Improved two-stage estimation to adjust for treatment switching in randomised trials: g-estimation to address time-dependent confounding

Improved two-stage estimation to adjust for treatment switching in randomised trials: g-estimation to address time-dependent confounding