Worldwide, depression remains one of the most widely diagnosed disorders, and the first line of treatment in many countries is antidepressant medications. While early reports showed promise, emerging evidence over the last several years has raised massive doubts. This evidence has questioned both the efficacy of these drugs and the adverse effects associated with them.
A 2019 review synthesizing the evidence on antidepressants was published in BMJ Evidence-Based Medicine. It was conducted by researchers Janus Jakobsen and Christian Gluud of Copenhagen University Hospital and Irving Kirsch of Harvard Medical School.
The researchers assert that while antidepressants show statistically significant differences when compared to placebo, the effect itself is so small as to be clinically meaningless. Given that the adverse effects of antidepressants are severe and pervasive, their use should be restricted until more is known about them.
Antidepressants, once hailed as the definitive treatment of depression, have suffered significant blows to their reputation. Recently, a change in the UK’s NICE (National Institute for Health and Care Excellence) guidelines, followed by the acknowledgment of their long-lasting harms by the former president of the Royal College of Psychiatrists, has brought their risks into sharp focus.
New reviews have noted that the withdrawal effects of antidepressants can last longer than a year. A recent study found that when ineffective antidepressants are augmented by antipsychotics, early death rates increase by 45%. Other researchers have pointed to rampant industry corruption in clinical trials of antidepressants.
In this evidence synthesis, the well-known researchers note that antidepressant use has increased exponentially across the world, and over 60% of the people taking them have been doing so for more than 2 years.
The researchers begin by reviewing what statistical significance means in antidepressant trials. Here they analyze the use of the popular Hamilton Depression Rating Scale (HDRS). In a clinical trial, whether a drug is effective or not for depression is often measured by the average drop points on this scale; it is supposed to represent a drop in the severity of depression symptoms.
The scale is mired in numerous controversies, however. Earlier a drop of 3 points on the scale was considered to be clinically significant; this was displayed on the NICE website but has since been removed because of numerous criticisms. Despite this, many studies still continue to use this benchmark to suggest that a drug is working. Some studies show that such a small change on the HDRS scale produces no change in the person’s condition and is undetectable in clinical practice. Others have argued that a change of 7 points is necessary for any clinical improvement to be identified.
Another problem is that trials often divide the 52-point scale into two binaries: people showing greater than 50 percent improvement on HDRS are called responders to drugs and those below that as non-responders. This is an arbitrary division obfuscating complex realities. For example, a person showing a 49% change is called a non-responder while 51% is considered to have responded to the drug. The mere difference of 2% puts them in completely different categories. At the same time, it puts people with a score of 0% change and 49% percent in the same category. The researchers write:
“Hence, when assessing such dichotomized outcomes, there is a considerable risk of overestimating benefit, but there is also a risk of not detecting a ‘true’ effect. Hence, dichotomized outcome results, such as ‘response’ or ‘remission’, should not be used to assess the statistical or clinical significance and should be interpreted with caution.”
The researchers note that several recent reviews of antidepressant studies have shown that the drugs have small statistically significant effects compared to placebo. At the same time, most of these reviews are non-systematic (according to the PRISMA checklist) and are thus considered less rigorous than systematic analyses. The researchers evaluate two recent reviews.
First, in 2017, the authors of this synthesis conducted a systematic review of the evidence for antidepressants. They found that while the difference between antidepressants and placebo was statistically significant, the effect size (1.94 HDRS points) was too low for clinical significance (3 HDRS which was the earlier NICE criteria) and far lower than “minimal improvement” (7 HDRS).
In other words, the magnitude of the difference between antidepressants and placebo was too small to matter. For long-term effects, the size was even smaller. Additionally, the measured adverse effects and chances of bias in many of these trials were both high.
The second review was published in The Lancet in 2018. It measured only short-term outcomes and similarly found statistically significant results for antidepressants, but also a really low effect size. Researchers report that only 18% of the trials in that review were at low risk for bias.
The researchers write that one of the major problems with the existing evidence, apart from the low effect sizes, is the high chance of bias in the trials. For example, the Lancet review also included head to head trials, which are especially vulnerable to industry sponsorship.
Moreover, we now know that patients can break blind in a trial because the adverse effects of antidepressants cue them to the fact that they are not receiving a placebo. Thus, even the small significant effect seen in the trails could be a result of an enhanced placebo effect. In other words, the participants break blind and begin to feel great about receiving the actual drug, which influences their depression rating. Lastly, many trial results cannot be easily generalized to the wider population as they include only a very specific type of patient.
For these reasons, even the small statistically significant effect might be inflated. For example, one study found that if the meta-analyses included an author working for the pharmaceutical company manufacturing the drug, it was 22-times less likely to “have negative statements about the drug than other meta-analyses.” It also found that trials at low-risk for profit-bias found no statistically significant effect for antidepressants.
While these results cast doubts on the efficacy of SSRIs, researchers of this review also note that both serious and non-serious adverse reactions have been downplayed. These range from sexual dysfunction and gastrointestinal problems to birth defects for SSRIs, and seizures, and even death for tricyclic antidepressants.
Withdrawal symptoms are also severe and long-lasting; these include hallucinations, stroke-like symptoms, panic disorder, rebound depression, and anxiety among many others. Some have insisted that this should be called antidepressant withdrawal rather than the more evasive discontinuation syndrome. Often, even when patients want to get off antidepressants, these symptoms make it difficult to stop. The authors write:
“Withdrawal symptoms might also explain why some studies have alleged to show that the risk of relapse seems to be reduced if antidepressants are continued instead of not continued. Withdrawal symptoms might be the reason why patients who do not continue antidepressants might do worse compared with patients who continue antidepressants.”
The researchers insist that given recent evidence, social determinants of health like unemployment and poverty should be addressed as causal factors in depression. Further, it appears that for many patients the priority is not simply a reduction in symptoms, but that they can join social activities and return to work. Given the low efficacy (clinically insignificant effect), risks of harm, and industry-bias, patients should be informed about other treatment options. The authors write:
“Antidepressants should not be used for adults with major depressive disorder before valid evidence has shown that the potential beneficial effects outweigh the harmful effects.”
Jakobsen, J.C., Gluud, C., & Kirsch, I. (2019). Should Antidepressants be used for Major Depressive Disorder? BMJ Evidence-Based Medicine, 25(4), 130-136. http://dx.doi.org/10.1136/bmjebm-2019-111238 (Link)