Dear colleagues, I have a question if you could please answer. During meta analysis using CMA software, I found out that a value of risk ratio and 95% CI was not converting to odds ration; log values were not symmetrical.
I got analysis done again from the author and he has sent me values without rounding off. But still I am getting error that log values are not symmetrical. Please, advise.
I don't know if I fully understand your question but I'll try. I believe the issue is a symmetry issue for your upper and lower limits on your 95% confidence interval. I am attaching a PowerPoint file that illustrates the importance of the symmetry setting under the Computational Options drop-down menu in CMA. (I realize that the CMA file is not set up exactly as what I believe you are trying to do but I think it is probably close enough.) Note that as the symmetry factors get closer to one, the more errors you will get when calculating a study's variance.
Dear Dr. Warren! Thank you ever so much for your useful input. However, I do not understand what value to enter in "symmetry for confidence intervals" dialog box?
I don't know if there is a right or wrong number to enter. I normally don't work with odds or risk ratios and I don't normally enter upper and lower limits for a 95% CI. Intuitively, I would think the symmetry factor should be as close to one as possible. I think the default setting is 1.10. I suggest that you play with the factor to identify the studies with the most asymmetrical 95% CI. They might have a round-off, data entry, or calculation error. I am sorry that I can't be of more help.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis