Memory impairment in Alzheimer's disease: replication and extension of the delayed word recall (DWR) test

Abstract

AbstractObjective: The original DWR test, which measured delayed free recall, was reported to have high predictiveaccuracy in discriminating Alzheimer's disease (AD) patients from control subjects (overall accuracy of 95%).Comparison of differential performance in free recall and recognition of the same material may be of clinical interest. In the present study a delayed recognition component was added to the DWR test and the utility of both measures in discriminating AD patients from control subjects was evaluated.Procedure: This extended version of the DWR test was administered to 66 patients meeting NINCDS/ADRDA criteria for probable AD and 42 control subjects.Results: In a comparison between 42 of these patients (MMSE range 18–29), and 42 age matched healthy controls, both the delayed free recall and recognition measures were highly accurate in distinguishing patients from controls. The free recall measure achieved 98% sensitivity, specificity and overall accuracy, while the recognition measure yielded 98% sensitivity, 95% specificity, and 96% overall accuracy. The recognition performance of all 66 patients, ranging in severity from very mild to severe (MMSE range 11–29), was also evaluated to determine its relationship, if any, to measures of global cognitive impairment. While therecognition measure correlated poorly with MMSE and CAMCOG there was a modest but significant correlation with the CAMCOG memory subscale.Conclusions: In this study of highly selected AD patients both the free recall and recognition measures were sensitive and specific indicators of AD compared to control subjects. Recognition performance appears to be more closely related to degree of amnesia than to degree of global cognitive impairment.