Rasch Analysis of the Pediatric Outcomes Data Collection Instrument in 720 Patients With Cerebral Palsy

Abstract

Background: The Pediatric Outcomes Data Collection Instrument (PODCI) was originally developed to carry out a functional assessment of children and adolescents (including patients with cerebral palsy), focusing on musculoskeletal health. Validated questionnaires are important for assessing the functional outcome of cerebral palsy, and are meant to have unidimensionality, proper item gap, no ceiling and floor effects, and no item redundancy. The advances in health measurements have led to the application of Rasch analysis to assess questionnaires. This study evaluated PODCI in patients with cerebral palsy using Rasch analyses. Methods: The study included a total of 720 patients with Gross Motor Function Classification System level I to III, 192 with unilateral involvement and 528 with bilateral involvement. Rasch analysis was performed to obtain information on (1) the information weight fit statistic to assess the unidimensionality and redundancy of the items in each domain; (2) average item calibration to detect the item separation; and (3) item map to evaluate the ceiling and floor effects. Results: The PODCI worked best in the sports/physical function domain. In information weight fit statistics, there were 4 items violating unidimensionality, which included "putting on his/her coat" in transfer/basic mobility and "getting together and do things with friends" in sports/physical function. There were 4 items with item redundancy. Inadequate item separation was observed in the transfer/basic mobility domain. A ceiling effect was found in all domains, except for the sports/physical function. Conclusions: The sports/physical function domain in PODCI generally satisfies the requirements of Rasch item response theory and is an appropriate measure of the function in cerebral palsy. Although some individual items do not fit well, the PODCI can be improved by eliminating the redundant items and by adding more difficult tasks to fill in the gaps. Level of Evidence: Level II.