palliative care earlier into the emergency department to help patients and families choose suitable care plans, thereby improving the quality of life for terminal patients; Ning Xiaohong etal. [34] also achieved significant results by implementing palliative care consultations in the emergency department. The findings of this study demonstrate that the level of a good death is significantly influenced by factors within the control of healthcare systems and providers. To enhance the dying experience for patients and their families, we recommend: ① Optimize the care system by integrating palliative care into the emergency department, improving staffing, spatial layout, and related training; ② Strengthen physical and psychological care by enhancing symptom management and psychological support, and adding specialized nurses; ③ Promote Do Not Resuscitate (DNR) decisions and improve skills in delivering bad news; ④ Provide bereavement support for families and advance the popularization of palliative care and death education.

By focusing on these key areas, EDs can transform from being solely places of rescue to also being sanctuaries of compassion and dignity at the end of life.