Olfactory dysfunction in general, and impaired odor identification in particular, have been reported in Alzheimer's disease (AD). Olfactory testing may be a useful diagnostic aid for AD, but the types of odor most commonly affected need to be identified. This study aimed to determine pattern and types of odor affected in AD with the goal of improving clinical applicability. 54 outpatients with mild to moderate AD and 40 age and gender-matched non-demented controls (NDC) were tested using British version of University of Pennsylvania Smell Identification Test (UPSIT; Sensonics, Inc., Haddon Heights, NJ) and data analyzed to identify an optimal subset of UPSIT to best differentiate AD patients from controls. AD subjects had significantly lower UPSIT total scores than NDC. Random Forest with backward elimination identified 12 UPSIT items which accurately differentiated AD patients compared to controls (sensitivity, 0.89 and specificity, 0.83, positive predictive value of 0.889, and negative predictive value of 0.833). The 12 smell items found to be most affected in AD subjects reflects important attributes such as safety and food, known to be affected in people with AD and that has the potential to impair activities of daily living. The 12 items of British UPSIT most affected in AD subjects provides a potential brief scale for early detection of AD in clinical settings. Independent replication is needed to validate these findings.
Keywords: Alzheimer’s disease; odor; olfaction; pattern of deficits; smell identification.