Help for Meta-analysis

#1
Hello all,

I am Farah, am new to this forum. I am a physician and my knowledge of stats is limited. We are conducting a meta-analysis and exploring the relationship between obesity (BMI) and mortality. The reference group we set is normal weight, whereas some studies included in the meta-analysis use obese/overweight group as reference. For example this one study is using obese group as reference and the factitious data is as under:

Normal weight : RR (95% CI)= 1.5 (1.0, 2.2)
Overweight= 1.0 (0.7, 1.5)
Obese= 1 (Reference)

Now if I want to change the reference group to normal weight and calculate RR for overweight, what would be the way, if I don't have the raw data.
Any help would be greatly appreciated.

TIA

EDITTED: Example data:

Normal weight : RR (95% CI)= 1.5 (1.0, 2.2)
Overweight= 1.0 (0.7, 1.5)
Obese= 1 (Reference)
 
Last edited by a moderator:

hlsmith

Not a robit
#2
Are all studies reporting RRs? ORs are invertible, meaning A vs B can be converted to B vs A by calculating 1/(A vs B).

Also, In am not sure of the setting you are looking at for this MA, but BMI predicting Mortality is the biggest culprit of collider stratification bias and paradoxical results (e.g., health weight more predictive of mortality).