Reply to mol: Comparing apples and oranges in TRD treatment: the importance of assumptions in network meta-analysis