Objective: The Delayed Word Recall (DWR) test has been shown to be highly sensitive and specific for Alzheimer’s disease (AD) versus normal controls, but its potential utility in differential diagnosis has not been established. The present study evaluated DWR sensitivity and specificity for AD versus Vascular Dementia (VaD), Fronto-Temporal Dementia (FTD), dementia with Lewy Bodies (DLB), and community dwelling controls.
Design: Cross-sectional evaluation of sensitivity and specificity.
Materials and Methods: Data from patients with established diagnoses (multi-disciplinary Consensus) from our Memory Clinic database were examined. All patients with MMSE<15 were impaired on both DWR free recall (DWRfr) and recognition (DWRrcg) and were not further analyzed. Total remaining patient n = 362. Screened and AGECAT-verified normal controls (n=56) were also evaluated. Results: As in previous studies, normal control specificity was very high (96%). On DWRfr 287/301 Probable AD cases (MMSE mean=21.26, range=15-28) were accurately classified (95% sensitivity), with sensitivity=86% on DWRrcg. Even in 87 AD cases where MMSE>23/30 (mean=24.89, range 24-29), DWRfr correctly classified 79/87 (91% sensitivity). Of 16 FTD cases, 12/16 were accurately classified on DWRfr (specificity=75%) and 14/16 on DWRrcg (specificity= 88%). A majority of VaD cases were misclassified on DWRfr (VaD DWRfr specificity= 35%, 13/37), and recognition memory was better maintained but overlapped substantially with AD (VaD DWRrcg specificity=65%, 24/37). There were only 8 DLB cases, with 5/8 correctly classified on DWRfr (specificity=63%), and only 3/8 on DWRrcg (specificity=38%). Interestingly, 3/8 DLBs were normal range on free recall but impaired on recognition, likely reflecting attentional deficits on recognition memory testing. This was quite uncommon in the other dementias.
Conclusion: Normal DWRfr performance is highly unlikely in AD, even in mild cases. Memory performance (DWR) was generally normal in FTD. In VaD, while recall and recognition deficits were less pervasive than in AD there was substantial overlap, particularly for free recall. In our small DLB sample, the attentional deficit characteristic of DLB was evident, with recognition memory poorer than free recall in 3/8 cases. The nature of memory deficits on DWR appears to be potentially useful in identifying AD, FTD and DLB, but limited in the differential diagnosis of VaD.
Back to Detection of dementia in clinical and community settings
Back to Symposia
Back to The IPA European Regional Meeting (1-4 April 2003) of IPA