Psychometric Validation of the LED Good Death Index in Patients with Terminal Cancer: A Mixed-Methods Study

Introduction

Ensuring a dignified and positive end-of-life experience for every individual has become a key focus in both societal discussions and government policies. Achieving this goal requires robust training programs and ongoing education for healthcare workers to conduct detailed appraisals of end-of-life services and make the best possible use of available resources [1]. What counts as a good death remains highly personal and is shaped by a person’s unique background, including cultural, social, political, and spiritual beliefs [2]. In East Asian societies, where Confucian and Buddhist philosophies have long influenced thinking, ideas about a good death contrast sharply with those commonly found in Western traditions [3–6]. Nearly all existing psychometric instruments used to measure a good death were originally built upon contemporary Western theoretical models. Although a few studies have introduced cultural adjustments, it is still important to create and apply tools that emerge directly from Eastern cultural understandings of this concept.

Meier and colleagues performed a systematic examination of prior qualitative and quantitative studies. They distilled several fundamental themes, including desired characteristics of the dying process, the absence of pain, emotional stability, family involvement, preservation of dignity, a sense of life completion, spiritual or religious dimensions, choices about treatment, overall quality of life, and the nature of interactions with medical staff [7]. A different comprehensive review of the literature, which incorporated views from people who were dying, healthy individuals, and care professionals, uncovered four recurring themes: (1) good death viewed as a form of control, (2) the misguided good death, (3) the good death under threat, and (4) outright rejection of the reality of dying. According to Cotterell and Duggleby (2016), discussions of a “good death” frequently carry a negative tone because they are seen as attempts to manage or regulate the dying experience for both the person facing death and those responsible for providing care [8]. Across many East Asian cultures, serious illness and death have historically been regarded as unlucky or taboo subjects, causing most people to steer clear of any conversation about “death” [9]. Clinical practice has repeatedly revealed that what patients themselves hope for in a good death often differs from the expectations of their relatives and medical team. This mismatch prompted the research group to explore the topic through patients’ eyes. Views on what makes a death good can shift dramatically depending on who is asked. Still, medical ethics places the highest value on patient autonomy, meaning the dying person’s thoughts and wishes must be given top priority. Each individual holds both the right and the duty to define a good death on their own terms. For this reason, patients should be empowered to make fully informed, autonomous decisions about accepting or declining medical treatments. Such autonomy can significantly improve their quality of life and help them prepare for death in a way that aligns with their personal idea of a good death.

Assessing whether a patient is moving toward a good death is a fundamental part of quality palliative care. Over time, various scales and inventories have been developed in East Asian countries to measure the quality of death. A well-known example is the Good Death Inventory (GDI) from Japan, which comprises 18 domains and 54 items [10, 11]. These domains are split into 10 core areas and 8 optional ones. The 10 core domains include (1) environmental comfort, (2) life completion, (3) dying in a favorite place, (4) maintaining hope and pleasure, (5) independence, (6) physical and psychological comfort, (7) good relationships with medical staff, (8) not being a burden to others, (9) good relationships with family, and (10) being respected as an individual. The eight optional domains cover (11) religious and spiritual comfort, (12) receiving sufficient treatment, (13) control over the future, (14) feeling that one’s life is worth living, (15) unawareness of death, (16) pride and beauty, (17) natural death, and (18) preparedness for death. Because it has 54 separate items, the GDI allows families to provide rich, detailed feedback on whether they feel the deceased person experienced a good death.

The development of the GDI relied on qualitative input from hospice nurses, a small-scale study of patients with advanced cancer, and observational data from patient settings. It has become a standard tool in palliative care research and everyday clinical work, always from the viewpoint of bereaved family members [12, 13]. In 2008, Miyashita reported that the GDI had good concurrent validity when compared with the Care Evaluation Scale and general care satisfaction scores. It also displayed solid internal consistency (alpha = 0.74–0.95) and reasonable test-retest reliability (ICC = 0.38–0.72) [10]. In parallel, the Good Death Score (GDS) was developed in Taiwan based on input from experts and university professors and is typically scored by palliative care staff [14]. The GDS is likewise used in both research and clinical palliative settings [15, 16]. Even so, while these regionally developed tools are widely in circulation, they have seldom been tested or validated through direct patient feedback.

A thorough review of the available literature shows that nearly all current good-death assessment instruments are filled out either by family members or by healthcare professionals. Instruments that let patients rate their own experience are almost nonexistent [17]. Although patient autonomy is universally upheld by medical professionals as a core ethical value, in East Asian settings, the autonomy of terminally ill patients is frequently overshadowed by family expectations and the traditional authority of physicians, largely because of deep-rooted Confucian values and the practice of filial piety [18]. Given the large population of terminal patients in East Asia, there is a genuine need for a culturally grounded psychometric instrument—built on Confucian principles and centered on genuine patient autonomy—that allows patients to evaluate their own dying process. Such a tool would benefit both day-to-day clinical care and future research.

The LED Good Death framework was constructed through extensive in-depth interviews with patients who had terminal cancer, followed by careful qualitative analysis carried out by a team of experts. This work was conducted within a research initiative funded by the Taiwan Ministry of Science and Technology [19]. LED is an acronym derived from the first letters of its three central themes: living in dying, experiencing the existential self, and dying in living [19]. Put simply, even when death is very close, people with terminal illness still need to affirm the value of their life and stay engaged with the present to reach a good death. This outlook echoes traditional Confucian thought, which stresses the importance of understanding life before death, as Confucius himself expressed: “Without knowing life, how can we know death” [20].

The purpose of this study was to provide healthcare teams with a practical scale that terminal patients can complete independently. This would give staff clearer insight into how patients feel about death and dying. In turn, caregivers would be better equipped to meet individual needs and guide patients with terminal illness toward achieving their own version of a good death.

Materials and Methods

Study design

This project was part of a larger 4-year research initiative supported by the Taiwan Ministry of Science and Technology and the Medical Research Department of MacKay Memorial Hospital. The main program was titled “Enhancing Medical Professionals’ Knowledge and Skill to Good Death of Dying Patients and Normal Grief of Their Family.” It ran from August 1, 2016, to July 31, 2020. Ethical clearance was obtained from the MacKay Memorial Hospital Human Research Ethics Review Board under reference number 15MMHIS113. Throughout the program, the Ministry of Science and Technology required annual progress reports documenting the research activities and results.

Data collection took place at the primary sites of MacKay Memorial Hospital, including the Taipei campus in Taipei City and the Tamsui campus in New Taipei City. MacKay Memorial Hospital is a major tertiary hospital with more than 2000 beds and offers full-spectrum cancer care as well as dedicated hospice and palliative services. Study participants were adult inpatients admitted to cancer wards or palliative care units, selected through purposive sampling. Every participant had to be at least 20 years old. Anyone under 20 or unable to read or write was excluded. Once informed consent was obtained, a research assistant assisted each participant in completing the questionnaire independently. The study was conducted in full accordance with the ethical principles of the Declaration of Helsinki.

Qualitative research method

In line with the qualitative work conducted by Fang et al. [19], the first stage of the qualitative phase involved recruiting 12 individuals diagnosed with terminal illness (life expectancy under 12 months) (Table 1) for semi-structured, in-depth interviews. Among them, 7 participants were female, and 4 were male. Their ages ranged from 29 to 73 years, with an average of 57.8 years. During the interviews, each patient answered five core questions: How would you characterize your daily life right now? What are your main thoughts and emotions about your present medical condition? What thoughts or feelings come to mind when you consider your impending death? In what ways would you like to use the time remaining to you? When you envision the actual moment of death, what kind of setting or atmosphere would you prefer? The latter part of each session was left open so patients could freely share any additional reflections. Before beginning, full consent was obtained to audio-record the conversations. Once completed, all recordings were transcribed verbatim and then processed using ATLAS.ti 7.5 software (Muhr T, ATLAS.ti Scientific Software Development, Berlin, Germany) with a hermeneutic phenomenological approach to analysis. To enhance the data’s credibility, two external reviewers—a senior palliative medicine specialist and a social worker—were invited to provide triangulation by cross-checking the material and verifying the research team’s interpretations. Following thorough analysis and validation, the findings revealed that patients living with terminal illness conceptualized their “good death” experiences through three main dimensions [19]. The first dimension, termed “living in dying (L)”, contained five specific aspects: (1) establishing a life pattern, (2) recalling past times, (3) confirming self-identity, (4) connecting with essential others, and (5) maintaining a doctor-patient relationship. The second dimension, “experiencing the existential self (E)”, included five aspects: (1) positive responses to sick experiences, (2) conflicting feelings at the border between life and death, (3) maintaining personal autonomy, (4) authentic companionship with others, and (5) settling in with religious or spiritual experiences. The third dimension, “dying in living (D)”, comprised five aspects: (1) having a complete understanding of the disease, (2) facing death with a rational attitude, (3) being aware of death through physical conditions, (4) planning after death, and (5) anticipating the situation of death [19].

Table 1. Relevant information of the 12 interviewees

Quantitative research method

The primary investigator, together with the co-investigator, developed 15 assessment items drawn directly from the 15 aspects identified in the qualitative phase, while also consulting supporting literature in the field. These 15 items were formatted on a four-point Likert-type scale where responses were scored as follows: 4 for “strongly agree”, 3 for “agree”, 2 for “disagree”, and 1 for “strongly disagree” (with total possible scores ranging from 15 to 60). The resulting instrument was designated the LED-Good Death Index (LED-GDI) (Table 2). Higher scores on this index reflected a heightened perception of a good death. To evaluate content validity, seven specialists from pertinent disciplines participated in an expert panel review. The group consisted of two palliative care physicians, one psychiatrist focused on palliative care, one senior nursing supervisor, one social worker, and one experienced research assistant. Panel members reviewed each item for its relevance and clarity of expression. Following this, 30 patients were recruited to trial the LED-GDI and confirm that every statement was straightforward and comprehensible.

Table 2. Contents and rating of the LED-GDI

All statistical analyses and descriptive summaries were executed using SPSS for Windows, version 18.0. Reliability checks and detailed item analyses were performed on the set of indicators. An independent-samples t-test was applied after dividing participants into high- and low-scoring groups to examine each item’s performance. Correlations were also computed between the mean score for each item and the total scale score. After these preliminary evaluations, an exploratory factor analysis (EFA) was conducted, using the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity to assess the appropriateness of factor extraction. Since the scale had originally been constructed from qualitative data and grouped into 15 items across three conceptual domains, the EFA results were used to reorganize the item sequence based on the observed statistical patterns. As a final step, confirmatory factor analysis (CFA) was conducted with AMOS software, version 24.

Results and Discussion

Demographic characteristics of the participants

The study successfully enrolled 144 patients. Participants had a mean age of 57.54 years (SD = 11.195), with 76 males and 68 females. In terms of religious background, 28 identified as Christian, 42 as Buddhist, 45 as Taoist, 5 as adherents of other faiths, and 23 reported no religious affiliation. Occupational backgrounds varied: seven were civil servants, 21 worked in industry, eight in commerce, 11 in service roles, 10 were self-employed, one was a student, 60 were retired, and 26 fell into miscellaneous categories. Education levels showed that 61 participants had completed junior high school or less, 45 had finished senior high school, and 36 held college-level qualifications or higher. The most frequent cancer types were malignancies of the upper digestive system and lower digestive system, each affecting 30 patients. The remaining cases included 19 patients with head and neck cancer, 18 with breast cancer, 17 with cancers involving the respiratory system, 16 with reproductive system cancers, and 14 with other cancer diagnoses (Table 3).

Table 3. Demographic characteristics of the participants (n = 144)

Item analysis

Item-level examination indicated that the LED-GDI produced an overall mean score of 49.43 (SD = 5.834). Among the individual items, the highest mean was recorded for Q8 (“I have a good relationship with the medical team”) at 3.54 (SD = 0.527). This was followed closely by Q10 (“I can feel good interaction and caring with others”) with a mean of 3.47 (SD = 0.566). The third strongest item was Q6 (“I can maintain good connections with individuals who are important to me”), scoring 3.45 (SD = 0.624) on average.

On the lower end, the three weakest items were Q2, Q3, and Q12. Q2 (“I can build new lifestyles”) received a mean score of 3.08 (SD = 0.762). Q3 (“I feel that my life has value”) averaged 3.15 (SD = 0.757). Q12 (“I know how to face the end of life”) had a mean of 3.16 (SD = 0.799).

When participants were divided into high- and low-scoring groups, t-test results revealed statistically significant differences on every single item. Additional correlation analysis confirmed that each item was significantly associated with the total LED-GDI score (Table 4).

Table 4. Item analysis of LED-GDI (n = 144)

Notes: *** Correlation is significant at the 0.001 level (two-tailed). ** Correlation is significant at the 0.01 level (two-tailed).

Exploratory factor analysis and reliability

The preceding qualitative interviews had shown that terminal patients tended to understand and describe their “good death” experiences according to three core viewpoints: L, E, and D. In the current quantitative phase, the KMO statistic reached 0.805 and Bartlett’s test of sphericity was highly significant (P < 0.001), confirming that the dataset was appropriate for factor analytic procedures.

To preserve the original three-dimensional framework (L, E, and D) identified earlier, the principal component analysis used 3 factors. Although the qualitative findings initially assigned Q4–Q5 to the E dimension and Q6 and Q8 to the L dimension, the exploratory factor analysis prompted several adjustments to item grouping across the three factors (Table 5).

Table 5. Rotated factors for principal components analysis of LED-GDI.

Reliability testing yielded a Cronbach’s alpha of 0.854 for the complete LED-GDI scale. The three subscales showed alphas of 0.766 for living in dying (L), 0.718 for experiencing the existential self (E), and 0.788 for dying in living (D). Furthermore, all three factors showed strong, significant correlations with the overall instrument score (Table 6).

Table 6. Reliability of the LED-GDI (n = 144)

Confirmatory factor analysis for construct validity

Confirmatory factor analysis was conducted using structural equation modeling to assess the proposed three-factor structure—living in dying (L), experiencing the existential self (E), and dying in living (D)—and the model’s overall fit. The tested model included all 15 items distributed across the three factors. Key fit indices were as follows: χ² = 208.701, df = 87, χ²/df = 2.399, P = 0.000, GFI = 0.847, AGFI = 0.789, and RMSEA = 0.09. Taken together, these statistics supported the validity of the three-factor structure and indicated an acceptable level of fit for the entire LED-GDI model (Figure 1) [21-23].

This research appears to be the pioneering effort in developing an instrument that enables patients with terminal illness to directly evaluate their own progress toward a good death experience. Upon examining previous literature, the team observed that most existing scales depend on family members providing retrospective judgments about the quality of death once the patient has already died, instead of allowing the patients to report their personal perceptions while still alive [10, 16, 24, 25]. Taiwan introduced the Patient Autonomy Act in January 2019, establishing a new legal framework that grants patients greater control over their healthcare choices. This change is especially meaningful because decisions such as declining futile interventions or opting for palliative support are now enshrined in enforceable legal rights [18]. In 1997, the Institute of Medicine described a good death as one “in which patients and their families are spared from suffering; their wishes are honored, within the context of clinical, cultural, and ethical considerations” [26].

According to Emanuel, patients often express deeper concerns about preserving dignity, avoiding excessive dependence, not becoming a burden to others, and retaining personal control compared with worries about physical pain or emotional distress [27]. Steinhauser similarly argued that addressing bodily needs is only the initial step; psychological, spiritual, and relational dimensions are equally important for both the patient and their loved ones [28]. A comprehensive “whole person” viewpoint is therefore necessary to fully appreciate the nature of suffering experienced by individuals at the end of life. In Taiwanese society, protecting personal dignity, facilitating a good death, helping families recognize the patient’s authentic preferences, and easing the sense of guilt or regret that survivors may feel are regarded as essential social obligations and fundamental expressions of human rights. The LED-GDI was created precisely to meet this need by offering patients a means to express their own expectations for a good death and enabling caregivers and families to gauge whether those expectations are being fulfilled.

The Good Death Inventory (GDI), originally developed by Japanese specialists in palliative medicine, remains a popular scale for assessing the quality of death [10]. It has been successfully adapted into Korean, Taiwanese, and Chinese versions, each demonstrating strong reliability and validity [16, 24, 25]. However, the GDI functions as a backward-looking measure completed by grieving family members to determine whether they believe the deceased had a good death [10, 11]. While useful for capturing family perspectives, it cannot reveal how the patient themselves felt about their dying process during the time they were receiving care. Relying on bereaved relatives’ views is consistent with longstanding Confucian traditions in East Asia, where direct conversations about death during serious illness are often avoided as culturally inappropriate [29, 30]. Nevertheless, contemporary clinical practice in end-of-life care is increasingly emphasizing the importance of hearing directly from patients about whether they sense they are approaching a good death, even within East Asian healthcare settings.

The scale developed by our team differs markedly from the GDI in both its design and its underlying philosophy. The LED-GDI emerged from in-depth conversations conducted directly with patients facing terminal illness and is structured around three primary domains. Its novelty lies not merely in being patient-completed but also in its distinct conceptual foundation that prioritizes continued meaningful living even in the final stages. Kübler-Ross [31] famously asserted that “the dying is, in fact, living,” reminding us that individuals with terminal conditions, although constantly aware of impending death, must still decide how to live their remaining time. For this reason, the first domain is labeled “living in dying” (L), acknowledging that patients continue to engage fully with life’s moments as long as they are alive. The second domain focuses on “experiencing the existential self” (E). Because patients remain alive while simultaneously confronting the reality of their approaching end, the third domain is termed “dying in living” (D) [18].

Three ideas expressed by Confucius in the Analects, recorded over 2400 years ago, closely align with the understanding of a good death uncovered in this study. The first states: “If you do not understand life, how can you understand death?” — underscoring Confucianism’s primary concern with how to live meaningfully. The second observes: “Life is lighter than a feather, and death is heavier than Mount Tai,” suggesting that a purposeful death carries greater weight than an aimless life. The third notes: “If you hear the Dao (truth) in the morning, you can die in the evening,” reflecting the belief that the search for life’s truths should persist until the very end [20]. These classical teachings are clearly embodied in the LED-GDI’s philosophical approach. To promote excellent care and support patients in reaching a good death, the mean total score on the LED-GDI in this sample was 49.43 ± 5.834. We therefore advise clinical teams to pay closer attention to any patient whose total score falls below 49, as lower scores may signal underlying distress contributing to existential suffering.

The study has several limitations. First, the development of the LED-GDI as a self-assessment tool for good death did not include comparisons with other established measures, such as quality-of-life instruments. As a result, it illuminates patients’ subjective sense of approaching a good death but does not address broader end-of-life concerns. Second, data collection occurred within a single hospital, which may have produced a relatively uniform pattern of end-of-life care practices. Third, the sample was restricted to patients with terminal cancer, leaving open the question of whether the LED-GDI would be suitable for individuals with other terminal conditions.

Subsequent research should strive to assess the concept of a good death concurrently from the perspectives of patients, family members, and healthcare providers to strengthen comprehensive end-of-life support. Additional validation is needed to establish whether the LED-GDI can function effectively as a good death measure for non-cancer terminal illnesses. We also encourage palliative care teams in other locations — particularly throughout East Asia — to trial the LED-GDI and explore its value in helping determine whether patients are progressing toward a good death.

Conclusion

The LED-GDI is an assessment instrument developed through direct patient interviews to help determine whether individuals feel they are attaining a good death. Its standout characteristic is the incorporation of existential ideas organized into three domains: L, E, and D. The scale has demonstrated acceptable reliability, suggesting that even patients with advanced cancer receiving palliative care can retain a meaningful awareness of their own existence. In the 21st century, particularly following the coronavirus disease 2019 pandemic, people have increasingly embraced a more self-directed outlook on both living and dying. Against this backdrop, the primary objective of clinical end-of-life care should be to ascertain whether patients with terminal illness can move toward a good death while continuing to live as fully as possible until their last moment.

Acknowledgments: We sincerely thank the patients with terminal cancer who participated in this study for expressing their opinions on a good death, even though they were facing death threats. We are grateful to three experts for their professional opinions: Dr. Chien-An Yao, head of the palliative care unit at National Taiwan University Hospital; Ms. Ching-Hui Chung, social worker and education lecturer at MacKay Hospice and Palliative Care Center; and RN. Chih-Ju Liu, nursing supervisor in cancer care units at MacKay Memorial Hospital.

Conflict of interest: None

Financial support: This study was supported by grants from the Taiwan Ministry of Science and Technology [MOST 105-2511-S-195-001-, MOST 106-2511-S-195-001-, and MOST 107-2511-H-195-001-] for 3 years and the Department of Medical Research at MacKay Memorial Hospital.

Ethics statement: None