Exploration of Diagnostic Validity of Autism Diagnostic Observation Schedule (ADOS) in Korean Population
Objectives: ADOS has been widely used as one of the gold standard instruments for the diagnosis of ASD. However, research on the use of the ADOS and its applicability in individuals with ASD across non-English speaking populations is limited.
Methods: We used pooled data of ADOS from an ASD genetic study, clinical trials of social skills trainings, research on the development of early screening instruments, and clinical records of Seoul National University Bundang Hospital referred to diagnostic evaluation from 2008 to 2017. The subjects were composed of individuals with ASD, their unaffected siblings, children with other developmental delay referred for diagnostic evaluation, and typically developing children volunteered via community channels. Either ADOS or module-T of ADOS-2 were administered and scored by professionals whose research reliability had been established. Autism Diagnostic Interview–Revised (ADI-R) was also administrated. Other measurements for ASD symptoms were administrated as well, including Social Responsiveness Scale (SRS), Social Communication Questionnaire (SCQ), and Korean version of the Childhood Autism Rating Scale (K-CARS). Clinical best estimate diagnosis was established by two board-certified psychiatrists based on DSM-IV-TR and DSM-5TM criteria. We classified the Best Estimate Clinical Diagnosis into ASD (including pervasive developmental disorder and Asperger’s syndrome) and non-ASD. Autism and ASD in ADOS classification and mild-moderate concern and moderate-severe concern in module-T of ADOS-2 were considered as ASD in the ADOS Diagnosis. Statistical analyses were performed to compare those two diagnoses using SPSS 22.0 and t-test and Cohen’s kappa were used to compare the two groups.
Results: Data from total 1,248 subjects (age 90.29±62.04 months, 912 males, 336 females) were collected (Module-T, n=86; Module-1, n=385; Module-2 n=260; Module-3 n=285; Module-4, n=232). The sensitivity and specificity of ADOS showed sufficient degrees (Module-T=100%, 94.55%; Module-1=100%, 89.36%; Module-2=97.83%, 97.40%, Module-3=98.44%, 96.82%;, Module-4=99.29%, 92.39%) across all modules. Positive Predictive Value (PPV) was 97.37 (91.18 to 98.90) and Negative Predictive Value (NPV) was 98.31 (94.94 to 100), depending on modules. The agreement between all the modules of ADOS and clinical diagnosis were at excellent levels (k=.948; Module-T, k=.926; Module-1, k=.937; Module-2, k=.945; Module-3, k=.950; Module-4, k=.927) and other measurements’ were at good to excellent levels (Cohen's kappa of ADI-R, K-CARS, SCQ and SRS with ADI-R =.844, .810, .625, .656, respectively). Individual item analyses showed significant differences in scores for all algorithm items on module 1 to 4 between ASD and non-ASD groups (p<.05). In module T, 'D2-Mannerism,' and 'D5-Repetitive Behavior' items were not statistically significant, falling between two groups.
Conclusions: The Korean version of ADOS showed high level of diagnostic validity in all modules. This study suggests that ADOS might be a valuable diagnostic instrument for individuals with ASD across countries with different languages and cultural backgrounds. Further analysis is required with larger samples to examine the validity of individual items of Module-T.