Questions on meta-analysis

A retrospective pregnancy safety study is planned to be conducted to assess a drug and major birth defects. The drug has multiple indications with different strengths. For indications for low versus high strengths, other drugs were included for the corresponding indication as comparators. As the drug has low use in pregnancy, combining all indications is likely to reach the target sample size earlier. Propensity scores, the probability of receiving the treatment based on observed covariates, will be generated for the treatment and comparators. Then matching will be implemented for the treatment and comparators. My question is whether a patient-level analysis for all indications should be conducted or two analyses (one for low strength and the other for high strength) should be conducted first and then a meta-analysis follows after that (pooling at the study level). Which method has more advantages than the other? Are there any special considerations? Thank you.

I avoid matching at all costs: https://hbiostat.org/bbr/propensity

Please add major and minor categories and tags to your post.

1 Like

Thank you, Frank. How do I add new categories such as “analysis strategies” and “exploratory and formal analysis” after the post is generated? Sorry for the silly question.

There is an edit button for all posts.

Frank, I looked everywhere around my post. Unfortunately, underneath my original post, I cannot find the edit button. There are only “link”, “bookmark” and “reply” after it. Did I miss something? I found the “edit” button at the end, which only allowed me to edit the last post, not my original post.

I think I saw one after clicking on the 3 dots or the wrench.

Only a trash can appears after I click on the 3 dots. The wrench is for sharing.