Boston Naming Test in Alzheimer's disease.

The 60-item Boston Naming Test (BNT) was administered to 55 subjects: 15 mildly-to-moderately demented patients meeting NINCDS-ADRDA criteria for "probable" Alzheimer's disease (AD), 15 age-equivalent normal control (NC) subjects, and--for purposes of validation--25 additional subjects with other forms of dementia (OD). A cutting score of 51 correctly classified 80% of AD patients and 86% of NC subjects. To facilitate rapid screening of confrontation-naming performance in these populations, three 30-item shortened versions of the BNT were constructed. Even and Odd Versions were equivalent for AD, NC, and OD subjects; high correlations between these two and the 60-item BNT permit easy extrapolation to a total BNT score. A new Empirical Version, derived from performance of our AD and NC reference groups, maintained most of the intergroup discrimination of the full BNT.