Presentation (CTI)

OBJECTIVES: Improving methods related to deduplication of clinical trial registry records in systematic reviews has the potential to enhance efficiency. The Bramer method, an established multi-step EndNote-based approach, is commonly used in deduplication but remains time-intensive. Automated deduplication tools offer speed but often lack transparency and accuracy. This study evaluates a new deduplication method that leverages unique trial registry numbers aiming to improve accuracy and efficiency. The performance of this new deduplication method is compared to the Bramer method, an automated tool (Deduplicator; Bond University), and gold standard manual deduplication using records from the WHO International Clinical Trials Registry Platform (ICTRP) and ClinicalTrials.gov (CTG).
METHODS: The new method extracts unique trial registry numbers and consolidates them into a single searchable field for deduplication. Two independent screeners applied the new method, the Bramer method, the automated Bond Deduplicator, and manual deduplication method to datasets from ICTRP and CTG. Processing time and error rates (i.e., missed or false duplicates) were used to compare performance.
RESULTS: The new method identified 280 duplicates out of 872 records, compared to 48 duplicates by the automated Bond Deduplicator method and 27 duplicates by the Bramer method. The new method was completed in 11 minutes, substantially faster than the Bramer method (60 minutes) but slower than the automated method (1 minute). Accuracy will be assessed by estimating error rates against the gold standard manual deduplication. Initial results suggest the new method may offer enhanced detection of duplicates and efficiency, addressing challenges such as incomplete field matching and detection failures.
CONCLUSIONS: Preliminary findings suggest the new method can streamline systematic review workflows in high-throughput environments. By improving detection of duplicates while maintaining practical efficiency, this method represents a potential advancement over traditional techniques, enhancing the robustness of trial registry data processing.