Abstract
BACKGROUND: One third of older people are estimated to die with dementia, which is a principal cause of death in developed countries. While it is assumed that people die with severe dementia this is not based on evidence. METHODS: Cohort study using a large secondary mental healthcare database in North London, UK. We included people aged over 65 years, diagnosed with dementia between 2008 and 2016, who subsequently died. We estimated dementia severity using mini-mental state examination (MMSE) scores, adjusting for the time between last score and death using the average annual MMSE decline in the cohort (1.5 points/year). We explored the association of sociodemographic and clinical factors, including medication use, with estimated MMSE score at death using linear regression. RESULTS: In 1400 people dying with dementia, mean estimated MMSE at death was 15.3 (standard deviation 7.0). Of the cohort, 22.2% (95% confidence interval 20.1, 24.5) died with mild dementia; 50.4% (47.8, 53.0) moderate; and 27.4% (25.1, 29.8) with severe dementia. In fully adjusted models, more severe dementia at death was observed in women, Black, Asian and other ethnic minorities, agitated individuals, and those taking antipsychotic medication. CONCLUSIONS: Only one quarter of people who die with dementia are at the severe stage of the illness. This finding informs clinical and public understanding of dementia prognosis. Provision of end-of-life services should account for this and healthcare professionals should be aware of high rates of mild and moderate dementia at end of life and consider how this affects clinical decision-making.