Commentary on: Assessing surrogates as trial endpoints using mixed models

In summary, I am enthusiastic about the methods proposed by Korn et al., but only with regard to their model (4). I believe that this model is appropriate both for ordered treatments and unordered treatments, for the reasons given above. I would therefore be very interested to know from the authors what happens to the results if they apply model (4) to the Daniels and Hughes' meta-analysis. Do their results change? Is the evidence for or against the Concorde study's consistency with the trials in the meta-analysis still inconclusive, when using model (4)?.