An umbrella review of 77 papers and 1.48 million participants puts the annual cost at £2.8 billion and finds two widely assumed remedies do not work.

About one in six hospital beds in the UK is occupied by a person with dementia, at a cost to taxpayers of £2.8 billion a year, and on current trends that will be one in four by 2040.

A review in Alzheimer’s & Dementia, published alongside a report titled Breaking the cycle of hospital admissions in people with dementia, examines what drives those admissions and what reduces them. It is described as the most comprehensive review of dementia hospitalisations conducted in the UK, drawing on 77 unique papers and 1.48 million participants.

The lead author is Professor Andrew Sommerlad of UCL Psychiatry.

The figures

People with dementia have a 42 per cent higher risk of acute care admission and up to a 35 per cent higher risk of readmission than their peers. Once admitted, they stay seven times longer.

A substantial part of that length of stay is not medical. People with dementia are five times more likely to remain in hospital when they are medically fit to leave, and delayed discharges alone cost £328 million a year. Nearly half — 45.8 per cent — develop delirium while in hospital.

The review attributes the delays to delayed discharge processes, poor communication between staff, and hospital environments unsuited to people with cognitive impairment. Eighty-seven per cent of hospital workers say dementia care is challenging because of poor resourcing.

Families absorb the difference. The report calculates that they provide more than 100 unpaid hours a week, which it equates to two and a half unpaid jobs, and 90 per cent of families say hospitals are frightening and cause confusion.

What the families describe

The review includes accounts from family members, and they are specific about what the resourcing figure means in practice.

“He was just in bed doing nothing. I ended up having to feed him and if we didn’t go, he wouldn’t get fed,” one family member said.

Another described a ward that advertised itself as dementia-friendly: “shocked that the ward said they were dementia-friendly, because it wasn’t in the slightest – the nurses just left his food and didn’t assist him with eating it”. A third described staff “only taking care of medical needs, nothing else, with no regard for creating a safe space and familiarity for her”.

These are not incidental to the numbers. A person with dementia who is not helped to eat deteriorates, and deterioration extends a stay — which is the mechanism by which an understaffed ward converts into the seven-fold difference in length of stay.

The two things that do not work

The most useful finding in the review is a negative one, and it is the part least likely to be quoted.

Case management and exercise — both widely assumed to be effective, and both funded on that assumption — do not reduce dementia-related hospitalisation.

An umbrella review is well suited to establishing this. By pooling existing systematic reviews rather than individual trials, it can show that an intervention with an encouraging early literature has not accumulated evidence of effect, which is harder to see from any single study. Finding that two popular interventions do not deliver is a result that redirects spending, and the report’s value may rest more on this than on its headline cost figure.

What the review recommends

Three measures are put forward as low-cost.

First, integrated multidisciplinary teams as the standard of care rather than an enhancement. Second, clinical pharmacists within those teams, which the report estimates saves £38 for every £1 spent — plausible in a population where polypharmacy is common and medication review prevents the falls and confusion that cause admissions. Third, advance care planning, which the report says can reduce admissions by up to 55 per cent.

Both of those last figures need reading carefully. “Up to 55 per cent” is an upper bound, which means at least one study found that and others found less; a return of £38 per £1 is a modelled estimate whose assumptions determine the answer. Neither is wrong, and neither is the number to plan with.

Advance care planning also works partly by changing what counts as the right outcome. A documented decision that a person does not want to be admitted for an acute event reduces admissions by honouring a preference, which is a different kind of success from preventing the event — and a good one, but not interchangeable with it in a cost calculation.

Sommerlad’s argument

“Despite public and political perception, these dementia-related failings in the UK health and care system are not intractable,” Sommerlad said. “The research clearly shows how important community-centric approaches are to reducing hospitalisations and minimising long stays, yet the NHS is not implementing them.”

“With one in three people over 90 expected to have dementia, it has never been more important to redirect resources to evidence-based care and improve communication to ensure families, hospital staff and carers are working as one joined-up team,” he said.

Who paid for it

The report was supported by Alzheimer’s Society, Dementia UK and the Geller Commission, and the Commission’s founder is quoted in the release.

“For too long, the crisis facing people with dementia in our hospitals has been hiding in plain sight – acknowledged but never resolved,” said Laurence Geller, the Commission’s chairman. “I founded the Geller Commission and in turn, was pleased to support this important research, because I believe that what gets measured gets fixed. Until now, we have lacked a complete picture of the true human and financial cost of these failures. This report provides that evidence, and with it, the basis for change. Ignorance is no longer an excuse for inaction.”

Advocacy funding of research into the thing being advocated for is normal and worth naming. It does not make the figures wrong — they come from 77 published papers, and the peer-reviewed component is in a mainstream journal — but a report commissioned to make a case will be framed to make it, and the negative finding about case management and exercise is a reasonable sign that the framing did not override the evidence.

The limits of the method itself are the ordinary ones. An umbrella review inherits the quality of the reviews it pools, and the comparisons between people with and without dementia are observational: longer stays and higher readmission risk reflect comorbidity, frailty and the availability of somewhere to be discharged to, as well as anything hospitals do or fail to do. The release names only Sommerlad and does not state any conflicts of interest.

---

Notes

1. Sommerlad, A., et al. “The effect of psychosocial and healthcare interventions on reducing hospitalization in people with dementia: an umbrella review.” Alzheimer’s & Dementia, 2026. DOI: 10.1002/alz.71323.
2. University College London. “One in six hospital beds filled by dementia patients.” EurekAlert!, 4 August 2026. https://www.eurekalert.org/news-releases/1138803.

Photo by cottonbro studio on Pexels.com