Principled selection of effect modifiers: Comments on 'Matching-adjusted indirect comparisons: Application to time-to-event data'.