Dear all,
I'm writing a meta-analysis about treatment of stimulants use disorders. I am pooling data from RCTs using a certain class of medications.
However, some of the described studies used more than one treatment group of interest. Here are two situations which I'm not exactly sure how to handle methodologically:
Trial 1: group 1: drug A, dose X. Group 2: drug B, dose Z. Group 3: placebo. (Here, both drugs A and B are from the same class, both within the scope of our review).
Trial 2: group 1: drug A, dose X. Group 2: drug A, dose Y. Group 3: placebo.
I see two options to deal with this:
A) Break that specific trial into two, and insert results as in separate trials: XXXa, 2015: group 1 x group 3 (placebo) and XXXb, 2015: group 2 x group 3 (placebo). - One Cochrane review I checked does that in situations like the one described in trial 1.
B) Merge the two treatment groups together in a single comparison: (group 1 + group 2) versus placebo. - The same Cochrane review does that in situations like the one described in trial 2.
Both are easy to do mathematically since I am working with a dichotomous variable.
Between A) and B), which is more appropriate to deal methodologically in each situation (1 and 2)?
Thanks!
Vitor - the way that I see it is that you could do either. The choice is yours. The main thing is to add some detail as to justification and rationale of the chosen approach over the other. The worst thing that could happen is that the reviewers suggest that you use the other approach to the one that you have adopted.
Thanks for your answer, Dean. I tried both methods and the estimates don’t really change much. I’ll search the literature to provide rationale. If you know of any interesting paper about this matter, I’d be happy to read it.
Dear Vitor,
At first glance, one could provide justification for either approach. Out of curiosity, I went back to the Cochrane Handbook: http://handbook-5-1.cochrane.org/. The scenario you outlined is addressed in 16.5.4 How to include multiple groups from one study.
Assuming that both dose X and Y of Drug A (Trial 2) are clinically relevant and eligible for inclusion, approach B seems to be the preferred option. It would be worthwhile revisiting how you defined your PICO, specifically the criteria you set for the types of intervention (I) and comparator (O) to include. (Refer to 5.3 Defining types of interventions: which comparisons to make? of the handbook).
- Badges
- Science topic
- Similar topics
- Chemistry
- Pharmacology
- Pharmaceuticals
- Drugs