Estimating hazard ratios using Parmar's methods

1 - 2.30pm


To provide reviewers with the knowledge and skills to use Parmar's methods to estimate hazard ratios if these are not explicitly presented in trial reports.


Many reviews e.g. of treatments for cancer, specify survival as the primary outcome of interest. The best statistic for comparing survival in two arms of a trial is the hazard ratio, but this is often not presented in trial reports. Nevertheless, it can often be estimated from other data reported, using methods described by Parmar et al

This workshop will focus on how to estimate the hazard ratio using these methods. We will present the methods for estimating hazard ratios from summary statistics reported in trial publications in a non-technical way, describing a step-by-step approach from reading trial reports to entering data into RevMan. The workshop will include presentations and hands-on practical experience of data extraction, manipulation and analysis. Participants will: extract statistical summary measures from published reports; extract and manipulate data from published survival curves; use these data to estimate a log hazard ratio and its variance; learn how to enter these into RevMan.

Target audience

Reviewers and students


Lesley Smith; Andrew Bryant

Lesley Smith

Andrew Bryant