Recently I started a discussion about modeling secular trends using flexible models in the context of cluster randomized trials. I’ve been motivated by a trial I am involved with that is using a stepped-wedge study design. The initial post focused o…