Reliability and Validity of Autism Diagnostic Interview-Revised, Japanese Version

Kenji J. Tsuchiya, Kaori Matsumoto, Atsuko Yagi, Naoko Inada, Miho Kuroda, Eiko Inokuchi, Tomonori Koyama, Yoko Kamio, Masatsugu Tsujii, Saeko Sakai, Ikuko Mohri, Masako Taniike, Ryoichiro Iwanaga, Kei Ogasahara, Taishi Miyachi, Shunji Nakajima, Iori Tani, Masafumi Ohnishi, Masahiko Inoue, Kazuyo Nomura, Taku Hagiwara, Tokio Uchiyama, Hironobu Ichikawa, Shuji Kobayashi, Ken Miyamoto, Kazuhiko Nakamura, Katsuaki Suzuki, Norio Mori, Nori Takei

To examine the inter-rater reliability of Autism Diagnostic Interview-Revised, Japanese Version (ADI-R-JV), the authors recruited 51 individuals aged 3-19 years, interviewed by two independent raters. Subsequently, to assess the discriminant and diagnostic validity of ADI-R-JV, the authors investigated 317 individuals aged 2-19 years, who were divided into three diagnostic groups as follows: autistic disorder (AD), pervasive developmental disorder not otherwise specified, and other psychiatric diagnosis or no diagnosis, according to the consensus clinical diagnosis. As regards inter-rater reliability, intraclass correlation coefficients of greater than 0.80 were obtained for all three domains of ADI-R-JV. As regards discriminant validity, the mean scores of the three domains was significantly higher in individuals with AD than in those of other diagnostic groups. As regards diagnostic validity, sensitivity and specificity for correctly diagnosing AD were 0.92 and 0.89, respectively, but sensitivity was 0.55 for individuals younger than 5 years. Specificity was consistently high regardless of age and intelligence. ADI-R-JV was shown to be a reliable tool, and has sufficient discriminant validity and satisfactory diagnostic validity for correctly diagnosing AD, although the diagnostic validity appeared to be compromised with respect to the diagnosis of younger individuals.