Background/Purpose:  Systemic autoimmune diseases (SADs) are heterogeneous conditions with peculiar characteristics that share several clinical features. It is suspected that SADs share similar molecular abnormalities, however, until now, the cross-sectional study of SADs has been hampered by the difficulty to establish common a clinical groundwork. A challenge of such studies is the generation of a CRF capable of capturing similarities and diversities without being redundant with the diagnosis themselves or, conversely, too detailed. We present a framework used within the Innovative Medicines Initiative Joint Undertaking (IMI JU) project PRECISESADS to develop a CRF that balances compactness and granularity to allow the identification of diseases clusters based on molecular features. A similar decisional process can also be applied when building the data structure for registries and the CRF content we developed can be used to align registries or studies data focussing on the transversal analysis of SADs.

Methods:  The following steps were performed: 1) Bioinformatics gave insight into the analysis plan and suggested the key rules for data collection. Briefly, unsupervised clustering analysis is planned; missing and redundant data are to be avoided as much as possible; yes/no answers are preferred to open entry fields; data should focus on elements necessary to implement the analysis and limit the number of extraneous (“nice to know”) elements. 2) A working group of experts on SADs (RA, SLE, SSc, SjS, UCTD/MCTD/APS) was established. A first broad set of transversal items to the different diseases (n = 130) was created and divided in 8 domains (constitutional symptoms, gastrointestinal, vascular, heart and lung, nervous system, skin and glands, muscle-skeletal, therapy). 3) The items were ranked and reviewed via a Delphi technique and the top ranking items were selected after convergence was reached. 4) The core items were discussed by all the members of the consortium to gain consensus among the stakeholders, and suggestions were gathered. 5) A final set of items was created, digitalized and pilot tested. 6) The final CRF was released along with explicit data definitions.

Results:  Convergence among experts was obtained after 3 tiers and a core set of 28 items was generated. This set was enriched by additional baseline demographic and enrolment data, comorbidities, essential laboratory tests and some disease-specific clinical data. The final CRF proved to be flexible and easy to compile. The average rate of missing data (median, IQR) was 1.9% (0.83 – 7.95). Missing data were: 1.66% (1 – 2.2) in the core set and 0.83% (0.66 – 1.91) for comorbidities. Higher missing rates were observed for lab results: 10% (1.28 – 17.32) and for the additional non-transversal data: 17.03% (14.57 – 24.25).

Conclusion: We describe a seamless procedure to build a core data set transversal to different SADs. This set may be used with few modifications or integrations as data standards for studies or registries that plan to analyze different SADs at once. The generalisation of this core data set may allow a better comparison across studies and contribute to evolving medical knowledge in the field of SADs. www.precisesads.eu

« Back to 2016 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/establishing-a-case-report-form-crf-for-systemic-autoimmune-diseases-studies/