Expectancy effects in antidepressant withdrawal studies

Zhang et al (2024) have published a systematic review and meta-analysis of the incidence of antidepressant withdrawal symptoms. More than 40% experienced such effects. This figure is higher than the 15% estimate from another recent systematic review (see previous post). As I pointed out in that previous post, so-called withdrawal symptoms also occur in the continuation arm of randomised controlled trials (RCTs), when such withdrawal symptoms wouldn't necessarily be expected in this group as participants are still taking their antidepressant. This could be said to demonstrate how expectation can influence the experience of antidepressant withdrawal, as participants may have expected adverse reactions on withdrawal, thinking they were being withdrawn from antidepressant even though they were in the blinded control group. Zhang et al report that the incidence in the discontinuation group is significantly higher than the continuation group. What I want to challenge is their claim that this significant difference therefore necessarily excludes such expectancy, or nocebo, effects as an at least partial, if not complete, explanation of antidepressant withdrawal symptoms.

I want to emphasise that I'm not saying such nocebo effects are not real. They are felt and experienced as true medication effects. It may well be difficult for people who experience withdrawal effects to understand that they could be nocebo effects. However, in my experience, people do generally appreciate that people may become dependent on antidepressant medication. It was the basis for me writing my BMJ letter that led to my special interest in antidepressant withdrawal (see eg. my book chapter). A drug that is thought to improve mood may well be expected to be difficult to give up because of, for example, a fear of relapse. Psychological dependence with antidepressant medication is not surprising (see eg. previous post). What people find difficult to accept is that such nocebo effects can be so powerful that they can cause the severe and longlasting effects that they do. But, again for example, if the taking of antidepressants is associated with the belief that the medication is correcting a brain problem, even though this is the wrong way of understanding how mental health problems are corrected, then it's not surprising that it may take some time to come to terms with managing without the drug because of the complex set of meanings that the medication has acquired. Undoing these beliefs is not easy, particularly perhaps if the experience of taking antidepressants initially seemed to help.

Of course using placebos in clinical trials of efficacy is designed to exclude placebo effects. In the same way in discontinuation trials, having a control group which continues antidepressant is designed to control for nocebo effects. How effective these control methods are in preventing placebo/nocebo effects depends on how well blinded the participants are from knowing to which group they have been allocated. There is considerable evidence that people are not completely blinded in antidepressant efficacy trials (see eg. previous post). As far as I know, there has been no attempt to measure unblinding in an antidepressant withdrawal study. If unblinding occurs in antidepressant efficacy studies, I think it is also likely to occur in antidepressant withdrawal studies. As the blind can be broken in antidepressant efficacy trials, it cannot be said that expectancy effects have been eliminated.  So my case is that it cannot be said that expectancy effects have been eliminated from antidepressant discontinuation RCTs, because I think there is also likely to be significant unblinding in these withdrawal studies as well.

Habituation to antidepressants is to be expected (see eg. previous post). It helps to explain why people take them for such long periods of time. Psychological mechanisms causing antidepressant withdrawal symptoms should not be dismissed. I have considerable doubts about antidepressants being more than placebo in their antidepressant effect (see eg. previous post). Those that argue that antidepressants cause organic physical dependence tend to say that the sense that antidepressants have stopped working, which can occur, sometimes colloquially called the "poop-out" effect, is evidence that there is tolerance with antidepressants. As I don't think antidepressants are "effective" in the sense of being more than placebo, this explanation doesn't make sense to me. I am at least consistent in my scepticism about the effects of antidepressants, which for their apparent benefit I put down to placebo, and for their withdrawal effects I am inclined to think could be due to nocebo. To emphasise again, this does not mean I am saying any experienced benefit for antidepressants is not real. Nor am I saying that the experience of antidepressant withdrawal is unreal. What brought me into the area of antidepressant withdrawal years ago was my critique of mainstream psychiatry for denying the reality of such symptoms. I just don't think that there's necessarily been much progress since in understanding the mechanisms of such withdrawal effects, and it worries me that psychological mechanisms seem to be being ignored, even within the Critical Psychiatry Network (see previous post).