Deathbed dreams and visions: what Christopher Kerr’s research actually shows

The source interview

Watch the complete original interview · The Diary Of A CEO · published October 19, 2026. The full English auto-caption track and description were reviewed; advertising segments were separated from the substantive discussion. Captions can misrender names and individual words, so bibliographic identities were checked against research sources.

At 03:23, Steven Bartlett says he has “I think 15 research papers” in front of him. He does not supply a complete numbered bibliography. Our library therefore identifies its scope explicitly: the publications in Kerr’s official research page, the ELDV section of Hospice Buffalo’s research page and its 2026 toolkit, supplemented by identifiable studies and contextual sources discussed in the interview. It is not represented as a verified reconstruction of every sheet on Bartlett’s desk or as Kerr’s complete career bibliography.

1. What is an end-of-life dream or vision?

In this research, an end-of-life dream or vision—ELDV—is a reported experience during the dying process. A dream occurs during sleep; a vision is reported while awake. Content may include deceased or living relatives, friends, animals, travel, familiar places, religious figures or unresolved experiences. The researchers are interested in frequency, content, subjective realism, comfort and meaning.

These are experiences before death. They should not be merged with near-death experiences recalled after a medical crisis and resuscitation, terminal lucidity, or dreams experienced by otherwise healthy bereaved people. Those are related but distinct questions with different populations and methods. Dreaming of a deceased relative is not, by itself, evidence that someone is about to die.

The hospice team’s own 2026 toolkit explicitly says its work does not establish an afterlife and excludes paranormal experiences and sudden deaths from the toolkit’s substantive scope. It also asks caregivers not to impose their own beliefs on a patient’s interpretation.

2. “Nearly 90%”: a real finding, but not a universal rate

The pilot research is a repeated-interview study of selected hospice patients—not a representative census of everyone who dies. Its published final sample is 59 patients. The official programme reports 88.1% experiencing at least one ELDV. This supports “common in this studied hospice group,” not “90% of all human beings see deceased relatives.” The experience category is broader than seeing the dead.

At 02:42, Kerr estimates that he has interviewed about 1,500 people by this point. That cumulative interview claim must not be substituted for the enrollment of the 2014 study or treated as one published 1,500-person trial. Likewise, repeated interviews, individual dreams and unique patients are different counting units.

The reported increase in comforting experiences featuring the deceased near death is an association within the research. It is not a validated countdown or clinical prediction tool. Patients who could consent, communicate and complete interviews are not interchangeable with people who were too ill, delirious, unconscious or cognitively impaired to participate.

3. “It felt real” is important—but answers a different question

The official research page reports that 99% described their experiences as seeming or feeling real. The interview sometimes uses absolute language such as “always” when describing realism. The defensible finding is that a very high proportion reported strong subjective realism—not that every account was objectively verified.

Feeling an experience intensely can make it emotionally consequential. It does not distinguish, on its own, among memory, dreaming, perception, neurological changes or a spiritual interpretation. A questionnaire about realism is not a measurement of an independently present deceased person.

Similarly, the study finding that experiences involving deceased loved ones were especially comforting does not show that all dreams were pleasant. The official website reports both comforting and distressing experiences; the interview itself discusses painful dreams and patients who struggled. Differences among headline distress percentages in the interview, website and toolkit should not be averaged into a single “true rate” without matching their denominators and definitions.

4. Delirium: neither dismiss everything nor overlook illness

At 15:58, Kerr explains that his studies screened participants for confusion. This is a real methodological strength: the researchers were not simply collecting every hallucination in a severely confused inpatient population and calling it an ELDV.

But screening a research sample does not provide a home diagnostic rule. A comforting, coherent account does not automatically exclude delirium, medication effects or another medical issue. Delirium can fluctuate and can be hypoactive—quiet, withdrawn or unusually sleepy—not just agitated and frightened. Clinical assessment matters, especially when a person’s attention or alertness changes.

A 2026 international clinician survey adds a useful reality check: 247 responses were analyzed, 18.6% reported never encountering patient reports of ELDVs, and only 9.3% reported relevant training. Clinicians identified missing protocols and standardized criteria as important challenges. This is a self-selected professional survey—not a patient-prevalence estimate or evidence that clinicians can always distinguish the experiences effortlessly.

The sensible middle position is to listen without ridicule, ask how the experience feels, and involve the care team when there is new confusion or distress. The patient does not have to defend a spiritual belief to receive compassionate symptom care.

5. Growth and bereavement: meaningful associations, not proven cures

The posttraumatic-growth studies explore positive psychological changes in people facing serious adversity. They provide a more nuanced picture than the assumption that physical deterioration means there can be no emotional or spiritual development. They do not show that visions cause growth, reverse disease, or guarantee resolution of lifelong trauma.

A later 2025 comparison of self-transcendence is especially important: among 72 hospice patients—35 reporting ELDVs and 37 not reporting them—there was no statistically significant difference on the self-transcendence measure (p = 0.310). This does not prove equivalence, but it means that this study did not demonstrate greater self-transcendence in the ELDV group. Positive narratives should not erase this null finding.

The caregiver work similarly asks whether a dying person’s experiences, and the family’s interpretation of them, relate to bereavement. The library distinguishes the earlier caregiver survey, the later attitudes/focus-group study and the separate study of growth during bereavement. A figure such as “750 families” is not evidence that 750 people all had their grief successfully treated.

These are largely observational reports, often retrospective. Selection, recall, pre-existing beliefs, family relationships and the quality of care can influence both the interpretation of a dream and the reported grief outcome. Positive associations are valuable clinical signals, but not randomized evidence that inducing visions improves bereavement.

No family has failed because a death was difficult or no comforting vision was reported. Grief has no single correct trajectory. The filmed examples in the interview are illustrative personal stories, not a complete account of all patients’ outcomes.

6. Children: moving accounts, limited generalizability

The interview’s closing section describes children’s experiences. The directly relevant published pediatric paper is explicitly a case study. A detailed account can reveal clinically important possibilities, but it cannot establish how often children have ELDVs or prove that children universally accept death without fear.

Children differ in developmental understanding, illness, communication, family context and distress. Statements such as “children die differently” are best read as the clinician’s observations, not a rule that eliminates the need for skilled pediatric palliative care. Families should not be encouraged to hide information, discontinue symptom relief or expect a particular spiritual experience.

7. Does this prove an afterlife? Kerr’s own qualification matters

At approximately 08:04, Kerr says he does not know what the experience means and would not be so bold as to say it speaks to an afterlife. Later, around 13:40 and 48:11, he expresses personal spiritual beliefs and discusses continuing or collective consciousness.

Those statements belong in separate categories: documented patient reports, scientific interpretation and personal belief. The interview contains all three. A hypothesis about consciousness does not become a result of a hospice interview study merely because a researcher finds it persuasive. Equally, uncertainty about mechanism does not make the patient’s experience worthless or settle religious belief against the patient.

The practical conclusion is narrower and stronger: these experiences can be important to the people reporting them and deserve sensitive clinical attention.

8. What patients prioritize—and what rankings cannot prove

The identifiable patient-goals study compared 80 hospice patients with 176 medical students using an 11-item survey. Students ranked seven goals differently from patients, including burden on family, time with loved ones, spiritual peace, home death, meaningfulness, longevity and resolving conflicts. That supports asking patients directly rather than assuming professionals already understand their wishes.

A ranking is relative to the choices offered. A low position for resolving conflicts is not evidence that nobody has regrets, and concern about being a burden is not a judgment that a person actually is a burden. The published abstract verifies the sample and differences; this review does not claim to have independently audited every rank in the interview’s displayed chart. The full publisher record is linked in the library.

Other qualitative work records personal growth, relationships, spirituality, vocational fulfillment and regret. These studies explore experiences; they do not supply a universal prescription for what makes a life worthwhile. The popular regrets account attributed to Bronnie Ware is also not a representative population survey.

The interview’s wording that someone “can’t die” without sleep should not be interpreted as a literal biological rule or a reason to force sedation. Reduced intake can accompany dying, but swallowing safety, hydration, mouth care and medication decisions require individual assessment. Home-care preferences vary, and dying at home is not always feasible or safest. Approximate US home-death and hospice-stay figures in the conversation are not presented here as verified current Canadian statistics.

9. Independent research: why the denominators matter

A systematic integrative review by Hession and colleagues—published online in 2022 and in a 2023 journal issue—identified 22 papers representing 18 studies. Its pooled estimates differed markedly by who was asked: 77% of patients (95% confidence interval 69–84%; patient denominator 119) versus 32% of bereaved family respondents (21–44%; denominator 2,444). Eighty percent of healthcare professionals in a separate pooled group reported witnessing or hearing about an ELDV in the previous five years. That is not an 80% patient prevalence.

The review supports taking ELDVs seriously, while showing how a headline percentage depends on the question, definition, setting, time window and respondent. It is not an independent experiment showing that spirits visit patients. Nor are 22 papers necessarily 22 independent groups of patients.

A large Japanese bereaved-family survey provides another perspective. Of 2,221 analyzed responses, 463 reported deathbed visions—21%, with a 95% confidence interval of 19–23%. Good-death scores did not significantly differ between those reporting visions and those not reporting them. This is not a direct contradiction of the hospice pilot: retrospective family reports and narrower deathbed-vision definitions differ from repeated interviews with patients about dreams and visions. It does challenge any universal “almost everyone, always beneficial” framing.

One particularly important result is easy to miss in the positive narrative: in the 228-caregiver study, validated Core Bereavement Items scores did not differ significantly between caregivers whose loved ones reported ELDVs and the other caregivers. Comfort and meaning were described, but that is not equivalent to a demonstrated reduction in overall grief severity.

A Brazilian case report also describes meaningful ELDV-like experiences co-occurring with delirium. One case cannot establish prevalence or a definitive classification, but it reinforces the clinical warning: pleasant content is not a reliable test for absence of delirium.

10. Dying can be less frightening than expected—not universally painless

The interview and its promotional description emphasize peaceful dying. That can offer real reassurance, but it should not become a promise. The US National Cancer Institute’s evidence summary reports pain in 30–75% of patients in the last days of life in the cited literature, while also explaining that most patients can obtain relief and average pain intensity often decreases near death. Its context is principally advanced cancer; those estimates are not universal rates for every disease or care setting.

The same clinical resource recognizes comforting encounters with deceased loved ones as a poorly understood phenomenon that appears distinct from delirium. It also describes terminal delirium as common and often hypoactive. Compassionate listening and medical symptom assessment are complementary, not competing, responses.

Discussion of accepting death must not be translated into refusing beneficial treatment, withholding analgesia or stopping prescribed medicines without clinical advice. Palliative care supports comfort and goals of care; it is not a judgment that a person has stopped mattering. Ask the treating team what can be relieved, what changes need attention and what the person wants.

11. Other claims in the interview: the actual papers and limits

Genetics and lifespan, around 43:00

The identifiable Science paper by Shenhar and colleagues estimates approximately 55% heritability of intrinsic lifespan variation after modelling confounding from extrinsic mortality and age cutoffs. It is a modelling and twin-data reanalysis, not a longevity-treatment trial. “55% heritability” does not mean 55% of your personal lifespan is fixed by genes, and does not show that prevention or lifestyle is pointless. Its final PDF records a February 20, 2026 correction to a funding grant number, not a change to the results.

Healthspan versus lifespan, around 42:00

The interview’s roughly 11-year US gap is not fully identified by a spoken citation. A relevant 2024 JAMA Network Open analysis of 183 WHO member states reports a US gap of 10.9 years in 2000 and 12.4 years in 2019. It is included as a contextual companion, not a confirmed exact interview citation. The gap is population life expectancy minus health-adjusted life expectancy, which uses disability weights. It is not a forecast that every individual spends their final 12.4 consecutive years seriously ill. The data end in 2019; the publication date does not make them 2026 observations.

Psychedelics, around 50:53

Two relevant randomized crossover trials—Griffiths and colleagues, 51 participants, and Ross and colleagues, 29 participants—reported improvements in depression or anxiety among selected patients with life-threatening cancer receiving psilocybin with structured professional support. These are contextual papers, not titles expressly named in the interview. Small samples, difficult blinding, expectancy and crossover limit interpretation, especially at longer follow-up. They do not establish that everyone loses fear, that a peaceful death is guaranteed, or that the intervention proves an afterlife. Transient distress and physiological adverse effects occurred; the findings do not support unsupervised use.

“53%” and dreams of deceased relatives, around 57:50

The relevant Pew Research Center survey found 53% of US adults said a deceased family member had visited them in a dream or in any other form. That is broader than “53% dream of dead relatives.” The separate figure for dreams in the past 12 months was 34%. These are self-reports in a general-adult survey, not dying-patient data or independent verification of contact with the dead.

Statements we could not assign to a specific published result

The interview mentions recently completed work about suffering and expectations without enough identifying information to map it confidently to a publication. Its cumulative interview totals, selected patient histories, observations about regrets and comments about longevity culture should not all be presented as individually verified study findings. No paper has been invented to fill a gap.

12. What families can take from this

• Listen first. “Would you like to tell me about it?” and “Was it comforting or upsetting?” make space for the person without imposing an explanation.
• Support meaning without asserting proof. It is possible to respect a spiritual interpretation without claiming science has established it.
• Tell the care team about new confusion, fluctuating alertness, agitation or distress. These may need clinical assessment even when an experience sounds coherent or comforting.
• Ask about the person’s own priorities. Relationships and burden concerns often matter; do not assume every patient ranks them the same way.
• Do not expect a perfect ending. Absence of visions, unresolved feelings or a difficult death is not a failure by the patient or family.

This article is research reporting, not a diagnosis or treatment plan. For symptoms or medication decisions, use the patient’s clinical team. For immediate danger or a medical emergency, contact local emergency services.

13. Read the papers and the results

Open the complete research library: individual results, limitations and original-paper access

Download the research guide (PDF) · Download research pack (ZIP) · CSV · Excel · JSON manifest

Methods and disclosure

The complete English auto-caption track, description and chapters of the 83-minute interview were captured and reviewed. Advertisements were separated from substantive claims. Bibliographic records were cross-checked against PubMed/Europe PMC, Crossref, publisher records and institutional sources; abstracts and accessible full texts were distinguished. Sample sizes, respondent types, overlapping research streams and null findings were checked. Available PubMed editorial-notice metadata and the Science PDF correction were reviewed; this is not a guarantee that every later notice has been captured. Only article-specific CC BY permissions justified local unchanged PDF copies. No full interview transcript or paywalled book/article text is republished.

Research and publication: October 19, 2026. No claim is made to have reproduced the studies or obtained unpublished patient data. No patient records, full interview transcript or unlicensed journal PDFs are republished.

Selected sources are cited directly in the text; the library carries the complete bounded catalogue, DOI links, access status and rights notes.