Predicted survival vs. actual survival in terminally ill noncancer patients in Dutch nursing homes.

Studies on the prediction of survival have mainly focused on hospital and hospice patients suffering from cancer. The aim of this study was to describe the predicted vs. the actual survival in terminally ill, mainly noncancer patients in Dutch nursing homes (NHs). A prospective cohort study was conducted in 16 NHs representative for The Netherlands. A total of 515 NH patients with a maximum life expectancy of 6 weeks, as assessed by an NH physician, were included. NH physicians were accurate in more than 90% of their prognoses for terminally ill--mainly noncancer--NH patients, when death occurred within 7 days. For a longer period of time, their predictions became inaccurate. In the category of patients who were expected to die within 8-21 days, predictions were accurate in 16.0%, and in the category of patients expected to die within 22-42 days, this was 13.0%. Predictions in these categories were mainly optimistic (patient died earlier) in 68.6% and 52.2%, respectively. The findings of this study suggest that accurate prediction of survival of (mainly) noncancer patients in NHs is only possible when death is imminent and seems to be dependent on an intimate knowledge of patients. Prognostication over a longer period of time tends to be less accurate, and, therefore, continues to be a challenging task for NH physicians.