I am conducting a meta-analysis on the therapeutic benefits of drug A - treated vs placebo. I would like to address the possibility of side effects/tolerability/likeliness of developing side effects. The side effects are categorized into to three subgroups: mild side effects, moderate side effects and severe side effects.

I have included 20 studies in my analysis. I want to visualize the proportion of patients who develop side effects. However, the follow-up time differs among the studies. Some report side effects that occurs within the first month whereas others report side effects first, second and six month. Further, the sample size differs greatly (I do not know it this is relevant).

Can I do some sort of weighted histogram (I don't know if this exist)? Or how can I investigate this matter further/visualize it?

  • $\begingroup$ Could you provide more details on exactly what the different studies reported? $\endgroup$ – Lejafar Jan 13 '18 at 16:29
  • $\begingroup$ It sounds as though you are looking for a forest plot. $\endgroup$ – mdewey Jan 15 '18 at 10:05

You can estimate a logit model where

1) developing side effects is the dependent variable,

2) follow-up time is one of independent variables,

3) each study is 1 observation,

4) the reported standard errors in the study are used to determine the weight of the observation (the standard errors will take account of the reported sample size).


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.