I have a quick question. I am now doing a meta-analysis on the fasting concentrations of hormones.
Aim: To determine what is the fasting concentrations of hormones?
I have the hormones in the following format:
Studies N Concentration SD
(pM)
S1 9 2.5 0.8
S2 6 2.9 0.3
S3 16 4.1 1.1
S4 12 1.2 0.3
S5 16 0.8 0.2
S6 9 3.5 1.3
But, I have one problem, S5 is a study which consisted of 3 substudies and the author claims that the concentration between three occasions was not significantly different and so the author reported a pooled mean concentration from the 3 substudies, which is now 0.8 (+/- 0.2), instead of providing the mean concentration for substudy1, substudy2 and substudy3. They have measured the same 16 participants repeatedly on three separate days.
So I am wondering how can I incorporate this values into my meta-analysis if I cannot manage to get the data for three individual substudies? The other studies are individual studies.
Thanks!
I think if you just mention it, you can use the pooled concentration. You can also send an e-mail to ask the author for the data you need.
If I understood correctly, I think you may use pool mean concentration since author aggregated all substudies in one. Otherwise contact author to get the best solution.
Possibly, the main concern here is the value for the SD reported for S5, and how it has been calculated. You highlight that "They have measured the same 16 participants repeatedly on three separate days" which, in some modelling frameworks, would imply dependence in the sample values that would need to be taken into account.
Hey I'd recommend you check the Cochrane Handbook for Systematic Reviews. There's a special chapter on pooled data. Helped me a lot for my last meta-analysis.
Generally, it's always a good idea to contact the authors directly, I mostly got a quick reply and most authors are very kind and helpful.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics