"Largely ineffective and potentially harmful"

An interview about the problematic success of antidepressant drugs

Dr Michael P. Hengartner is graduate psychologist and researcher at the Zurich University of Applied Sciences in Switzerland. He was research associate at the Psychiatric University Hospital of Zurich from 2009 till 2014. Since 2015 he is also teaching psychosocial medicine at the University of Zurich and psychopathology at the University of Applied Sciences and Arts of Northwestern Switzerland. His major research areas are clinical psychology, social psychiatry, and the frequency (epidemiology) of mental disorders.

Important note of the author: The discontinuation of antidepressant drug treatments can lead to withdrawal symptoms or an increase in severity of the symptoms. Please do not take such steps without the supervision of an experienced physician of trust.

Dr Hengartner, how did you become interested in the research on antidepressant drugs?

Michael P. Hengartner: I have been interested in research methods and the theory of science for a very long time. How reliable is research as altogether? But in particular also the research in psychology and psychiatry? I was further interested in statistics and the ways in which some researchers are misusing or misinterpreting statistical methods.

And then we obviously had the replication crisis in psychology: Why do the results of so many studies not replicate? In addition to that there seems to be a "publication bias" everywhere. That is, positive results are overestimated and published, while negative results vanish in the file drawer. Many scientists try to advertise their field of research - and that is why they are biased.

I also took a closer look at studies on the efficacy of psychotherapy. Then I found the critical reports on the efficacy of antidepressant drugs by the British psychiatrists David Healy and Joanna Moncrieff, by the American psychologist Irving Kirsch, and by the Danish physician and director of the Nordic Cochrane Center Peter Gøtzsche. I was shocked about the extent of methodological biases in these antidepressant trials.

In your recent review article you are drawing the conclusion that antidepressant drugs - in particular the popular drugs of the types SSRI and SNRI - are largely ineffective and potentially harmful. Can you explain that briefly?

Michael P. Hengartner: The major problem of the studies on antidepressant drugs is that they are flawed by design. Already before the actual trials begin, much has been done such that they will yield a statistically significant effect in favor of the antidepressant drug. Yet, randomized and placebo-controlled studies should avoid such biases.

One example is that participants who are strongly responding to the placebo substance are excluded before the trials officially begin. Why are researcher doing that? Because this increases the likelihood that the investigated drug will eventually seem to have a higher efficacy than the pharmacologically ineffective placebo pill.

Treating Withdrawal Symptoms

Can you explain that in a bit more detail?

Michael P. Hengartner: Patients with a diagnosed depression often already have a prescription for another drug. To avoid that a study’s results will be biased by that medication, they will get a placebo a few days before the outset of the randomized trial. This is called the placebo-washout. The idea is to wash the old drug out of the body. Those showing an improvement in that period, that is, those who are strongly responding to the placebo, are excluded from the actual trial.

Another problem is that the discontinuation of the old medication can lead to withdrawal symptoms, such as mood swings, irritability, anxiety, or sleep problems. The few days of placebo-washout right before the actual trial are often too short for these symptoms to disappear. Now one must understand that these very symptoms are also used to measure the severity of depressive disorders.

That means that some patients in the placebo group have higher depression scores at the beginning of the trial due to the discontinuation of their original medication. Conversely, in participants receiving the active drug an improvement occurs right from the beginning, because the antidepressant drug stops the withdrawal syndrome immediately. So if in trials of four to six weeks duration patients in the drug group show a slightly higher symptom reduction than patients in the placebo group, this is possibly unrelated to the antidepressant property of the investigational drug, but instead resulting from the abatement of the withdrawal syndrome.

One cannot distinguish the true drug effect on the depressive symptoms from that on the withdrawal symptoms in the data. Some patients in the placebo group will have withdrawal symptoms right after discontinuation of the previous medication, that is during the washout phase, but others just after days or a few weeks, that is during the trial period. We know from studies that between 10 and 50 percent of the patients can suffer from such withdrawal symptoms.

A similar problem is that patients in some studies additionally receive tranquilizers, such as benzodiazepines. Their doctors are free to treat restlessness or sleeping problems, which are typical side effects of antidepressant drugs, with such tranquilizers. The effect attributed to the new antidepressant medication therefore might simply be due to the prescription of these other drugs.