[Binswanger's disease or multi-infarct dementia? Diagnostic keys in vascular dementia]

To develop a model that assists in the classification of patients with vascular dementia in two of its most relevant subtypes: multi-infarct dementia (cortical involvement) and Binswanger's disease (subcortical involvement).

Retrospective analysis of 50 consecutive patients with Binswanger's disease and 50 patients with multi-infarct dementia studied throughout a 12-year period (1986-1998) in a tertiary university hospital. The statistical model was obtained by multivariant analysis (logit model) and its accuracy was estimated by means of ROC curve analysis.

The variables included in the regression model were: arterial hypertension (p < 0.03), cardioembolic disease (p < 0.05), and presence of cortical (p < 0.01) and lacunar (p < 0.01) infarctions in CAT. The specificity of the model was 90%, the sensitivity near 60%, and the positive likelihood ratio 5.8. The area under the ROC curve was 0.80 (0.73-0.87).

Despite the availability of consensus diagnostic criteria for vascular dementia, their use is very commonly restricted to the research field. We have developed a classification model for patients with vascular dementia in two relevant subgroups: multi-infarct dementia and Binswanger's disease. This model can contribute to make identification of these patients in the daily clinical practice easier.