Background: Most of the research concerning place of death focuses on terminally ill patients (cancer patients) while the determinants of place of death of the elderly of the general population are not intensively studied. Studies showed the influence of gender, age, social-economical status and living arrangements on the place of death, but a facet not taken into account so far is the influence of the availability of nursing homes.
Methods: We conducted a survey of deaths, between January 1999 and December 2000 in a small densely populated area in Belgium, with a high availability of nursing homes (within 5 to 10 km of the place of residence of every elderly). We determined the incidence of total mortality (of subjects >60 years) from local official death registers that we consulted via the priest or the mortician of the local parish, to ask where the decedent had died and whether the deceased had lived in a nursing home. We compared the distribution of the places of death between parishes with a nursing home and with parishes without nursing home.
Results: 240 women and 217 men died during the two years study period. Only 22% died at home, while the majority (78%) died in an institutional setting, either a hospital (50%) or a nursing home (28%). Place of death was influenced by individual factors (age and gender) and the availability of a nursing home in the 'own' parish. The chance of in-hospital death was 65% higher for men (95% Confidence Interval [CI]: 14 to 138%; p = 0.008) and decreased by 4% (CI: -5.1% to -2.5%; p < 0.0001) for each year increase in age. Independent of gender and age, the chance of in-hospital death was 41% (CI: -60% to -13%; p = 0.008) lower in locations with a nursing home.
Conclusion: Demographic, but especially social-contextual factors determine where elderly will end their life. The majority of elderly in Flanders die in an institution. Age, gender and living situation are predictors of the place of death but the embedment of a nursing home in the local community seems to be a key predictor.