They then go on to say that the study demonstrated that many people can come off antidepressants safely, whereas I’m not convinced that’s what the study necessarily shows. Of course there are problems with generalising the results of a controlled trial to the real world, and subjects who discontinued their medication in the trial were given placebo, rather than no medication. It would have been interesting to have had a third arm in the study looking at the relapse rate of a no medication group. And to know what happened to the placebo group after the end of the trial. Furthermore, the study does not show whether people can manage without antidepressants longer-term after a year.
The Lewises also say that “Neither researchers nor patients knew which group people were allocated to”. This is not correct, as they report information about unblinding in a supplementary appendix to their paper. They found that “71% (162/228) in the discontinuation group and 47% (108/232) in the maintenance group correctly guessed their randomised group at any time before being unblinded”. People in the trial were better at guessing their allocation if the antidepressant was stopped. Trials typically turn a blind eye to the methodological problems of unblinding (see eg. my eletter). It is possible that the statistically significant advantage found for continuing antidepressant could be due to nocebo amplification through unblinding, in a similar way to which the statistical advantage for the effectiveness of antidepressants could be due to placebo amplification (see eg. previous post).
People are understandably fearful about stopping antidepressants (see eg. previous post). The subjects in the ANTLER trial were said to feel well enough to consider stopping the pills. From the information in the supplementary appendix, some seemed to have stopped taking any antidepressant when coming out of the trial, both in the maintenance and discontinuation groups.
It is not easy to distinguish withdrawal symptoms from relapse. Anxiety about treatment withdrawal can lead to both relapse and withdrawal symptoms (see eg. Moncrieff et al, 2021). Discontinuation problems from antidepressants should not be minimised (see eg. another previous post) and the ANTLER study confirms that long-term outcome of the treatment of depression is not always as good as might be hoped (see previous post). This is a different message from the headline in The Conversation article.