Estimation of historical control rate for a single arm de-escalation study: Application to the POSITIVE trial.