E014 Migration of the large complex dataset of the UK Juvenile Dermatomyositis Cohort Biomarker Study: challenges and lessons

Abstract

Abstract Background/Aims In rare disease research, longitudinal prospective cohort studies are important tools to generate invaluable real-world evidence to inform treatment choices and understand outcomes. The Juvenile Dermatomyositis Cohort Biomarker Study & Repository (JDCBS) is a multi-centre, prospective longitudinal cohort study with linked biobank, established in 2000, with 17 active UK-Centres contributing data and samples. It supported over 60 projects generating new evidence, treatments and biomarkers in the field of juvenile myositis. The database in use until March 2023 used aged technology which did not sufficiently meet current needs of the study. REDCap is secure web application for hosting online databases, able to support online longitudinal data capture for multi-centre research studies. The aim of this project was to build and deliver a future-proofed environment and migrate the large JDCBS dataset into the REDCap platform. Methods Clinical, laboratory and outcome data at clinic visits were previously captured through a web interface which loaded data into an MS-Access back-end database. In this project, an internationally agreed set of data elements for use in research in juvenile myositis was adopted and a new data entry platform built, using REDCap and a Trusted Research Environment hosted at Great Ormond Street Digital Research Environment. A multidisciplinary team was formed to deliver migration, including clinicians, study investigators, systems analysts, and health informatics data scientists. Careful comparison of fields in the existing database to those in the internationally agreed dataset before delivery of the new REDCap database and migration of the legacy data took place. Results Key steps to deliver the REDCap interface and migrate the data included: 1. Agreement of data elements to be included, dropped or added, compared to the legacy dataset; 2. Design of study forms and fields within REDCap; 3. Testing new interface; 4. Script design and testing for migration; 5. Data cleaning of legacy dataset; 6. Training 15 collaborators and REDCap accounts management for 63 collaborators; 6. Migration of legacy dataset and quality control of migrated data; 7. Post-migration data checking and feedback on use of the new platform; 8. Summary reports of process, data quality and audit trail of project. In total, 670 patient records and 7431 clinical forms, with data from 338 data fields, were successfully migrated to the new REDCap database. A median value of 11 clinical forms (IQR: 6-18), and 3,718 data points (IQR: 2,028-6084) were migrated per patient. Conclusion Migration of a multi-dimensional longstanding study data set is a complex multi-step task requiring multi-disciplinary engagement of stakeholders. Key elements to achieve success include early agreement of data fields for inclusion, strict definitions of fields, regular engagement with database users, providing robust training, enabling feedback, and acceptance that research datasets should strive to be parsimonious in fields collected. Disclosure D. Cancemi: None. R. Hamilton: None. S. Varakliotis: None. H.D. Nguyen: None. N. Sabire: None. N. Martin: None. L. McCann: None. C. Papadopoulou: None. L. Wedderburn: Other; LW has non-renumerated collaboration with Novartis Inc, relevant to this work.