Psychometric properties and efficiency of the computerized adaptive testing system for measuring self-care performance in taiwanese children with developmental disabilities

Objective: To investigate the psychometric properties and efficiency of the computerized adaptive testing system for measuring self-care performance (CAT-SC). The psychometric properties included intrarater and interrater reliabilities, concurrent validity, minimal detectable change, minimal important difference, and responsiveness. Design: Criterion standard study. Setting: A teaching hospital. Participants: A convenience sample of 60 caregivers of children with developmental disabilities was recruited at the initial assessment, and 95% of the children (N=57) were followed-up at 3 and 6 months. Interventions: Not applicable. Main Outcome Measures: The CAT-SC and the Pediatric Evaluation of Disability Inventory Chinese version were used at the initial assessment. We assessed the CAT-SC and asked caregivers to rate children's changes in self-care performance with a separate question rated on a 15-point Likert-type scale at the 3- and 6-month follow-ups. Results: The CAT-SC had excellent intrarater and interrater reliabilities (intraclass coefficient=0.99 and 0.92, respectively), high concurrent validity with the Pediatric Evaluation of Disability Inventory (r=0.91-0.92), and trivial to small responsiveness at 3- and 6-month follow-ups (effect size=0.02 and 0.12, standardized response mean=0.08 and 0.33). The minimal detectable change of intrarater reliabilities and the minimal important difference at the first follow-up were 0.22 and 0.17 logits, respectively. The administration time of the CAT-SC was about 5 minutes. Conclusions: The results of our study validated the sound psychometric properties and good efficiency of the CAT-SC. Moreover, the values of the minimal detectable change and minimal smallest change can be used as a reference for clinicians and caregivers to interpret children's progress.