Validation of the Korean Childhood Autism Rating Scale-2: Clinical Study Notes and Study Results

Study Background and Objectives of the K-CARS2 Validation

• Study Title: Validation of the Korean Childhood Autism Rating Scale-2.
• Authors: SoHyun Lee, PhD (Department of Special Education, Ewha Womans University); Sun-Ah Yoon, PhD (Department of Psychiatry, Seoul National University Hospital); Min-Sup Shin, PhD (Department of Psychiatry, Seoul National University Hospital).
• Core Objective: The study aimed to evaluate the psychometric properties (reliability and validity) of the Korean version of the Childhood Autism Rating Scale, Second Edition (K-CARS2).
• Instrumentation Scope: The validation covered both the Standard Version (K-CARS2-ST) and the High Functioning Version (K-CARS2-HF).
• Context of Autism Spectrum Disorder (ASD):     * ASD is a neurodevelopmental disorder affecting early development.     * Global prevalence reported by the US Centers for Disease Control and Prevention is $1$ out of $54$ children.     * In South Korea, prevalence was found to be $2.64\%$ in a study of elementary students in a single city, and remains above $2.0\%$ even with DSM-5 criteria adjustments.
• Need for the K-CARS2 in South Korea:     * Timely and accurate diagnosis is essential for favorable prognosis and alleviating adverse developmental effects.     * Previous instruments in South Korea were often used immediately after translation without verifying reliability or validity.     * Existing instruments like the original K-CARS (introduced in 1996) tend to under-identify High-Functioning Autism (HFA).     * In one study, $21$ out of $23$ children diagnosed with ASD by the Autism Diagnostic Observation Schedule (ADOS) did not meet the autism cutoff on the original CARS.     * The K-CARS2-HF was urgently needed to address the "spectrum" of support needs, particularly for children in general education who often go undiagnosed due to high academic achievement and cultural lack of awareness.

Methodology and Participant Demographics

• Total Sample Size: $N = 340$ participants.
• Recruitment Details:     * Participants were recruited from 2016 to 2017 ($2$ years) from pediatric clinics, child/adolescent psychiatry departments, social welfare centers, special schools, and parent advocacy groups in Seoul and Gyeonggi Province.     * Inclusion criteria: Age above $2$ years; official diagnosis by board-certified psychiatrists based on DSM-IV or DSM-5; middle-class socioeconomic status; Korean race and ethnicity.
• Group Breakdown:     * K-CARS2-ST Group: $n = 145$ (ages $2-36$; $M = 11.77$; $78.0\%$ male). Assignation requirement: IQ $< 80$.     * K-CARS2-HF Group: $n = 65$ (ages $6-36$; $M = 11.95$; $81.5\%$ male). Assignation requirement: IQ $\geq 80$.     * Control Group: $n = 130$ total.         * Clinical Control Group: $n = 62$ (diagnoses including ADHD ($45.2\%$), Communication Disorders ($29.0\%$), Emotional Disorders ($16.2\%$), Learning Disorders ($4.8\%$), Developmental Delay, and Cerebral Palsy).         * Typical Development Group: $n = 68$ (no clinical diagnosis).
• Sample Power Calculation: Using G-power, an effect size of $0.5$, confidence level of $0.05$, and power of $0.80$ indicated a minimum of $70$ cases per group (accounting for $10.0\%$ dropout).

Assessment Instrumentation and Procedures

• K-CARS2 Structure: 15 items evaluated on a seven-point scale ($1$ to $4$ points). Includes the Questionnaires for Parents or Caregivers (QPC) to supplement clinical observation.
• Intelligence Tests:     * Korean Leiter International Performance Scale-Revised (K-Leiter-R): For ages $2-7$ (non-verbal).     * Korean Wechsler Intelligence Scale for Children-Fourth Edition (K-WISC-IV): For ages $8-15$.     * Korean Wechsler Adult Intelligence Scale-Fourth Edition (K-WAISE-IV): For ages $\geq 16$.
• Validation Tools:     * Ewha Checklist for Autistic Children (E-CLAC): $25$ pathology items used for concurrent validity (Cronbach’s $\alpha = .88$).     * Korean Gilliam Autism Rating Scale-2 (K-GARS-2): $42$ items including stereotyped behavior, communication, and interaction.
• Rater Training: Seven examiners (one psychologist, six special educators) with master's degrees or higher. They underwent $6$ hours of intensive training and regular reliability checks by professional clinicians.
• Translation Process: The first author translated the items, followed by back-translation by a US-based professor. The modified version was reviewed and finalized by the original US authors and publishers.

Reliable Psychometric Outcomes

• Internal Consistency:     * K-CARS2-ST: Cronbach’s $\alpha = .77$.     * K-CARS2-HF: Cronbach’s $\alpha = .96$.     * Corrected item-total correlations ranged from $0.70$ to $0.95$ for ST and $0.66$ to $0.92$ for HF.
• Inter-rater Reliability: Calculated using Intraclass Correlation (ICC). Reliability ranged from $0.57$ to $0.86$ for ST and $0.57$ to $0.91$ for HF.
• Test-Retest Reliability: Conducted with a $3$-week interval ($n = 30$ for ST, $n = 15$ for HF).     * K-CARS2-ST: ICC = $.98$ (range $0.89-1.00$).     * K-CARS2-HF: ICC = $.99$ (range $0.80-1.00$).

Validity and Factor Analysis Results

• Concurrent Validity: Significant positive correlations found with existing tools.     * Correlation with E-CLAC: K-CARS2-ST ($r = .60$); K-CARS2-HF ($r = .61$).     * Correlation with K-GARS-2: K-CARS2-ST ($r = .47$); K-CARS2-HF ($r = .64$).
• Construct Validity (Factor Structure):     * K-CARS2-ST (Four-factor structure): Explains $43.90\%$ of variance.         * Factor 1: Social relations and flexibility issues.         * Factor 2: Restricted interest and stereotyped behavior.         * Factor 3: Sensory issues.         * Factor 4: Anxiety issues.     * K-CARS2-HF (Three-factor structure): Explains $63.83\%$ of variance.         * Factor 1: Social interaction and communication.         * Factor 2: Sensory and emotional regulation issues.         * Factor 3: Cognitive ability issues.
• Discriminant Validity (Accuracy Rates):     * Cut-off scores used: $30$ points for ST and $26.5$ points for HF ($T$-score equivalent of $42$).     * Classification accuracy against typical development group: $100\%$ for ST; $98.5\%$ for HF.     * Sensitivity and Specificity for ST: Both $100$ and $100$ ($96.0$ specific value cited elsewhere).     * Sensitivity and Specificity for HF against clinical control: $92.1\%$ and $92.3\%$.     * Area under the ROC curve: $.98$ ($95\%\,CI = .97-0.99$) for both groups.

Discussion and Clinical Implications

• Comparisons to Global Data: The internal consistency values (ST: $.77$, HF: $.96$) align with or exceed findings in Spanish ($.90$) and Japanese ($.91-0.93$) versions.
• Factor Structure Significance: Unlike some previous studies, this study successfully validated items related to social communication. The focus on sensory specificity and intellectual response consistency distinguishes the K-CARS2 and aligns it with DSM-5 criteria.
• Practical Utility in Korea:     * Addresses the lack of standardized tools for HFA identification.     * Critical for early detection and service eligibility.     * Relevant for social issues like mandatory military service requirements in South Korea, where accurate HFA diagnosis is essential for appropriate placement or exemption.
• Study Limitations:     * Did not re-verify if participants diagnosed via DSM-IV would still meet DSM-5 criteria.     * Lack of inter-rater agreement data between different classes of raters (professionals vs. parents).     * Regional bias: Samples were only from Seoul and Gyeonggi Province, limiting nationwide generalization.     * Absence of comparison data with ADOS-2 or ADI-R due to their lack of standardization in South Korea.
• Final Conclusion: K-CARS2 is a reliable and valid instrument for the clinical diagnosis of ASD in South Korea across the full functioning spectrum.