percentages or adjusted odds ratio on revman help

[glassman365]

Hi
Hi, I'm new to meta-analyses and revman and would really appreciate your help!

I have data from an article I wish to add to revman.

They have given an adjusted odds ratio, percentages of the event, and the total number of participants in each group.

I thought that if I had the adjusted odds ratio I could just add it to revman. However, I cannot see this option. I seem to be able to just add events and totals.

But I cannot add events as the numbers as I just have percentages. Also, they will not be adjusted for confounders.

I'm not sure what to do here. tried searching online but could not find anything.

Any advice would be greatly received

 

[hlsmith]

What data did you already have entered (counts)? Now you want to add adjusted odds ratios?

 

[glassman365]

What data did you already have entered (counts)? Now you want to add adjusted odds ratios?

Hi

Other studies have provided means and sds, so that was straightforward enough.

This study only has those percentages and an adjusted odds ratio. Not separate odds or counts.

Not sure what I should do

 

[glassman365]

24% improved in the treatment group out of 145 participants in this group
1.9% improved in the control group out of 89 participants in this group

Adjusted odds ratio given is 15.7

 

[hlsmith]

Well you can either convert to a common unit or not, and if one controls for covariates and the others do not that poses more issues. If you can find a common effect unit, you can include it once and than not at all, but note this in the write-up.

 

[glassman365]

Would there be no way for me to just input the adjusted odds ratio somehow in revman?

Sorry newbie here.

 

[hlsmith]

So you have how many studies reported with means, std, and n? And these are unadjusted?


Now you have a study with percentages for groups and odds ratios? So if you had percentages could you create the data (given they are all categorical variables (predicts)), and rerun their model yourself for fun.


Can you think of a common effect measure to get them both on without losing too much info. in the process?

 

[glassman365]

So you have how many studies reported with means, std, and n? And these are unadjusted?

Now you have a study with percentages for groups and odds ratios? So if you had percentages could you create the data (given they are all categorical variables (predicts)), and rerun their model yourself for fun.

Can you think of a common effect measure to get them both on without losing too much info. in the process?

I have 6 studies with follow up means, sd's, and n's. I presume unadjusted. I have entered these into Revman as continuous data.

I was trying to enter the study in question as dichotomous data, when I saw a problem as I did not have the number of events, just percentages. So I looked at a way of just entering the adjusted OR but that doesn't seem possible.

So, do you mean there may be a way to convert the data I have to get an effect size?

 

[hlsmith]

I have only played around with Revman a couple of times awhile ago. I believe you can get them all to a "Cohen's d", but you still need to report the study with adjusted effects. No sure if Revman will do the conversion for you.

 

[glassman365]

I have only played around with Revman a couple of times awhile ago. I believe you can get them all to a "Cohen's d", but you still need to report the study with adjusted effects. No sure if Revman will do the conversion for you.

Thanks. I didn't think it was possible with the data I have available. But I'll definately have a look.