Somewhere tonight, a woman in her fifties is setting an alarm for 3 a.m. Not for a flight. For her father, who wanders when he wakes and has twice tried to leave the house in the dark. She will be at her desk by nine, or she will not, and if she is not, her employer will record it as absence and her country’s statistics will record nothing at all.
That invisible night shift is the single largest item in the global cost of dementia. And because it appears on no health budget, most health systems behave as if it does not exist.
A trillion-dollar system with no line item
The World Health Organization estimates that dementia cost economies around the world US$1.3 trillion in 2019. About half of that cost reflects care provided by informal carers, meaning family members and close friends. WHO says these carers provide on average five hours of care and supervision a day, and that women provide around 70 per cent of care hours.
Put those facts together and a different picture emerges from the usual one. Dementia is not mainly a medical expense. It is a labour-market event. It removes working-age people, overwhelmingly women, from paid work or pushes them into reduced hours, at exactly the career stage when their earnings and pension contributions would be highest.
WHO also counts 57 million people living with dementia in 2021, more than 60 per cent of them in low- and middle-income countries, with nearly 10 million new cases a year. Many of those countries have almost no formal long-term care sector at all. The family is not a supplement to the system. It is the system.
Pricing the night shift
It helps to put rough numbers on a single household, with the caveat that these are illustrative, not measured.
Take WHO’s average of five hours of care and supervision a day. Over a year that is more than 1,800 hours, close to a full-time job. Value those hours at even a modest local care-worker wage and the figure for one family runs into the tens of thousands of dollars in a high-income country, every year, for a condition that commonly lasts many years after diagnosis.
Now look at what the carer gives up. Suppose she is 52, earning an average salary, and cuts to half-time for eight years. The direct loss is four years of full salary. The indirect loss is larger: slower promotion, lower pension contributions in her highest-earning decade, and a real chance she never returns to full-time work at all. Her employer loses an experienced worker and pays to recruit and train a replacement. The state loses income tax and, later, may pay more in pension support.
None of this appears in the dementia line of the health budget. Add it up across millions of households and the WHO total stops looking abstract. It is a transfer from the formal economy to kitchens and bedrooms, recorded nowhere.
Why “free” care is the most expensive option
Unpaid care looks cheap because no invoice arrives. But the costs are displaced rather than avoided.
They show up in lost tax revenue when carers leave work. They show up in the carer’s own health, because chronic sleep loss and stress do not stay confined to the person with dementia. They show up in hospitals, when an exhausted family can no longer cope and the only available “care setting” is an emergency admission. And they show up decades later, in the poverty of older women whose pension records have a ten-year hole in them.
A health ministry that counts only what it pays for will conclude that dementia is under control. A finance ministry that looked at the whole picture would reach a different conclusion.
Why the invisibility persists
The incentive structure explains why this stays hidden. Health budgets are built around services delivered and paid for. Labour ministries count people in and out of work but rarely ask why. Pension systems record contributions, not the reasons for gaps. Each ministry sees a fragment, and no one owns the whole.
Families, meanwhile, have strong reasons not to demand more. Caring for a parent is bound up with love, duty and, in many cultures, shame at the idea of handing it over to strangers. That makes family care resilient. It also makes it politically silent. People who are exhausted and awake at 3 a.m. are rarely at public meetings.
There is a gendered compounding effect too. Because women are more likely to have taken career breaks for children, they often earn less than male partners by midlife, so when a household decides who reduces hours to care, the arithmetic points at her again. Two unpaid shifts, one at each end of adult life, add up to the pension gap that shows up in old-age poverty statistics.
The Japanese lesson is about who pays, not how much
Japan faced a version of this problem in the 1990s. Older people were being admitted to hospital and kept there because their families could no longer care for them, a phenomenon officials called “social admission.” It was expensive, inappropriate and deeply unpopular.
In 2000, Japan introduced a mandatory long-term care insurance scheme, with everyone aged 40 and over paying premiums. Its explicit goals, as described in the Japan Medical Association Journal and by the country’s own social security research institute, included reducing the burden on families and avoiding unnecessary institutionalisation. People aged 40 to 64 are covered only for specific age-related conditions, including early-onset dementia.
The important design choice was not generosity. It was recognising long-term care as a separate insured risk, paid for collectively, rather than as a gap between the health system and the family. Once care is a line item, it can be planned, priced, regulated and improved. When it is invisible, it can only be endured.
The strongest objection
The case against formalising care deserves a serious answer. Critics make three points. First, public long-term care insurance is expensive, and ageing populations make it more so each year; Japan itself has repeatedly adjusted premiums and co-payments to keep its scheme affordable. Second, paying for care can crowd out family care that would have happened anyway, so the state ends up funding something it previously got for nothing. Third, many low- and middle-income countries simply lack the fiscal room or the workforce for a Japanese-style scheme.
Each point partly holds. Formal care costs money, and some substitution of paid for unpaid care is inevitable. But the crowding-out argument assumes unpaid care is free, which is the very error this piece is about. If a paid carer lets a daughter keep her job, the state recovers part of the cost in tax and avoided hospital admissions. The fiscal argument is stronger for poorer countries, and the answer there is not a full insurance scheme but cheaper steps: training community health workers in dementia support, respite through existing community structures and protecting carers’ access to their own health care.
The real question is not whether to spend but where the spending lands. Right now much of it lands in emergency departments and hospital beds, which are the most expensive places to look after a person with dementia who needs supervision rather than surgery.
Prevention is a real economic lever, with a caveat
The 2024 Lancet Commission on dementia identified 14 modifiable risk factors across the life course, including hypertension, hearing loss, smoking, physical inactivity, air pollution, high LDL cholesterol and untreated vision loss. It estimated that addressing them could, in theory, prevent or delay up to 45 per cent of cases.
That figure is a population-attributable estimate, not a promise. It assumes risk factors are fully eliminated and that the relationships are causal. But even a fraction of it would matter economically, and several of the levers, like hearing aids and blood pressure control, are cheap compared with years of full-time care.
The trade-off is political. Prevention pays out slowly and to someone else’s budget. The minister who funds hearing screening at 50 will not be in office when the dementia cases fail to appear at 80.
Second-order effects nobody budgets for
There is a subtler cost as well. The same people who shoulder dementia care are often the people who would otherwise staff the formal care workforce, teach, nurse, or run small businesses. In countries already short of care workers, pulling experienced mid-career women out of paid employment makes the formal system harder to build, which pushes more care back to families. It is a loop.
New treatments add another wrinkle. The amyloid-targeting drugs approved in recent years in the United States and elsewhere offer modest slowing of decline for some people in early disease, and require diagnosis, scans and infusion capacity. Whatever their clinical value, they will not remove the need for care. If anything, they increase demand for early diagnosis, which brings more families into the system sooner and makes the question of who supports them more urgent, not less.
What an economic approach to dementia would change
If governments treated dementia as an economic issue rather than a clinical niche, a few things would follow:
- Measure carer hours. National surveys should count unpaid care time the way labour surveys count paid work.
- Protect carers’ jobs. Flexible work rights and carer leave are cheaper than losing experienced workers entirely.
- Fund respite as infrastructure. A few hours of relief a week can be what keeps a family caring rather than collapsing into a hospital admission.
- Credit care in pensions. If care is work, it should build retirement entitlements.
- Put prevention in the finance ministry’s forecast, where its returns are visible.
The implications differ by actor. For finance ministries, the first step is simply to model dementia care as a cost to labour supply and tax revenue, not a health line. For health ministries, it is building dementia into primary care and hearing and blood pressure programmes that already exist. For hospital leaders, it is recognising that a share of delayed discharges are really failures of community care, and pushing for funded alternatives rather than more beds. For employers, carer leave and flexible hours are a retention policy, not a perk. For funders and founders, the gap is less in diagnostics than in the unglamorous middle: respite services, carer training and tools that help families coordinate care without burning out.
What to watch
In the next year or two, watch whether more governments add unpaid care questions to national labour and time-use surveys, because what gets measured gets budgeted. Watch how countries with ageing populations reform long-term care financing, and whether they treat carers’ pensions as part of that reform. Watch WHO’s monitoring of its global dementia action plan, which set targets for national dementia plans and carer support that many countries have not met. And watch whether prevention programmes for hypertension and hearing loss are funded and evaluated with dementia outcomes in mind. Those are slow signals, but they are the ones that show whether the invisible night shift is finally being counted.
The takeaway
The cost of dementia is not coming. It is here, and it is being paid mostly in the hours, incomes and health of women who are not on any payroll. The choice facing governments is not whether to pay for dementia care. It is whether to keep paying for it in the most inefficient way available: silently, through families, until they break.
If you have seen a policy that genuinely supports family carers, what was it? We would like to hear about it in the comments.
