Background: With an aging population, and most deaths due to a nonmalignant cause, there is urgency to review the nature of end-of-life care (EoLC) to minimize gaps in service provision. Early introduction of EoLC benefits patient and carers, so identification of those at risk of dying 6 to 12 months before death is highly desirable.
Objective: To identify the most predictive patient characteristics of a risk of death within 6 to 12 months as a precursor to developing a user-friendly primary care screening tool.
Design: Retrospective case–control study.
Setting/Subjects: Australian general practice. Cases were patients aged =70 years who died in the previous 5 years. Controls were matched for age and gender. Exclusion criteria were: no available practice records for the 18 months before death (cases) and data collection (controls); no corroborated evidence of death.
Measurements: Supportive and Palliative Care Indicators Tool (SPICT) indicators of deterioration in medical records.
Results: There were 215 deaths and 267 controls. The most predictive patient characteristics of a risk of death within 6 to 12 months are: deteriorating performance status, weight loss, persistent symptoms, request for palliative care or treatment withdrawal, impaired activities of daily living, falls ± fractured hip, neurological deterioration, advanced lung disease, and estimated glomerular filtration rate <30 mL/min/1.73 m2 with deteriorating health. Our predictive model has a sensitivity and specificity of 67% and 87%, respectively, with a predictive accuracy of 78%.
Conclusions: This model predicts risk of death within 6 to 12 months with acceptable reliability in a general practice setting and has the potential to be incorporated into clinical practice and electronic records.