Setting
The University of Western Ontario Dementia Study, which is a registry of cases of dementia seen for secondary and tertiary assessment in a university memory disorders clinic with extensive follow-up data and histopathological confirmation of clinical diagnoses.

Patients
One hundred twenty-nine patients with definite or probable AD, 12 patients with definite or probable VaD, and 36 patients with definite or probable mixed dementia.

Methods
Patients were grouped as having an early, moderate, or advanced stage of disease according to the extended scale for dementia (ESD). The ESD was subdivided into cognitive domains, and the domain scores were compared for each stage of disease by diagnostic category with the use of a 2-way analysis of variance with repeated measures.

Results
As expected, the scores in all domains decreased significantly with increasing severity. There was a significant difference in the decline in memory among the diagnostic groups (P=.02) that was mostly attributable to the difference between AD and mixed dementia (P=.03), with the difference between AD and VaD only approaching significance (P=.06). There was a similar finding for praxis. The interaction between diagnosis (AD and VaD) and severity was significant only for memory (P=.02), showing a less severe memory deficit at onset but a proportionately more rapid progression in VaD and arithmetic ability (AD and mixed dementia [P=.03]).

Conclusions
Alzheimer disease, VaD, and mixed dementia evolve similarly as assessed using cognitive domains obtained by subdivision of the ESD in a patient population derived from a memory clinic and by analyzing VaD as a single entity. Only memory impairment evolves differently between AD and VaD, with this depending on the severity. Memory is more severely impaired in the early stage of AD; however, with increasing severity of dementia, memory impairment in VaD accelerates and catches up with AD at the level of moderate impairment. The differences between AD and mixed dementia are greater than those between mixed dementia and VaD, suggesting an important role for the ischemic component of mixed dementia. Simple neuropsychological tools (eg, the ESD) may be incapable of distinguishing between AD and VaD, and more focused instruments may be required. Inherent bias in case selection may prevent extrapolation of these results to VaD in general, but the neuropsychological criteria for VaD may need to vary, depending on the severity.