it is likely that these two conditions will remain difficult to separate, we plan further analyses to explore these possibilities.Withdrawal symptoms can only occur in the discontinuation group, but participants reported some new and worsening symptoms while continuing to take antidepressants.We found no evidence that the hazard ratio for relapse varied across the 12-month follow-up period, whereas one would expect withdrawal symptoms to cluster around the time after the medication was terminated.We found that the difference in withdrawal symptoms between the groups was largest at 12 weeks.Kuschpel inquires about the severity of relapses in our trial.In addition to our prespecified definition of relapse, we used internationally agreed-upon International Classification of Diseases, version 10, criteria for relapse of depression and found similar results to those in our primary analysis (see Table S11 in the Supplementary Appendix, available with the full text of our article at NEJM.org).We recorded relapses that may have occurred at any time in the 3 months preceding the assessment and may have resolved by the time the participant was assessed on our secondary outcomes, such as the PHQ-9.We therefore would have expected a smaller betweengroup difference regarding secondary outcomes because they only assessed how the participant was feeling at the time of the assessment.Our finding that people in the discontinuation group were more likely to guess their allocation could be due to the clinical effect of discontinuation and is not, in our view, an indication that our findings were invalid.