Background: Early detection of dementia will be important for implementation of disease-modifying treatments in the near future. We aimed to investigate the diagnostic validity and reliability of the Japanese version of the revised Addenbrooke's Cognitive Examination (ACE-R J) for identifying mild cognitive impairment (MCI) and dementia.

Methods: We translated and adapted the original ACE-R for use with a Japanese population. Standard tests for evaluating cognitive decline and dementing disorders were applied. A total of 242 subjects (controls = 73, MCI = 39, dementia = 130) participated in this study.

Results: The optimal cut-off scores of ACE-R J for detecting MCI and dementia were 88/89 (sensitivity 0.87, specificity 0.92) and 82/83 (sensitivity 0.99, specificity 0.99) respectively. ACE-R J was superior to the Mini-Mental State Examination in the detection of MCI (area under the curve (AUC): 0.952 vs. 0.868), while the accuracy of the two instruments did not differ significantly in identifying dementia (AUC: 0.999 vs. 0.993). The inter-rater reliability (ICC = 0.999), test-retest reliability (ICC = 0.883), and internal consistency (Cronbach's α = 0.903) of ACE-R J were excellent.

Conclusion: ACE-R J proved to be an accurate cognitive instrument for detecting MCI and mild dementia. Further neuropsychological evaluation is required for the differential diagnosis of dementia subtypes.