In the original publication, the abstract incorrectly states treatment side-effects and outcomes were compared at 2 days after the target quit date. The correct timing is 21 days.

The authors regret this error.