Antidepressants Are Not More Effective for Severe Depression, Study Finds

Peter Simons

A new study, published in Acta Psychiatrica Scandinavica, found that antidepressant efficacy was not dependent on severity. The researchers analyzed the individual-level data from Japanese double-blind, randomized, controlled trials (RCTs) of current-generation antidepressants. They found that antidepressants do not appear to be more effective for severe depression than they are for mild depression.

Photo Credit: Pixabay

The study was conducted by prominent antidepressant researchers: Toshi Furukawa at Kyoto University, Stefan Leucht at Technische Universitӓt München, and Andrea Cipriani at the University of Oxford, among others. These three authors were also involved in the Lancet Psychiatry study earlier this year that found antidepressants to be effective.

Furukawa and the other researchers identified 11 RCTs of current-generation antidepressants conducted in Japan and requested individual participant-level data. Only 6 of the studies complied with this request, so the current analysis is based on this sample. A total of 2,464 participants were included. The studies compared duloxetine (Cymbalta), escitalopram (Lexapro), mirtazapine (Remeron), paroxetine (Paxil), and bupropion (Wellbutrin) to placebo.

The authors used individual patient-level data in their analysis so that they could compare antidepressant efficacy based on the spectrum of initial depression severity (rather than lumping participants into broad groups). Their analysis found that there was no significant difference in efficacy based on initial severity.

The authors note that their finding is contrary to previous findings that antidepressants are more effective in severe depression. They suggest that because antidepressant efficacy over placebo is such a small effect, improvement can easily be confounded with other factors.

Antidepressant medications have generally been found to be just slightly better than placebo in efficacy studies—enough to be statistically significant. However, the clinical significance of this slight difference has been questioned.

The effect, which equates to about a 2-point difference on the 52-point Hamilton Depression Scale, may be unnoticeable either clinically or in terms of quality of life. For example, a 2-point difference could be created by answering two questions about work problems or anxiety as a three instead of a four on a Likert scale. In fact, the Hamilton Depression Scale has been criticized in the American Journal of Psychiatry as having “poor content validity” and “poor interrater and retest reliability.”

On the other hand, some researchers have found that antidepressants are more effective for people with more severe depression. As a result, some clinical practice guidelines recommend providing antidepressant drugs as a first-line treatment for severe depression. However, the new finding by Furukawa and colleagues challenges the assumption that antidepressants are more effective for severe depression than they are for mild to moderate depression.

There were many limitations of the study: it included only studies from Japan; only half of the identified studies were included in the data; many types of antidepressants were not included, and very severe or suicidal patients were not included in the studies. In fact, the researchers describe the participants as “highly selected.” This means that the study cannot answer questions about whether antidepressants could prevent suicidality in severe depression.



Furukawa, T. A., Maruo, K., Noma, H., Tanaka, S., Imai, H., Shinohara, K., . . . Cipriani, A. (2018). Initial severity of major depression and efficacy of new generation antidepressants: Individual participant data meta-analysis. Acta Psychiatrica Scandinavica, 137(6), 450-458. doi: 10.1111/acps.12886 (Link)

Support MIA

MIA relies on the support of its readers to exist. Please consider a donation to help us provide news, essays, podcasts and continuing education courses that explore alternatives to the current paradigm of psychiatric care. Your tax-deductible donation will help build a community devoted to creating such change.

Select Payment Method
Personal Info

Credit Card Info
This is a secure SSL encrypted payment.

Billing Details

Donation Total: $20


  1. The “mental health professionals” are an odd bunch. They believe inanimate objects, pills, can know how depressed a person is, and function “more effective for people with more severe depression.” They also believe their inanimate pills know how old people are.

    The antidepressants have a black box warning pointing out they cause suicides and violence in children and adults 24 and younger. Yet the “mental health professionals” believe somehow their inanimate antidepressants are beneficial for adults aged 25-65? Why have the “mental health professionals” believed inanimate objects have some sort of intellect, for decades?

    The inanimate antidepressants are dangerous, mind altering, mania and even psychosis, as well as violence and suicide inducing drugs, in people of all ages and “depression” or lack thereof levels, of course. They are only “wonder drugs” from the stand point they help doctors create “mental illnesses” in people for profit, which is the opposite of what all doctors promised their patients they’d do, “first and foremost, do no harm.”

    Let’s hope some day the disapproving and judgmental “mental health professionals” (the picture nicely depicts this attitude) will some day learn to respect the intellect and reputation of their patients, more than that of their inanimate, neurotoxic drugs.

  2. I’m glad to see this here, because it can’t be said often enough. Some other authors had already figured it out, though I can’t remember exactly what they said…it had to with problem that arises when success is measured by the distance between response to placebo and response to the drug. In very severe cases, there is not the typical placebo response. This alone caused a wider distance between placebo response and response to drug, which was misinterpreted or misrepresented as being caused by greater drug effect in that group.

    Despite that debunking, trusted experts continue to rattle off the cherished “but they do work in the most severe cases.”

    But that is just STUPID. What other drug exists that only works in severe cases?
    What acid dissolves diamonds but not chalk?
    What steamroller crushes Samsonite suitcases but not paper bags?

    • I’ve never understood that either: how can an aspirin do nothing for a mild headache but take away all pain when you have a crushing headache? And how does the drug know that your pain is only mild or moderate and not severe? Where is the line between mild/moderate/severe?

      At least aspirin does work most of the time, unlike antidepressants.

  3. Actually, the purpose of this study seems to me they wanted to prove that antidepressants ALWAYS work, for every type of depression, not just for severe cases, rather than to confess they don’t really have a clinical effect in severely depressed people. That’s what I read in their conclusion:

    “We may expect as much benefit from antidepressant treatments for mild, moderate or severe major depression. Clinical practice guidelines will need to take these findings into consideration.”

    • I read the text in the study more carefully and what it actually says is that antidepressants work equally well in “mild, moderate and severe MAJOR depression”. So they are actually only talking about MAJOR depressive disorder but then go on to separate three types of it. This is very confusing because depression itself is already divided into three categories which are also referred to as mild, moderate and severe. People reading the article will think that the drugs work for all types of depression and that they are just as effective for each of those types, not just for severe depression.

      • The authors of this study are similar to the recent Cipriani study except in a different order, and its probably very good research.

        But the commentary is always at odds with the research. So in the Lancet we have Cipriani et al saying, in response to the social media furore, “some coverage in the media and social platforms was inaccurate—in particular, there was an undue focus on the binary and polarising question of clinical significance. People can always manipulate information to fuel controversy”. People can indeed. They seem to suggest that it was a travesty that so many got sidetracked by that disappointing average effect size of 0.3 reported in their recent paper, when they should have been ignoring the vast majority of the data just to focus on the tail-end of recovered patients where indeed there were more people on AD’s. Their league tables on that subcategory of recovered patients are inconsistent and raise more questions than answers. I don’t know how you manipulate an effect size, by the way.

        This particular study on severity says that there is no variation with severity and clinical guidelines should reflect this. This would appear aimed at NICE, who suggest that its only for the most severe depression that you can expect a good result for AD’s. So other authors and defenders of AD’s ( Eriksson Hieronymus) ,have pitched in with “The alleged lack of efficacy of antidepressants in non‐severe depression: a myth debunked”. Gentlemen, the myth debunked is that severity has anything to do with the LACK of efficacy. You cannot change the data on efficacy by learning that severity makes no difference. These 2 papers together provide good evidence that:
        (1) AD’s have a very weak effect size of 0.30.
        (2) Some AD’s are even worse than others.
        (3) Severity has nothing to do with it.

        In total defence of Prof Tim Kendall and NICE, they boldly expressed the truth as they knew it at the time – kudos to them. AD’s weren’t much good, but just possibly could be in severe depression. Fair enough – but we now know better, although we suspected that severity was a red herring all along.

        • I’m still confused. It is my understanding that this newest paper focuses on severe depression only (what they call “major depression”) but somehow they distinguish between mild, moderate and severe types of “severe depression”. This is ridiculous because when depression is “major” it already means that it is severe. Unless I’m missing something here?

          Personally, I don’t believe the drugs do much good even for severe depression (not in the long run anyway) but that’s not even the main issue here. My point is their terminology is unclear and confusing, and so are their conclusions. Wasn’t it Cipriani himself who said that “all debate is now over” and “we now have the FINAL answer” back in February? If his answer was final, then he shouldn’t be saying different things now. The Lancet study admitted antidepressants don’t work for mild depression so they can’t change their mind now just a few months later and then also add that the guidelines for prescribing should be changed.

          • The definition of Major Depression does indeed seem confusing, but, actually, does it matter?
            (1) AD’s pretty inneffective in “Major Depressive Disorder”, effect size 0.3 (Cipriani, Furukawa et al , Feb 2018).
            (2) AD effectiveness in “Major Depression” unaffected by severity (Furukawa, Cipriani et al, June 2018)
            So for Major Depression, these highly professional researchers have shown that the evidence suggests, on average, clinically insignificant benefit in all degrees of major depression.

            If The RCPsych, SMC etc want to do a backflip and say that AD’s actually only work for non-major depression, that would not stand up to evidence either.

        • Of course, not. But in theological society everything which is connected to death is an illness. You all should read Revisioning psychology, I know that, this book is necessary bible for everyone who don’t know nothing about roots of psychiatry. This book and Manufacture of madness, are pure logic. Neccessary knowledge about ideologies like transcendentalism, nihilism (LAING) theology/christianism, spiritualism – neglectance, materialism and many others, which are obstacles before getting to know what psyche actually means. We – psychological poeple are victims of theology. Church has Vatican and power and money, psyche has got only asassins.

      • It’s also important to remember that “works” is defined here only as “temporarily makes ‘symptoms’ less bad than they were.” It does not imply feeling good, nor does it imply long-term improvement in prognosis or outcomes. Whitaker has done a great job showing that even if these drugs “work” for the short term, they tend to make things worse in the long run. And for many, even the short-term results can be disastrous, even if they are judged “helpful” on the average.

  4. It is useless to write about antidepressants.Write about JAMES HILLMAN and his revolutionary books which is Suicide and the soul and Re – visioning psychology or, Manufacture of madness – those books are greate weapons. Society does’t know about them, they know only bible which has got nothing to do with psyche,remove word satan from the bible and put a word “psyche” in the same place, you will know everything, psyche means =evil . This is a sabotage and betrayal of them —-so called, normal, christians and so on Normalcy, means nothing good.
    The word “normal” comes from the Greek norma, which was a carpenter’s square, that right-angled tool for establishing straightness.

    more of Hillman attitude

  5. How can any medication work when this sort of thing is happening.

    And this:

    It is all about underhanded abuse by someone related and inhumane in the person’s life. Unless this is seen and addressed there can be no fiddling with brain chemistry, which only causes imbalances that can make it go away.

    You can also ward off the offenders by a counter attack in the Mind

    And remember the best revenge is when their foul games fail and you can jump the hoops without being negatively affected. Spoil their game and give psychiatry the boot as well.