You can either perform a meta-analysis using patient-level data from RCTs or using summary measures (from published papers).
Ths principle is quite similar in both cases, as a meta-analysis will allow to estimate the overall/pooled treatment effect. One of the main thing to consider in a meta-analysis is to study if there is a variation of the tratment effect between trials: it's called treatment effect heterogeneity.
Here is a nice paper in oncology, that can apply for other fileds: