Skip to content


  • Oral presentation
  • Open Access

Assessing and improving the reliability of meta-analyses of hazard ratios derived from published time-to-event data

  • 1,
  • 1,
  • 1,
  • 2 and
  • 1
Trials201314 (Suppl 1) :O93

  • Published:


  • Public Health
  • Standard Error
  • Systematic Review
  • Primary Outcome
  • Estimation Method


Trial design and conduct is informed by systematic reviews and meta-analyses (MAs). For time-to-event outcomes, they are typically based on hazard ratios (HRs). Ideally, these are calculated from individual patient data (IPD). Alternatively, they can be extracted from published aggregate data (AD) or estimated from other statistics or from Kaplan-Meier (KM) curves. As reporting and the reliability of each estimation method is likely variable, we investigate how HRs from AD and IPD compare.


We used 18 of our published IPD systematic reviews and MAs. For our primary outcome of overall survival, we extracted or estimated logHRs and standard errors (SEs) from eligible trials wherever possible. We compared HRs from AD and IPD at the trial and meta-analysis level, using Bland-Altman methods to assess agreement.


Of 245 unique trials eligible for the meta-analyses, HRs were available from IPD for 199 (81%), and from AD for 155 (63%) including 75 (48%) based on KM curves. Trial logHRs from AD were biased towards benefit compared to IPD (5% on HR scale, p=0.009), and SEs were inflated (+0.23 SDs, p<0.001). Bias was not influenced by estimation method (p=0.84). Similar bias was seen in MA logHRs from AD (3.5% on HR scale, p=0.031), and again inflated SEs (+0.87 SDs, p<0.001). The level of bias varied by meta-analysis.


Although MAs of AD-derived HRs can be biased, their use is widespread and needed when the IPD approach is infeasible. We will present results on causes of these biases and guidance on overcoming them.

Authors’ Affiliations

MRC Clinical Trials Unit Hub for Trials Methodology Research, London, UK
Centre for Reviews and Dissemination, York, UK


© Tierney et al; licensee BioMed Central Ltd. 2013

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Please note that comments may be removed without notice if they are flagged by another user or do not comply with our community guidelines.