FRI0648 Directly comparing latent functional ability in adolescents with jia using the chaq and haq: an item response theory analysis. (12th June 2018)

• Record Type: Journal Article
• Authors: Shoop-Worrall, S. J. W.; Voshaar, M. O.; van de Laar, M.; Thomson, W.; Hyrich, K. L.; McDonagh, J. E.; Verstappen, S. M. M.
• Is Part Of: Annals of the rheumatic diseases. Volume 77(2018)Supplement 2
• Page Start: 845
• Page End: 846
• Publication Date: 2018-06-12
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2018-eular.3670
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Measuring and comparing functional ability in adolescents with JIA is challenging due to the use of multiple questionnaires, including the proxy-completed Childhood Health Assessment Questionnaire (P-CHAQ), the adolescent version (A-CHAQ) and the Health Assessment Questionnaire (HAQ). Item response theory (IRT) allows items on multiple questionnaires to be linked to an underlying continuous variable. This allows scores to be corrected for characteristics of the administered items, thus making them comparable between different questionnaires. Recently, a common reporting metric for functional ability was developed in a combined dataset of 16386 patients with various inflammatory rheumatic diseases, including 1029 paediatric patients with JIA. Objectives: i) To cross-validate the item response models using three functional ability questionnaires in adolescents with JIA. ii) To assess agreement between overall function scores obtained from the different questionnaires. Methods: Adolescents aged 11 to 17 with JIA were enrolled to a UK, multicentre inception cohort, the Childhood Arthritis Prospective Study (CAPS). In a sub-study, adolescents were asked to complete the A-CHAQ and HAQ and their guardians the P-CHAQ. Adolescents were selected if at least two of the questionnaires had been completed simultaneously within the first year following diagnosis. Fit of the item response models was assessed by comparing model expected item scores with those observed in CAPS for each item (i.e. residuals). An item response model was considered to adequately describe item response behaviour of CAPS patients when the mean of the residuals was <±5%. Agreement of overall questionnaire IRT scores were then compared using limits of agreement (2SD) and intra-class correlations. Results: Of 303 adolescents, 61% were female and median age at JIA diagnosis was 13 years (range 11 to 17). Raw HAQ scores consistently fell below both CHAQ scores. IRT model fit in the CAPS population was good, with 1% of item residuals >±5%. When modelled using IRT, the mean differences in overall scores approximated zero, with narrow limits of agreement, at 15 (PCHAQ vs ACHAQ), 12 (PCHAQ vs HAQ) and 10 (ACHAQ vs HAQ), on a 0–100 scale. High intra-class correlations between overall scores were evident (range 0.83 to 0.90). There was therefore high agreement between IRT-modelled scores obtained for different questionnaires. A scale characteristic curve (figure 1) illustrates the relationship between the expected scores for CHAQ and HAQ questionnaires with the latent functional ability variable. Conclusions: IRT models for functional ability previously developed in a mixed population of adult and paediatric patients with inflammatory arthritides are applicable to adolescents with JIA in CAPS. IRT scores across CHAQ and HAQ measures had high agreement. IRT scores for functional ability can therefore be used in clinical practice and research to directly compare scores on the CHAQ and HAQ. This will be important as adolescents transfer from paediatric to adult rheumatology. Disclosure of Interest: None declared