VALIDATION OF THE NOVEL DEMENTIA ASSESSMENT SCALE FOR ALZHEIMER’S DISEASE: A THREE-DIMENSIONAL SCORING ALGORITHM RELEVANT FOR CLINICAL PRACTICE

It is beneficial if a single neuropsychological test can simultaneously evaluate the states of Activities of Daily Living (ADL), cognitive function and Behavioral and Psychological Symptoms of Dementia (BPSD) for the Alzheimer's disease (AD) patients. However, a reasonable algorithm is needed for interpreting these different qualities as a patient state. The novel test consists of 13 items rating the states of ADL, cognitive function and BPSD with 9 levels of the Likert scale where higher scores are better. Raters ask the caregivers about their patient's recent episodes by a semi-structured interview and the method of limit. By a clinical trial with 312 subjects, the construct validity, concurrent validity, test-retest validity and responsiveness of the scale were statistically confirmed. The standard tests for the analyses were MMSE, NPI-D, DAD, CDR SOB and Global CDR. The scores related with ADL, cognitive function and BPSD of patients were plotted on a three-dimensional axis coordinate (See an example figure) and the Euclidian distance from the origin (0, 0, 0) to the score coordinates (ADL, Cognitive function, BPSD) were calculated for evaluations. The Euclidian distances are highly correlated with the scores of MMSE, NPI-D, DAD, CDR sum of box (CDR SOB) and Global CDR of which the correlation coefficients are 0.736, -0.600, 0.717, -0.830 and -0.826, respectively. The specificities and sensitivities to discriminate the severities of AD may be reasonable enough to use practically. For example, the sensitivity and the specificity of probable MCI versus other severities were 0.928 and 0,684, respectively. If the baseline severity was CDR0/0.5, CDR1 or CDR2, the changes of the three dimensional distance between the baseline and 12 weeks were statistically significant.