People later diagnosed with dementia before the age of 65 were already earning measurably less than their matched peers up to 15 years before the diagnosis was recorded, according to a retrospective, population-based cohort study published on 8 July 2026 in Neurology, the journal of the American Academy of Neurology. The study, led by Ave Kivisild at the University of Eastern Finland with colleagues in Kuopio and Oulu, covered all early-onset dementia patients seen in the Kuopio and Oulu University Hospital catchment areas between January 2010 and December 2021: 793 people, half of them women, with a mean age of 59.6 years, whose diagnoses were re-checked against clinical records. Each was matched by age and sex to ten controls, 7,926 in total. Cumulative productivity loss, measured as the difference in registered taxable income against controls and adjusted for education and other conditions, came to 74,577 euros per patient (95% confidence interval 46,423 to 102,732). Eino Solje, the senior author, put the annual figure at roughly 12,000 euros per person and called the loss "strikingly large".
The timing varied by subtype, which is the part that will read most sharply to families who have spent years wondering when the trouble actually started. In the 179 people with frontotemporal dementia spectrum disorders, a statistically significant gap opened 11 years before diagnosis (4,799 euros per patient per year; 95% CI 433 to 9,166) and widened to 16,116 euros in the diagnosis year. In the 421 people with Alzheimer disease, the gap appeared later, six years out (2,767 euros; 95% CI 18 to 5,515), then rose to 11,431 euros at diagnosis. Among the 46 people with alpha-synucleinopathies, a group that includes dementia with Lewy bodies and Parkinson's disease dementia, the trajectories were more variable and no difference reached significance until the year of diagnosis (11,284 euros). In the remaining 147, whose diagnoses included vascular cognitive disorders and mixed dementias, losses were consistently high across the whole follow-up rather than emerging at a particular point.
The authors attribute part of the pattern to diagnostic delay, meaning years of symptoms that were noticed at work and at home but not yet named in a clinic. The lower bound of the confidence intervals for the earliest years sits close to zero in several subtype analyses, which means the earliest signal is the least certain part of the finding. The alpha-synucleinopathy group is small enough that an absence of an early difference cannot be read as an absence of early change. This context follows earlier work from the same Finnish groups, published in Neurology in July 2024, which found a crude early-onset dementia incidence of 20.5 per 100,000 people aged 30 to 64 per year in the same two regions, with amnestic Alzheimer disease the most common subtype.
The study was funded by Roche Oy, the Finnish subsidiary of Roche, a disclosure the journal records alongside consultancy and lecture fees received by Solje from several pharmaceutical companies. The design carries its own limits: the data were assembled after the fact, income was the only measure of work capacity, and no cognitive testing was done during the pre-diagnostic years, so no link can be drawn between a given year's earnings and a given year's brain change. The findings also sit inside Finland's registry, tax and sickness benefit systems, and the euro figures should not be read across to countries with different employment protection or disability arrangements. Solje said future work should pair employment records with repeated neuropsychological testing.