Another Failed Study of “Personalized” Depression Treatment

Due to lackluster antidepressant study results, researchers test if subgroups of depressed patients show greater improvement.

Peter Simons

An effort to identify subgroups of people diagnosed with depression who will see a substantial benefit from antidepressants has come up empty. The study was published in the top-tier journal JAMA Psychiatry.

The research team was led by Marta Maslej at the University of Toronto and included prolific antidepressant researchers such as Benoit Mulsant, Toshiaki Furukawa, and Andrea Cipriani. They sum up their results as follows:

“In conclusion, our results do not support the widely held assumption that individual differences underlie the variability in the association between total depression scores and antidepressant treatment.”

Studies of antidepressant effectiveness generally find that the average effect of the drug is clinically insignificant when compared to the placebo effect. Indeed, when published and unpublished trials are included, 49% of the studies found that antidepressants were no better than placebo. Researchers have suggested that the placebo effect accounts for all, or almost all, of the drug’s impact on symptoms.

However, clinicians have argued that antidepressants may be more effective for some people and less effective for others. This could result in an insignificant overall effect but still indicate that some people may benefit from the drug. If such potential populations could be identified, then depression treatment could be “personalized” so that only those likely to respond well would receive the drug.

One suggestion is that people with more severe depression may benefit more from taking an antidepressant—although other researchers have not found evidence to support this possibility. One way to test this hypothesis is to assess treatment effect heterogeneity (TEH). If antidepressants work very well for one group (such as those with severe depression) and do not work well for another group, then the drug’s effect is heterogeneous (variable within the population). Researchers tested the heterogeneity of treatment effects in a new study by conducting a meta-analysis of 91 randomized, controlled trials (18,965 participants).

They compared TEH between those receiving the drug and those receiving the placebo. If particular factors could account for treatment success—such as depression severity—then the drug group should be significantly more heterogeneous than the placebo group.

The researchers, however, found that the groups were not significantly different.

“Variability in response did not differ significantly between antidepressants and placebo,” they write, adding that “Variability was not associated with baseline MDD severity or publication year.”

In an accompanying editorial, Alex Luedtke and Ronald C. Kessler further elaborate on the study’s implications. Luedtke is a statistician at the University of Washington, while Kessler is a health care policy researcher at Harvard Medical School.

Luedtke and Kessler note that Maslej et al.’s study is the fifth analysis of the same dataset. Four of the analyses (including this one) came up negative (Mad in America covered one of these studies here). One study used an alternative statistical method and found a positive result—and was soon retracted because the method was inappropriate. When an appropriate method was used, that study also found a negative result.

Although the five meta-analyses, including data on tens of thousands of participants each, failed to find any evidence for “personalized” antidepressant treatment, Luedtke and Kessler appear undeterred. They write, “While such meta-analyses can provide compelling evidence for HTE if results are positive, they provide little useful evidence for the absence of HTE if the results are negative.”

Even after five studies have failed to find any evidence supporting the assumption that personalized depression treatment is possible, they suggest that the negative results should not deter future research searching for groups who will reliably have a clinically meaningful response to existing antidepressants.

Maslej et al. agree. “This study cannot rule out the possibility of treatment effect heterogeneity,” they write.

Despite the repeated failure to find TEH for antidepressants, a recent meta-analysis of psychotherapy studies did find such an effect, indicating that psychotherapy could be successfully personalized.



Maslej MM, Furukawa TA, Cipriani A, Andrews PW, Sanches M, Tomlinson A, . . . & Mulsant BH. (2021). Individual differences in response to antidepressants: A meta-analysis of placebo-controlled randomized clinical trials. JAMA Psychiatry. Published online February 17, 2021. doi:10.1001/JAMA psychiatry.2020.4564 (Link)

Luedtke A, & Kessler RC. (2021). New directions in research on heterogeneity of treatment effects for major depression. JAMA Psychiatry. Published online February 17, 2021. doi:10.1001/JAMA psychiatry.2020.4489 (Link)


  1. Stupidity is doing the same thing over again and expecting different results.

    Still, I have trouble swallowing the push for more professional therapy or calling most therapies evidence based when the evidence base shows largely that it’s the relationship that matters, not any particular method. It also seems that these relationships could be cultivated without the exchange of payment or authority but that the ease of going into time limited session and dumping our problems on someone else without any concern about how they will process or handle all that they receive has effectively turned support into a commodity removed from genuine concern in either direction. I think we need to seriously question the cultural values that lead us to think that big problems could have such simple answers as drugs and therapy, especially as the dial of distress continues to trend upward and these methods are obviously not treating anything but rather making folks feel better temporarily, when they have any effect at all.

    I fail to see the goal of these studies, for either the drugs or the various therapies. It’s such a reductionistic approach to human distress.

    • You failed because you overlooked the big bucks in store for providers and manufacturers alike. Much of this manufacturing labor is done in search of the big financial killing, should the Philosopher’s Antidepressant be eventually discovered by the company that can successfully ride the waves of adversity in the search for this Miracle Medication.

    • In psychiatry when the first SSRI doesn’t work they try a second. If that doesn’t work they try a SNRI, then they try another. The end result is a serotonin addiction. Trying to quit results in withdrawal which psych manipulates people into thinking means the drugs are good and needed. A neuroleptic drug (which blocks serotonin and dopamine) is added to the serotonin drug. This occurs because the serotonin drug doesn’t help but does cause mania, psychosis, agitation and other negative effects. When the first neuroleptic doesn’t help, a second and or third are tried. When these drugs result in Tardive dyskinesia or akininsia another deadly addicting drug with no scientific long term benefits is added.

      To you this appears to be insanity that doesn’t work. But to the people getting rich and gaining social status by stigmatizing others it is “working”. Those with power also see it as working since now all the faults of society can be falsely blamed on diseased people having poor genetics and biology.

  2. Well, personalized treatment for “depression” is possible, but it’s not going to be with drugs alone. Since “depression” is a syndrome and not a disease unto itself, it’s necessary to find out what’s inducing it, in order to provide proper and appropriate treatment.

  3. I don’t quite understand the conclusion “This study cannot rule out the possibility of treatment effect heterogeneity.”

    They had 2 groups (meds & placebo) and they observed these groups wrt TEH i.e. they looked for instance whether severe depression groups (or other preidentified subgroups) benefit more from the medication than placebo. Ok, so the point is because there may be infinitely many subgroups to look for in the vast landscape of behavioral continuity (or just a lot so that you may never be finished) you can never rule out the possibility that there may be a subgroup where TEH may be observed?

    • The ultimate subgroup is a single person. If you break it down far enough, you end up with individualized “treatment,” which is what should be happening anyway, if any “treatment” at all is considered appropriate.

      Of course, there is “heterogeneity” in a group you put together based entirely on how they FEEL at a given moment in time. The idea that people who feel depressed are all the same is the most basic deception psychiatry is based on.