Session Type: Poster Session B
Session Time: 9:00AM-11:00AM
Background/Purpose: The Juvenile Arthritis Multidimensional Assessment Report (JAMAR) is a questionnaire developed to comprehensively assess Juvenile Idiopathic Arthritis (JIA) patients. Despite being translated into 54 languages, there is still limited literature about it. The length of the questionnaire could have been influencing its clinical practicality. The purpose of this study is to answer the following questions:
- “Which are the most informative questions?”;
- “How well do the collected data correlate with other clinical variables?”;
- “Are there discrepancies between the perceptions of patients and parents?”;
Methods: We included 71 children with JIA according to ILAR criteria, all of them receiving treatment and we followed them up for a year. JAMAR questionnaires were answered by both children and parents at baseline, 6 and 12 months. Also, a thorough clinical examination was performed in every visit: all the joints were clinically assessed for swelling, tenderness, and limited range of motion, and Juvenile Arthritis Disease Activity Score (JADAS), disease activity state, parents and patients assessment through Visual Analogue Scale (VAS), physician’s VAS, Erythrocyte Sedimentation Rate (ESR) and C-reactive protein (CRP) were recorded. We applied state of the art machine learning methods in order to find the most relevant questions in JAMAR. Additionally, we utilized tensor decomposition to identify relevant patient clusters. Furthermore, we correlated these critical questions with clinical and biological parameters recorded. We have compared the discordance rate between patients vs parents responses in 5 of JAMAR parameters as previously reported (Vanoni F, et al. Pediatr Rheumatol Online J. 2016;14(1):2). We explored the relation between discordance and demographic and clinical variables.
Results: A total of 374 JAMAR questionnaires are analyzed with our Machine Learning algorithms. First, we identify a small group of questions as the most relevant for patients and parents. The identified questions exhibit better correlations with the JADAS scores than the non-relevant ones. Second, 96% of the pairs (child-parent) are discordant for at least one item, but the differences are small and VAS well being is the only score with a statistically significant difference (P < 0.0001). We observe a higher rate of activity in the patients exhibiting discordant evaluations with their parents. In addition, the observation patient-parent agreement in Juvenile Arthritis Functionality Scale (JAFS) is better than Pediatric Rheumatology Quality of Life Scale (PRQL).
Conclusion: In this study, we revisit the JAMAR questionnaire by applying modern data mining techniques in a longitudinal dataset. Our results suggest that a small number of questions in the JAMAR questionnaire provide significant information and correlate well with the JADAS scores. We argue that this reduced set of questions could make the data collection easier by trading off the number of questions for frequency and ease of self-reported data collection.
To cite this abstract in AMA style:Quesada-Masachs H, Faloutsos M, Ghose S, Marsal S, Modesto C, Quesada-Masachs E. A Data Science Evaluation of the Juvenile Arthritis Multidimensional Assessment Report (JAMAR) Questionnaire for Improving Management of JIA Patients [abstract]. Arthritis Rheumatol. 2020; 72 (suppl 10). https://acrabstracts.org/abstract/a-data-science-evaluation-of-the-juvenile-arthritis-multidimensional-assessment-report-jamar-questionnaire-for-improving-management-of-jia-patients/. Accessed November 25, 2020.
« Back to ACR Convergence 2020
ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-data-science-evaluation-of-the-juvenile-arthritis-multidimensional-assessment-report-jamar-questionnaire-for-improving-management-of-jia-patients/