The utility of naming tests in the diagnosis of Alzheimer's disease

Objectives: To evaluate the sensitivity of the CAMCOG naming subscale to an emia in Alzheimer's disease (AD) patients identified as anemic on the Bosto n Naming Test (BNT); and to compare the sensitivities of BNT, category (sem antic) fluency (CF), and episodic memory (Delayed Word Recall, DWR) in dete cting AD. Method: Data from 140 probable AD patients (NINCDS/ADRDA) were analysed. An emic AD patients were identified (BNT) and the sensitivity to anemia of the CAMCOG naming subscale was calculated. Sensitivity (to AD) and specificity of the BNT, CF and DWR were compared in a subgroup of 60 patients age-matc hed to controls. Results: On BNT 81% of ADs were classified as anemic (87% specificity). Of these anemic ADs, 23 scored 6/6 on CAMCOG naming (sensitivity = 80%), and s ensitivity fell to 30% using a 5/6 cut-off. Correlation between BNT and CAM COG naming yielded Kendall's tau = 0.48 (p<0.0001). Compared to age-matched controls BNT correctly classified 70% of ADs and 87% of controls (overall accuracy = 77%). On CF 68% of ADs and 83% of controls were correctly classi fied (overall accuracy = 72%). BNT and CF were moderately correlated (r = 0 .49, p<0.0001). DWR correctly classified 93% ADs and 98% controls (overall accuracy = 95%). Conclusions: Given its brevity, we recommend the inclusion of a split-half form of the BNT in screening test batteries for the detection of AD. The ob served pattern of cognitive impairment is consistent with current models in which semantic memory deficits occur in AD only when the neuropathology ex tends beyond the entorhinal/hippocampal region to the temporal neocortex.