Is it meaningful to combine relative risks and hazard ratio data in meta-analysis? Trial data I have extracted takes the form of either 2x2 data (from which the RR is easily calculated) or HR with associated 95%CIs. I'm not interested in the time-to-event aspect of the HR metric for the meta-analysis and wondered whether it was meaningful to pool HR and RR as both representing ratios of a dichotomous outcome.

This differs from existing questions on CrossValidated in that others do not specifically address the combination of RR and HR

  • $\begingroup$ That is the correct approach! $\endgroup$ – Michael R. Chernick Mar 8 '17 at 0:25