Rasch analysis of items from two self-report measures

Abstract

The article by Stout et al. provides an excellent discussion of the International Classification of Functioning, Disability and Health (ICF) model, as it attempts to clarify use of the terms capacity, capability, and performance. These are important concepts when comparing results in studies. The authors present these concepts with a broad brush, indicating that they were designed before the ICF model. In evaluating articles such as this one, readers must understand the differences among the concepts of capacity, capability, and performance. Each functional question is structured so that the responses relate to one concept or another. While it is true that the Pediatric Outcome Data Collection Instrument (PODCI) is not purely about capability or performance, most of the questions of the PODCI ascertain the child’s actual performance of a task within the past week. The authors do not share with the reader which version of the Functional Assessment (FAQ-22) was used, but indicate that it is rated on a five-level Likert difficulty scale and available retrospectively. The original version of the FAQ-22 required a yes or no for the tasks the patient is able to do. A recent version (available online at http://www.gillettechildrens.org/ fileupload/GFAQsurveypreview2.pdf) has responses for: (1) the child’s ability to perform the following activities in addition to walking; and (2) if performed as part of routine season-appropriate play. The reference to the FAQ-22 notes, ‘The five-level Likert response scale used for the 22 skill items was ‘easy’, ‘a little hard’, ‘very hard’, ‘can’t do at all’, and ‘too young for activity’. Those who responded ‘too young for activity’ were coded as ‘can’t do at all’ for this analysis.’ 2 One can only assume the same referent version is used in the current analysis. The questions concern the child’s ability to do tasks. Whether considered capacity questions or capability questions, they clearly are not performance questions. Personal factors (including sex, age, coping styles, attitude, and motivation), in addition to environmental factors and comorbid conditions, can have an impact upon actual performance of tasks in daily life. Two articles referenced in the Stout et al. article show clearly that capability exceeds performance, one showing average 17% greater values for capability, the other showing 90% correlation between the capability and performance measures, with performance lower. The resulting ‘upper-range’ coverage by the FAQ-22 noted by Stout et al. may be partially due to type of question posed. There was an a priori decision to drop numerous questions from the PODCI, especially from the sports and physical function subscale, due to the different question structure. Several of the current authors published another paper which included the dropped questions, using confirmatory factor analysis and item response theory in children with cerebral palsy, results showing that the sports and physical function subscale covers upper-range mobility, which Stout et al. indicate is covered by the FAQ-22, though their discussion section concedes upper-range function coverage would improve with inclusion of these questions. In an article comparing the PODCI with the School Function Assessment (primarily a measure of performance), three PODCI questions were found to have the greatest correlation with performance on School Function Assessment. The two questions from the transfers and basic mobility and sports and physical function subscales having the greatest correlation with the School Function Assessment are dropped in the current study by Stout et al. It would be useful to compare in this population the full PODCI mobility scales of transfers and basic mobility and sports and physical function with the current combined tool, using the performance questions from the FAQ-22. I would agree with the comments in the discussion that the information detailed is specific enough to be of practical use and may be very helpful in clinical practice. But there is not