Development of an item bank to measure factual disease and treatment related knowledge of rheumatoid arthritis patients in the treat to target era.

Abstract

To develop a Disease and treatment associated Knowledge in RA item bank (DataK-RA) based on item response theory. Initial items were developed from a systematic review. Rheumatology professionals identified relevant content trough a RAND modified Delphi scoring procedure and consensus meeting. RA patients provided additional content trough a focus group. Patients and professionals rated readability, feasibility and comprehensiveness of resulting items. Cross-sectional data were collected to evaluate psychometric properties of the items. Data of 473 patients were used for item reduction and calibration. Twenty items were discarded based on corrected item-total point biserial correlation <0.30. Confirmatory factor analysis with weighted least squares estimation on the polychoric correlation matrix suggested good fit for a unidimensional model for the remaining 42 items (CFI 0.97 TLI=0.97, RMSEA=0.02, WRMR=0.97), supporting the proposed scoring procedure. Scores were highly reliable and normally distributed with minimal ceiling (1.8%) and no floor effects. 75% of tested hypotheses about the association of DataK-RA scores with related constructs were supported, indicating good construct validity. DataK-RA is a psychometrically sound item bank. DataK-RA provides health professionals and researchers with a tool to identify and target patients' information needs or to assess effects of educational efforts.