Abstract
OBJECTIVE: Cognitive screens to diagnose mild cognitive impairment (MCI) or dementia require supervision and cannot be easily administered in primary care. Here, we validated the Self-Administered Gerontocognitive Examination (SAGE), investigating the alternate version equivalence, the convergent validity using neuropsychological tests, and its diagnostic accuracy. PATIENTS: Thirty-two MCI patients and 34 with dementia were recruited from a memory clinic in the Netherlands, and 69 healthy controls over the age of 50. METHODS: The 4 alternate versions of the SAGE were compared. Receiver operating characteristic (ROC) analyses were performed, comparing the controls to the MCI and dementia groups. Associations between SAGE scores and standard neuropsychological tests were examined. RESULTS: No performance differences were found between the alternate versions. Performance differences were found on the SAGE between the 3 groups, with fair to good areas under the curve. A cutoff score of <18 had the best diagnostic accuracy for controls versus dementia, <20 for controls versus MCI and <19 for controls versus cognitively impaired. SAGE scores correlated with standard neuropsychological tests. DISCUSSION: The SAGE is a valid tool for distinguishing cognitively unimpaired individuals from people with dementia or MCI.