Is it possible to merge effect sizes expressed as hazard ratio and odds ratio in the same dataset? Of course, the construct underlying the outcome of interest is the same, but in some articles was explored with logistic regression and in other articles was explored with Cox regression.
Dear Luca, It Is better to convert all in the same metrics. Or, altenatively, if you have enough data in the papers you can obtain HR from p-values and observed events. You can find bere a nice paper on this : https://trialsjournal.biomedcentral.com/articles/10.1186/1745-6215-8-16
I think formalists would be hesitant to merge ORs and HRs. Yes, they are both relative measures of effect, but ORs are estimations at a specific time point while HRs are averages of time-dependent effects over the follow-up. Logistic and cox regression modelling are based on different assumptions and merging the outcomes will introduce uncertainty about how the results should be interpreted.
One way to handle it is to present results of pooled ORs and HRs separately and then reason about their similarity/differences in the discussion section in your manuscript.
I don't merge them, different concepts/ assumptions.
Present them separately. Suhail A. Doi
They are two different entities and I would not merge them but present them separately.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis