20.4 — Dementia

Around 45 percent of dementia cases worldwide are attributable to modifiable risk factors.

That estimate, from the Lancet Commission, is the most important thing in this chapter — because dementia is the diagnosis people fear most in later life, and the widespread assumption is that nothing can be done about it.

Something can. Not everything, and not with certainty, and substantially more than most people believe.

What dementia is

A progressive decline in cognitive function, sufficient to interfere with daily life.

And it is a syndrome, not a disease — several different diseases produce it.

It is not normal ageing. Occasional word-finding difficulty and slower processing are normal. Getting lost in familiar places, repeating questions within minutes, and being unable to manage familiar tasks are not.

Mild cognitive impairment — measurable decline without significant functional impairment. Around 10 to 15 percent progress to dementia per year, and a meaningful proportion revert to normal — which is worth knowing, because the diagnosis is frequently heard as an early death sentence.

The types

Alzheimer's disease — 60 to 70 percent.

Pathology: amyloid-beta plaques between neurons and tau tangles inside them (Chapter 1.3), with progressive neuronal loss beginning in the medial temporal lobes.

Which is why episodic memory fails first (Chapter 11.6) — the hippocampus is where the pathology starts.

Course: gradual onset over years. Recent memory first — repeating questions, misplacing things, forgetting recent conversations while remote memories remain intact. Then word-finding, navigation, planning and judgement. Then personality change and, eventually, physical decline.

Vascular dementia — around 20 percent.

From strokes, or from small vessel disease.

Characteristically stepwise rather than gradual, though small vessel disease produces a gradual course. Executive function and processing speed are affected earlier than memory, and gait disturbance is common.

And it shares its risk factors with stroke (Chapter 18.6), which is why controlling blood pressure is a dementia intervention.

Mixed Alzheimer's and vascular is very common, particularly over 80 — and pure single pathology is the exception at that age.

Dementia with Lewy bodies — around 10 to 15 percent.

And it is frequently misdiagnosed, with consequences.

Features: marked fluctuation in attention and alertness — described by families as good days and bad days, or being "away" for hours; detailed, recurrent visual hallucinations, typically of people or animals, and frequently not distressing; parkinsonism; and REM sleep behaviour disorder (Chapter 20.3).

And the critical point: severe sensitivity to antipsychotics.

Giving a standard antipsychotic for the hallucinations can cause severe rigidity, deterioration and death.

Which makes recognising it genuinely life-saving, and it is why the hallucinations are managed with cholinesterase inhibitors — which work unusually well in this type — rather than with antipsychotics.

Frontotemporal dementia — around 5 to 10 percent, and higher in those under 65.

And it does not present with memory problems.

The behavioural variant presents with personality and behaviour change — disinhibition, apathy, loss of empathy, compulsive behaviours, and changes in eating, often toward sweet foods.

Which is why it is frequently misdiagnosed as depression, midlife crisis or a psychiatric disorder for years, and why relationships and careers are frequently destroyed before a diagnosis is made.

The language variants present with progressive difficulty with speech or with word meaning.

Younger onset, typically 45 to 65, and a stronger genetic component — around 30 to 40 percent have a family history.

Other causes: Parkinson's disease dementia; Huntington's disease (Chapter 2.8); alcohol-related brain damage; HIV; and prion disease, which is rapid.

The reversible causes

And these are why anyone with cognitive decline gets a set of blood tests, because missing them is missing a treatable condition.

Depression — "pseudodementia". Characteristically the person complains of memory problems, gives "I don't know" answers, and has a relatively abrupt onset — whereas in dementia the family complains and the person minimises. Treatable.

Hypothyroidism (Chapter 12.3).

Vitamin B12 deficiency (Chapter 7.1).

Normal pressure hydrocephalus — the treatable triad (Chapter 11.10).

Subdural haematoma — particularly in older people and after minor injury (Chapter 5.2).

Medication — and anticholinergic burden is the largest single culprit (Chapter 11.9). Also benzodiazepines, opioids and sedating antihistamines.

Alcohol.

Infection, particularly in the acutely confused.

And delirium, which is not dementia — see below.

Delirium

And separating the two matters enormously.

	Delirium	Dementia
Onset	Hours to days	Months to years
Course	Fluctuates markedly	Slowly progressive
Attention	Impaired — the core feature	Preserved until late
Consciousness	Altered	Clear until late
Reversible	Usually	No

Delirium is an acute confusional state caused by something else — infection, drugs, dehydration, pain, constipation, urinary retention, electrolyte abnormalities, hypoxia, or surgery.

It affects 20 to 30 percent of older hospital inpatients, and it is associated with longer stays, worse outcomes and increased subsequent dementia risk.

And the hypoactive form — quiet, withdrawn, drowsy — is commoner than the agitated form and is missed far more often, because a quiet patient does not attract attention.

Prevention works: orientation, ensuring glasses and hearing aids are used, sleep hygiene, early mobilisation, hydration, avoiding unnecessary catheters, and reviewing medication. Multicomponent programmes reduce delirium by around a third.

Diagnosis

History from the person and, crucially, from someone who knows them — because insight is frequently lost.

Cognitive testing — brief bedside tools, and more detailed neuropsychological assessment where needed.

Blood tests for the reversible causes.

Imaging — to exclude structural causes and to look for the pattern of atrophy or vascular damage.

And newer biomarkers are changing this. Amyloid and tau can now be measured in cerebrospinal fluid and, increasingly, in blood — and blood-based tau tests are approaching the accuracy needed for clinical use. PET imaging of amyloid and tau exists and is expensive.

Which raises a genuine question: a biomarker can identify Alzheimer's pathology in someone with no symptoms — and until there is effective treatment, whether that helps is not obvious.

Treatment

Cholinesterase inhibitors — donepezil, rivastigmine, galantamine.

They increase acetylcholine, which is depleted because the cholinergic neurons of the basal forebrain degenerate early (Chapter 11.2).

They produce modest symptomatic improvement — roughly equivalent to delaying progression by 6 to 12 months — and do not alter the underlying disease.

They work unusually well in dementia with Lewy bodies, which is another reason for identifying it.

Memantine — an NMDA receptor blocker, for moderate to severe disease.

Anti-amyloid antibodies — and this is where an honest assessment is needed, because the reporting has been polarised.

Lecanemab and donanemab remove amyloid from the brain, which is demonstrable on imaging, and they slow cognitive decline by around 25 to 35 percent over 18 months.

Whether that translates into a difference a patient or family would notice is genuinely debated. The absolute difference on cognitive scales is small.

And they carry real risks: brain swelling and microbleeds — ARIA — occurring in a substantial minority, occasionally fatal, and much commoner in carriers of the APOE4 variant, who are also those at highest risk of the disease.

They require regular infusions and repeated MRI monitoring.

The honest summary: these are the first drugs to alter the disease process rather than only the symptoms, which is a genuine scientific milestone — and their clinical benefit is modest and their risks are real. Regulators have reached different conclusions, which reflects the balance rather than any disagreement about the data.

And they support the amyloid hypothesis without vindicating it entirely — removing amyloid helps somewhat, which suggests amyloid is part of the story rather than the whole of it.

Non-drug approaches, which matter more day to day:

Cognitive stimulation therapy has evidence comparable to the drugs.

Exercise — for physical function, mood and behaviour.

Managing behavioural symptoms without antipsychotics wherever possible.

And this is important: antipsychotics in dementia increase mortality, and they are used far more than they should be, frequently for behaviour that has an addressable cause — pain, constipation, infection, boredom, fear, or an environment the person cannot navigate.

Looking for the cause of distressed behaviour, rather than sedating it, is the standard of care and it is inconsistently delivered.

Prevention

And this is the section that justifies the chapter's opening.

The Lancet Commission's modifiable risk factors, roughly by contribution:

Hearing loss — the largest single modifiable factor in midlife. And hearing aid use is associated with reduced risk, with a randomised trial showing slowed cognitive decline in those at higher risk. Which makes treating hearing loss a dementia intervention.

Less education in early life — cognitive reserve.

Smoking.

Depression.

Social isolation.

Traumatic brain injury.

Physical inactivity.

Air pollution.

Diabetes.

High blood pressure in midlife — and midlife specifically, which is why treating hypertension at 50 matters for the brain at 75.

Obesity in midlife.

Excess alcohol.

Untreated vision loss — added in the most recent update.

And high LDL cholesterol in midlife.

Which produces a practical list that overlaps almost entirely with cardiovascular prevention (Chapter 18.1): do not smoke, control blood pressure, stay active, stay socially engaged, treat hearing and vision loss, manage diabetes, limit alcohol, and protect your head.

Cognitive reserve — education, occupational complexity, bilingualism and mentally demanding activity — appears to allow people to tolerate more pathology before symptoms appear. It does not prevent the pathology; it delays the point at which it shows.

What is not supported: brain training games, which improve performance at those games with little transfer; and most supplements marketed for brain health.

Living with it

Diagnosis matters even without a cure — for planning, for accessing support, for explaining behaviour that families otherwise find bewildering, and for avoiding harmful medication.

Advance care planning while capacity remains — lasting power of attorney, advance decisions, and stating preferences. This is the single most useful practical action after a diagnosis, and it is frequently deferred until it is too late.

Practical adaptations: routine, familiar environment, labels and signs, simplified choices, and addressing sensory impairment.

Driving — must be reported to the licensing authority in most jurisdictions.

And carer support, which is where the burden actually falls. Carers of people with dementia have high rates of depression, and their health outcomes are measurably worse. Respite, support groups and practical help are as much part of treatment as anything given to the patient.

End of life — dementia is a terminal illness, which is frequently not acknowledged. Recognising that allows a shift toward comfort rather than repeated hospital admissions and interventions that do not help, and it is one of the areas where advance planning makes the largest difference to how someone's last year goes.

What the next page fixes

Chapter 20.5 covers multiple sclerosis — a condition whose treatment has changed more dramatically over thirty years than almost anything else in neurology.