Objectives: Alzheimer's disease (AD) is associated with variable but shortened life expectancy. Knowing expected survival time may empower people with AD and their families, but clinicians currently have limited predictive information. Our objective was to identify determinants of survival in a cohort of people with mild to moderate AD and test these on a separate validation cohort.
Methods: We followed a representative cohort of 158 people for 42 months and identified independent determinants of shorter survival. From these we constructed the Survival in Alzheimer's Model (SAM), and tested this on a validation cohort.
Results: Baseline constructional apraxia, age and gait apraxia independently predicted shorter survival: about half of those scoring 2 on the SAM survived > or =3.5 years compared to 85% of those scoring 0.
Conclusions: The SAM is a potentially useful tool for clinicians who previously had very limited specific and quantitative prognostic information to tell AD patients and carers. This model predicted survival from age, constructional and gait apraxia. This may be because constructional and gait apraxia are relatively free from educational or cultural bias and thus are better indicators of severe neuropathology than global cognitive tests. Alternatively, they may increase falls or immobility, or represent disease sub-types with worse prognoses. Oncology services are able to inform patients and their families about 5-year survival rates. This step towards such provision in AD is new and of potential importance to patients and their carers.