Evaluating the Completeness of ClinicalTrials.gov.

Background: To date, although studies have been conducted to assess compliance with listing clinical trial information, to our knowledge there is nothing in the literature examining the completion and accuracy of clinical trial site information on ClinicalTrials.gov .

Methods: We compared clinical trial information originating from ClinicalTrials.gov to a widely subscribed and well-established commercial clinical trial database, Informa Pharma Intelligence's Trialtrove, as Trialtrove includes information from ClinicalTrials.gov among its more than 30,000 sources. We assessed breast cancer, non-small cell lung cancer, type 2 diabetes mellitus, and pain clinical trials submitted to ClinicalTrials.gov . We compared the number of trials associated with each disease indication for each database, and we conducted a head-to-head comparison of certain data fields of clinical trial information found in both databases for clinical trials with identical National Clinical Trial (NCT) numbers.

Results: As of January 17, 2017, Trialtrove captured 31% more clinical trials (10,786 trials) in the selected disease indications than did ClinicalTrials.gov (7,419 trials) using identical Medical Subject Headings (MeSH) terms for breast cancer, non-small cell lung cancer, type 2 diabetes mellitus, and pain. Clinical trial site information was identical in 48% of clinical trials, while country information was identical in 82% of clinical trials, and patient enrollment figures identical for 86% of clinical trials. Clinical trial status, Phase, and start year differed across the 2 datasets.

Conclusion: This study provides a baseline to compare the completeness of trial information required by the NIH and HHS with respect to reporting practices on ClinicalTrials.gov . While one cannot determine which database is more accurate: ClinicalTrials.gov or Trialtrove, wide variation exists in clinical trial site and country information for trials with identical NCT numbers suggesting that caution should be used when relying solely on ClinicalTrials.gov to assess the clinical trial landscape.