I am trying to show the releationship between a drug and an outcome in a particular setting. Using a systematic review, I have found 18 studies addressing this topic. Among those, the highest quality evidence comes from observational studies (n=130) and the lowest quality comes from case series (n=5). I did not include individual case reports.

Application of this drug is not terribly common, and there are no controlled studies. For the most part, the inclusion and exclusion criteria for the studies were comparable.

I'd like to find a statistically responsible way to combine these studies. If these were RCTs, I could easily pool the data into a meta-analysis. Is there some kind of reliable way to methodologically analyze this data without calling it a meta-analysis?

  • $\begingroup$ If you use study-level data it will anyway be a meta-analysis... I recommend you to pool point effect estimates using standard errors and a suitable random-effect model (eg with inverse variance weighting), and then explore potential moderators with meta-regression. $\endgroup$ – Joe_74 Apr 24 '18 at 9:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.