How should cost data in pragmatic randomised trials be analysed?

Abstract
No abstract available