In our meta-analysis, most included studies have two case groups (treated and untreated) and compare them with a single control group of healthy persons. In our paper, we want to study if early diagnosis and treatment can have a statistically significant impact on the prevention of some negative outcomes. We were thinking that the best way to do it is a subgroup analysis but the common control group is an obstacle. We have about 3-6 papers depending on the outcome, and everyone has both case groups. The ways that we found in order to go forward with the subgroup analysis according to Article Methods for including information from multi-arm trials in p...

are:
  • splitting the shared group
  • approximate adjustment of the shared group
  • exact adjustment

So which of these should we use? Are they correct methodologically? Is another most correct way to study if the treatment is statistically significant with the negative outcomes?

More Antonios Siargkas's questions See All
Similar questions and discussions