Epistemonikos Database of Randomized Trials: methods and preliminary results of a comprehensive, living database of trials
Epistemonikos Database of Systematic Reviews (EDSR) is the largest database of systematic reviews in health. It combines a systematic approach, large-scale human collaboration, and multiple technologies.
Considering the relevance of randomized trials for health decision-making, we are applying a similar methodology to develop a comprehensive, open database of trials.
Objectives: To describe the methods used and the number of trials identified in the Epistemonikos Database of Randomized Trials (EDRT).
Methods: To develop and maintain the database, we screen multiple electronic databases, trial registries, and preprint servers on a regular basis. Applying different technologies, including artificial intelligence and other machine-based approaches, we select potentially relevant trials, which are then validated by a network of collaborators. To complement the electronic searches, randomized trials identified in systematic reviews are indexed in EDRT. All records are deduplicated and studied by automated and manual methods. To test the comprehensiveness of the database, we will generate a sample based on the relative recall method, which uses a set of primary studies from published systematic reviews as a reference standard. The sample will be composed of all the systematic reviews identified in the EDSR with a publication date during May 2023. To assess performance, we will determine if the trials within the sample are in the database when systematic reviews were detected. Trials included in the database before detection will be defined as contained. Trials added after detection will be defined as not contained, because missing studies may have been added after the manual detection of the systematic review’s list of studies, which is part of the EDSR search strategy.
Results: Up to March 2023, we have captured more than 730,000 trials, which will be freely available in a single place in August 2023 through a simple and an advanced search interface. The updated results, including the estimated comprehensiveness of the database, will be presented during the Colloquium.
Conclusions: EDRT intends to provide a comprehensive, reliable, free-access database of randomised trials relevant for health. Patient, public, and/or healthcare consumer involvement: none.