Dementia detection using visual memory performance

Trevor F Hyde, Marquette University

Abstract

Research has demonstrated that testing of long-term visual memory is extraordinarily sensitive to detecting early stage dementia. However, to date no prospective studies have demonstrated adequate diagnostic validity within a representative geriatric population. The current study examined the differential diagnostic utility of a visual memory procedure designed specifically for use with a geriatric population. Participants were drawn from a continuous sample of new referrals for differential dementia diagnosis to the neuropsychology, neurology, and geriatric medicine clinics at the Clement J. Zablocki VA Medical Center in Milwaukee. All participants underwent standard diagnostic examinations, and were classified into three diagnostic categories: neurologically intact clinical controls (NICC; n = 23), Vascular Dementia (VaD; n = 19), and Senile Dementia of the Alzheimer's Type (SDAT; n = 20). Depression and anxiety were examined as independent variables across diagnostic conditions. Nondecisional patients and patients not fitting into one of the three categories listed above were excluded. In addition, a convenience sample of 29 neurologically intact non-psychiatric non-substance abusing geriatric controls was recruited from a continuous sample of geriatric primary care outpatients. A new visual memory instrument, the Visual Learning and Retention Test (VLRT) was added to the standard examination. It differs from available instruments in that it utilizes variable levels of exposure and heterogeneous stimuli with systematically manipulated features that reflect modern understandings of neurocognitive processes. This allows performance comparisons along dimensions that typically contribute only to "noise" variance; i.e., outcome measures on the instrument are designed to reflect performance effects of specific stimulus features and patterns of learning, retention, and errors. The instrument incorporates methodological refinements to reduce floor effects and non-memory related performance demands. Comparisons were made between performance on the VLRT and on the Brief Visuospatial Memory Test-Revised (BVMT-R; Benedict, 1997), an established test of visuoconstructive learning and recall. Results indicated that the VLRT has excellent construct and predictive validity, and is less sensitive to the effects of demographic and psychiatric variables than the BVMT-R. Both the ability to detect dementia and differential diagnostic utility were determined to be comparable to the BVMT-R when used independently. Using the two instruments synergistically produced a dramatic increase in diagnostic classification over the use of either individually.

This paper has been withdrawn.