I'm writing a systematic review and there are both crude and adjusted OR for the primary outcome. Which should be used for further analysis? Or should I mention both?
It all depends on the question you are trying to answer with the systematic review. Adjusted OR (vs crude OR) is adjusting or attempting control between multiple independent variables (confounders). As such, it may well answer your question better.
This is a tricky question, Lowilius Wiyono . While I generally agree with Steven C Y Ng , it is important to note that the known confounders accounted for in adjusted ORs or RRs will vary between studies (which I assume to be observational in your case). The problem with considering only the crude ORs or RRs is that you may carry over the explicit effect of known confounding in your review results. An issue with considering only their adjusted counterparts is you may likely observe increased heterogeneity, especially if the list of covariates controlled for across studies are only barely overlapping. It would be nice if you can acquire the raw data of all included studies (but this is unlikely) or you can ensure, at the very least, that the controlled known confounders are similar between studies (and the baseline characteristics are also similar between them) to justify using the adjusted ORs or RRs alone. Otherwise, sub-group analyses based on differences of adjusted covariates and/or in baseline sample characteristics may be considered. Mentioning both crude and adjusted measures seems to be a good compromise, but this will increase complexity of your review.
I wish to add that the sole purpose of performing a systematic review would be to collate the best supportive evidence published so far and pool the results by doing a meta-analysis if possible. As such, it would therefore be best to select only those studies which would best answer your research question eg selecting only those studies with adjusted ORs. It would be rather complicated if you were to include subgroup or stratified analysis in your systematic review/meta-analysis.
I would prefer Adjusted OR (AOR) and explain about common confounders in the SR narratives.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Writing