Seriously Misleading Testimony by Psychiatry Professor in Oslo District Court About the Effect of Antipsychotics

Lawsuits are a means to obtain much-needed changes in an inhumane psychiatry, which routinely violates basic human rights and national laws about forced treatment without facing any consequences.

I am an expert witness in an ongoing lawsuit against the Norwegian State and testified on 20 November 2024 in Oslo district court on behalf of the plaintiff, Inger-Mari Eidsvik, and her lawyers from Føniks Advokater, Stine Moen, Gro Hermansen and Professor Mads Andenæs from the Law Faculty of the University of Oslo. They were assisted by Marie Hellesylt and Louise Bardsen from the Oslo law firm Wikborg Rein who intervened on behalf of the International Commission of Jurists in Norway.

We all work pro bono because we consider the case to be very important. If we lose, the plan is to appeal to the European Court of Human Rights in Strasbourg.

Like in most other countries, Norwegian law stipulates that forced drugging can only be used when, with high probability, it can lead to a cure or substantial improvement in the patient’s condition, or if the patient avoids a substantial worsening of the disease. Moreover, the beneficial effect must clearly outweigh the harms (euphemistically called “disadvantages of possible side effects”).

Norwegian former Supreme Court Attorney Ketil Lund and I explained in 2016 in a law journal why forced medication with antipsychotics cannot be justified because the effect is poor while they have considerable harms.[1] In 2019, the Ombudsman concluded about a concrete case that it violated the Psychiatry Act to use forced treatment with an antipsychotic.[2] Among other things, the Ombudsman mentioned our argument that it is misleading to assert that antipsychotics can prevent relapses in a quarter of the patients, because what is called recurrence when the patients no longer receive the drug is very often withdrawal symptoms.

However, Norwegian psychiatrists have not changed their practice. I noted in my testimony that the effect of antipsychotics is considerably smaller than the least clinically relevant effect, which psychiatrists have established is 15 points on the Positive and Negative Syndrome Scale (PANSS).[3] In submissions to the FDA of placebo-controlled trials of newer antipsychotics, the average effect was only 6 points on PANSS,[4] even though it is easy for scores to improve quite a bit if someone is calmed down by a major tranquilliser—as these drugs were originally called—and express their abnormal ideas less frequently.[5]

Erik Johnsen, professor of psychiatry in Helse, Bergen, gave testimony after me via a video link and I saw most of it. He said that 80% of patients who receive an antipsychotic for the first time have at least a 20% effect, while around 50% have at least a 50% effect.[6] He pointed to documentation from over 400 individual studies of acute treatment of psychosis.

Since Johnsen claimed an effect of 20% in 80% of the patients, which contrasted with my testimony, the judge asked what the 20% meant but she did not get a meaningful reply.

Johnsen also said that the effect was better when antipsychotics were used as forced treatment because it was possible to support the patient during treatment. I had said the opposite. Patients who are forced to take an antipsychotic sometimes hide it in their mouth and spit it out when no one is watching, and when they get out of hospital, some stop taking the drug cold turkey, which is dangerous and increases their risk of dying. Moreover, in the placebo-controlled trials, the patients give informed consent, which means there is no coercion, and as they are positive to using antipsychotics, this will likely exaggerate the measured effect because of expectation bias, as the intended blinding is often broken because of the conspicuous side effects of the drugs.

Johnsen also claimed that the worse the symptoms are, the better the effect. This is a mathematical artefact. As I have explained,[7] when the change in symptoms is regressed on initial symptom severity, this corresponds to looking at (x−y) = ax +b, where x is initial value and y the final value. Since x appears on both sides of the equation, 50% of the variation is already explained. This means that a false relationship will be obtained even when the two factors are totally unrelated.

The 2017 meta-analysis by Zhu et al.

Johnsen’s claim of an effect of 20% in 80% of the patients is invalid. It comes from the most flawed meta-analysis I have seen in my entire career.[8] I had never seen a meta-analysis based on single treatment arms from randomised trials, all of which had at least two treatment arms. The whole idea with randomised trials is to have a control group, and if we wish to know what the effect of a drug is, we will need to compare it with the effect in a group treated with placebo, but the authors did not include a single placebo-controlled study.

In a single-arm meta-analysis, the drug effect includes the spontaneous remission that will occur even without treatment. It is a meta-analysis of the before-after values of symptom severity, a method strongly recommended against because it is fatally flawed. The authors concluded that “the response rate of first-episode patients with schizophrenia to antipsychotics is rather high,” but they had no idea about what the response rate was, as there was no placebo comparator.

There were many other issues. The authors found that the response was greater when the disease severity was greater, which, as noted, is a mathematical artefact.

They included studies without a control group, which they called “open-label” studies. In such studies, blinding is impossible, which will tend to exaggerate the measured effect. Only 12 of their 17 included studies were blinded. The authors wrote that “The test for subgroup differences of response rate between blinded studies and open-label studies was just not statistically significant” (P = 0.055). But the test did not have much power, and they acknowledged in the Discussion that blinded studies reported smaller effects.

Only 5 studies reported 20% or 50% reductions from baseline, and the authors therefore imputed the missing values, which is problematic. They wrote that details were often missing about randomisation procedures and that the dropout rate was high (39%). They claimed there “was no important selective reporting” but did not explain how they could know this. What we do know is that selective reporting is the norm, not the exception,[9] and in psychiatric drug trials, about half of the deaths and half of the suicides have been left out in published trial reports.[10]

The authors wrote that their results were more positive than those in another meta-analysis where only 23% reached a 50% reduction from baseline.[11] This meta-analysis had psychiatrist Stefan Leucht as first author, and he was last author on the single-arm meta-analysis. He was also first author on the study that determined the least clinically relevant effect of antipsychotics.

Leucht declared extensive conflicts of interest: “In the last three years Stefan Leucht has received honoraria for lectures from Eli Lilly, Lundbeck (Institute), Pfizer, Janssen, BMS, Johnson and Johnson, Otsuka, Roche, Sanofi Aventis, ICON, Abbvie, AOP Orphan, Servier; for consulting/advisory boards from Roche, Janssen, Lundbeck, Eli Lilly, Otsuka, TEVA; for the preparation of educational material and publications from Lundbeck Institute and Roche.”

The authors wrote that a study sponsored by the US National Institute of Mental Health from 1964 showed that 61% of the patients on antipsychotics became much improved versus only 22% on placebo. These percentages appear nowhere in the article they cite.[12] The article shows the results in a graph from which I measured that 69% versus 34% were very much or much improved. Curiously, the only numerical information in the article didn’t appear before the summary at the end, and one of the percentages was not even stated precisely:

“Over 75% showed marked to moderate degrees of improvement … only 23% of the placebo group were rated as showing marked or moderate improvement.” Now, the difference between drug and placebo was even higher than in the graph, and a new terminology was introduced, marked or moderate improvement.

I have explained why this study is totally untrustworthy:[13]

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 psychosis pills are effective. It was a trial of 344 newly admitted patients with schizophrenia who were randomised to phenothiazines such as chlorpromazine, or to placebo. The investigators reported, without offering any numerical data, that the drugs reduced apathy (indifference) and made movements less retarded, the exact opposite of what these drugs do to people, which the psychiatrists had admitted a decade earlier. 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 tranquillisers but antischizophrenic drugs. Their study contributed to shaping the erroneous beliefs that schizophrenia can be cured with drugs and that psychosis pills should be taken indefinitely.

The 2017 meta-analysis by Leucht et al.

This meta-analysis[14] is also highly problematic. The authors included 167 placebo-controlled trials. All the trials were seriously flawed because the patients were already receiving an antipsychotic before they were randomised. Thus, patients randomised to placebo were exposed to a cold turkey, which is very harmful and may elicit a withdrawal psychosis, which is not a true psychosis, as it would not have happened without the cold turkey.

Despite this flaw, the authors only found minor effects, “At least a ‘minimal’ response occurred in 51% of the antipsychotic group versus 30% in the placebo group, and

23% versus 14% had a ‘good’ response.” They reported that the number needed to treat to benefit one patient was 6, but as there were only 9% more patients with a good response, the correct number is 11 ( = 1/9%).

They claimed that quality of life (effect size 0.35) and functioning (0.34) improved, but these data are unreliable, given that the placebo group patients were harmed by drug withdrawal effects. In surveys of their experiences, patients report the opposite. In a survey of 200 patients, over 80% reported that side effects had negatively impacted their work and social functioning; 56% had stopped taking the drug; and side effects commonly reported as having led to stopping treatment were “feeling like a ‘zombie’” (22%), feeling drowsy/tired (21%), and weight gain (20%).[15]

A survey of 2,031 Australians showed that people thought that antipsychotics, antidepressants, electroshock and admission to a psychiatric ward were more often harmful than beneficial.[16] The social psychiatrists who had done the survey were dissatisfied with the answers and argued that people should be trained to arrive at the “right opinion.”

The authors of the 2017 Leucht meta-analysis said, with reference to the 1964 study, that 61% versus 22% of patients were much improved, which appear nowhere in the cited article.

Many of the trials they included were of very poor quality: “The reports often did not indicate full details about sequence generation or allocation concealment. Descriptions of the methods and success of blinding were frequently insufficient as well. The data confirmed the high dropout rates in current schizophrenia studies (mean 37.2%, SD 20.5). Older studies were poorly reported, making it often impossible to extract outcome data (50% of the studies had a high risk of selective reporting).”

They found that small studies reported bigger effects than large studies, which means they should have abstained from lumping all the studies.

Interestingly, they reported that the mean effect was only 9.7 on PANSS, which is considerably smaller than the least clinically relevant effect, which Leucht himself determined is 15. So, they should have concluded that the drugs didn’t work.

To summarise, the Zhu 2017 meta-analysis and the Leucht 2017 meta-analysis can best be described as garbage in, garbage out. Amazingly, the latter review filled 16 pages, which is a very long article, in one of the flagship journals of psychiatry, the American Journal of Psychiatry.

I shall write more about the pivotal Norwegian court trial later.

References

[1] Gøtzsche PC, Lund K. Tvangsmedisinering må forbys. Kritisk Juss 2016;2:118-57.

[2] Gøtzsche PC. Forced drugging with antipsychotics is against the law: decision in Norway. Mad in America 2019; 4 May.

[3] Leucht S, Kane JM, Etschel E, et al. Linking the PANSS, BPRS, and CGI: clinical implications. Neuropsychopharmacology 2006;31:2318-25.

[4] Khin NA, Chen YF, Yang Y, et al. Exploratory analyses of efficacy data from schizophrenia trials in support of new drug applications submitted to the US Food and Drug Administration. J Clin Psychiatry 2012;73:856–64.

[5] Moncrieff J. The bitterest pills. Basingstoke: Palgrave Macmillan; 2013.

[6] Rettssak mot staten for tvangsbruk. NRK 2024; Nov 22.

[7] Gøtzsche P. Depression severity and effect of antidepressant medications. JAMA 2010;303:1597.

[8] Zhu Y, Li C, Huhn M, et al. How well do patients with a first episode of schizophrenia respond to antipsychotics: A systematic review and meta-analysis. Eur Neuropsychopharmacol 2017;27:835-44.

[9] Chan A-W, Hróbjartsson A, Haahr MT, Gøtzsche PC, Altman DG. Empirical evidence for selective reporting of outcomes in randomized trials: comparison of protocols to published articles. JAMA 2004;291:2457-65.

[10] Hughes S, Cohen D, Jaggi R. Differences in reporting serious adverse events in industry sponsored clinical trial registries and journal articles on antidepressant and antipsychotic drugs: a cross-sectional study. BMJ Open 2014;4:e005535

[11] Leucht S, Leucht C, Huhn M, et al. Sixty years of placebo-controlled antipsychotic drug trials in acute schizophrenia: systematic review, Bayesian meta-analysis, and meta-regression of efficacy predictors. Am J Psychiatry 2017;174:927-42.

[12] Cole JO. Phenothiazine treatment in acute schizophrenia; effectiveness: the National Institute of Mental Health Psychopharmacology Service Center Collaborative Study Group. Arch Gen Psychiatry 1964;10:246-61.

[13] Gøtzsche PC. Mental health survival kit and withdrawal from psychiatric drugs. Ann Arbor: L H Press; 2022, page 34.

[14] Leucht S, Leucht C, Huhn M, et al. Sixty years of placebo-controlled antipsychotic drug trials in acute schizophrenia: systematic review, Bayesian meta-analysis, and meta-regression of efficacy predictors. Am J Psychiatry 2017;174:927-42.

[15] Doane MJ, Sajatovic M, Weiden PJ, et al. Antipsychotic treatment experiences of people with schizophrenia: patient perspectives from an online survey. Patient Prefer Adherence 2020;14:2043-54.

[16] Jorm AF, Korten AE, Jacomb PA, et al. “Mental health literacy”: a survey of the public’s ability to recognise mental disorders and their beliefs about the effectiveness of treatment. Med J Aus 1997;166:182-6.