Background A key aim of the End of Life Care Strategy is to enable more people to die in their preferred place; this is often their own home. However, hospital remains the most common place of death, particularly for people dying from causes other than cancer.

Aims To explore end of life care preferences and determine predictors of achieving preferred place of death, home death and hospital death.

Methods The Office for National Statistics drew a proportionally allocated stratified sample of deaths registered in two PCTs between October 2009 and April 2010. Coroner-registered deaths and deaths under the age of 18 years were excluded. The VOICES questionnaire was sent to each informant (n=1422, usually bereaved relative) 6–12 months after the death. As planned, data were weighted to account for response bias.

Results 473 returned questionnaires (33%). 35% of respondents reported that the deceased had said where they wanted to die (mainly home 68%). This was apparently recorded for 36% of those expressing a preference. Only 1% was reported to have changed their mind about where they wanted to die. Cancer patients were significantly more likely to die at home than those who died from any other condition (p<0.001) and significantly more likely to have a record of preferences (OR=5.86). Having a record of preferences was significantly associated with both achieving preference (OR=21.0) and home death (OR=15.9) when adjusting for age, sex, deprivation, knowledge of imminent death, relationship of informant, cause of death and number of home services used. Conversely, a reduction in the odds of having recorded preferences was associated with hospital death (OR=0.20).

Conclusions Although, out of all causes, cancer deaths were most likely to occur at home, having a record of preferences was the only independent predictor of achieving preferred place of death.