CDRJ（臨床認知症評価尺度日本版）の臨床応用における 注意点 : Japanese Alzheimer’s Disease Neuroimaging Initiative （JADNI）研究の体験をもとに.

The 432 subjects were tested for the CDR-J by the examiner of the J-ADNI. Our study reexamined the CDR-J data to clarify the errors. The results revealed that there were the following 5 types of errors. 1. Error of insufficient information in the answers for the episodic memory task was observed in 135 subjects (31.3%) among the 432 subjects. The examiner of the J-ADNI failed to follow the test instructions which caused this type of errors. 2. Illegible hand-writing in the answers for the episodic memory task was observed in 24 subjects (5.6 %) among the 432 subjects. This error type could be avoided it the examiner exercised precautions. 3. Repetition error in the name and address memory task was observed in 19 subjects (4.4%) among the 432 subjects. The examiner neglected the test in structions, which caused this type of errors. 4. A blank space error refers no answer without comment. A blank space error of one or more test items among the 79 test items of the CDR-J was observed in 188 subjects (43.5%) among the 432 subjects. Some examiners of the J-ADNI do not believe that a blank space error on a test item makes it impossible to rate the result of the CDR-J, which caused a blank error. 5. The 432 subjects had been rated by the examiner of the J-ADNI before our study which revealed that for 297 subjects (68.8%) among the 432 subjects, there were one or more errors in the 4 above types of errors and a part of the CDR-J was inappropriately performed. Accordingly, these 297 subjects could not to be rated. Subtracting the 297 subjects from the 432 subjects gives a total of the 135 subjects. These 135 subjects were again rated by us and a rating error was observed in 69 subjects (51.1%) . The rating errors of the 69 subjects were caused by byexaminer of the J-ADNI since he failed to follow the rating criteria in the CDR-J Rating Table. A total of 366 subjects (84.7%) among the 432 subjects had one or more errors among the 5 types of errors, since the 297 subjects plus the 69 subjects with the rating error is the 366 subjects. The responses of those 366 subjects were considered invalid data, while the responses of the remaining 66 subjects were considered valid data without the 5 types of errors. Points of note regarding the CDR-J: 1. One or more of the 5 types of errors were observed in 366 subjects (84.7%) among the 432 subjects of the J-ADNI project. Researchers using the CDR in research or clinical practice must endeavor honestly to prevent those 5 types of errors. When the examiner follows instructions, and understands the premise and the rating criteria of the CDR, the 5 types of errors will be avoided. 2. Scientific papers using the CDR show only the rating score of the CDR, but not the details of the results for the CDR. In the J-ADNI project, the 297 subjects who had one or more of the first 4 error types, and inappropriate performance of a part of the CDR-J were rated by the examiner of the J-ADNI, though they should not be rated. Subtracting the 297 subjects from the 432 subjects gives the 135 subjects. The 135 subjects were rated by us and a rating error was observed in 69 subjects (51.1%) among the 135 subjects included. Authors of scientific papers using the CDR should rate the CDR after he ensuring that the 4 first error types are absent and the CDR was correctly performed. They must also ascertain that there were no rating errors. [ABSTRACT FROM AUTHOR]