Picture the leaflet that arrives with a diagnosis in the family, or the poster in the waiting room. Move more. Eat better. Watch the blood pressure. Stay connected. All of it is decent advice, and none of it mentions the damp in the rental, the eight kilometres to the nearest pool, the bus that stops running at six, the school years that ended at fourteen because the family needed the wage. A team led by Curtin University has now put a number on that silence. In a Viewpoint published in The Lancet Public Health on 9 September 2026, Jennifer Dunne, Mario Siervo, Kaarin Anstey, Blossom Stephan and colleagues took the 61 potentially modifiable dementia risk and protective factors that appear in the published population attributable fraction literature and sorted each one by the level at which it operates: individual, interpersonal, community, societal. Forty-nine of the 61, about 80 per cent, were individual. Two were interpersonal (marital status and social isolation). Three were community (air pollution, neighbourhood safety, living in a deprived area). Seven were societal.
It matters what kind of study this is, because the number is easy to over-read. This is not a trial, not a cohort, and not a new risk calculation. There were no participants and no follow-up period. It is a Viewpoint: an argument built by applying the socioecological model, a long-established public health framework, to evidence that already existed. The 61 factors came from a 2024 systematic review and meta-analysis of population attributable fractions led by Stephan in The Lancet Healthy Longevity, which pooled published estimates and found low education, physical inactivity, hearing loss, hypertension and obesity to be the most prominent contributors. The classification work in the new paper was done by two authors reading independently and then resolving where they disagreed, as reported by Earth.com. The funder listed on the release announcing the paper is Dementia Australia.
So the headline figure is a fact about a literature, not a measurement of blame. That is the authors' point rather than a flaw in it. Research has concentrated where measurement is easiest, on things that sit in a body and a calendar of habits, and the tally of factors now mirrors that concentration. Dunne's framing in the university's own announcement is careful: if prevention focuses predominantly on individual behaviour, she says, "we risk asking people to make healthier choices without addressing the conditions that make those choices possible". Stephan, who holds the Chair in Dementia at Curtin's enAble Institute and led the 2024 PAF review the new paper draws on, puts the next move bluntly: "The next step is turning this evidence into action."
Read honestly, the paper does not demote the personal. The 49 individual-level entries include smoking, drinking, inactivity, poor sleep, obesity, high blood pressure, diabetes, hearing loss, depression and head injury, and several of those carry the largest attributable fractions anyone has managed to estimate. Nothing in this analysis suggests a family should stop caring about blood pressure or hearing. What it suggests is that the list has been sorted in a way that hides its own shape: education, housing, healthcare access, neighbourhood safety and clean air appear thinly, or as single entries, despite operating upstream of half the habits further down the page.
For a family living with a diagnosis, the useful thing here is permission to stop reading prevention advice as a verdict on a life. Someone who left school early in a polluted industrial town, who worked shifts, who could not get a hearing test for two years, has not failed a personal test of discipline. The Curtin analysis says the categories of dementia prevention were drawn in a way that made that person's circumstances nearly invisible, and that the fix belongs partly with governments, planners and health systems rather than entirely with the person reading the leaflet. The paper cites the scale of what is coming: roughly 57 million people worldwide living with dementia now, projected to reach 139 million by 2050, with about 446,500 in Australia and more than a million expected by 2065.
What happens next is a policy question, and the paper is honest that it is only the first step. Counting factors is not the same as weighing them, and a Viewpoint cannot tell a housing minister how many cases better housing would prevent. It can, however, change what "the evidence supports doing now" is allowed to mean: not only what a person does at 7am, but what the street, the school and the air did first.