SAT0730-HPR Development of psychometrically equivalent short forms to measure disease and treatment associated knowledge in rheumatoid arthritis: application of item response theory (IRT)

Background Patient education can be used to support and enable people with Rheumatoid Arthritis (RA) to optimise health and wellbeing. It has been recommended as an integral part in management of RA in order to undertake self-management activities or to adhere to treatments. Disease-specific knowledge can be measured with a Patient Knowledge Questionnaire (PKQ). Because PKQs in RA are outdated, de Jonge et al. developed the Disease and Treatment Associated Knowledge in Rheumatoid Arthritis (DataK-RA) item bank and provided preliminary evidence to support its construct validity. It was developed with input from patients and rheumatology experts. DataK-RA contains 42 multiple choice items with 2–4 response alternatives per item and was calibrated using the two parameter item response model for dichotomous responses. IRT scores are corrected for item characteristics, which allows scores to be compared between measures that include different items. IRT models also provide detailed information about the precision of scores at different levels of knowledge. Various methods are available that can help select optimal items to be administered to patients, given certain criteria. Objectives The objective of this study was to develop two DataK-RA short forms using linear optimal test design. Methods The open source excel add in “solver” was used to program a linear optimisation algorithm to develop two short forms. The algorithm was instructed to optimise precision (i.e. reliability) of the scores for both short forms, subject to the constraints that: 1) each item could only be included in one short form 2) each short form should include 15±1 items, 3) reliability for each short form should be >0.70 for all patients who are within 1 SD of the mean of knowledge scores, and 4) scores on each short form should be similarly precise, defined as maximum allowable difference in information of 0.15. Results Two short forms were derived from the DataK-RA item bank that satisfied all content constraints. The short forms include respectively 15 and 16 unique items. Reliabilities across different score levels ranged from 0.71–0.80 for both short forms, and the maximum difference in information between the short forms was 0.13. Conclusions DataK-RA is a new and promising tool that can be used by healthcare providers to measure disease and treatment related knowledge in patients with RA. The short forms can be used in pre/posttest intervention studies in which disease related knowledge is one of the outcomes Because each short form includes unique items and IRT scores are adjusted for item characteristics, the application of these short forms will allow users of DataK-RA to avoid learning effects commonly associated with using the same items at two occasions. Furthermore, the equal and high reliabilities of both forms ensure that the observed score distributions for both versions will have similar variances. Disclosure of Interest None declared