proper use of meta-analysis in manuscript submitted for publication in medical journal?

[coloradosurgeon]

Hi there -
Help! I am a physician performing a peer review of a manuscript submitted for publication in a medical journal. (Disclaimer before reading further - I have very little statistics knowledge) The authors have performed a meta-analysis of six RCTs involving the use of a specific type of suture called fast-absorbing gut. Each of the six studies included in the meta-analysis however compares the fast-absorbing gut suture to different things - another type of suture called polypropylene, "skin glue" (cyanoacrylate), and fast-absorbing gut suture of varying thickness. My understanding is that a meta-analysis can only be performed on a set of RCTs that are all comparing the same two things, for example all six RCTs must compare gut suture with skin glue, and therefore the authors have used an incorrect statistical method in their manuscript. Is this correct? Thank you in advance for your help!

 

[hlsmith]

Correct. If they are trying to pool six studies with different comparator groups, that likely goes beyond heterogeneity and is inappropriate.

There is a process called network analysis where you compare to treatments to each other even if they have not been directly compared to each other, since they both have been compared to another treatment (likely placebo). Though i doubt they a re doing this.

if you have status with the journal and would like me to peer-review it for analytics you can DM me. Otherwise good luck and good job investigating something you didnt know and not just letting the manuscript skate by. Reviewers are gatekeepers for good science!

 

[coloradosurgeon]

Correct. If they are trying to pool six studies with different comparator groups, that likely goes beyond heterogeneity and is inappropriate.

There is a process called network analysis where you compare to treatments to each other even if they have not been directly compared to each other, since they both have been compared to another treatment (likely placebo). Though i doubt they a re doing this.

if you have status with the journal and would like me to peer-review it for analytics you can DM me. Otherwise good luck and good job investigating something you didnt know and not just letting the manuscript skate by. Reviewers are gatekeepers for good science!

Thank you very much for taking the time to reply to my question and confirming my intuition on this manuscript. The journal likely has an analytics editor on staff that I will recommend review this paper as I am only an occasional reviewer for this particular journal. Thanks again!