Meta-analysis for Pre vs Post (Which SD to use)

[azizfar]

Hello everyone,
I have extracted data from longitudinal studies that have collected data on pre and post time points. All studies have same time period.

So I have got
N_pre
Mean_Pre
SD_Pre
N_post
Mean_Post
SD_Post

In order to compute the standardised or unstandardised Mean differences, I need to use the denominator SD. What I know is I need the SD of mean difference. Unfortunately this is not mentioned available for many studies. Also it may be computed based on (r=correlation coefficient if it is provided, or other statistics like t etc.)

But I have seen some papers that ignore the above rule and just use SD_Pre or SD_Pooled: (SD_Pre and SD_Post). I am so confused, are we allowed to do that?

Can anyone please guide.

Thank you.

Kind regards

Aziz

 

[Kseroks]

I'm dealing with a similar issue. I'm currently trying to get clinicians to help me with correlation coefficients for the parameters I'm investigating. However this is proving difficult. I'm interested in what others have to say about the subject.

 

[azizfar]

I sent an email to 23 authors and so far received 3 emails.
Unfortunately. It is not easy to get reply from authors who published this data 10-15 years ago.

 

[Kseroks]

I received 1 reply, though at least he sent original data from which I could calculate a correlation coefficient. I would like to rely on more than one source though. I may have a contact who can provide additional material from past studies, but he's of course really busy. You are also right that authors of published meta-analyses often don't explain how exactly they calculated their numbers, I wonder why..