Recall of three words after five minutes: its relationship to performance on neuropsychological memory tests.

This paper examines the statistical relationship between a memory screening test (recall of three words after 5 minutes, 3W5M) and an extensive battery of neuropsychological memory tests. Subjects were 142 consecutive referrals for neuropsychological examination and were given 3W5M in addition to measures of memory, language, and intelligence. Performance on 3W5M was categorized as either "impaired" or "unimpaired" in a manner consistent with judgement used in clinical settings. Linear discriminant function analysis was used to predict 3W5M performance from neuropsychological test scores. Results indicated that 3W5M was a relatively good screening measure, but that its sensitivity and utility as a measure of memory was significantly influenced by the nature of the neurological disorder. Significantly higher diagnostic accuracy rates were obtained in patient groups with clinically obvious memory impairment (e.g., dementia, pure amnesia) than in groups with less obvious memory impairment (e.g., closed head injury, psychiatric disorders). Implications for clinical memory testing, and for the relationship between screening and battery approaches to assessment, are discussed.