This is a follow-up to:
“my more substantive complaint is that the drug-skeptics focus too narrowly on commercial drug trials. In the process, they ignore other systematic outcome research and, more importantly, miss any sense of how the body of knowledge adds up. In my view, the notion that antidepressants have no direct main effect on depressive episodes is untenable.”
Peter Kramer, in Q&A with me for Psychiatry at the Margins
Jack Goulder’s guest post “Merchants of Psychiatric Doubt” generated a flurry of reactions and responses, unsurprisingly. The reaction of a few people, some publicly and some privately, focused on the fact that Goulder had cited the 2022 Stone et al. meta-analysis of FDA trials of antidepressants in major depression, with its claim of trimodal response distributions and a subset demonstrating a substantial antidepressant effect beyond a placebo effect, as primary evidence in support of antidepressant efficacy. Various people for various methodological reasons disagree with the robustness or the validity of these conclusions of the meta-analysis, and some readers take Goulder’s discussion to imply that anyone doubting the Stone et al. conclusion is now accused of being a merchant of psychiatric doubt.
I don’t read Goulder’s post as making that sort of claim. I personally do think that the evidence for a substantial antidepressant effect beyond placebo is convincing, but I also agree that the methodological considerations around the data are anything but simple or settled.
Any one particular piece of evidence is almost always imperfect or limited, and both the Stone et al. and Cipriani et al. meta-analyses are imperfect in their own ways. While it is tempting to invoke these meta-analyses as the ultimate and sufficient evidence for antidepressant efficacy, what really works, in my opinion, is to zoom out and look at the totality of converging lines of evidence, from animal research to RCTs, from observational data to meta-analyses, from clinical observation to patient experience, and from the treatment of major depression and dysthymia to the treatment of anxiety disorders and OCD. And it is this converging body of knowledge that allows us to say with confidence that they are doing something clinically worthwhile, even as we debate the exact magnitude of differences from placebo, unimodal vs. multi-modal trajectories of improvement, the role of pharma bias, the burden of adverse effects, etc., etc.
Philosopher of science Nancy Cartwright and colleagues talk about the idea of “the tangle of science.” Science is a messy, interconnected web of theories, methods, models, experiments, and interventions in diverse contexts. And this tangle is why science is reliable and robust in the way it is. A tangle is considered “virtuous” when it is rich (diversity of methods and evidence), entangled (the pieces relate to each other), and long-tailed (many of the pieces in the tangle also become a part of tangles of other successful scientific endeavors).
With antidepressants, we have consistent meta-analytic evidence of drug-placebo differences across a range of depressive and anxiety disorders. We have data comparing antidepressants to other interventions (psychotherapy, exercise, psilocybin, ECT, etc.) and indirect comparisons of effect sizes to other medical interventions. We have some degree of evidence around neuropsychological shifts in emotional information processing, neuroplasticity, and neuroticism. We have evidence of behavioral effects from animal models. There is a longstanding history of successful veterinary use (how many dogs around the world are on fluoxetine?). And we have decades of clinician experience and patient testimony. The tangle robustly supports that an effect exists.
Any psychiatrist, psychologist, or researcher who believes that antidepressants do not work, that they do not have a degree of efficacy that can justify clinical use, that they are ineffective across the board for depressive and anxiety disorders, has simply lost the plot. Something has seriously gone wrong with their ability to assess how the body of knowledge adds up. That person could be very intelligent, they may even have made valuable contributions to the field, but on this point, they are wrong, and wrong in a way that warrants a stern response.
There are many issues that are unsettled around antidepressants and where disagreement is reasonable. I also don’t want to take “sides” here. Mainstream psychiatrists make all sorts of embarrassing claims all the time; some still try to deny that antidepressant withdrawal is a substantial clinical concern, and it is important to call those out as well. In fact, moderate forms of skepticism are closer to the truth. Exaggeration of antidepressant efficacy, over-reliance on them in clinical treatment, and neglect of adverse effects are all genuine concerns and actual realities on the ground. And the history of antidepressants gives us many reasons to be wary, with examples of malicious industry influence, publication bias, p-hacking, selective reporting, false promotion, and public miseducation via chemical imbalance language.
I myself have argued that public expectations around the efficacy and safety of these medications need to be tempered:
But the evidence for the efficacy of antidepressants has survived all that.
For someone skeptical that antidepressants have any clinical value, I believe they either have to commit to broad-sweeping skepticism where they are doubtful that anything works for depressive and anxiety disorders, including psychological and social interventions, a skepticism that is ultimately self-defeating and paralyzing, or they have to recognize that they are being hypocritical and engaged in isolated demands for rigor.
In 2026, to publicly doubt that antidepressants work at all is, frankly, embarrassing. It is embarrassing for the person who claims so, embarrassing for the journalists who write it up, and embarrassing for the newspapers that publish it. And if you are confused and wondering who in 2026 is saying that, well, examples are not hard to find.
We tolerate, even encourage, strong forms of skepticism in many academic contexts. We create spaces where philosophers can doubt the reality of the external worlds or the existence of other minds. We encourage scientists to challenge the foundational assumptions of our best theories. We look for anomalies in our scientific paradigms. There is a role for all that, and there is a circumscribed role for strong antidepressant skepticism too in some contexts of inquiry, but public communication comes with a responsibility, and we have to recognize when skepticism goes too far and becomes barren and corrosive. After all, we do not look kindly on scientists who publicly doubt the role of human factors in global warming. The absolute skeptics of antidepressant efficacy may remain unpersuaded, but it is time for the rest of us to move on, to better and more productive clinical and scientific questions.
See also:









Can we say that antidepressants work for some people and not for others? If so, wouldn't a clinician's time be better spent trying to discern who needs to be on an antidepressant and who doesn't?
After I developed stupor, intention block, episodes of delirium, and autonomic dysfunction from SNRIs, and long QT syndrome from antipsychotics, it was easy and tempting to think that I’d been written off for chemical sedation as “schizophrenic and incapable of insight” and/or “borderline, thus stupid and incapable of considering the humanity of others”. But it’s been half a decade since I tapered off all the meds that weren’t working and stayed on the ones that were, and now I’m far more concerned about misrepresentations of psychiatric conditions as incurable, untreatable, and incapable of mysterious spontaneous remissions. I took meds, I tapered off meds, I did a lot of self-directed therapy (DBT and also CBT aimed at OCD) and generally worked hard to improve.
And now I’m better enough that a friend who’s trained as a psychologist was shocked to learn I’d been frankly delusional and suicidally depressed less than a decade ago.
I think the impression of futility that unwarranted rejection of even the possibility of antidepressants working for some people is pretty concerning. I held on grimly throughout my 20s because I’d read the suggestion that sometimes, schizoaffective and bipolar disorders level out and attenuate in one’s 30s and 40s. It turned out true for me. I don’t think I’d be here if I hadn’t been able to read and understand the research.