Falsified data from apixaban trials find their way into meta-analyses

  • Garmendia CA & al.
  • JAMA Intern Med
  • 4 Mar 2019

  • International Clinical Digest
Access to the full content of this site is available only to registered healthcare professionals. Access to the full content of this site is available only to registered healthcare professionals.

Takeaway

  • Almost half of meta-analyses of apixaban include falsified data from the ARISTOTLE trials.
  • In a third, the inclusion yields a “considerable change” in outcome.

Why this matters

  • The ARISTOTLE clinical trial data were falsified in some cases, yet the published results have been included in several meta-analyses that are used for patient care and policy decisions, say the authors.
  • The falsified data in ARISTOTLE were submitted from a clinical site in China; according to an analysis, without these data, the significant mortality benefit vanishes, yet was still reported in related literature. 

Key results

  • The authors identified data from ARISTOTLE in 22 meta-analyses, published 2012-2017.
  • 46% would have had different findings with omission of ARISTOTLE data.
  • 32% of estimates of apixaban effect would change:
    • 97% of these would no longer favor apixaban.
    • 3% would favor control.
  • Subgroup analyses also were affected: 44% would have different conclusions with omission of ARISTOTLE data.

Study design

  • The ARISTOTLE trial evaluated apixaban, and in a review of publications with falsified data, ARISTOTLE had the most.
  • These authors followed Cochrane rules for systematic review to assess meta-analyses that included data from at least 1 ARISTOTLE publication.
  • Funding: Federal Employee Education and Assistance Fund/National Treasury Employee Union; NIH. 

Limitations

  • The limitations related to the included meta-analyses.

Please confirm your acceptance

To gain full access to GPnotebook please confirm:

By submitting here you confirm that you have accepted Terms of Use and Privacy Policy of GPnotebook.

Submit