Hidden duplicates: 10s or 100s of Indian trials, registered with ClinicalTrials.gov, have not been registered in India, as required by law

Background This study's primary goal was based on the fact that since 15 June 2009 it has been mandatory to register regulatory trials running in India with Clinical Trials Registry-India (CTRI). Were all such trials, registered with ClinicalTrials.gov (CTG) after 2009, that included India as a location, also registered with CTRI? We first had to determine how to correctly identify a trial that was registered in both the registries, but that lacked the relevant secondary ID. Therefore the secondary goal of this study was to identify the best method to do this. Methods We used a control set of 1013 trials that cross-referenced a record in the other registry. We used two algorithms to-in a blinded fashion-identify CTRI matches for the 1013 CTG records. 80% of the predictions were correct. Using the same methodology, we identified matches for the CTG trials without known CTRI matches. We then used a logistic regression model to predict which of these matches were correct. Results (i) 3664 CTG records listed India as a location, but did not list any CTRI ID, and were not identified by any CTRI records either. (ii) The best single field to find a CTRI match for a CTG trial was the title field. (iii) Between 50 and 300 of 581 relevant CTG trials were not registered with CTRI. Conclusions This is the first study to use hidden duplicates to determine that the law on trial registration has been broken (in India). Similar studies need to be done for trials run in other countries.