Dear Christopher,

If your scales cannot be converted and unified, a good choice will be to use standardized mean difference (e.g. Hedges' g). Mathematically it will always makes sense, however you have to be sure that your outcome scales represent some specific dictinct phenomenon (here: CDRS - depression scale vs MMSE - cognitive impairment?).

Best,

Jakub

Dear Christopher

It is very nice to answer your query indeed

All papers for which metanalysis is done usually contain many variables including the complications we are trying to make metanalysis for them.

Each complication should be taken individually and separate metanalysis is done for it.

In this way, we can do metanalysis for each complication. Otherwise there is no group singled metanalysis.

Good luck in your endeavor.

Wisam

Dear Jacob

Indeed I understood your question. Not only this but I have peer reviewed such articles indeed for Saudi Medical Journal. It was a very nice metanalysis about the use of Circulage operation in twin pregnancy and the results showed no benefit in preventing preterm labor!

At that time one of the studies among the other 5 was not provided with the number of infants who born after or before- I don't really remember - 34 weeks of gestation.

So I suggested taking the whole group as preterm labor and why I should do ramification which further complicates their results?

They accepted my suggestion and really it was a very nice experience I have done in my life.

Hope that you got the Idea

Actually, that is why Cochrane now tries to unify the protocol of papers which will be subject to metanalysis. I have already registered on their web site

Best Wishes

Good Luck in your endeavor

Wisam Akram

From Baghdad

Dear Christopher,

If I understand your question correctly, I hope my answer will help you.

1. If the main outcome on the use of the scale is preset, it should be followed.

2. Consider summarizing the studies using the same scale. For example: Four studies were included (A, B, C, D), of which A, B, D used MMSE, and A, C used CDRS. You can consider using the mean difference to combine the MMSE outcome and CDRS outcome separately.

3. If there are multiple scales that meet the analysis conditions. A more meaningful scale can be selected from a clinical perspective or the core outcome. By asking clinical experts, consulting the core outcome opinions, and combining professional knowledge.

4. If different articles use different outcome indicators (A: MMSE, B: CDRS, C: MMSE, etc.), the standardized mean difference can be calculated.

Best wishe.

Ruowen

Dear Jacub

Indeed your reply indicates you are professional in this field of statistical analysis. however, I have never seen such a type of metanalysis paper. Maybe I should dig more into this field to discover more. thanks indeed for your comprehensive reply.

Wisam