N01 Using Item Response Theory to improve the efficiency of the Crohn and Colitis Knowledge score for patients with Inflammatory Bowel Disease

Abstract

Abstract Background Patient knowledge varies widely in inflammatory bowel disease (IBD), but knowledge of the disease and its management is essential for patients with chronic disorders and can improve health-related quality of life, coping ability, self-care skills, and adherence and can positively impact several outcomes including disease activity and direct and indirect costs. To assess patients’ knowledge is available the Crohn’s and Colitis Knowledge Score (CCKNOW), but can be biased by subjects’ abilities, attitudes, or other latent traits not considered in its validation process. The purpose of this study was to reassess the index developed for measuring IBD patients’ knowledge using the Item Response Theory (IRT). Methods IBD patients with a recent (<18 months) diagnosis, stable therapy, and no previous IBD-related surgery have been asked to answer the 30 items described in the validation of the CCKNOW questionnaire. Measures of between-item relationships (scalability) and unidimensionality (single major trait) were estimated prior to applying a 2-PL logistic IRT model across all 30 items. To evaluate a questionnaire with improved item information, low discriminatory items were removed by order of discrimination value, with the lowest discriminating item<1.7 removed before the 2-PL model was re-estimated with the remaining items. The process was then repeated twice to remove poorly discriminating and inefficient scaling items. The difficulty parameters have also been estimated in the 2-PL logistic IRT model. Cronbach’s alpha (α) was used as a guide to measuring whole-scale reliability. Results 49 patients (17 females, 21 Crohn’s Disease) were enrolled, with a median (IQR) age of 37 (29-47) years, a median disease duration of 334 (212-426) days and a median number of outpatients visit performed of 1 (1-3). Only 6 subjects involved in the study were graduated; the median level of health-related quality of life (S-IBDQ) was 47,7 (30,3-60,9). About 50% of subjects in both groups were in disease remission. Based on discrimination ability, item number 4, 10, 11, 22, 29, and 30 could be excluded because demonstrated a discrimination ability below 1.7. The Cronbach’s alpha for the 24 items questionnaire indicates a suitable internal consistency (α = 0.84). Based on difficulty parameters, the most difficult items were the 25th and 14th, and the easiest were the 19th and 21st. Conclusion This is the first study systematically to interrogate a commonly used index for patients’ IBD knowledge, using robust psychometric and statistical methodologies. IRT is a valid and robust psychometric methodology, which may be used to analyse patient questionnaires. The revised CCKNOW questionnaire provides optimum trait information.