End-of-life experiences: a multicenter longitudinal study on characteristics and prevalence in terminally ill patients

Abstract

Background

Evidence suggests that transcendent end-of-life experiences (ELEs) are common and impact the dying; however, there is a shortage of longitudinal studies.

Objectives

To evaluate the prevalence and characteristics of ELEs in terminally ill patients.

Methods

This multicenter longitudinal mixed‑methods study collected demographic, clinical, and ELE‑related data—including prevalence, content, and comfort/distress ratings—from patients with a Palliative Performance Scale (PPS) ≤ 50% across three palliative care units in Brazil. Data were gathered through repeated interviews until death, discharge, or ongoing hospitalization by the study’s end. Descriptive statistics were performed.

Results

Among 85 patients, 52% were female (44) and mean age was 61 (range 28–93); religiosity/spirituality was common (63% moderately/very religious, 98% believed in God, 65% in afterlife). Functional status declined progressively during follow‑up, with mean PPS decreasing from approximately 40 early in observation to 10 near death. ELEs were reported across varying levels of impairment. Of these patients, 34% experienced ELEs, with a median interval of 29.5 days from the first ELE to death (range: 1–186 days). Nearly half of the ELEs (49%) occurred during sleep, and 80% were perceived as real. The majority of ELEs involved deceased loved ones (52%), with 36% being described as comforting and 27% as distressing. Additionally, 78% of ELEs were reported by non-delirious patients.

Conclusions

ELEs exhibited consistent patterns in timing, phenomenology, and content, supporting the possibility that they represent an inherent feature of the dying process. Cultural and psychosocial contexts may shape both the interpretation of ELEs and their emotional valence.

Introduction

Over 56.8 million people worldwide require palliative care annually, including 25.7 million nearing the end of life [1]. Palliative and end-of-life care adopts an integrative approach, addressing physical, psychological, and spiritual aspects of patient care [2, 3]. Despite strong evidence of its importance in serious illness [4, 5], spirituality remains underemphasized, with limited clinical and academic exploration of patients’ inner experiences during the dying process [6–8].

Spiritual experiences reported by people who are close to death are common. Although such experiences have been systematically described in the literature since the 19th and 20th centuries, scientific interest has only recently resurged [7]. The term End-of-life experiences (ELEs) was proposed by Fenwick et al. [9] and encompass a broad range of spiritual or transcendent inner experiences that occur during the dying process [7, 9]. Reported across cultures and historical periods, these experiences have been documented in studies without consistent association with demographic characteristics, beliefs, illness, or circumstances of death [8, 10–13].

The spectrum of ELEs includes diverse phenomena, such as end-of-life dreams and visions, transitions between realities, deathbed coincidences—events reported at the time of death, such as distant relatives or friends perceiving the dying person or unusual environmental changes—and episodes of terminal lucidity [9, 14], brief periods of clarity in individuals with cognitive impairment shortly before death [15–17].

Dreams and visions are the most extensively studied form of ELEs. These frequently involve deceased loved ones and themes of preparation for a journey or transition [6, 18, 19]. Such experiences are typically deeply comforting and reassuring for both dying individuals and their families [6, 8, 20–22]. While a small number of patients report distressing experiences, often linked to past traumas or unfinished business [23–25], these experiences can be transformative, helping to address these challenges, alleviate guilt, or foster forgiveness [26, 27]. Overall, these phenomena provide profound personal and spiritual comfort, offering insight and contributing to patients’ post-traumatic growth [28]. However, despite their clinical relevance, distressing experiences remain underexamined and inadequately addressed in clinical practice [25].

Clinically, ELEs are often overlooked or dismissed as hallucinations or delirium caused by medication or an unstable clinical condition [10, 29]. However, unlike delirium, which involves fluctuating impairments in cognition, attention and disrupted sleep-wake cycle [30, 31] - ELEs usually occur in patients with preserved consciousness and are clearly remembered [6, 18]. They differ from typical hallucinations in the peace and acceptance they evoke, as well as their clarity, detail, and organization [18, 23, 32], along with the realism they convey [6]. Notably, studies show no significant link between opioid use and these experiences [33–35], and even when patients experience both ELEs and fluctuating states of delirium, ELEs remain distinct, providing subjective meaning and comfort [6].

In a dual-phase design study by Fenwick et al., clinicians across three UK palliative care settings were surveyed whether patients or relatives had reported specific ELE phenomena to them; such reports were common (91% of retrospective; 68% of prospective), most often involving deathbed visions of deceased loved ones or religious figures (62%; 48%) and experiences of resolving unfinished business (41%; 35%), usually occurring in clear consciousness (70%; 89%), being distinguished from medication- or dying-related confusion (76%; 79%), and being perceived as spiritually comforting for both patients (92%; 82%) and relatives (86%; 79%) [9].

Related findings have been reported in other settings. A study of 75 American nurses identified deathbed communications with deceased individuals in up to 95% of patients, with 89% passing away peacefully [29]. A cross-cultural study of 1,708 American and Indian nurses and physicians found that 28% of patients reported visions, with 62% occurring on their final day, often involving a deceased person (USA) or a religious figure (India) [35]. In Brazil, a multicenter study of 133 clinicians revealed that 71% had observed or heard reports of ELEs. The most frequently reported instances involved visions of deceased acquaintances or religious figures guiding the dying (88%), followed by visions of deceased relatives providing comfort (81%), deathbed coincidences (76%), and terminal lucidity (60%). Most clinicians considered ELEs transpersonal experiences (78%), and distinguishable from fever or drug induced hallucinations (69%) [36]. In a further study, in the US and Europe, involving 187 clinicians, relatives, and caregivers, 124 reports describing lucid episodes in dementia patients with severe cognitive impairment were evaluated. Most episodes (89%) lasted up to 24 h, and 79% involved a return of clear, coherent, near-normal verbal communication. Furthermore, 56% of patients died within two days of a lucid episode, whereas episodes persisting beyond 24 h were associated with longer survival [37].

Although research on ELEs has increased in recent years [38], significant gaps persist. First, most studies are retrospective and rely on caregivers and clinicians’ reports, potentially introducing bias. Second, only few studies have examined ELEs from the perspective of dying individuals themselves, and among these, most have focused exclusively on dreams and visions, rather than the broader spectrum of ELEs. Third, many of these studies have excluded patients with delirium during follow-up in order to distinguish ELEs from delirium more clearly. Although this strategy has methodological value, it also limits understanding of how these phenomena may co-occur in real clinical settings, which is highly relevant for bedside assessment and care. Fourth, the literature remains geographically concentrated in North America and Europe, underscoring the need for studies in culturally and religiously diverse settings [7, 39]. To date, few cross-cultural investigations have systematically examined ELEs phenomenology, reinforcing the importance of considering how social, cultural, and contextual factors may shape these experiences [40]. In South America, only two studies have been conducted, one in Argentina [41] and one in Brazil [36], both limited to healthcare professionals.

These limitations are clinically relevant because religiosity and spirituality are strongly intertwined with worldviews, values, coping, and health-related outcomes [42], and beliefs in life after death are widespread globally [43, 44]. For example, around half of U.S. adults (53%) reported have been visited by a deceased family member in a dream or some other form [45]. In Brazil, anomalous perceptions involving deceased persons have also been reported as highly prevalent in the general population [46]. Such cultural and religious frameworks may influence not only the frequency of ELEs but also the meanings attributed to them, their emotional valence, and the way they are interpreted in clinical settings. Accordingly, culturally sensitive research should not be restricted to dreams and visions alone, but should also include other forms of ELEs, including experiences that individuals interpret as contact with what they regard as sacred [47].

To address these gaps, we conducted a multicenter, longitudinal, mixed-methods study that prospectively examined ELEs among terminally ill patients in an culturally diverse South American setting. Its objectives are to: (1) determine the prevalence of ELEs among terminal ill patients; (2) analyze their content; and (3) assess the level of comfort they provide; (4) and explore their co-occurrence with delirium.

Methods

Design

This study forms part of a broader investigation of ELEs in Brazil involving patients, family caregivers, and healthcare professionals. It was designed as a multicenter, mixed-methods longitudinal study using a convergent design [48], with concurrent data collection and independent quantitative and qualitative analyses, followed by comparative integration to examine convergence and divergence.

To address a key gap in the literature, ELE data were collected prospectively and in parallel from patients, family caregivers, and healthcare professionals, enabling cross-informant comparison to strengthen validity. To enhance comparability across respondent groups, we used the same core ELE assessment questions, with minor wording modifications to reflect each informant’s perspective: caregivers reported patient ELEs they observed or that were reported by the patient, whereas healthcare professionals were interviewed using corresponding questions about patient ELEs they had witnessed or that were reported to them during follow-up. Additional questions and questionnaires were also administered to caregivers and clinicians.

Participants and settings

This study included terminal ill patients receiving palliative care from three centers of excellence in Southern and Southeastern Brazil. These centers comprised the Barretos Cancer Hospital in São Paulo, a leading cancer hospital with a 50-bed inpatient unit exclusively for palliative care, and two university-affiliated hospitals: the CuidATIVA – Regional Center for Palliative Care at the Federal University of Pelotas in Rio Grande do Sul, providing home-based palliative care for up to 10 oncology patients at a time, and an eight-bed inpatient unit at the Technical and Scientific Center for Palliative Care of the Clinical Hospital, Faculty of Medicine, University of São Paulo.

All participants had advanced, incurable diseases and were followed in either home-based (Pelotas) or inpatient settings (Barretos and São Paulo). Advanced home palliative care was considered equivalent to inpatient care, as both provided professional support until death. Inclusion criteria were: (1) age over 18 years; (2) ability to provide informed consent; and (3) a Palliative Performance Scale (PPS) score of at most 50% - indicating a mean of 37 days until death [49, 50]. Patients were excluded if they had delirium at enrollment or language or communication impediments. Patients who developed delirium during follow-up continued participation.

The sample size was estimated using a one‑sample proportion test to compare the expected ELE prevalence in our cohort with the reference proportion reported by Kerr et al. [6]. Assuming a two‑sided α = 0.05 and 80% power, G*Power (v3.1.9.7) indicated that, based on a preliminary prevalence of 74% in our study compared with the 88% reference prevalence reported by Kerr et al. [6], a minimum of 65 participants was required.

Measures

• Demographics: Age, sex, race/ethnicity, marital status, religious affiliations, education, income, and primary diagnosis.

• Religious and Spiritual Beliefs: Five items assessing self-reported religiosity (rated on a 4-point Likert scale) and beliefs in God, life after death, reincarnation, and dualism (soul’s existence), based on “yes,” “no,” or “not sure” responses. These measures were previously used in the only ELE Brazilian study [36, 51].

• Palliative Performance Scale (PPS-Version 2): This clinician-rated tool is widely used in palliative care to summarize patients’ overall performance and care needs and as tool to estimate survival among patient populations across care settings. It is scored from 100% (full ambulation and normal activity) to 0% (death) in 10% increments, based on a best-fit assessment across five observable domains: ambulation, activity level/evidence of disease, self-care, oral intake, and level of consciousness. Lower scores indicate greater functional decline and dependence [49]. In this study, it was used the 2009 Brazilian Portuguese version to characterize participants’ functional status and to describe functional trajectories in relation to ELE occurrence [52].

• Confusion Assessment Method (CAM): The CAM is a highly sensitive and specific tool for identifying delirium (acute confusional state) [53], and has been validated for use in palliative care settings [54]. According to the CAM algorithm, a delirium diagnosis requires the presence of both the first criterion (acute onset or fluctuating course) and the second criterion (inattention), along with either the third (disorganized thinking) or fourth criterion (altered level of consciousness) [53]. The Portuguese version, which has demonstrated a sensitivity of 94% and specificity of 96%, was used in this study [55].

• ELEs Questionnaire: In the absence of a validated global scale for ELEs, this study utilized an adapted version of Kerr’s End-of-Life Dreams and Visions survey [6], translated into Brazilian Portuguese with the author’s permission. The first author conducted the translation, which was reviewed by the senior author; the final version was reached through iterative revisions and consensus. This survey was expanded to include perceived contact with what participants considered sacred; sacred was understood as experiences highly valued and important in relation to an individual’s religious/spiritual identity. Because the COVID-19 pandemic substantially restricted access to terminally ill patients, the items were not pilot-tested or content validated by experts due to feasibility constraints. The questionnaire examined the presence of ELEs, their occurrence (sleep/wakefulness), content, frequency, perceived realism, and emotional valence (comforting/distressing). Open-ended questions, aligned with previous research [23, 56], explored patient descriptions, feelings, meanings, and interpretations of ELEs. Given participants’ terminal condition and vulnerability, all closed- and open-ended items were administered by the first author in interview format.

Data collection and procedures

• Data Collection: Data were collected between September 2021 and October 2022 by the first author, who was trained in administering the survey instruments. Collection periods varied across palliative care units (PCUs), ranging from 45 days in the smaller units (São Paulo and Pelotas) to 90 days in the largest unit (Barretos).

• Procedures: Patients who met the inclusion criteria were invited to participate. Those who agreed signed an informed consent form and were asked about the presence of ELEs at least three times per week until their outcome (death, discharge, or continued hospitalization by the study’s end). For patients who were discharged or remained hospitalized, death dates were later obtained from the PCU.

• Initial Assessment: Participants completed a sociodemographic questionnaire, along with questions regarding their religious and/or spiritual beliefs. Clinical information and the presence of ELEs were also assessed, using the question: “Are you having dreams, visions, or contact with what is sacred to you?“. Affirmative responses triggered the administration of ELEs questionnaire with closed and open-ended questions designed to elucidate the characteristics of these experiences. These interviews were recorded and transcribed for analysis.

• Follow-Up: The PPS and CAM were regularly monitored. Participants were continuously screened for the emergence of ELEs. Positive responses to the aforementioned ELE inquiry prompted the application of the same ELEs questionnaire used at baseline to characterize the reported experiences (Fig. 1).

Fig. 1.

Data collection process

Data analysis

Prior to analysis, the dataset was audited for data-entry accuracy by cross-checking all entries against the original paper survey forms. Eligibility for the primary analysis required that participants report at least one reported dream, vision, or perceived contact with the sacred during follow-up.

Descriptive statistics were computed (means, medians, and standard deviations for continuous variables; frequencies for categorical variables). Interview transcripts were examined using a phenomenological approach and Bardin’s content analysis framework (pre-analysis, material exploration, and treatment/interpretation of results) [57]; these qualitative findings will be reported in a subsequent publication.

During pre-analysis, we observed that not all endorsed experiences fulfilled the criteria for ELEs, indicating that the screening question was highly sensitive but insufficiently specific. To mitigate this aspect, qualitative data derived from open-ended responses were triangulated with quantitative data from closed-ended questions in the ELEs questionnaire. Each experience was evaluated using a conservative, multi-rater procedure: the first author and two collaborators independently reviewed all transcripts, compared their classifications, and analyzed them in conjunction with the quantitative data. Disagreements were resolved through discussion; unresolved cases were reviewed by a fourth investigator (senior author).

Experiences were classified as ELEs only if they satisfied all of the following criteria: (1) Content usually reported in ELEs (e.g., deceased relatives or pets; traveling/“going somewhere”; encounters with living persons; revisiting meaningful past events); (2) Realism (real and vivid experiences, e.g., not just as an illusion or ordinary dream); (3) Meaningfulness (personally significant experiences grounded in the patient’s life narrative, often associated with meaning and belief).

ELEs could be reported directly by the patient or by a family caregiver or healthcare professional, based on their observations or patient accounts obtained during follow-up. To ensure temporal consistency and preserve a clear focus on experiences occurring in proximity to death, analyses were restricted to ELEs occurring within 90 days prior to the baseline interview or during follow-up. This restriction was necessary because, at baseline, some patients reported experiences that had occurred before study enrollment. Limiting inclusion to the 90-day pre-baseline period reduced heterogeneity in recall timing and avoided incorporating more remote events that may have been less representative of the terminal phase. This timeframe was also considered clinically appropriate because participants had already been admitted to hospice care, and thus had an expected survival of approximately six months, reflecting the clinical context of proximity to death and its attendant complexity.

Results

Patients characteristics

The final sample comprised 85 patients. The mean age was 61 years (range: 28–93). The majority of patients were female (44/52%), white (40/47%) or mixed race (34/40%), and married (43/51%). Most patients had low income (61/71% ≤ 3 minimum wages), and about half (40/47%) had only elementary education. The majority of participants were Catholic (50/59%) or evangelicals (34/40%). The most common clinical condition was cancer (72/92%), particularly those of the gastrointestinal tract (20/25%), lung (14/17%), and breast (13/15%) (Table 1).

Table 1.

Sociodemographic and clinical characteristics of the patients (N = 85)

Characteristic	Group	N (%)
Age (years)	Average (± SD)	61 (± 13)
PCU Location	Barretos	71 (84)
	Pelotas	7 (8)
	São Paulo	7 (8)
Sex	Female	44 (52)
Ethnicity	White	40 (47)
	Mixed Race	34 (40)
	Black	8 (9)
	Asian/Indigenous	3 (4)
Marital Status	Married/Common-law	43 (51)
	Widowed	15 (18)
	Single	14 (16)
	Separated/Divorced	13 (15)
Education	Elementary School	36 (42)
	High School	31 (36)
	Higher Education/Graduate	14 (17)
	None	4 (5)
Monthly Family Income	Up to 1 MW	25 (29)
	1–3 MW	34 (40)
	3–6 MW	16 (19)
	More than 6 MW	8 (9)
	No income	2 (2)
Religious Affiliationa	Catholic	50 (59)
	Evangelical	34 (40)
	Spiritist	13 (15)
	None	8 (9)
	Umbanda	3 (3)
Clinical Condition	Cancer	78 (92)
	Gastrointestinal	20 (25)
	Lung	14 (17)
	Breast	13 (15)
	Prostate	7 (8)
	Brain	5 (6)
	Others	19 (21)
	Cardiovascular Diseases	5 (6)
	Neurological Diseases	2 (2)

^aThe sum was greater than 100% because participants could mark as many religious affiliations as they wished

Minimum Wage (MW)

Most patients were from Barretos Cancer Hospital (71/84%). As the data collection was during the COVID-19 pandemic, many patients delayed seeking medical attention despite severe symptoms. As a result, they arrived at this PCU seeking a cure but were diagnosed with advanced terminal cancer and admitted for palliative care, with an average hospital stay of 15 days until death.

Regarding outcomes, 35 (41%) patients were followed until death, 29 (34%) were discharged during the follow-up period, 19 (22%) remained hospitalized at the end of data collection, and two patients withdrew from the study.

Participants were highly religious or spiritual, with 54/63% identifying as moderately or very religious, 83/98% believing in God, 55/65% in life after death, 69/81% the duality of body and soul, and 35/41% in reincarnation (Table 2).

Table 2.

Religious and spiritual beliefs

Questions	Responses	N (%)
How religious do you consider yourself?	Not religious	6 (7)
	Slightly religious	25 (30)
	Moderately religious	41 (48)
	Very religious	13 (15)
Do you believe in God?	Yes	83 (98)
	No	2 (2)
	No opinion	-
Do you believe in life after death?	Yes	55 (65)
	No	11 (13)
	No opinion	19 (22)
Do you believe in reincarnation?	Yes	35 (41)
	No	31 (37)
	No opinion	19 (22)
Is a human being composed of a body and a soul/spirit?	Yes	69 (81)
	No	4 (5)
	No opinion	12 (14)

ELEs prevalence and features

A total of 104 interviews were conducted with a mean of 1.7 interviews per person. Of the 85 patients interviewed, 29 (34%) experienced at least one ELE, resulting in a total of 48 reported ELEs. In cases where family members or healthcare professionals provided the accounts, they occasionally lacked complete details about the experiences; consequently, detailed data on ELE features were not available for all reported experiences in Table 3.

Table 3.

End-of-Life Experiences features (N = 48)

ELEs Features		N (%)
Asleep or awake?	Asleep	16 (49)
	Awake	7 (21)
	Both	10 (30)
Total		33 (100)
What time of day?	Morning	4 (12)
	Afternoon	5 (16)
	Evening	6 (19)
	Early morning	10 (31)
	More than once a day	7 (22)
Total		32 (100)
Who/what are you seeing?	Deceased friends/relatives	25 (52)
	Deceased pets	1 (2)
	Living friends/relatives	2 (4)
	Living pets	2 (4)
	Other people	2 (4)
	Religious figures	1 (2)
	Meaningful past experiences	-
	Other transcendent experiences	15 (31)
Total		48 (100)
How many times in the past 24 h?	Once	8 (26)
	2 to 4 times a day	10 (32)
	More than 4 times a day	2 (7)
	None	11 (35)
Total		31 (100)
Do they seem real to you?	No	6 (20)
	Yes	24 (80)
Total		30 (100)
Do you remember these experiences?	No	3 (10)
	Yes, a little	16 (53)
	Vividly	11 (37)
Total		30 (100)
Comforting or distressing?	Extremely distressing	6 (18)
	Distressing	3 (9)
	Neither distressing nor comforting	8 (25)
	Comforting	9 (27)
	Extremely comforting	3 (9)
	Both comforting and distressing	4 (12)
Total		33 (100)

About half of occurred during sleep (16/49%), while 7 (21%) happened while the patients were awake, and 10 (30%) took place in both states. These events were most commonly reported during the night and early morning hours (16/50%). Notably, 24 (80%) ELEs were perceived as real, and 27 (90%) participants reported recalling the experiences, with 53% remembering them ‘a little’ and 37% remembering them ‘vividly’ (Table 3).

For patients who experienced ELEs, the median time from the first ELE to death was 29.5 days (range: 1 to 186 days), with a mode of 8 days.

Content of ELEs

Patients shared descriptions of their ELEs, with illustrative examples shown in Table 4. The majority of reported experiences (25/52%) involved interactions with deceased friends or relatives. A minority of accounts described encounters with living friends, relatives, or pets (n = 2), and unrelated individuals (n = 2). Similarly, religious figures were mentioned in only one instance. The subgroup labeled “Other transcendent experiences” (15/31%) comprised heterogeneous reports, including perceived contact with spiritual entities, descriptions of shifting between different realities, and sensations of divine presence and protection (Table 4).

Table 4.

Illustrative examples of patient-reported End-of-Life Experiences

	Examples
Deceased Friends/Relatives/Pets	Betty (Age 67): “My mother, who passed away years ago, has been appearing in my dreams recently. Every night, I dream of her and my father, who is also deceased. I see them, and I’m not sure if it’s to bring me peace or to help me in some way [DBD 36]. Robert (Age 72):“He saw his father in a dream, spoke with him, and told him to wait a little longer. Last Saturday, he called out for his father, and when people passed by, he pointed and said, ‘There’s my father.’ That night, he managed to get out of bed unassisted, though he was very frail. He gestured and spoke softly, saying, ‘Calm down, Dad, I’ll come later, but not now’ [DBD 7]. Sarah (Age 64):“I dreamed of being in a closed space with people singing, dancing, and talking. From a distance, I saw someone who resembled my father. I tried to approach him, but as I moved closer, he moved further away. Relatives—some alive, some deceased—appeared, including my grandmother, who encouraged me to forgive and close the cycle. These issues are serious, and I still haven’t resolved them [DBD 170]. Brian (Age 63):“Today, I dreamed of him, though he passed away over ten years ago. He was in USA. He looked happy and fat, and he greeted me before walking away. It felt incredibly real [DBD 7]. Maya (Age 74): “First, I felt someone beside me and heard them say I wasn’t well. I woke up and saw no one, but later, I dreamed of my deceased mother. I only remember her face, perfectly as it was when she was alive [DBD 34].
Living Relatives	Stephanie (Age 74): “She dreamed of a beautiful green field full of flowers. She saw our cousin Mary, dressed in white, reaching out to her, which gave her a sense of well-being” [DBD unknown]. Linda (Age 61): “I dreamed my niece had a baby girl. She was so beautiful and brought me immense joy. When I woke, I still felt happy, as though the dream symbolized hope and new life for me [DBD 25].
Other People	Jamie (Age 61): “He told me that he had dreamed of several people coming to take him, and at that moment, he could see them approaching. He said, “They are bad, and they have come to take me.” Shortly after, he began bleeding and was unable to speak further. A spiritual leader from the local spiritist center, came to talk to him. Afterward, I noticed that his heart had stopped. I felt a sense of relief, a peaceful feeling, as if he had truly said his goodbyes. I looked outside, and the wind was unusually strong, making the tree branches sway vigorously. Then it all calmed down. He was calm” [DBD 1]. Sophie (Age 39): “I saw him again, running around here. It seemed like he was trying to get my attention. He was near his mother—it’s funny how much they resembled each other. Physically and even in his gestures, he reminded me more of Mary than his father. I didn’t know who he was at first; he had straight hair… Yes, the dream felt very real. It’s curious because this person doesn’t even exist yet. Perhaps they could exist someday, but as of now, Mary doesn’t have a child. For me, this dream was extremely comforting, as I’ve been feeling deeply sad for quite some time, and nothing good seemed to happen. Then this appeared. The meaning of this dream, to me, is that I believe I’m being supported across all three realms: the realm of the living, the realm of the deceased, and the space of those who are yet to be born” [DBD 18].
Religious Figures	Bob (Age 54): “I dreamed that there was a struggle between good and evil over who would take me. It lasted a long time because I would wake up, wait about five minutes, and then return to the same place in the dream—exactly the same place. Then, all of a sudden, the struggle ended. The side with God took me, and I went with them. After that, it was over… I went with God. I felt scared because nothing like that had ever happened to me before. But later, everything felt normal again. In the dream, there was God, there were angels, there were demons—everything. It felt like a courtroom, with one side wanting me and the other opposing it” [DBD 72].
Other Transcendent Experiences	Steve (Age 45): “In these dreams, I feel a profound sense of God’s presence. I feel that God has poured the Holy Spirit upon us, upon our family. I feel that both I and my family have been surrendering ourselves, changing our routines—from our prayers to our way of being, thinking, and showing solidarity with others. God has been speaking to us, guiding us to follow this path, assuring us that we will reach where we are meant to be. God has a purpose for us” [DBD 54]. Katherine (Age 29): She said, “Mom, I had a dream. I just had a dream… I dreamed that I died, you know? And you cried a lot, cried so much, Mom. I don’t know what it means, this dream.” That’s how she told me, my daughter… She said that in the dream, she was walking, walking, and then someone was calling her, calling her. And there were flowers too, there were flowers” [DBD 14]. Vernon (Age 88): “He said, ‘Oh, no, but everything is going to be fine because a doctor with a mustache from the spiritual realm came to talk to me. He said everything is going to be fine, everything is okay. He prayed with me, and then he left’ [DBD 8].

Comfort levels of ELEs

Patients rated their ELEs on a 5-point comfort-distress scale (1- extremely distressing; 5 extremely comforting). Among ELEs, 12 (36%) were rated as either extremely comforting or comforting, 9 (27%) as either distressing or extremely distressing, and 4 (12%) as both comforting and distressing. Notably, 8 (25%) of the ELEs were categorized as neutral, being neither distressing nor comforting (Table 3).

Patient’s cognitive state at the time of interview

The majority of the ELEs (36/78%) were reported by cognitively intact patients with a negative CAM score. Conversely, only 10 (22%) ELEs were reported by patients with a positive CAM score, suggestive of delirium. In these delirium-associated ELEs, it was relevant to note the degree of attention observed during the interview, since delirium is a fluctuating condition and consciousness in terminal ill patients may vary over time. As most interviews occurred shortly after the ELE occurrence, the patient’s immediate mental state provided important context for interpreting the report.

Among these cases, 7 (70%) involved patients who were mildly (2/20%) or markedly (5/50%) attentive during the interview. Notably, 9 (90%) of these experiences were reported by lethargic patients, 5 (62%) by patients with psychomotor retardation, and 3 (33%) by patients with an altered sleep-wake cycle. In contrast, most of these delirium-associated ELEs were not accompanied by disorganized thinking (8/80%), disorientation (8/89%), or memory problems (8/89%). Only one experience (1/11%) involved hallucinations (Table 5).

Table 5.

Characteristics of delirium-associated ELEs

Confusion Assessment Method - CAM		ELEs N (%)
Acute change of mental state	Yes	10
Attention focus	Absent	1 (10)
	Mildly Present	2 (20)
	Markedly Present	5 (50)
	Uncertain	2 (20)
	Total	10 (100)
Disorganized thinking	No	8 (80)
	Yes	2 (20)
	Total	10 (100)
Level of consciousness	Vigilant	1 (10)
	Lethargic	9 (90)
	Total	10 (100)
Disorientation	No	8 (89)
	Yes	1 (11)
	Total	9 (100)
Memory problems	No	8 (89)
	Yes	1 (11)
	Total	9 (100)
Perceptual disorders	No	8 (89)
	Hallucinations	1 (11)
	Total	9 (100)
Psychomotor agitation/retardation	No	2 (25)
	> abnormal psychomotor activity	1 (13)
	< abnormal psychomotor activity	5 (63)
	Total	8 (100)
Alteration of the sleep-wake cycle	No	6 (67)
	Yes	3 (33)
	Total	9 (100)

Longitudinal changes in palliative performance and confusion assessment during follow-up

The study’s design enabled examination of how functional decline—a proxy for phases of the dying process—corresponded with the timing and pattern of ELE reports, an area rarely addressed in prior literature. Mean PPS among patients reporting ELEs was 40 across follow-up time points 1–5, signifying that patients were predominantly bedridden and required substantial assistance. Mean PPS subsequently declined to 20, consistent with bed confinement and near-total dependence. A transient improvement to 40 was followed by a decrease to 30 and a return to 40, after which PPS values showed pronounced fluctuations between time points 7 and 21. A further decline to 10 was observed between time points 15 and 21, reflecting profound deterioration with minimal intake and reduced consciousness consistent with imminent death.

ELEs were reported across varying levels of functional impairment, except at time point 6. Overall, this trajectory underscores the heterogeneous reporting patterns, likely influenced by differences in follow-up duration, intermittent missing data due to patient clinical instability, and the possibility that some events were either unreported or not classified as ELEs.

Figure 2 shows the temporal distribution of ELE reports at each assessment point, stratified by delirium status according to the CAM algorithm, with patients classified as CAM-positive (delirium) or CAM-negative (no delirium). Most reports came from patients without delirium, except at time points 7 and 9, when reports were observed only in patients with delirium.

Fig. 2.

Temporal distribution of end-of-life experiences according to delirium status

Discussion

This study examined the frequency, themes, and associated comfort of ELEs among terminally ill patients. It addresses key gaps in the literature through its multicenter, prospective design and its focus on patients’ first-person phenomenological reports rather than proxy accounts. To our knowledge, it is also the first investigation of ELEs among Brazilian patients, offering important cross-cultural insights to the field.

During follow-up, 29 patients (34%) reported at least one ELE during the follow-up period. This prevalence is lower than rates reported in studies conducted with dying populations in the United States (US) [88% [6], 66% [18] and 50% [28] ]; Switzerland [90% [58] ]; India [63% [56] ], and Sweden [64% [19] ]. By contrast, studies based on family caregiver reports have described prevalence estimates more comparable to ours (21–40%) [20, 21, 34, 59, 60].

Direct comparisons across ELE studies should be made cautiously due to methodological heterogeneity, particularly interview frequency: studies using daily interviews and observations until death report the highest prevalence [e.g., end-of-life dreams and visions [6] and spiritual experiences [58]. Our lower prevalence likely reflects differences in follow-up duration, interview frequency, and screening, because only 41% of our sample was followed until death and ELEs increase in the days preceding death [7, 9, 22]; however, ELE prevalence in our study sample was identical among those followed to death (34%, 12/35) and those not (34%, 17/50).

Underreporting may also have contributed to the observed prevalence. While death anxiety is notably high among patients with advanced cancer [61] and often correlates with religious struggle and dignity-related distress [62], this existential vulnerability may be exacerbated by the clinical environment. In highly medicalized settings, where transcendent experiences are often met with skepticism, patients facing significant psychological and spiritual distress may be particularly wary of disclosure. The fear of being judged or pathologized acts as a primary barrier; consistent with prior ELE research, both patients and caregivers frequently avoid reporting such experiences due to concerns about being perceived as “insane” or having their accounts dismissed by healthcare professionals [19, 60, 63–65].

In line with prior work, Brazilian patient’s ELEs were most frequent within eight days of death [6, 14, 22, 29, 34, 66]. Similarly, to the American study [6], Brazilian ELEs occurred predominantly during sleep (49% in this investigation and 45% in the US study), or both states (30% vs. 39%, respectively). Common themes included encounters with deceased loved ones (52% vs. 46% in US), and a strong sense of realism (80% and 99% in US). Nevertheless, the US study documented a higher frequency of ELEs featuring living loved ones (17% compared to 4%) and a greater proportion of comforting or extremely comforting experiences (60% vs. 36%), while our study found more distressing or extremely distressing ELEs (27% versus 19%).

Comparisons with India [56] also suggest both shared and divergent features. In the Indian study, more experiences occurred during both wakefulness and sleep (66% versus 30% in the current investigation), and fewer reporting a strong sense of realism (68% vs. 80%). Similarities were observed regarding the capacity to recall the experiences either somewhat or vividly (92% vs. 90%) and the timing of the occurrences, primarily at night (53% vs. 50%). Importantly, both Indian [56] and Brazilian studies found higher rates of distressing experiences (84% and 27%, respectively) compared to U.S. studies (19%) [6].

The literature suggests that distressing experiences are correlated with prior traumas, unresolved psychological conflicts, and complex interpersonal dynamics [23–25], highlighting the importance of psychosocial factors in influencing individuals’ approaches to their mortality [58, 67]. Brazil and India, two developing countries, ranked low on global quality-of-death indices (42nd and 67th, respectively) [68], may have societal attitudes toward death that influence how patients process their inner experiences as they near the end of life. Nevertheless, the combined contribution of these psychosocial factors—together with the impact of the COVID-19 pandemic on trajectories of mortality acceptance and the management of ELEs—remains insufficiently understood.

Experience appraisal appears central: recent reviews highlight that how people interpret transcendent or non-ordinary experiences significantly influences their mental health outcomes. Doubting or negatively appraising these experiences can lead to distress, fear, and shock, highlighting the importance of cultural expectations and cognitive models in shaping these outcomes. While cultural and cognitive factors are key, more research is needed on how appraisals work and how stable belief systems might offer protection. Therefore, assessment should clearly distinguish between these experiences themselves and their appraisal [69, 70]. Consistent with seminal work suggesting ELEs may become increasingly comforting as death approaches [6], the relatively high proportion of participants with PPS 40 (greater impairment but potentially more time until death) may partly account for the higher proportion of distressing ELEs observed.

Evidence indicates that belief in life after death is associated with enhanced mental health outcomes, encompassing reduced anxiety, depression, and symptom severity in phobias and somatization [71–73]. It seems to foster tranquility, alleviating the negative effects of health and financial challenges [71], and aligns with a belief in an equitable world [74]. Among terminally ill individuals, it was associated with lower levels of despair, hopelessness, suicidal thoughts, and the desire for hastened death [75]. However, it is noteworthy that despite our cohort demonstrating a high degree of religiosity and/or spirituality, with 65% endorsing afterlife belief, these convictions alone appear insufficient to offer a comprehensive comfortable framework for interpreting ELEs. It points to the need of further investigation of other factors that moderate the impact of spirituality in ELE appraisals and impact.

Most ELEs were reported by non-delirious patients (78%). Nevertheless, notably, half of the ELEs reported by delirious patients occurred in those who were lethargic and exhibited reduced alertness and diminished engagement with their surroundings, indicative of hypoactive delirium [30]. Notably, despite their altered state of consciousness, most cases showed no signs of disorganized thinking, disorientation, or memory impairment (over 80%).

Both delirium (88% prevalence) [31] and ELEs (50–90% prevalence) [7] are common in the dying process. Although distinct phenomena [6, 18, 30], they can co-occur, with delirium potentially altering the perception and usual positive impact of ELEs [76]. This has an important clinical implication - accurate differentiation is crucial: misdiagnosing ELEs as delirium deprives patients of meaningful experiences, while untreated delirium could hinder access to the comfort usually associated with ELEs. Given the limited understanding of end-of-life mental states [11, 58, 67], further research is needed to explore their interplay and its implications for consciousness.

Although caution is needed when making direct comparisons of ELE prevalence due to methodological variations across studies, the consistent features of ELEs across cultures — occurrence during both wakefulness and sleep, heightened sense of reality, vivid recall, and emotional significance —suggest that these phenomena may be inherent to the dying process. This aligns with research on transcendent experiences, which reveals consistent core features regardless of cultural background or clinical conditions [11, 46, 77, 78].

Nonetheless, frequency, content, and comfort level of ELEs may vary in accordance with cultural and religious/spiritual (R/S) frameworks. For instance, a cross-cultural study with healthcare practitioners in the U.S. and India revealed that ELEs in the U.S. often involved deceased individuals, while in India, they were more frequently associated with religious figures, with 29% of participants expressing negative emotional responses to these visions [35]. Hindu patients in India exhibited a higher prevalence of ELEs (44%) compared to Muslim (21.2%) [34]. In secular Sweden, no religious themes were found in ELEs [19]. In the Indian study, although 84% of participants initially found their dreams distressing, 95% reported feeling comforted after discussions with the research team [56].

Overall, these findings underscore the importance of clarifying how cultural contexts shape the interpretation and emotional impact of ELEs and point to clear clinical implications. Healthcare professionals should recognize ELEs as common and potentially natural aspects of dying and respond with compassionate, holistic, non-pharmacological support. Further cross-cultural research using standardized screening instruments and consistent methodologies—and explicitly examining the roles of religiosity, spirituality, and culturally embedded attitudes toward death—is needed to distinguish universal from culturally specific features of ELEs and to guide proper clinical management especially of distressing experiences, thereby improving end-of-life care.

Strengths and limitations

Key strengths of this study include its multicenter, prospective design, the use of the validated Confusion Assessment Method (CAM) to assess delirium, and a phenomenological approach grounded in patients’ first-hand accounts rather than proxy reports. In addition, the sample was sociodemographically and religiously diverse, including participants of White and mixed-race backgrounds and multiple religious affiliations.

However, some limitations may affect interpretation and generalizability. Firstly, the sample size of patients experiencing ELEs was relatively small. Secondly, while representative of patients in high-quality palliative care settings, the predominantly Christian sample limits the generalizability of these findings to broader populations. Thirdly, the data collection period at each center was insufficient to follow up most patients until death, which may have contributed to the lower prevalence of ELEs observed. Fourthly, the absence of a global standardized ELE measure required use of a non-validated adapted questionnaire; pandemic constraints precluded pilot testing and psychometric evaluation of content. Fifthly, the study did not include multiple CAM assessments throughout the day, which could have provided a more accurate evaluation of the patients’ mental state. Sixthly, clinical variables were not rigorously controlled, although data on diagnoses, staging, and medications used by patients during hospitalization were collected.

Establishing an expert panel to refine the definition of ELEs and developing a validated assessment scale would significantly enhance both research and clinical understanding of these experiences. Future studies should extend data collection periods and assemble multidisciplinary research teams to conduct more robust investigations that accurately assess and control the clinical variables influencing the end-of-life context.

Conclusions

Our findings suggest that consistent patterns in the timing of occurrence, nature and content of ELEs indicate that these experiences may be an inherent part of the dying process. The results also underscore the influence of cultural and psychosocial factors on how ELEs are interpreted and the emotional responses they elicit. While beliefs in life after death may offer some solace, they do not fully explain the emotional depth of ELEs.

Further research should investigate how attitudes toward death, spirituality, and religious beliefs influence ELEs interpretation and emotional valence. Cross-cultural studies employing consistent methodologies, such as continuous follow up until death, daily interviews, and improved screening tools are needed to differentiate universal elements of ELEs from culturally specific variations. Such work will help inform more culturally sensitive and effective end-of-life care across diverse populations and support appropriate clinical management of distressing experiences.

Abbreviations

CAM

Confusion assessment method

ELEs

End-of-life experiences

PCU

Palliative care unit

PPS

Palliative performance scale

R/S

Religious/spiritual

Authors’ contributions

TOS contributed towards conception and design, data collection, analysis and interpretation of results and manuscript preparation. BSRP, BG and AMA contributed to the analysis and interpretation of results and manuscript revision. All authors have read and approved the final version to be published and agree to be accountable for this work.

Funding

Conceptualization, design, data collection of this study was supported by Homero Pinto Vallada Institute, Helene Reeder Foundation. Researcher’s scholarships were supported by Salvador Health Department, Coordination for the Improvement of Higher Education Personnel - Brazil, Finance Code 001 and CNPq (Brazilian National Council for Scientific and Technological Development).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Institutional Review Board of four institutions: the research proponent center - Hospital Universitário da Universidade Federal de Juiz de Fora – (CAAE: 36150620.8.1001.5133), Hospital Universitário da Universidade Federal de Pelotas (CAAE: 36150620.8.2004.5317), Hospital de Câncer de Barretos (CAAE: 36150620.8.2001.5437), and Hospital das Clínicas, Universidade de São Paulo, Faculdade de Medicina, São Paulo (CAAE: 36150620.8.2005.0068). All participants provided informed consent to take part in the study. The study was conducted in compliance with the WMA Declaration of Helsinki and its principles for ethical conduct of medical research with human participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.