Background/Purpose: Collaborative research is compromised by heterogeneity of data collection in observational rheumatoid arthritis (RA) databases. Therefore a EULAR taskforce has been convened to develop a minimum core dataset (MCD) of data items (i.e. Ôwhat to collectÕ) and instruments for data collection (i.e. Ôhow to collectÕ) to 1) harmonize future data collection 2) act as a common data model to which existing databases can be mapped 3) serve as a template for standardized data collection for RA research in routine clinical practice.

Methods: The task force comprised a study steering committee, a task force working group and a pan-European expert panel. The project involved a multi-step process (Figure): 1) a hierarchical literature review to identify data items and instruments of existing RA cohorts and registers 2) an online survey to capture information on perceived importance of extracted items and instruments for possible inclusion 3) two face-to-face (F2F) meetings of the working group with discussion and voting on content (items) and structure (instruments) of the MCD. The voting of the F2F meetings were confirmed and consolidated by a ratification survey and work performed by the steering group between the two F2F meetings.

Results: Published articles from 67 different European registers and cohorts were included for data extraction. The number of patients recruited in each register ranged from 130 to more than 50,000. A total of 40 different items and 125 instruments were identified in literature; 7 items felt to be missing were added by the steering group. A total of 90 experts from 28 different European countries, including patients (18%), health professionals (18%), physicians (55%) and researchers or other experts (10%) participated in the online survey. 27/47 (57%) items were regarded to be important for inclusion in a MCD by >80% of responders. At the first F2F meeting 22/47 items were voted to be INCLUDED, 24/47 to be EXCLUDED in a MCD; for 2 items no consensus was reached. Ratification survey and second F2F meeting revealed consensus to include 21/47 items and their instruments (Table). Remaining work in the task force pertains to instruments for two items (ÔglucocorticoidsÕ and ÔcomorbiditiesÕ).

Conclusion: Based on the multistep process, a first draft of a MCD was developed which has to be tested for feasibility in clinical settings and applicability to answer important research questions. Figure: Flowchart of the multistep process of the project Table: Content of minimum core dataset