Predicting lewy body pathology in a community-based sample with clinical diagnosis of Alzheimer's disease

Accurate antemortem prediction of Lewy body pathology in patients with dementia is problematic. This study generates a model that better predicts Lewy body pathology in community-based patients with clinical Alzheimer's disease. Lewy body pathology was detected in 80 of 152 participants (52.6%) with an initial diagnosis of probable Alzheimer's disease. In a stepwise logistic regression model, female gender, lower education, being married, bradykinesia, hallucinations, and absence of irritability predicted the greatest likelihood of Lewy body pathology. The predictive model correctly diagnosed Lewy body pathology with an estimated sensitivity of 75%, specificity of 68%, and accuracy of 72%; the area under the receiver operating characteristic curve was 0.75. In a community-based autopsy sample, this predictive model confirmed parkinsonism and hallucinations as important predictors of Lewy body pathology in patients with clinical Alzheimer's disease. The model also identified other demographic and clinical characteristics that might enhance the prediction of Lewy body pathology.