Editor’s Note: Over the next several months, Mad in America will publish a serialized version of Peter Gøtzsche’s book, Mental Health Survival Kit and Withdrawal from Psychiatric Drugs. In this blog, he discusses the science behind psychiatric drugs and the flaws and bias that corrupt drug trials. Each Monday, a new section of the book will be published, and all chapters will be archived here.
Flawed trials have led the psychiatrists astray
The rating scales used in placebo-controlled trials of psychiatric drugs for measuring the reduction of symptoms have made the psychiatrists believe that the drugs work and that the effect is specific for the disorder being treated. However, such results do not say anything about whether the patients have been cured or can live a reasonably normal life. Furthermore, the effects measured with these scales are unreliable. Virtually every single placebo-controlled drug trial in psychiatry is flawed.4,54
Since the trials are flawed, systematic reviews of trials are also flawed, and guidelines are flawed. Even the drug approval process is flawed. The drug regulators don’t pay enough attention to the flaws. They don’t even ask the drug companies for the many missing data or appendices which, according to the indexes the companies provide, should have been included in their applications.55
Cold turkey in the placebo group
In the vast majority of the trials, the patients were already on a drug similar to the one being tested against placebo. After a short wash-out period without this drug, the patients are randomised to the new drug or placebo. There are three main problems with this design.
First, the patients who are recruited for the trials are those who have not reacted too negatively on getting such a drug before.52 They will likely therefore not react negatively to the new drug, which means that the trials will underestimate the harms of psychiatric drugs.
Second, when patients who have tolerated a psychiatric drug are randomised to placebo, they will likely react more negatively to this than if they had not been in treatment before. This is because psychiatric drugs have a range of effects, some of which can be perceived as positive, e.g. euphoria or emotional numbing.
Third, the cold turkey withdrawal that some patients in the placebo group go through harms them. It is therefore not surprising that the new drug seems to be better than placebo. Introducing longer wash-out periods does not remove this problem. If people have been permanently brain damaged before entering the trials, wash-out periods cannot compensate, and even if that is not the case, they could suffer from withdrawal symptoms for months or years.7,56,57
Thousands of trials of neuroleptics have been carried out, but when my research group recently searched for placebo-controlled trials in psychosis that only included patients who had not received such a drug earlier, we found only one trial.58 It was from China and appeared to be fraudulent. Thus, all placebo-controlled, randomised trials of neuroleptic drugs in patients with schizophrenia spectrum disorders were flawed, which means that the use of neuroleptic drugs cannot be justified based on the evidence we currently have.4
The first trial that was not flawed was published on 20 March 2020,59 70 years after the discovery of the first neuroleptic, chlorpromazine, that Rhône-Poulenc marketed in 1953 with the trade name Largactil, which means broad activity. However, even 70 years wasn’t enough for the psychiatrists to come to their senses. They were not yet ready to draw conclusions from their results, which their abstract demonstrates:59
“Group differences were small and clinically trivial, indicating that treatment with placebo medication was no less effective than conventional antipsychotic treatment (Mean Difference = -0.2, 2-sided 95% confidence interval -7.5 to 7.0, t = 0.060, p = 0.95). Within the context of a specialised early intervention service, and with a short duration of untreated psychosis, the immediate introduction of antipsychotic medication may not be required for all cases of first episode psychosis in order to see functional improvement. However, this finding can only be generalised to a very small proportion of FEP [first episode psychosis] cases at this stage, and a larger trial is required to clarify whether antipsychotic-free treatment can be recommended for specific subgroups of those with FEP.”
I have translated what this means for those of us who do not have guild interests to defend:
“Our study was small, but it is unique because it only included patients who had not been treated with a neuroleptic before. We found that neuroleptics are not needed for patients with untreated psychosis. This is great progress for patients, as these drugs are highly toxic and make it difficult for them to come back to a normal life. Based on the totality of the evidence we have, the use of neuroleptics in psychosis cannot be justified. Neuroleptics should only be used in placebo-controlled randomised trials of drug-naïve patients.”
The authors of a 2011 Cochrane systematic review of neuroleptics for early episode schizophrenia pointed out that the available evidence doesn’t support a conclusion that antipsychotic treatment in an acute early episode of schizophrenia is effective.60 This is one of the very few Cochrane reviews of psychiatric drugs that can be trusted.4,54 There are huge problems with most Cochrane reviews, e.g. Cochrane reviews in schizophrenia routinely include trials in a meta-analysis (which is a statistical summary of the results of several trials) where half of the data are missing.4 This is garbage in, garbage out, with a nice little Cochrane logo on it, as Tom Jefferson said in an interview in the article, “Cochrane – a sinking ship?”61
To find out for how long patients should be advised to continue taking their drugs, so-called maintenance studies, also called withdrawal studies, have been carried out. These studies are highly misleading because of cold turkey effects. A large meta-analysis of 65 placebo-controlled trials found that only three patients needed to be treated with neuroleptics to prevent one relapse after one year.62 This looks very impressive, but the result is totally unreliable. The apparent benefit of continued treatment with neuroleptics decreased over time and was close to zero after three years. Thus, what was seen after one year was iatrogenic harm, which was described as a benefit.
When follow-up is longer than three years, it turns out that discontinuing neuroleptics is the best option. There is only one appropriately planned and conducted maintenance trial, from Holland. It has seven years follow-up, and patients who had their dosages decreased or discontinued fared much better than those who continued taking neuroleptics: 21 of 52 (40%) versus 9 of 51 (18%) had recovered from their first episode of schizophrenia.63
Leading psychiatrists interpret the maintenance studies of neuroleptics and depression pills to mean that these drugs are highly effective at preventing new episodes of psychosis and depression, respectively,4 and that the patients should therefore continue taking the drugs for years or even for life.
Danish researchers tried to repeat the Dutch study, but their trial was abandoned because the patients were scared about what would happen if they did not continue taking their drugs. A psychiatrist involved with the failed trial told me about another, recent withdrawal trial, carried out in Hong Kong.64 The researchers treated first-episode patients with quetiapine (Seroquel) for two years; discontinued the treatment in half of the patients by introducing placebo; and reported the results at ten years. They found that a poor clinical outcome occurred in 35 (39%) of 89 patients in the discontinuation group and in only 19 (21%) of 89 patients in the maintenance treatment group.
I immediately suspected that the trial was flawed, as this result was the exact opposite of the Dutch result, and that they had tapered off the neuroleptic far too quickly, even cold turkey. As there was nothing about their tapering scheme in the article, I looked up an earlier publication, of the results at three years.65 They didn’t taper at all; all patients randomised to placebo quit their neuroleptic cold turkey.
The ten-year report was highly revealing: “A post-hoc analysis suggested that the adverse consequences of early discontinuation were mediated in part through early relapse during the 1-year period following medication discontinuation.”64
The investigators defined a poor outcome as a composite of persistent psychotic symptoms, a requirement for clozapine treatment, or death by suicide. They called their trial double-blind, but it is impossible to maintain the blind in a trial with cold turkey symptoms, and it is highly subjective whether there are any psychotic symptoms and whether clozapine should be given.
I’m much more interested in whether the patients return to a normal life, and a table showed that after ten years, 69% of those who continued taking their drug were employed versus 71% in the cold turkey group, a quite remarkable result considering the iatrogenic harms inflicted on the latter group.
I consider this trial highly unethical because some patients commit suicide when they experience cold turkey effects. Robert Whitaker has demonstrated that this trial design is lethal.1,66 One in every 145 patients who entered the trials for risperidone (Janssen), olanzapine (Eli Lilly), quetiapine (AstraZeneca) and sertindole (Lundbeck) died, but none of these deaths were mentioned in the scientific literature, and the FDA didn’t require them to be mentioned. The suicide rate in these clinical trials was 2-5 times higher than the norm.
It is no wonder that AstraZeneca (seller of quetiapine) was happy to fund a trial in Hong Kong that was seriously flawed in favour of their drug.64
The investigators’ attempt at explaining away what they found is breathtaking. They wrote that their result in the third year raised the suggestion that, “there might be a time window or critical period during which a relapse might be course-modifying.”
The plausibility of the existence of such a time window between year two and three is zero. As it is highly variable when or if a patient relapses, there cannot exist any such time window. The psychiatrists deliberately harmed half of their patients, but they concluded they did nothing wrong and that their patients, or their disease, or a “time window” are to blame.
Lack of blinding
Because of the conspicuous side effects of psychiatric drugs, trials labeled double-blind are not double-blind. Quite a few patients—and their doctors—know who is on drug and who is on placebo.4 It takes very little unblinding in a trial before the small differences recorded can be explained purely by bias in the outcome assessment on a subjective rating scale.4
In trials supposed to be double-blind, investigators may report positive effects that only exist in their imagination. This occurred in a famous trial funded by the US National Institute of Mental Health in 1964, which is still highly cited as evidence that neuroleptics are effective. It was a trial of 344 newly admitted patients with schizophrenia who were randomised to phenothiazines such as chlorpromazine, or to placebo.67
The investigators reported, without offering any numerical data, that the drugs reduced apathy and made movements less retarded, the exact opposite of what these drugs do to people, which the psychiatrists had admitted a decade earlier.3 The investigators claimed a huge benefit for social participation (effect size of 1.02) and that the drugs make the patients less indifferent to the environment (effect size 0.50). The drugs do the opposite.
They also claimed, with no data, that 75% versus 23% were markedly or moderately improved and suggested that the drugs should no longer be called tranquillizers but antischizophrenic drugs. Their study contributed to shaping the erroneous beliefs that schizophrenia can be cured with drugs and that neuroleptics should be taken indefinitely.1
Neuroleptics do not have clinically relevant effects on psychosis. Despite the formidable biases—cold turkey, lack of blinding, and industry funding that involves torturing the data till they confess,4,51—the published outcomes have been very poor.4 The smallest clinically relevant effect would be a change of about 15 points on the Positive and Negative Syndrome Scale (PANSS)68 commonly used in the trials. Yet, what was reported in placebo-controlled trials of recent drugs submitted to the FDA was only 6 points69—even though scores easily improve when someone is knocked down by a tranquillizer and expresses abnormal ideas less frequently.9
Depression drugs don’t work either. The smallest effect that can be perceived on the Hamilton scale is a 5-6 point change,70 but a change of only about 2 is obtained in flawed trials.71,72
Some meta-analyses have found that the effect of depression pills is larger if the patients are severely depressed,71,73,74 and the pills are generally recommended for severe and sometimes also for moderate depression. However, the reported effects are very small for all depression severities, e.g. 2.7 points for patients with a baseline Hamilton score above 23 which is considered very severe depression,74 and 1.3 for milder degrees.71
Moreover, it is likely just a mathematical artefact that the effect seems to be slightly larger in severe depression.75 Since the baseline scores for severe depression are larger than for mild depression, any bias will influence the measured result more in patients with severe depression than in those with mild depression. If we assume the unblinding bias is 10% when estimating the effect in the drug group and, for the simplicity of the example, that there is no bias in the placebo group and no improvement between baseline and the final visit, then a Hamilton baseline score of 25 would still be 25 after treatment. But because of the bias, there would be a 2.5-point difference between drug and placebo. If the baseline is 15, that difference would only be 1.5.
The small effect of depression pills measured in flawed trials disappear if the placebo contains atropine, which has similar side effects as the pills, e.g. dry mouth.76 Such trials were done many years ago when the depression pills were tricyclics. Many psychiatrists say that these are more effective than newer depression pills (but also more dangerous, which is why they are rarely used). Despite this, the effect in seven trials with atropine in the placebo only corresponded to 1.3 points on the Hamilton scale.76 The “effect” of the newer drugs is around 2.3, or almost double as much.71,72
It is very easy to make almost any substance with side effects “work” for depression, including stimulants.77 Three of the 17 items on the Hamilton scale are about insomnia, and that issue alone can yield six points on the scale,53 or three times as much as the “effect” in biased trials. And if a person goes from maximum anxiety to no anxiety, eight points can be earned.
What do doctors want to achieve with drugs? Above all, to avoid suicides and deaths from other causes. Moreover, to get the patients back to a normal life, with good social contacts.
Sometimes, this cannot be obtained. Most patients who get a diagnosis of depression live depressing lives, e.g. are married to the wrong person, have a bullying boss, a tedious job, or a chronic disease, and it is hardly the job of doctors to try to get them out of this predicament. Moreover, there are no pills for this, but these people are nonetheless routinely prescribed depression pills, which are seen as the “solution” to life’s troubles.
A score on a rating scale doesn’t tell us whether the patient is well. Over a thousand placebo-controlled trials of depression pills have been carried out, but I have not seen any that measured whether the patients were cured by a drug, i.e. whether they came back to a normal productive life. If any such trials existed, we would have known about them. Unless they showed that the drugs made the situation worse and were therefore buried in company archives.4
According to the American Psychiatric Association’s disease manual, DSM-5, major depression is present when the patient exhibits 5 or more of 9 symptoms that “cause clinically significant distress or impairment in social, occupational, or other important areas of functioning.” Given how the disorder is defined, it makes no sense that drug trials don’t use these outcomes.
One trial with such outcomes was not appropriate, as it was a withdrawal trial that only told us that the cold turkey withdrawal harms that the psychiatrists inflicted on patients were bigger for some drugs than for others.78 Unsurprisingly, patients on fluoxetine (the sponsor’s, Eli Lilly’s product) could endure a short treatment interruption where the patients got placebo because this drug has an active metabolite with a very long half-life. Paroxetine has a short half-life, and even after missing just one dose, a statistically significant increase in harms occurred, which worsened during the next five days.
Lilly’s trial was grossly unethical. The abstinence symptoms after paroxetine withdrawal were severe, which was expected based on clinical observations and a previous, similar study also sponsored by Lilly.79 The patients experienced “statistically significantly worsened severity in nausea, unusual dreams, tiredness or fatigue, irritability, unstable or rapidly changing mood, difficulty concentrating, muscle aches, feeling tense, chills, trouble sleeping, agitation and diarrhoea during placebo substitution.”78
In Lilly’s previous trial,79 roughly one-third of the patients on paroxetine or sertraline experienced worsened mood, irritability and agitation, and had an increase in the Hamilton score of at least eight, which is the difference between being mildly depressed and severely depressed.74
Lilly sacrificed the patients for a marketing advantage. Many patients suffered from an abstinence depression caused by the cruel trial design, and the various harms they experienced increase the risk of suicide, violence, and homicide.4 This was known long before the trial was carried out.2,4,80
Unsurprisingly, “Patients treated with paroxetine reported statistically significant deterioration in functioning at work, relationships, social activities and overall functioning.”78
- If you are being asked to participate in a clinical trial with a psychiatric drug, you will need to investigate very carefully what it is about and whether it is ethically acceptable.
- Ask the doctor for all documentation, including the full trial protocol and investigator’s brochure, which might be the only place where the harms have been listed, also from animal experiments.
- Pay attention to conflicts of interest. Will the doctor or the department benefit financially from conducting the trial?
- Will the raw data in anonymised form be made available to the investigators and to independent researchers, allowing them to do their own analyses? Will all patients who ask for these data get them?
- Ensure you get written confirmation before you make a decision. If the data will not be made available, or your doctor becomes uncomfortable when you ask, you should refuse to participate and publish your experiences as a deterrent for bad research practice and to warn other patients.
To read the footnotes for this chapter and others, click here.
Mad in America hosts blogs by a diverse group of writers. These posts are designed to serve as a public forum for a discussion—broadly speaking—of psychiatry and its treatments. The opinions expressed are the writers’ own.