I am performing a meta-analysis to pool the medical costs of an intervention. The studies reported the mean difference with 95% confidence interval which I extracted them and entered to the CMA software as you can see in the attached file. The problem is that the software could not calculate SMDs for 3 studies (rows with red numbers in attached file) and revealed an error by CIs. Could anyone help me how to manage the problem?
Thanks a lot
Benyamin
Your problem is that the means for those three studies should lie halfway between their upper and lower limits. However, they do not. This tells me that the mean, lower limit, and/or upper limit value(s) is (are) incorrect for those three studies. CMA will tolerate a little measurement error (i.e., a default setting of 10%). This setting can be adjusted under the "symmetry for confidence intervals" menu selection.
Is it possible that the means/CIs for these three studies were obtained after data transformation and then back transforming the results?
Dear Muhammad M Hammami
In two studies, the model for cost used the ‘regression method’ applied to log-transformed values in order to account for the skewness of the underlying data.
Is it the reason of the problem?
Thanks for your attention.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts