Recognising the advanced phase
Advanced dementia is characterised by profound memory loss, minimal or absent meaningful speech, complete dependence for personal care, incontinence, and loss of the ability to walk independently. Swallowing difficulty, recurrent aspiration pneumonia, urinary tract infection, pressure injury and progressive weight loss are the complications that mark the final phase.
Prognostication in dementia is genuinely difficult and less accurate than in cancer, which is one reason people with dementia have historically had poorer access to palliative care. But the clinical markers are recognisable: recurrent infection despite treatment, an eating and swallowing problem that does not resolve, unintentional weight loss, repeated hospital transfers, and new pressure injuries 1. When several of these appear together, the person has entered the last phase of their illness, and the care plan should say so.
Tube feeding in advanced dementia
This is the decision families find hardest. The evidence base needs to be described precisely, because it is consistent but not of the highest grade. There has never been a randomised trial of tube feeding in advanced dementia, and there almost certainly never will be. The 2021 Cochrane review identified 14 non-randomised studies, all at high or very high risk of bias, and found no evidence that enteral tube feeding improves survival, quality of life, pain, nutrition or carer outcomes — with moderate-certainty evidence that it increases the risk of pressure injury 76. A propensity-matched study of nursing home residents found percutaneous endoscopic gastrostomy associated with 2.27 times the odds of developing a new pressure ulcer and reduced healing of existing ones 77. A meta-analysis of 12 studies found higher mortality among tube-fed patients (odds ratio 1.79) with no improvement in survival duration or nutritional status 78.
Set against that absence of demonstrated benefit are burdens that are not in doubt: the tube frequently requires physical or chemical restraint to stop it being pulled out, it does not remove the risk of aspiration, and it removes both the pleasure of taste and the human contact of being hand-fed. Australian, American and European geriatric and nutrition societies all advise against initiating enteral feeding in advanced dementia on this basis.
The alternative is careful hand feeding — sometimes called comfort feeding or assisted oral feeding — offering food and fluid the person enjoys, in the amounts they will take, at their pace, accepting that intake will decline. This is not withholding care. It is a different, better-evidenced form of care that keeps the person in contact with people and with pleasure.
The conversation families need is usually not about the evidence. It is about the meaning of feeding. Providing food is what families do; stopping feels like abandonment. Naming that directly — “I know that offering food is how you show love, and hand feeding is exactly that” — is more useful than another citation.
Antibiotics and infection
More than half of people with advanced dementia receive antibiotics in their final weeks, often repeatedly, and the evidence that this improves either survival or comfort in the terminal phase is weak 89. Antibiotics carry their own burden — cannulation, hospital transfer, diarrhoea, resistance, and the distress of intervention.
This is not an argument for never treating infection. A urinary tract infection causing delirium and distress is worth treating for comfort alone. A recurrent aspiration pneumonia in someone who is dying is a different proposition. The question to ask each time is: what is this treatment for, and will it make this person more comfortable? If the answer is yes, treat. If the answer is that it might buy a few more weeks of the current state, that decision belongs to the person’s advance care directive and their substitute decision-maker.
Deciding in advance, and recording it, prevents these decisions being made at 3am by a locum who has never met the resident.
What comfort-focused care looks like in practice
-
Pain assessed with an observational tool. PAINAD or Abbey, used routinely rather than only when someone looks distressed. People with advanced dementia cannot report pain and are systematically under-treated 4974.
-
Regular rather than as-needed analgesia. Waiting for a person who cannot ask to demonstrate distress before medicating is not good pain management.
-
Mouth care. One of the highest-yield comfort interventions at the end of life, and one of the most neglected. Dry mouth causes more distress than dehydration.
-
Anticipatory prescribing. Subcutaneous medicines charted in advance for pain, breathlessness, agitation, nausea and secretions, so that a symptom at 2am does not require a hospital transfer.
-
Avoiding burdensome transfers. Hospital transfer in the final days causes distress, delirium and disorientation, and rarely changes the outcome. Treating in place, where the staff and surroundings are familiar, is usually better care.
-
Presence. Familiar voices, music the person loved, touch, and someone in the room. This is care, not a substitute for it.
-
Family support. Anticipatory grief in dementia often begins years before death. Bereavement support after death should be offered, not assumed unnecessary because the death was expected.
A note on voluntary assisted dying
Voluntary assisted dying is lawful in all Australian states, with legislation and eligibility criteria that differ between jurisdictions. In every Australian scheme, a person must have decision-making capacity at the time of the request and throughout the process. This means that a person with dementia who has lost capacity cannot access voluntary assisted dying, and an advance directive cannot be used to request it.
Families sometimes raise this, and the answer deserves to be given plainly and compassionately rather than deflected. Questions about the law in a particular state should go to that state’s voluntary assisted dying care navigator service, and questions about a specific person’s situation to their treating doctor.
Sources cited on this page
- 1 Australian Institute of Health and Welfare. Dementia in Australia — Residential aged care. (2021–22 ACFI data: 54% of ~242,000 permanent residents had dementia.) View source
- 3 Australian Bureau of Statistics. Causes of Death, Australia, 2024 — media release: Dementia is Australia’s leading cause of death (>17,500 deaths; 9.4% of all deaths). View source
- 49 Husebo BS, et al. Efficacy of treating pain to reduce behavioural disturbances in residents of nursing homes with dementia: cluster randomised clinical trial. BMJ. 2011;343:d4065.
- 74 Manietta C, et al. Algorithm-based pain management for people with dementia in nursing homes. Cochrane Database Syst Rev. 2022.
- 76 Davies N, et al. Enteral tube feeding for people with severe dementia. Cochrane Database Syst Rev. 2021. (No randomised trials exist; 14 non-randomised studies, all at high or very high risk of bias from confounding.)
- 77 Teno JM, et al. Feeding tubes and the prevention or healing of pressure ulcers. Arch Intern Med. 2012. (Propensity-matched nursing home cohort: PEG associated with 2.27 times the odds of a new pressure ulcer and reduced healing of existing ulcers.)
- 78 Lee YF, et al. The efficacy and safety of tube feeding in advanced dementia patients: a systematic review and meta-analysis. J Am Med Dir Assoc. 2020. (12 studies, 5,666 patients; mortality OR 1.79, 95% CI 1.04–3.07.)
- 89 Marra AR, et al. Antibiotic use during end-of-life care: a systematic literature review and meta-analysis. Infect Control Hosp Epidemiol. 2020. (72 studies; in 48 of them more than half of patients received antibiotics.)
General information only, reflecting the interpretation of Umbrella Aged Care’s GPs of the published evidence as at September 2026. It is not individual medical advice, does not create a doctor–patient relationship, and must not be used to start, stop or change any treatment. Evidence and Australian regulatory and PBS arrangements change — always confirm current advice with the treating GP, pharmacist or specialist.