Where You Live Matters: Regional Gaps in Dementia Diagnosis and Care

Within a single country, the care a person living with dementia receives depends a good deal on where they live. In England in June 2026, the share of people aged 65 and over thought to have dementia who had a recorded diagnosis ranged from 53.7% to 92.6% across 100 local NHS areas. In the United States, the chance of receiving a dementia diagnosis varied about twofold by region, and similar gaps appear in prescribing, support after diagnosis and hospital care.

For a family noticing changes in someone they love, the route from first worry to a diagnosis, a named contact, and help at home or in a crisis should not hinge on which side of a health-board boundary they live. The published figures show that it often does. Some of the difference reflects how common dementia is locally; much of it, the studies suggest, reflects how services are organised and staffed, and that part can be changed. This study brings together sub-national data from England, Scotland, the United States, Canada, Australia, Japan and Finland.

The chance of a diagnosis depends on where you live

England publishes one of the most detailed local pictures in the world. Each quarter, NHS England counts everyone on GP dementia registers and divides the number aged 65 and over by a modelled estimate of how many people in that area would be expected to have dementia. Nationally, 516,230 people had a recorded diagnosis on 30 June 2026, and the estimated diagnosis rate for people aged 65 and over was 66.2%^[1]. That national figure hides wide local differences. Across the seven NHS regions, the rate ran from 61.5% in the South West to 70.1% in the North West. Across 32 integrated care boards, it ran from 56.4% in Herefordshire and Worcestershire to 76.8% in South Yorkshire. Across 100 smaller sub-ICB locations, it ran from 53.7% in one part of Norfolk and Suffolk to 92.6% in one part of Staffordshire and Stoke-on-Trent^[1].

Put plainly, in the lowest area around half of the people expected to have dementia had a diagnosis on record. Only 53 of the 100 local areas reached two-thirds (66.7%) or more. Because the denominator is modelled, every local rate carries uncertainty: NHS England publishes a range of 48.0% to 58.8% around the lowest figure and 82.9% to 101.1% around the highest^[1]. The gap between the two ends is far wider than that uncertainty. A rate close to or above 100% means the register exceeds the modelled estimate, so at the top end it is the prevalence model, as much as local diagnosis, that is being tested. See our study of dementia diagnosis rates for the national trend.

The United States shows the same pattern with a different method. Using Medicare claims, Julie Bynum and colleagues compared the number of new diagnoses of Alzheimer’s disease and related dementias in each hospital referral region with the number expected from the local population’s age, risk factors and general intensity of medical practice. Crude new-diagnosis rates ranged from 1.7 to 5.4 per 100 across regions. After adjustment, diagnostic intensity ranged from 0.69 to 1.47, equivalent to a twofold difference in the likelihood of receiving a diagnosis. The variation was greatest for Black and Hispanic older people and for those aged 66 to 74^[2]. In the authors’ words, “Where one resides influences the likelihood of receiving an ADRD diagnosis.”

Some of the gap is real difference in how common dementia is

Not every regional difference is a service failure. In the US Health and Retirement Study, a national survey that classifies dementia using validated methods, estimated dementia prevalence among older adults in 2000 was 8.6% in the West, 10.0% in the Midwest, 11.1% in the Northeast and 14.6% in the South. By 2012 prevalence had fallen in every region but remained highest in the South, at 10.7%, against 7.0% to 7.8% elsewhere. The authors found that these regional differences in change were largely explained by education and health status^[4].

Australia’s 2021 census, which asked people about long-term health conditions including dementia, shows how geography can interact with other inequalities. Among non-Indigenous Australians aged 45 and over, the age-standardised prevalence of reported dementia was 7.4 per 1,000; among Aboriginal and Torres Strait Islander peoples it was 16.2 per 1,000. The odds of dementia fell with increasing remoteness for non-Indigenous people, but not for Aboriginal and Torres Strait Islander people^[5]. Because these are self- or family-reported figures, lower reporting in remote areas may partly reflect less contact with diagnostic services rather than less dementia.

A region with more dementia should have more diagnoses; the question is whether diagnoses keep pace with need. England’s diagnosis rate and the US study both compare recorded with expected cases, and both still found wide gaps^[1][2]. Where distance is the barrier, see our study of dementia in rural and remote areas.

Prescribing varies more than clinical need would suggest

Antipsychotics are sometimes prescribed to people living with dementia for distress or agitation, but guidelines restrict their use. England’s June 2026 data count how many people on the dementia register without a recorded psychosis diagnosis received an antipsychotic prescription in the previous six weeks. Our calculation from the published counts gives 7.0% for England as a whole (34,753 of 498,700 people). Across 106 local sub-ICB areas the figure ranged from 2.5% to 13.4%, a more than fivefold spread; the middle-ranked area was at 6.4%, with one area in ten below 4.0% and one in ten above 9.7%^[1].

The same data show a similar spread in medicines intended to treat the symptoms of Alzheimer’s disease. The share of people on the register prescribed donepezil, rivastigmine or galantamine in the previous year was 24.8% in England overall, but ranged from 11.7% to 39.6% across local areas^[1]. These figures are not adjusted for age, type of dementia, severity or whether people live in care homes, so some of the spread is about who lives where. It is hard to see how all of it could be.

North American studies point to how care is organised as a driver. In 2018, among 273,004 long-stay US nursing home residents with dementia, 20.9% used antipsychotics. Homes with poor staffing ratings had higher odds of antipsychotic use (adjusted odds ratio 1.10), and resident and facility factors accounted for much of the variation between states and regions^[6]. In Ontario, Canada, a study of 16,896 nursing home residents in 2022–23 found that 39% were prescribed an antipsychotic, and that the home a person lived in mattered: the home-level median odds ratio was 1.65, and residents’ own characteristics, including dementia and aggression, did not explain the differences between homes^[7]. Across Canada, data from the Canadian Institute for Health Information show that the share of long-term care residents on antipsychotics without a psychosis diagnosis fell from 27.2% to 20.2% between 2014–15 and 2019–20, then rose to 24.5% in 2023–24, with every reporting province and territory rising “by at least 2%” since the pandemic began^[8]. Our study of antipsychotic prescribing covers this in depth.

Support after diagnosis, and the specialists who give it, are unevenly spread

Scotland is one of the few countries that publishes, area by area, whether people are offered support after a diagnosis. Its national standard aims for everyone newly diagnosed to be referred for, and receive, at least a year of post-diagnostic support. In 2023/24, 50.6% of the people estimated to be newly diagnosed with dementia in Scotland were referred for that support. Across NHS Boards the share ranged from 65.3% in Dumfries and Galloway to 31.2% in Tayside and 29.7% in Orkney, where there was no consultant psychiatrist available to diagnose dementia for the first half of the year^[3].

Being referred is only the first step. Of everyone referred in 2023/24, 81.2% went on to receive at least a year of support. Here the ranking almost reverses: Orkney and the Western Isles reached 100%, while Dumfries and Galloway, the Board with the highest referral rate, had the lowest share completing a year of support, at 47.6%. Public Health Scotland links this to the COVID-19 pandemic and staffing problems with its former commissioned service^[3]. See our study of post-diagnostic support for the wider evidence.

Behind these figures sit uneven numbers of specialist staff. Across US hospital referral regions, the median density of dementia specialists (neurologists, geriatricians and geriatric psychiatrists) was 28.8 per 100,000 people aged 65 and over, with an interquartile range of 19.3 to 43.6. Against the authors’ estimate that 33 to 45 specialists per 100,000 are needed, 34% to 59% of older Americans lived in areas with a potential shortfall, which varied by state and rurality^[9].

In Japan, national policy requires at least one medical centre for dementia in every secondary healthcare area. A 2026 analysis identified 509 such centres, but 18 of 344 areas (5.2%), covering 105 municipalities, still had none. Municipalities that did not name a centre in their local dementia care pathway were mainly in geographically or climatically disadvantaged places, such as heavy-snowfall regions and peninsulas^[10].

Hospital care and care near the end of life also follow geography

Where someone lives shapes what happens when their health worsens. A study of 143,891,769 hospital admissions among Black and White US Medicare beneficiaries from 2000 to 2018, across 3,037 counties, found that 13.2% of admissions involved a dementia diagnosis. The shared risk of admission with dementia was higher in southern and eastern states. Black beneficiaries’ relative risk of admission with dementia was significantly higher than White beneficiaries’ in 90.2% of counties, compared with 44.7% of counties for admissions without dementia^[11].

Planning ahead for care at the end of life varies too. In Finland, among 6,090 long-term care residents aged 65 and over who died of chronic progressive diseases in 2019, advance directives were more common in urban than rural municipalities. The share with a do-not-resuscitate order ranged from 70.9% to 95.0% between municipalities, and the share with other advance directives from 27.1% to 70.0%^[12]. The authors warn this “may result in unequal care and service use despite similar clinical conditions.” Among the sources we checked, none published routine regional statistics on end-of-life care specifically for people with dementia.

What the evidence asks of us

The evidence points to a few clear asks of commissioners, health systems, funders and governments.

• Publish local data, and publish it in a usable form. England and Scotland can show these gaps only because they count diagnoses, prescribing and support area by area. Most countries we looked at do not publish comparable sub-national dementia statistics. A 2026 review of national dementia policies across 38 OECD countries found “a gap between policy and action, with varying degrees of implementation and support for improving diagnosis”^[13]. Without local data, that gap cannot be seen.
• Judge areas on need met, not activity. The best measures compare what happens with what would be expected, as England’s diagnosis rate and the US diagnostic-intensity study do^[1][2]. Prescribing and support measures should be adjusted for age, dementia type and care setting, so that low-performing areas cannot hide behind their populations and high-need areas are not wrongly blamed.
• Close the gap from the bottom. NICE guidance in England says people living with dementia should have “a single named health or social care professional who is responsible for coordinating their care”^[14]. Scotland’s figures show that a guarantee is only as good as the staff behind it: where link workers or psychiatrists were missing, referrals or completed support fell sharply^[3]. Workforce planning should follow the shortfalls the data reveal, including in rural and remote areas^[9][10].
• Treat wide prescribing variation as a quality signal. NICE recommends psychosocial and environmental approaches as the first response to distress, and antipsychotics only in narrowly defined circumstances^[14]. The US nursing home study concluded that its findings suggest “the need for evidence-based protocols to guide the use of these medications”^[6]. A fivefold spread between local areas is a reason to look closely at the areas at both ends.
• Look at place and inequality together. In the US, regional and racial differences in hospital admission with dementia compound each other, and in Australia the link between remoteness and dementia differs for Aboriginal and Torres Strait Islander peoples^[5][11]. Plans to reduce variation should be designed with the communities most affected.

The World Health Organization estimates that 57 million people were living with dementia worldwide in 2021; its global action plan names diagnosis, treatment and care, and support for carers among its action areas^[15]. Those aims are not met if they are met only in some places.

Language: this piece uses person-first language ("people living with dementia", "people with dementia"), as most dementia organisations and people with lived experience prefer, and uses clinical terms such as "Alzheimer's disease" or "mild cognitive impairment" only where a study, dataset or diagnosis refers to that specific condition.

Why we're publishing this

Dementia charities, carers’ groups and local campaigners often know from experience that services differ from one area to the next. This study puts published figures behind that experience. The gaps described here are gaps in services, not in the people who need them. It is part of our dementia and health research library.

How to read this data

Different measures, not one league table. England’s diagnosis rate is recorded diagnoses on GP registers divided by a modelled estimate of expected prevalence, so it depends on the model as well as on diagnosis. Scotland’s referral measure also uses a modelled estimate of new diagnoses, and its 2023/24 figures are provisional. US figures come from Medicare claims and the Health and Retirement Study, Australia’s from census self-report, Canada’s from long-term care assessments, and Finland’s from care assessments of people who died. None of these can be compared directly with each other, and we do not rank countries.

Our own calculations. The English prescribing ranges are our calculations from counts NHS England publishes for local sub-ICB areas at 30 June 2026. They are crude shares, not adjusted for age, dementia type, severity or care-home residence. The file lists 106 sub-ICB rows for these measures, while the diagnosis rate file lists 100.

Associations, not causes. Studies that link variation to staffing, education or healthcare use show associations; they do not prove that one factor causes the differences. Odds ratios are reported as odds.

Gaps in the evidence. We could not find comparable published sub-national dementia statistics for many countries, including the Nordic registers, and Australia’s official dementia statistics were not reachable when we checked.

Use this data

Free to cite with attribution to Understanding Aging. Charities, carers' groups, campaigners and journalists are welcome to reuse these figures and charts. For the underlying figures as a spreadsheet, get in touch.

Important: This article summarises published research and statistics for general information. It is not medical, care or financial advice and cannot be used to diagnose, treat or predict anything about an individual. If you have concerns about your own or someone else's memory or health, please speak to a GP or another qualified health professional.

Sources

1. NHS England. Primary Care Dementia Data, June 2026 (including diagnosis rate by NHS organisation and sub-ICB measures, CSV files). NHS England, 2026. digital.nhs.uk/…/primary-care-dementia-data/june-2026
2. Bynum JPW, Benloucif S, Martindale J, O’Malley AJ, Davis MA. Regional variation in diagnostic intensity of dementia among older U.S. adults: an observational study. Alzheimer’s & Dementia, 2024. europepmc.org/article/MED/39149970
3. Public Health Scotland. Dementia post-diagnostic support: performance against the Local Delivery Plan standard, 2023/24. Public Health Scotland, 2026. publichealthscotland.scot/publications/dementia-post-diagnostic-support
4. Ailshire JA, Walsemann KM, Fisk CE. Regional variation in U.S. dementia trends from 2000–2012. SSM – Population Health, 2022. europepmc.org/article/MED/35855971
5. Clarke AJ, Ahmed M, Katzenellenbogen JM, et al. The intersection of rurality and dementia prevalence in Australia for Aboriginal and Torres Strait Islander and non-Indigenous peoples. Medical Journal of Australia, 2025. europepmc.org/article/MED/40313188
6. Rataj A, Alcusky M, Baek J, Ott B, Lapane KL. Geographic variation of antidementia and antipsychotic medication use among US nursing home residents with dementia. Medical Care, 2024. europepmc.org/article/MED/38833712
7. Dash D, Rahim A, Siu HY, et al. Variation in psychotropic and non-psychotropic prescribing in nursing homes: a multi-level EMR-based retrospective cohort study. Journal of the American Geriatrics Society, 2026. europepmc.org/article/MED/41709403
8. Choosing Wisely Canada. Rising Rates: Antipsychotic Use in Canada’s LTC Homes (using Canadian Institute for Health Information data). Choosing Wisely Canada. choosingwiselycanada.org/news/new-report-rising-rates
9. Liu JL, Baker L, Chen AY, Wang JJ. Geographic variation in shortfalls of dementia specialists in the United States. Health Affairs Scholar, 2024. europepmc.org/article/MED/39081720
10. Baba Y, Tsuda S, Ito K, Awata S. Geographical and climatic factors driving regional variation in medical centres for dementia in Japan: a nationwide geocomputational analysis. BMJ Open, 2026. europepmc.org/article/MED/42601097
11. Mock L, Mork D, Audirac M, Braun D, Zanobetti A, Delaney SW. Hospitalization risk with and without dementia by region and race in the United States. American Journal of Epidemiology, 2026. europepmc.org/article/MED/42017264
12. Forsius P, Jämsen E, Finne-Soveri H, Aaltonen M. Advance directives in Finnish long-term care: do sociodemographic characteristics and regional factors matter? Scandinavian Journal of Public Health, 2026. europepmc.org/article/MED/40653659
13. Kim S, Rauet-Tejeda J, Milstein R, Lee J, Llena-Nozal A. Evolving national dementia policies in the OECD: prevention, diagnosis, and care. Alzheimer’s & Dementia, 2026. europepmc.org/article/MED/42222904
14. National Institute for Health and Care Excellence. Dementia: assessment, management and support for people living with dementia and their carers (NG97), recommendations. NICE, 2018. nice.org.uk/guidance/ng97/chapter/Recommendations
15. World Health Organization. Dementia fact sheet. WHO. who.int/news-room/fact-sheets/detail/dementia