Read this topic

Basic Neurosciences · Paper A

Alzheimer's disease: epidemiology and risk

The idea

Age is the risk factor for Alzheimer's disease that dwarfs every other, and the rest of the epidemiology is a list of things that move the date forward or back. Prevalence of dementia roughly doubles every five years after sixty-five, so about one person in fourteen over sixty-five and one in six over eighty is affected, and Alzheimer's disease accounts for about two-thirds of that. Around it sit the modifiers: sex (about two-thirds of cases are women), genes (APOE ε4 and the chromosome 21 story, taught on their own cards), and the fourteen modifiable factors that the 2024 Lancet Commission estimated account for about forty-five per cent of dementia worldwide. The pathology, the imaging, the drugs and the course of the illness live on neighbouring cards; this one owns who gets it and what could have moved the date.

How it works

The cause here is epidemiological rather than molecular: risk is the product of how much pathology has accumulated and how much damage the brain can absorb before function fails. Pathology accrues with time, which is why age dominates. Jorm, Korten and Henderson (1987, Acta Psychiatrica Scandinavica) pooled prevalence surveys and found dementia prevalence doubling roughly every five years after sixty-five, a pattern every later survey has reproduced. The capacity to absorb damage is what Stern (2002, Journal of the International Neuropsychological Society) called cognitive reserve: more education, occupational complexity and mental activity delay the clinical onset for a given load of pathology, and the same reserve explains why, once symptoms appear in a well-educated person, decline can be faster, because the underlying load is already high. Everything else in the risk list acts on one of those two quantities. Genes act on the rate of accumulation: each copy of APOE ε4 raises risk several-fold, and the extra copy of the amyloid precursor gene on chromosome 21 in Down syndrome brings Alzheimer pathology forward to the forties. Vascular factors (hypertension, diabetes, smoking, obesity, high LDL cholesterol), head injury and heavy drinking add damage of their own. Hearing loss, vision loss, social isolation, physical inactivity and depression are argued to work mainly by lowering reserve and engagement, though depression may also be an early symptom rather than a cause. The 2024 Lancet Commission (Livingston and colleagues) counted fourteen such modifiable factors and estimated that, if all were removed, about forty-five per cent of dementia cases worldwide could be prevented or delayed. Two cautions keep that number honest: it is for dementia as a whole, not Alzheimer's disease alone, and it is a population attributable fraction, which assumes the associations are causal.

How to spot it

A daughter asks whether her mother's dementia at eighty-four means she will get it too, and the honest answer starts with age, then APOE ε4, then the difference between a risk allele and a familial mutation.
"He was a professor and it seemed to come from nowhere": a rapid decline after late onset in a highly educated man, which is cognitive reserve running out rather than an unusual illness.
A public health stem gives fourteen factors and asks what proportion of dementia they account for, and the number wanted is about forty-five per cent from the 2024 Lancet Commission.
A referral notes memory decline at forty-five in a man with Down syndrome, and the detail that does not fit a sporadic case is his age, explained by an extra copy of the amyloid precursor gene.
A practice nurse asks whether to start HRT to protect a sixty-eight-year-old's memory, and the Women's Health Initiative Memory Study answers no.
A community survey reports higher Alzheimer's prevalence in a farming district than in the nearby city, and the explanation offered is schooling and access to care rather than the land.
An MCQ compares incidence in Ibadan and Indianapolis and asks what the difference implies about APOE ε4 across populations.

Tell it apart

Age vs every other factor — age multiplies risk across decades; no other single factor in a sporadic case comes close, so a stem asking for the strongest risk factor wants age.
Prevalence vs incidence — prevalence (about one in fourteen over sixty-five) counts everyone living with the illness; incidence counts new cases per year and rises with age just as steeply.
Female excess vs longevity — women make up about two-thirds of cases; part of that is living longer, and the remainder is not fully explained, so the honest stem answer is that the excess persists after allowing for age.
APOE ε4 vs the autosomal dominant genes — ε4 is a common risk allele that raises probability; mutations in APP, PSEN1 or PSEN2 cause early-onset familial disease outright and account for a small minority of cases; the genetics and pathology cards carry them.
Cognitive reserve vs brain reserve — cognitive reserve is the efficiency and flexibility built by education and activity; brain reserve is the physical quantity of neurons and synapses; both delay onset for a given pathology load.
Risk factor vs prodrome — late-life depression predicts dementia, but part of that is early illness presenting as low mood, so a stem that calls depression a proven cause has overstated the evidence.
Relative risk vs population attributable fraction — the Lancet Commission's forty-five per cent is the share of cases attributable to fourteen factors across a population; it says nothing about one person's odds.
Rural residence vs urban — a meta-analysis by Russ and colleagues (2012, International Journal of Epidemiology) found rural living associated with higher Alzheimer's prevalence (odds ratio 1.5), stronger for rural upbringing, with education and healthcare access the likely explanation rather than the countryside itself.

Watch for

⚠️ Recommending hormone replacement therapy to prevent dementia: in the Women's Health Initiative Memory Study (Shumaker and colleagues, 2003, JAMA), oestrogen plus progestin in women over sixty-five roughly doubled the rate of probable dementia (hazard ratio 2.05); HRT is not a preventive treatment.
Quoting the forty-five per cent as if it applied to Alzheimer's disease alone, or to an individual: it is a population attributable fraction for all-cause dementia from the 2024 Lancet Commission.
Confusing the two ways education protectsit delays onset by raising reserve; it does not reduce the pathology, which is why decline after diagnosis can be faster in the well educated.
Assuming the APOE ε4 effect is the same in every population: it appears weaker in people of African ancestry, and the Indianapolis-Ibadan study (Hendrie and colleagues, 2001, JAMA) found lower Alzheimer's incidence in Yoruba in Nigeria than in African Americans in Indianapolis despite similar ε4 frequency.
Treating anti-inflammatory drugs or statins as preventive: observational studies suggested benefit, randomised trials of NSAIDs and of statins did not show reduced dementia, and neither is recommended for prevention.
Reading the female excess as purely longevity: after adjusting for age, women still carry a somewhat higher risk, and the reasons are not settled.
Taking the sex ratio, the survival of three to ten years, or the death from bronchopneumonia as belonging here: those sit on the Alzheimer's course and management card; here the question is who gets it and why.

Use it again

Rank first, then modifyany stem asking for the single strongest risk factor wants age, then family history or APOE ε4, then everything else.
Sort each named factor into what it doesopens the tap (genes, vascular damage, head injury, alcohol) or lowers the rim (less education, hearing and vision loss, isolation, inactivity), and remember depression may be either a cause or an early symptom.
Attach every number to its populationone in fourteen over sixty-five is prevalence, doubling every five years is the age gradient, forty-five per cent is a population attributable fraction for all dementia, two-thirds are women.
Hand off at the boundarypathology and the amyloid cascade to the Alzheimer's pathology card, APOE dose effects and chromosome 21 to the Alzheimer's genetics card, imaging to the dementia imaging card, stages and treatment to the course and management card.
On prevention stems, the answer is mid-life risk reduction (NICE NG16, 2015, and the Lancet list), never HRT, NSAIDs or statins.

In detail

Hook — the professor who seemed to fall off a cliff

A retired professor of eighty-one is brought by his wife. Nothing was wrong a year ago, she says, and now he cannot follow a newspaper. His decline looks brutal and abrupt. Once delirium and a vascular cause have been excluded, the decline is probably neither brutal nor abrupt: it is what Alzheimer's disease looks like in a brain that spent sixty years building reserve, so the pathology had to climb a long way before it showed, and when it finally showed there was a great deal of it. You can read the whole epidemiology of this illness off that one presentation, provided you have a picture to hang it on.

Plain definition — who gets it and what moves the date

The epidemiology of Alzheimer's disease describes how the illness is distributed across a population and which characteristics raise or lower an individual's chance of developing it. The dominant variable is age. In the United Kingdom about 982,000 people were living with dementia in 2024 (Alzheimer's Society modelling), projected to reach 1.4 million by 2040, and Alzheimer's disease accounts for about two-thirds of them. Everything else on the list, sex, genes, education, vascular health, hearing, mood and social contact, moves the age at which the illness declares itself, forward or back.

The critical caveat — one number per population, never per person

Every figure on this topic belongs to a population and a definition, and the exam punishes moving them. (the proportion alive with the illness at one time) is about one in fourteen over sixty-five, with some surveys nearer one in eleven, and about one in six over eighty. (new cases per year) climbs with age just as steeply. The 2024 Lancet Commission's forty-five per cent is a population attributable fraction (the share of cases across a whole population that would disappear if fourteen factors were removed, assuming they cause the illness), for all dementia rather than Alzheimer's disease alone, and it says nothing about any one person's odds.

First principles — the bucket, the tap and the rim

Picture a bucket under a tap that has been running slowly since middle age. The water is Alzheimer pathology accumulating over time. The rim is cognitive reserve (the brain's to keep functioning despite damage, built by education, occupational complexity and mental and social activity, described by Stern in 2002). Dementia is the water passing the rim. From that picture, the age gradient follows at once: the longer the tap runs, the higher the water, and Jorm, Korten and Henderson showed in 1987 that dementia doubles roughly every five years after sixty-five.

Now move the parts. Raise the rim with education and the water takes longer to spill, so onset is later; but the level at spill-over is higher, so the professor declines fast once he declares. Widen the tap with APOE ε4, hypertension, diabetes, smoking or high LDL cholesterol and the bucket fills sooner. Turn the tap on from birth, as an extra copy of the amyloid precursor gene on chromosome 21 does in Down syndrome, and pathology is near-universal by the forties. Throw in a jug, for example a traumatic brain injury or years of heavy drinking, and the level jumps.

The rim can be lowered as well as raised. Hearing loss, vision loss, isolation and inactivity all lower it, so less reserve is left to spend. The bucket also tells you where it fails: it predicts that everyone who lives long enough will overflow, and many people in their nineties never do; and it has nothing to say about sex, though two-thirds of cases are women and longevity explains only part of that.

Structured walk-through — sorting a risk list by what each item does

Consider the fourteen factors the Lancet Commission named in 2024: less education, hearing loss, high LDL cholesterol, depression, traumatic brain injury, physical inactivity, diabetes, smoking, hypertension, obesity, excessive alcohol, social isolation, air pollution and untreated vision loss. Sort them. The tap group adds pathology or vascular damage: cholesterol, diabetes, smoking, hypertension, obesity, alcohol, head injury, and probably air pollution. The rim group lowers reserve or engagement: education, hearing, vision, isolation, inactivity. Depression sits in both columns and in a third: it may be a cause, it may lower engagement, and it may simply be the illness presenting early as low mood, which is why a stem calling it a proven cause has overstated the evidence.

Then test the sorting against the awkward findings. Rural living carries higher Alzheimer's , 1.5 in the by Russ and colleagues (2012), and about 2.2 for a rural upbringing; schooling and access to care, both rim factors, are the likely explanation. was lower in Yoruba in Ibadan than in African Americans in Indianapolis in the study led by Hendrie (2001), with similar APOE ε4 frequency in both, so environment and vascular load evidently change what the tap gene does. Neither finding breaks the bucket; both tell you which part to look at.

Look-alikes & differentials — figures that get swapped

for : one in fourteen over sixty-five is prevalence.
Alzheimer's disease for all dementia: the Lancet forty-five per cent and the 982,000 are all-cause figures.
A risk allele for a causal mutation: APOE ε4 raises probability; APP, PSEN1 and PSEN2 mutations cause early-onset familial disease and are a small minority of cases, taught on the genetics card.
Reserve for prevention: education delays onset and does not stop pathology.
A prodrome for a cause: late-life depression predicts dementia partly because it is early dementia.
An observational signal for a trial result: NSAIDs, statins and HRT all looked protective in observational data; randomised trials did not confirm it, and the Women's Health Initiative Memory Study (Shumaker and colleagues, 2003) found oestrogen plus progestin doubled probable dementia in women over sixty-five.

Diagnosis — where epidemiology hands over

Risk does not diagnose anyone. When memory decline is suspected, NICE NG97 (2018) sets the path: history from the person and an informant, a validated cognitive assessment, blood tests to exclude reversible causes, and structural imaging unless dementia is already well established; the imaging findings belong to the dementia imaging card and the clinical stages to the course and management card. What epidemiology contributes at diagnosis is the prior: a woman of eighty-six with a rural childhood, little schooling and an ε4 allele starts from a very different baseline from a fifty-year-old, and a fifty-year-old with symptoms should make you ask about Down syndrome or a familial mutation.

Management — moving the tap and the rim in mid-life

Prevention is mid-life work, and NICE NG16 (2015) frames it as reducing the vascular and lifestyle risks that also cause heart disease: treat hypertension and diabetes, stop smoking, keep alcohol down, and stay physically and socially active; correcting hearing and vision loss is the Lancet Commission's addition, not NG16's. Nothing on the shelf of hopeful drugs has earned a place: HRT increased dementia in a randomised trial, and NSAIDs and statins did not reduce it. Once the water has passed the rim, management is NG97 territory, with cholinesterase inhibitors and memantine taught on the course and management card; the commonest cause of death, bronchopneumonia after aspiration, is taught there too. NG97 and the Commission last checked 6 September 2026.

Exam pearl — the two numbers everyone swaps

In one breath, why do one in fourteen and forty-five per cent describe different things, and why is neither a number about one patient? The first is over sixty-five; the second is the share of all dementia attributable to fourteen modifiable factors across a population; a person's risk is the bucket they are standing next to.

In one line — the bucket

Age fills the bucket, reserve raises the rim, genes and vascular damage widen the tap, and the fourteen modifiable factors are the places where, in mid-life, the tap or the rim can still be moved.