Recognising That Somebody Is Dying

Recognising that somebody is dying: the question that changes everything

The Australian Institute of Health and Welfare examined where people with life-limiting conditions actually die. Among those who received specialist palliative care, 67 per cent died as admitted hospital patients. Among those who did not, 43 per cent died in residential aged care and 20 per cent in hospital.

Then there is the detail buried in the same analysis. Of 35,158 specialist palliative care patients who died in hospital, among the four-fifths who had not been in residential aged care during their last year, 63 per cent were admitted in their final week of life.

Read that again. Nearly two thirds of people who died in hospital, under the care of a specialist palliative care service, arrived there in the last seven days.

Some of those admissions were right — symptoms that genuinely needed inpatient management, a family who could not continue, a crisis nobody could have planned around. But a substantial proportion are the signature of something else: a death that was recognised too late to arrange anything else. By the time somebody realises the person is dying, the only available option is an ambulance.

Recognising dying earlier does not change whether somebody dies. It changes almost everything else — where they die, who is with them, whether their pain is managed before it becomes a crisis, whether anybody asked them what they wanted, and whether the people around them got to say goodbye.

For people with disability this is harder than it is for anybody else, and the reasons are specific. This guide is about those reasons, and about the tools that help.

We support NDIS participants across New South Wales, including people approaching the end of their lives. We are not a palliative care service and nothing here is medical advice. Prognosis is a clinical judgement, and the tools described here are structured prompts to trigger a conversation with a clinician, not instruments for deciding anything by yourself.

Deteriorating is not the same as dying

A useful distinction, because the two get conflated and they call for different responses.

Deteriorating means function is declining and the trajectory is downward. It may run for years. It is a signal to plan.

Dying means the person is in the final phase — days to a short number of weeks. It is a signal to act.

The failures happen at both ends. A person who is deteriorating but not dying gets treated as though nothing has changed, so nothing is planned. A person who is dying gets treated as though they are merely deteriorating, so they are sent to hospital for investigations that will not change anything.

And in disability there is a third failure, which is the most common of all: a person who has been deteriorating for three years is treated as though that is simply their disability, because they have always needed a lot of support and the gradual increase never registered as a trajectory.

Why this is harder in disability

Five reasons, and they compound.

One: there is no clear baseline to decline from. Recognising deterioration normally means comparing somebody to how they were. For a person with lifelong severe disability, “how they were” already involved full assistance with everything. A person who has always been fully dependent cannot become dependent, so the clearest signal the rest of medicine relies on is unavailable.

Two: the decline is slow and nobody reads it longitudinally. Progress notes are written shift by shift and read shift by shift. A person who has lost six kilograms over two years, who now sleeps four hours a day more than they did, who has had three chest infections this winter instead of one, is on an obvious trajectory — and that trajectory exists nowhere except across two hundred separate shift entries that nobody has ever looked at together.

Three: repeated crises that resolve teach everybody the wrong lesson. Somebody with severe cerebral palsy and recurrent aspiration pneumonia may have eight admissions over six years and come home from every one. The ninth is different, and nothing about the previous eight signalled which one it would be. The pattern of recovery is itself what prevents recognition.

Four: diagnostic overshadowing applies to dying as much as to anything else. New drowsiness is the medication. Reduced intake is the dysphagia. Weight loss is the feeding. Withdrawal is behavioural. Each has a plausible disability-related explanation, and each is also what dying looks like.

Five: the trajectory is the hardest kind to see. Which brings us to the four shapes.

The four trajectories

Most people die along one of four patterns, and knowing which one applies tells you how much warning there will be.

Sudden death. No decline, no warning. Nothing to recognise.

A clear terminal phase after a period of relative function. This is the classic cancer pattern: reasonable function maintained, then a recognisable decline over weeks to months. This is the trajectory palliative care services were designed around, and it is the easiest to recognise — which is why 71.4 per cent of people accessing voluntary assisted dying in New South Wales in 2024–25 had cancer, and why cancer patients are over-represented in palliative care caseloads generally.

Episodic decline with acute crises. Organ failure, advanced respiratory disease, motor neurone disease, advanced multiple sclerosis. The person has a crisis, is treated, recovers to slightly below where they were, and continues. Then one crisis does not resolve. The difficulty is that each crisis looks survivable because the previous ones were, and the one that is not looks identical until it is.

Prolonged gradual decline — the dwindling trajectory. Frailty, advanced dementia, and the picture common in people with severe lifelong disability and multiple comorbidities. Function declines very slowly over years, with no identifiable inflection point. There is no moment at which the person visibly starts dying.

That fourth trajectory is where most people with significant disability sit, and it is the one the system handles worst, because every tool for recognising dying was built around the second one.

The surprise question

The single most useful instrument in this field takes four seconds and requires no clinical training.

“Would I be surprised if this person died in the next twelve months?”

And then, closer in:

“Would I be surprised if this person died in the next days or weeks?”

If the answer is no — if their death in that period would not surprise you — then the person may be approaching the end of their life, and that is the trigger to do something about it.

What makes this work is that it bypasses prognostication entirely. It does not ask you to predict anything. It asks you to notice what you already believe, which people are frequently much better at than they are at estimating survival.

Use it deliberately. At every plan review. At every annual health assessment. When a person comes home from hospital. When a worker says something has changed. The question costs nothing and the cost of not asking it is the 63 per cent figure above.

SPICT, and the version written for everybody else

The Supportive and Palliative Care Indicators Tool (SPICT) is the structured version. It works on two axes: general indicators of deteriorating health, and indicators specific to particular conditions. A person is considered SPICT positive with at least two general indicators plus one serious health condition, which suggests they may be at risk of dying within twelve months and that care should be reviewed and planned.

And there is a version that matters enormously for this sector: SPICT-4ALL, written in plain language for people who are not clinicians. Family members, support workers, house managers. It asks the same questions without requiring medical vocabulary.

That is a genuinely useful thing, and almost nobody in disability services knows it exists. A support worker who has known somebody for eight years holds information no clinician has, and SPICT-4ALL is a structure for turning that knowledge into something a GP can act on. Both versions are available through the SPICT website, and the Australian ELDAC project publishes them alongside toolkits for primary care, home care, residential aged care and dementia.

The general indicators of deteriorating health, broadly, are the ones to learn:

• Declining function — doing less, needing more help, spending more of the day in bed or in a chair
• Increasing dependence on others for everyday activities
• Repeated unplanned hospital admissions, particularly clustering
• Significant unintentional weight loss, or a persistently low body weight
• Persistent symptoms despite treatment of the underlying conditions
• Deteriorating response to treatment, or treatments that are no longer achieving what they did
• The person or their family choosing to reduce, decline or stop active treatment
• The person saying they are approaching the end of their life, or wishing to focus on comfort

That last one deserves emphasis. People sometimes say it, in their own way, and it gets reassured away. A person who says “I’m not going to get better, am I” is opening a conversation, and the most common response is to close it.

Continued deterioration despite intervention, and multiple recent hospital admissions, are both specifically flagged as reasons to consider that somebody may be at the end of their life.

The last days of life

Different question, shorter timeframe, and worth knowing because the signs are consistent and because recognising them changes what should happen in the next few hours rather than the next few months.

When somebody is entering the final days, the picture typically includes:

• Profound weakness and fatigue, becoming bed-bound
• Reduced consciousness — sleeping most of the time, increasingly difficult to rouse
• Minimal or no interest in food and fluid, and difficulty swallowing
• Difficulty taking oral medication, which is why anticipatory medications by another route matter
• Changed breathing — irregular, with pauses, sometimes deep then shallow in cycles, sometimes noisy from secretions the person can no longer clear
• Cool, mottled or discoloured hands, feet and knees as circulation withdraws to the core
• Reduced urine output, and darker urine
• Withdrawal — less interest in surroundings, in talking, in people
• Sometimes a brief period of unexpected alertness shortly before death, which families find bewildering and should be warned about

None of these is definitive alone. Together, in somebody with a life-limiting condition, they are the pattern.

And the specific disability difficulty: several of these overlap with the person’s baseline. Somebody who has always been largely non-responsive, who has always had minimal oral intake, who has always had noisy respiration, does not show a change in the way an otherwise well person does. Which is why a written baseline — what this person looks like, sounds like and does when they are as well as they get — is one of the most valuable documents in their file, and almost nobody has one.

Prognostic uncertainty, honestly

Clinicians are not good at predicting how long somebody has, and nobody should pretend otherwise. Prognoses are frequently wrong in both directions, and families who are given a number and then watch it pass lose trust in everything else they have been told.

The resolution is not better prediction. It is planning that does not depend on prediction, which palliative care calls parallel planning: preparing for deterioration and death while simultaneously continuing to live and to treat what is treatable.

In practice that means:

• Advance care planning done early, while the person can participate, rather than when a prognosis arrives
• Anticipatory medications in the house before they are needed
• A plan for what happens at 3am, agreed in advance
• Not waiting for certainty, because certainty arrives after it is useful

A plan that works whether somebody dies in six weeks or three years costs nothing extra and removes the need to guess. A plan that waits for a prognosis produces the ambulance.

What this means for a disability provider

None of what follows requires clinical authority, and all of it is the part only you can do.

Write the baseline down. What this person is like at their best. How much they eat, how they breathe, how alert they are, what they do, what they weigh. Dated. Because every judgement about change depends on it, and nobody else holds it.

Read the notes longitudinally. Somebody — a house manager, a coordinator — needs to read six months of progress notes in one sitting, every six months, and write a one-page summary of what has changed. This single habit catches more than any assessment tool, because the information is already there and nobody has ever looked at it together.

Track the admissions. Dates, reasons, length of stay, and the function the person came home with compared to the function they went in with. A pattern of admissions that are closer together and recoveries that are less complete is the clearest signal there is.

Weigh the person and record it. Monthly. It is free and it is the most objective marker available.

Ask the surprise question at every review, out loud, and record the answer.

Say it when you notice it. A worker or coordinator who tells a GP “I have known this person for nine years and something is different in a way I have not seen before” is providing clinical information. It should be said directly, and it is frequently not said at all because people feel it is not their place.

Raise palliative care early rather than late. The referral conversation is covered in its own guide here. The relevant point for this guide is that early involvement is associated with better symptom control and quality of life, and that waiting until somebody is clearly dying wastes the part where it helps most.

Do not let a slow trajectory become invisible. The deterioration that happens over three years is the one most likely to be missed, and people with significant disability are overwhelmingly on that trajectory.

The conversation this is all for

Recognising that somebody is approaching the end of their life is not an end in itself. It exists to make four things possible, and all four are foreclosed by recognising it late.

Asking the person what they want, while they can still say.

Advance care planning, including whether they want to go to hospital at all.

Symptom management set up before a crisis, which means anticipatory medications in the house and a plan for the night.

Dying somewhere the person would have chosen, with people who know them.

That last one is what the 63 per cent figure is really about. Those were not failures of medicine. They were, in large part, failures of planning that happened because nobody had said out loud what everybody could see.