A dying woman dreams her husband is waiting for her. A man sees a white horse galloping along a beach he's never visited. These are not folklore — they're the kind of accounts a new study out of Italy collected directly from hospice staff, and they sit at an uncomfortable intersection: too consistent to dismiss as noise, too subjective to treat as evidence of anything beyond the brain doing something unusual on its way out.
The 2026 study, published in Death Studies, surveyed 239 hospice volunteers and palliative care professionals in Italy about what patients had described seeing near the end of life — staircases, open doorways, and overwhelmingly, visits from people who had already died (Vice). Most of these encounters were comforting. Some weren't — a handful brought up unresolved relationships or painful memories instead of reunion and peace.
A Methodological Catch That Matters
Here's the detail that should slow you down before you treat this as a finding about patients: the Italian researchers didn't interview the patients. They interviewed the caregivers about what the caregivers remembered patients telling them (Vice). That's secondhand testimony, filtered through memory and caregiver interpretation, not a direct phenomenological account. It's a legitimate way to gather data at scale, especially when patients themselves are often too close to death to be interviewed repeatedly, but it's a different evidentiary category from the work it builds on.
That earlier work is Christopher Kerr's research program out of Buffalo hospice care, and the bibliographic record is worth being precise about. The foundational 2014 study in the Journal of Palliative Medicine enrolled 66 patients and retained 59 in the final analysis, reporting that 88.1% had at least one dream or vision, with experiences involving deceased loved ones becoming more frequent and more comforting as death approached (bibliographic detail via managingexpectations.net). A follow-up qualitative study the next year worked from 63 patients and organized the reports into recurring themes — comforting presence, preparation to leave, waiting loved ones, unfinished business (managingexpectations.net).
Notice what that record itself flags as a limitation: these are selected, communicative hospice patients giving subjective repeated interviews, not a population-wide incidence rate, and the comfort these visions bring plus their proximity to death do not establish anything about supernatural causation or predictive accuracy (managingexpectations.net). That's the kind of self-imposed guardrail I'd like to see more research programs state explicitly rather than leave implicit.
The Phenomenology Is the Point, Not a Placeholder for Neuroscience
It's tempting to reach for a neurological explanation — oxygen deprivation, medication effects, the dying brain doing something structurally interesting — and the Vice piece gestures at this, noting that caregivers weren't always sure how much to attribute to delirium versus something else (Vice). But none of the sources in front of me offer a tested neurological mechanism for end-of-life dreams and visions specifically — that's a meaningfully different research question from, say, terminal lucidity in dementia patients, which I covered a few weeks back and which has its own separate (and equally unresolved) mechanistic puzzle involving transient changes in neural signaling (NZ Herald). These are adjacent mysteries, not the same phenomenon, and it's worth keeping them separate rather than folding "brain does something unexpected near death" into one undifferentiated bucket.
A recent narrative review in Philosophy, Ethics, and Humanities in Medicine offers a different frame worth sitting with — that phenomenology, the philosophical tradition concerned with lived experience rather than measurable physiology, might be a more honest lens for this material than clinical bioethics alone, precisely because it treats what a dying person reports experiencing as data in its own right rather than noise to be explained away (Springer Nature Link). That review focuses on South Asian end-of-life care, but the underlying argument travels: clinical attention can drift toward physiological parameters and away from the person having the experience.
What Would Actually Move This Forward
None of the current sourcing gets us to prevalence rates we can trust across settings, or a mechanism anyone has tested. What would help: direct, repeated patient interviews conducted prospectively (the Kerr model) replicated outside a single Buffalo hospice system, and caregiver-survey work like the Italian study paired with — rather than substituted for — primary patient accounts. Until then, the honest summary is this: people dying in hospice care very often report vivid encounters with the dead, caregivers find these accounts worth taking seriously rather than medicating away, and we don't know why it happens or what, if anything, it tells us about consciousness at the edge of life. That's not a small finding. It's just not the finding some of the headlines want it to be.
