Since 2000 I have been reading the literature about clinical trials for psychiatric medications. While some people do benefit, I think there is substantial doubt whether antidepressants and antipsychotics are generally any better than placebos.

Many colleagues take for granted that psychiatric medications are proven to work. I will offer some information to question this assumption. When we scrutinize the evidence a different picture emerges.

Clinical Drug Trials

a) Being shown results from selected clinical trials can be misleading. Multiple trials are conducted on a particular drug and only a small percentage will be reported in the literature. These will mainly be the successful trials, and we may not know of the many other trials. So if we hear of a good result, it could be there are more trials that found no significant benefit compared with placebo; this is common. Adding all the trial results together we might find the drug is no better than placebo. Hence scientific papers that review multiple clinical trials or conduct ‘meta-analyses’ on the collective results are more reliable.

Some examples of large combined studies:

  • Kirsch and Sapirstein (1998), in a meta-analytic review of nineteen studies involving 2,318 people, showed that 75 percent of the response to antidepressants was duplicated by placebo. They speculated that the remaining 25 percent of the positive antidepressant effect may be attributable to the un-blinding power of side effects. [My explanation: this means the obvious side-effects of the medication usually reveal to staff and patients who are receiving the medication and who get the placebo; this biases studies in favour of the drug.] Adding to the critique, Kirsch, Moore, Scoboria, and Nichols (2002) analyzed the efficacy data submitted to the US Food and Drug Administration (FDA) for the six most widely prescribed antidepressants approved between 1987 and 1999. Approximately 82% of the response to medication was duplicated by placebo control groups—57% of the studies failed to show a drug-placebo difference. When a difference was found, the drug/placebo difference was only, on average, 1.8 points on the clinician-rated Hamilton Depression Rating Scale (HDRS). FDA memoranda intimated that the clinical significance of such a small difference was questionable (Laughren, 1998).  [My comment: The HDRS has a maximum score of 53, and commonly people will need a score of more than 20 to enter a trial. Studies on antidepressants using a clinician-rated measure are more likely to find significant changes than when using a client-rated measure, suggesting clinicians report improvements that clients do not experience.] [Ref: Barry L Duncan’s book, Heart & Soul of Change, 2010; Ch 7 Psychiatric Drugs and Common Factors: An Evaluation of Risks and Benefits for Clinical Practice.]
  • Kirsch et al.(2008) provide further evidence that the belief in antidepressant efficacy is scientifically unfounded. Meta-analytically examining all trials submitted to the FDA for the licensing of four popular SSRIs (antidepressants), the authors found no clinically significant differences between placebo and the drugs, with the exception of the most distressed in the severely depressed group. Even this negligible difference was found to be due not to the drug, but to a decreased response to placebo. [Ref: Barry L Duncan, above.]


  • The World Health Organization’s studies of outcomes for schizophrenia in developed versus non-developed countries found that symptoms tend to dissipate within 5 years when medication is never used or when clients wean themselves off. Surprisingly, outcomes were better in nondeveloped countries, where most individuals did not take medication—people spent less time in hospitals with lower rates of relapse and were more likely to be employed and socially connected (de Girolamo, 1996). Harding, Zubin, and Strauss (1987) tracked 269 clients admitted to Vermont hospitals with a diagnosis of schizophrenia 32 years after their first admission. They found that about two-thirds of these former patients showed no signs of schizophrenia and had long since stopped their medications. In the Soteria Project (1971-1983) persons diagnosed with schizophrenia and randomly assigned to residential treatment with minimal use of antipsychotic medication had better 2-year outcomes than those assigned to “usual hospital” treatment (Bola & Mosher, 2003). More recently, a 15-year follow-up study of persons diagnosed with schizophrenia found that a larger percentage of those un-medicated showed better global functioning and intervals of recovery than their medicated counterparts (Martin & Jobe, 2007)—65% of those taking antipsychotics were experiencing psychosis compared to only 28% of those not medicated. Antipsychotics can remain part of the discussion, but should not be privileged. [Ref: Barry L Duncan, above.]


b) Clinical trials can be designed to get favourable results. There are endless critics of the methodology of clinical trials. For example: “Richard Smith, who resigned as editor-in-chief of the British Medical Journal because of rampant industry influence in academic research, explains that the number one aim of industry sponsored trials is to find favourable results for the company drug (Smith, 2003). He notes a host of strategies that help accomplish this goal, including comparing the industry drug against another known to be inferior, comparing a low dose of a competitor’s drug to prove efficacy and high dose to prove less toxicity, using multiple endpoints, then picking the one that casts the drug in the best light, or conducting subgroup analyses and selecting for publication those that are favourable. According to Smith, the design, conduct, analysis, and publication of clinical trials are, essentially, marketing issues.” [Ref: Barry L Duncan, above.]


c) The pharmaceutical industry in America has the most political lobbyists of any industry, far bigger than even the gun lobby. Why do they need to lobby so much if science is on their side?


Resource Information on Drug Company Behaviour


1. Below is a link to a British documentary video (approx. 60 minutes) detailing one drug company’s behaviour to increase its sales of the antidepressant known in Australia as Paroxetine. It also highlights the risk of increased suicide from using SSRIs in young people.

2. The link below allows you to listen to past radio programs by Barry Duncan on “Mind Matters,” which deals largely about the mental health industry, including drug companies.




Human beings and their biology are infinitely variable. It is not beyond possibility that occasionally a psychiatric medication has the effect the makers intend, and a placebo would not have benefitted. However, given the information from large review studies, it does not seem the norm.


If psychiatric medications are seldom better than placebos, are they worth the risk of the serious side-effects? Our society has become a society that believes them, so I do not expect they will disappear. However, as Barry Duncan says, they can “remain part of the discussion, but should not be privileged.”


Acknowledgement: I would like to thank all the doctors who take professional risks by speaking out and writing about the poor science and poor medicine in psychiatry. Psychiatrists Peter Breggin and David Healy are two authors the public may know, but there are many.


(Go to essay 6.)