LAUSR

Abstract Background hospitalisations are sentinel events for people with dementia. How patterns of unplanned hospital admissions change among people with dementia after diagnosis is relatively unknown. Objective to describe patterns of unplanned hospital admissions of people with dementia from diagnosis until death/study end. Methods retrospective cohort study using mental healthcare provider data of people diagnosed with dementia in London, UK (1995–2017), linked to mortality and hospital data. The primary outcome was the rate of unplanned hospital admissions after diagnosis until death/study end. We calculated the cumulative incidence of unplanned hospital admissions. The rates of unplanned hospital admissions and the percentage of time spent as an inpatient were stratified by time from first dementia diagnosis. Results for 19,221 people with dementia (61.4% female, mean age at diagnosis 81.0 years (standard deviation, SD 8.5)), the cumulative incidence of unplanned hospital admissions (n = 14,759) was 76.8% (95% CI 76.3%–77.3%). Individuals remained in the study for mean 3.0 (SD 2.6) years, and 12,667 (65.9%) died. Rates and lengths of unplanned hospital admissions remained relatively low and short in the months after the dementia diagnosis, increasing only as people approached the end of life. Percentage of time spent as an inpatient was <3% for people who were alive at the study end but was on average 19.6 and 13.3% for the decedents in the last 6 and 12 months of life, respectively. Conclusions the steep rise in hospitalisations before death highlights the need for improved community care and services for people with dementia who are approaching the end of life.