Background: end-of-life care is not always in line with end-of-life preferences, so patients do not always die at their preferred place of death (PPD). This study aims to identify factors associated with patients' PPD and changes in PPD. 

Methods: we prospectively collected data on PPD at four time points within 6 months from 230 acutely hospitalised older patients who were part of the control group in a stepped-wedge randomised controlled trial. Associations between patient characteristics and preferences were calculated using multivariable (multinomial) logistic regression analysis. 

Results: the mean age of participants was 80.7 years. 47.8% of the patients had no PPD at hospital admission. Patients previously admitted to hospital preferred to die at home (home versus no preference: odds ratio [OR] 2.38, 95% confidence interval [CI] 1.15-4.92; home versus healthcare facility: OR 3.25, 95% CI 1.15-9.16). Patients with more chronic diseases preferred the healthcare facility as their PPD (healthcare facility versus no preference: OR 1.33, 95% CI 1.09-1.61; healthcare facility versus home: OR 1.21, 95% CI 1.00-1.47). 32 of 65 patients changed their preference during follow-up, and most of these had no PPD at hospital admission (home versus no preference: OR 0.005, 95% CI ≤0.001-0.095) and poorer self-rated well-being (OR 1.82, 95% CI 1.07-3.08). 

Conclusions: almost half of the patients had no PPD at baseline. Previous hospital admission, having more chronic diseases and living alone are associated with having a PPD. Introducing PPD could make older people aware of PPD and facilitate optimal palliative care.