Background Heterogeneity in terms of data collection is thought to limit comparability and pooling of data in Rheumatoid Arthritis (RA) clinical cohorts and registers across Europe. Defining such standards should be informed by which data items are currently collected, how they are defined and collected, and their perceived utility.
Objectives To identify overlaps, discrepancies, and perceived usefulness of the currently collected data in existing RA clinical cohorts and registries.
Methods A survey was developed and sent to 29 European RA registers and clinical cohorts. The survey requested information on: IF specific data items are collected, HOW and with what Frequency they are collected, how often data are MISSING, and if these items are regarded as USEFUL for research purposes.
Results 25 cohorts/registers with different inclusion criteria and definition of RA completed the survey, totaling 189,043 patients (range: 130 to 55,000). 6/25 cohorts (24%) were established before year 2000, 60% were multi-center collections. The figure displays frequency of collected data items. There were differences in HOW items were defined (Composite Scores (CoSc): 80%DAScrp/DAS4items, 40%SDAI/CDAI, 8%Ritchie Index; Joint Counts (JC): 71%28JC, 5%66JC; pain 25%general pain, 35%pain related to arthritis) and HOW they were collected (JC: 54% total JC, 46% individual joints; pain: 85% VAS, 20% numeric rating scale; Comorbidities: 57%patient self reported, 29%standardized questionnaire; 10%free text; 33%records). Items with higher proportions of missing information were EQ-5D, and fatigue. DMARD history and serology were collected mainly at baseline only, whereas items of disease activity and severity such as JC, CoSc, pain, and HAQ were collected at almost every follow up visit.
CoSc (92%), disease duration (92%), biologic DMARDs (91%) and HAQ (91%) were most frequently used for research. Among all items collected, the responders ranked disease activity, treatment, and JC as the most essential, compared to extra-articular manifestations, smoking and imaging which had the lowest perceived utility. At the same time, smoking, imaging, co-morbidities and adverse event data were the top-ranked data items responders felt was missing from their data collection.
Conclusions Even though common core items such as disease activity, treatment and comorbidities have been identified, the mode of data collection is heterogeneous. To facilitate future data sharing and combined analytical approaches, data collection across Europe needs to be better harmonized.
Disclosure of Interest : H. Radner Grant/research support: IMI BT-Cure Project, K. Hyrich: None declared, W. Dixon: None declared, J. Askling Grant/research support: IMI BT-Cure Project