I have a 5 study dataset with 2 IRRs, 2 ORs and 1 HR, all with CIs given. Can I do a log transformation and then conduct a meta analysis. What is a plausible way to conduct meta analysis in this case

  • 1
    $\begingroup$ It is hard to compare the three, which is one reason to only use individual subject-level meta analysis where you can recompute whatever index you want. $\endgroup$ Commented Oct 21, 2021 at 11:30
  • $\begingroup$ You can refer to this earlier post for OR to RR conversion (and viceversa): stats.stackexchange.com/questions/131416/… HR are roughly an estimator of RR, so you might consider them equivalent However, if follow-up time varies, probably the most robust approach is to compute IRR $\endgroup$ Commented Oct 22, 2021 at 10:44


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.