Conditional estimation using prior information in 2‐stage group sequential designs assuming asymptotic normality when the trial terminated early