Naked Science Forum

Life Sciences => Physiology & Medicine => Topic started by: leophie on 17/08/2022 20:25:22

Title: Question about meta-analysis: more than one exposure-outcome per study?
Post by: leophie on 17/08/2022 20:25:22

Good day  :)

I am conducting a meta-analysis and wondering if anyone can please help me with a query about included studies.

So, I have begun data extraction. However, in a few of the papers, there is more than one exposure-outcome combination, and all of the combinations meet the eligibility criteria for my study.

How should I handle this?

Can I include all of the exposure-outcome combinations?

Here is an example of what I mean. My study eligibility criteria is antithrombotics exposure and mortality outcome.

One of the papers has both warfarin and (separately) heparin as the exposure, and both report on mortality.

Both combinations meet the eligibility criteria.

Would I be able to include both on the forest plot, separately, like "Author (a)" and "Author (b)"?

I think this might affect the between-study heterogeneity (I2), since both derive from the same study, but it looks like the two are separate studies.

Please can you advise on this?

Best wishes
leophie

Title: Re: Question about meta-analysis: more than one exposure-outcome per study?
Post by: alancalverd on 17/08/2022 23:31:13

I would find it suspicious if somebody combined the studies of two causes (e.g. warfarin and heparin) with a common effect (mortality).To my mind, meta-analysis is only scientifically useful if it includes all the unpublished but valid data, and is strictly confined to a single cause. In other words, I generally eschew it, and prefer a single well-conducted and fully-reported study that gives me access to all the method and results so I can make up my own mind as to whether a well-defined hypothesis is supported by experiment.