Brief Alerts

Neuropsychological Testing for Dementia

Sources: Stuss DT, et al. Do long tests yield a more accurate diagnosis of dementia than short tests? A comparison of 5 neuropsychological tests. Arch Neurol 1996;53:1033-1039; Meiran N, et al. Diagnosis of dementia. Methods for interpretation of scores of 5 neuropsychological tests. Arch Neurol 1996;53:1043-1054.

Two helpful papers from the same group of authors compare the relative sensitivity, specificity, and diagnostic value of three short tests--the Minimental Status Exam (MMSE), the abbreviated six-item Orientation-Memory-Concentration (OMC), a 10-item Mental Status Questionnaire (MSQ)--and two longer tests, the Dementia Rating Scale (DRS) and the Ottawa Mental Status Examination (OMSE). The latter tests required 2-3 times the time required to apply the MMSE. All five tests were applied, and subsequently repeated at an average of 14.6 months later, to 236 demented persons as well as 47 (17%) unequivocally non-demented ones. Demented patients included largely those suspected of Alzheimer type (DAT) and/or vascular dementia (VaD), although all patients had memory problems. Final diagnosis depended on full evaluation, including neurological, behavioral, laboratory tests, and psychometric tests. Follow-up examinations were obtained at approximately 14.6 months.

Despite the fact that some tests were more detailed and took longer to apply, all tests at final evaluation turned out to be highly correlated and apparently interchangeable in diagnosing dementia. Compared against other tests and the full evaluation program, the Minimental Status Examination showed the highest probability of defining dementia with a specificity of 0.90 and sensitivity of 0.85. Risk factors for poorer scores in all tests included education of eight years or less and increased age. Sex turned out not to be a factor. None of the tests was determined to be useful in distinguishing between DAT and VaD.

In the second of their two papers, the authors provide a large number of statistics supporting their conclusions but not of direct importance to the message. However, they do provide an important point, which is that if one arbitrarily puts an MMSE cut-off point of 23 of 24 as defining possibly demented/not demented, odds ratios indicate that a MMSE of 20 is 98% accurate for dementia. By contrast, a score of 28 can assuredly be defined as normal.