Precision and negative predictive value of links between and PubMed

AMIA Annu Symp Proc. 2012;2012:400-8. Epub 2012 Nov 3.


One of the goals of translational science is to shorten the time from discovery to clinical use. Clinical trial registries were established to increase transparency in completed and ongoing clinical trials, and they support linking trials with resulting publications. We set out to investigate precision and negative predictive value (NPV) of links between ( and PubMed. has been established to increase transparency in clinical trials and the link to PubMed is crucial for supporting a number of important functions, including ascertaining publication bias. We drew a random sample of trials downloaded from and performed manual review of retrieved publications. We characterize two types of links between trials and publications (NCT-link originating from MEDLINE and PMID-link originating from link precision is different based on type (NCT-link: 100%; PMID-link: 63% to 96%). In trials with no linked publication, we were able to find publications 44% of the time (NPV=56%) by searching PubMed. This low NPV shows that there are potentially numerous publications that should have been formally linked with the trials. Our results indicate that existing trial registry and publisher policies may not be fully enforced. We suggest some automated methods for improving link quality.

Publication types

  • Research Support, N.I.H., Intramural

MeSH terms

  • Clinical Trials as Topic*
  • Databases, Factual*
  • Humans
  • National Library of Medicine (U.S.)
  • PubMed*
  • Registries*
  • Translational Research, Biomedical
  • United States