Cochrane Recommends Antidepressants for Anxiety in a Garbage In, Garbage Out Review

In early 2025, Cochrane published a seriously flawed review of antidepressants for generalised anxiety disorder.¹ The review was praised in the BMJ by journalist Jacqui Wise under the headline, “Antidepressants are effective for managing anxiety, finds Cochrane review,”² which looks like the title of a press release from a drug company.

Wise allowed the Cochrane authors to praise themselves, their review and the drugs. The senior author, Giuseppe Guaiana, said that their research showed that the drugs are “highly effective” at treating generalised anxiety disorder.²

This is totally false. All studies reported change in symptom levels on the Hamilton Anxiety Scale.³ The drugs were better than placebo, mean difference -2.79 (95% confidence interval -3.49 to -2.08). But a difference is only a difference if it makes a difference, which wasn’t the case. The possible scores go from zero to 56. For comparison, the effect of depression drugs on depression is 2 on the Hamilton Depression Scale,⁴ which goes from zero to 53,⁵ and the least clinically detectable difference to placebo is 5-6.⁶

Wise also dutifully repeated the misleading, obligatory Cochrane gobbledygook, which no one understands, that there was high certainty evidence that depression pills have a benefit over placebo.

This was also totally false. For many key outcomes – change in symptom levels, a reduction of at least 50% in the Hamilton anxiety score, total dropout rate, rate of treatment response defined by study authors, remission rate, and total number of participants reporting adverse effects – the Cochrane authors wrote that “Visual inspection of the funnel plot suggested that there may have been publication bias.”¹ In addition, most studies used the last observation carried forward method, which inflates the drug effect.⁷

This is not “high certainty evidence.” It is garbage in, garbage out. The Cochrane authors nonetheless trusted the biased garbage. It is difficult to understand and even more difficult to accept Cochrane’s motto, “Trusted evidence.”

The Cochrane abstract noted that the tiny effect they found corresponds to a number needed to treat (NNT) of seven to benefit one patient. This is extremely misleading, as the effect is too small to be clinically relevant. The authors arrived at this result by dichotomising the outcome, reporting a risk ratio of 1.41, measured as at least a 50% reduction on the anxiety scale.

One of the world’s finest biostatisticians, the late Douglas Altman, statistical advisor for the BMJ for many years, has warned against dichotomising ranking scale data and continuous variables.⁸ This statistical hocus pocus is very popular with the drug industry and their paid allies because they can then show that psychiatric drugs “work.”

Moreover, the number needed to treat with a depression drug to benefit one patient is an illusion. It doesn’t exist.⁹ The harms of the drugs are much more prevalent than their benefits. For example, depression pills harm the sex life in about half the patients,¹⁰ which means that the number needed to treat to harm one patient is only two.

The Cochrane authors reported that depression drugs improved the quality of life, mean difference 6.51 (95% confidence interval 4.95 to 8.07). This result is also extremely misleading. Only four of the 37 studies (1013 participants) provided data for this analysis. Quality of life is selectively reported to the extreme, even within clinical study reports.¹¹ Depression drugs do not improve quality of life; they worsen it.^7,11,12

Citing the Cochrane abstract, Wise wrote that antidepressants were comparable with placebo in the total number of dropouts; had fewer dropouts because of a lack of efficacy; and more dropouts because of adverse effects.

It is wrong to split dropouts into two groups this way, as they are not independent.¹² When more people drop out on pills than on placebo because of their harms, which usually set in quickly, there are fewer people left that can drop out because of lack of effect. Thus, virtually any harmful and ineffective drug could be touted by saying that fewer patients drop out because of lack of effect.

Journal publications have fewer drop-outs on drug than internal company reports, and the published data can be misleading and even false.^7,12 When we used internal company reports submitted to drug regulators, we found that 12% more patients with depression dropped out on drug than on placebo (95% confidence interval 7% to 18%).¹² Thus, the patients prefer a placebo for a depression pill, and I would expect a similar result if we repeated our analysis for anxiety trials.

Conflicts of interest

When I read scientific articles, I always look for conflicts of interest first, but in the Index that runs over 5 pages and lists an excessive 104 different statistical analyses, I overlooked them. They are on the second-last of the review’s 206 pages.¹ The Cochrane authors had not listed any conflicts in relation to drug firms, but they did not tell their readers that the studies they included, 37 placebo-controlled trials (12,226 participants), were heavily influenced by drug firms. I worked out, based on the table, “Characteristics of the included studies,” which took up 80 pages, that 26 trials were industry funded.

There was no information about funding for the remaining 11 trials. Since psychiatrists are not likely to produce matching placebos themselves or to ask a pharmacy to do it for them, I assume that these 11 trials were also industry funded. For 10 of them, the Cochrane authors noted: “Declarations of interest among primary researchers: not specified,” which is suspicious. There should always be conflicts of interest statements in reports of drug trials. For the 11^th study, some of the authors had received money from drug firms; five were employees and shareholders of Eli Lilly and Company; and one was a former employee.

When the trial authors had declared their financial conflicts of interest in relation to drug companies, they were overwhelming. The record was 73 conflicts listed for just one trial.

The Cochrane authors did not provide an overview of their data sources. I found out that in 30 cases, the data came from published trial reports, in six cases from the US trial register, and in one case from the GlaxoSmithKline trial register. All these data sources are incomplete. The authors had not acquired a single clinical study report from drug regulators, although they are far more trustworthy than publications.^7,12,13

I would not normally read beyond such revelations, as I would already know that the review is highly unreliable. The Cochrane authors not only paid no attention to these issues; they dismissed them. They wrote in the Discussion that “we did not downgrade the certainty of the evidence for sponsorship bias as there were no substantial concerns.” This statement underlines what has been documented many times before, that Cochrane is too beholden to the drug industry.¹⁴

The Cochrane authors subscribed to the by now infamous and discredited biological model of psychiatric disorders that claims they are caused by a “chemical imbalance” in the brain.¹⁵ Under the heading, “How the intervention might work,” they noted that the drugs may allow for reconstruction of the pre‐episode neural networks. This mumbo jumbo is devoid of any meaning.

The simple fact is that anxiety is caused by things making people anxious, not by some hypothesised – never proven and not even remotely likely – pre-episode defect in neural networks. If a lion attacks us, we get frightened and produce stress hormones, but it wasn’t the stress hormones that made us scared.¹⁶

What were the harms in the Cochrane review?

There were little data on harms. The Cochrane authors stated that the studies did not report data on falls, hypotension, memory impairment, or numbers with withdrawal symptoms. They could not pool data on suicides or deaths, which is not surprising, as about half of the deaths and half of the suicides in psychiatric drug trials have been left out in published trial reports.¹⁷

The authors trivialised the suicide risk: “antidepressants can be associated with worsening of suicidal ideation, especially in younger people.”¹ As the relation is causal, it is not can be but is; it is not an association, which needs not be causal; and it is far worse than just suicidal ideation. Depression drugs not only double the suicide risk, both in children and adults,^7,18  they also double suicides.¹⁹

Peter Tyrer, professor of community psychiatry, who was not involved with the review, warned in the BMJ news piece that patients have difficulties in stopping the drugs because of withdrawal problems.² He added that the main reason depression drugs are preferred to benzodiazepines is the dependence risk, so “we just seem to have shifted the problem of adverse effects from one class of drugs to another.”

Indeed. Usage of depression drugs replaced usage of benzodiazepines so that the total consumption remained about the same.²⁰ Psychiatrists denied for decades that depression pills can lead to dependence, but the withdrawal symptoms are very much the same as those for benzodiazepines.²¹ A systematic review found that half of the patients suffer from withdrawal symptoms after depression pills, and in half of them, they are very severe.²²

An online survey of patients’ experiences of withdrawing from depression drugs showed that in 40%, withdrawal symptoms had lasted over two years; 80% were moderately or severely impacted by them; and one in four were unable to stop their depression drug.²³

The Cochrane authors did not provide a single meaningful comment about three of the worst harms of depression pills: severe withdrawal symptoms, suicide, and disruption of people’s sex lives.

Cochrane denigrated psychotherapy

The Cochrane authors treated psychotherapy stepmotherly. They wrote that the few studies that have compared psychotherapy to pharmacotherapy have shown similar efficacy and that current guidelines generally recommend that psychotherapy can be used if pharmacotherapies are ineffective.

This is misleading. Psychotherapy is highly effective for anxiety, which several Cochrane reviews have demonstrated; it can cure some patients, which drugs cannot; and it has an enduring effect that outperforms that of drugs.^16,24 In addition, psychotherapy can halve the risk of a new suicide attempt in patients at high risk of suicide, those admitted to hospital after a suicide attempt.²⁵

Any future for Cochrane?

This Cochrane review is not one bad apple. Most Cochrane reviews are flawed and there are two main reasons for this. Cochrane authors and editors are keen to protect guild interests, which are often the same as the industry’s commercial interests, and they are not sufficiently critical even though the source data they review are flawed.^14,16,24

A 2021 Cochrane review of depression pills in children with depression²⁶ was also industry-friendly “Garbage in, garbage out.”²⁷ The first author, Sarah Hetrick, is editor in the Cochrane group that published the review. Although she found “small and unimportant” effects, she argued that the drugs might be recommended “for some individuals in some circumstances.” Such wishful thinking can be used about all ineffective treatments. Moreover, the abstract noted that “escitalopram may ‘at least slightly’ reduce odds of suicide‐related outcomes.” This is impossible. The truth is that these drugs double the risk of suicide in children.²⁴

At a webinar in April 2021, Professor Ken Stein, Director of the Evidence Synthesis Programme at the UK National Institute of Health and Care Research (NIHR) criticised Cochrane substantially for much the same reasons as I had done and emphasised that Cochrane authors should be iconoclastic.²⁸ Stein mentioned the failing scientific integrity of Cochrane reviews and said that, “This is a point raised by people in the Collaboration to ensure that garbage does not go into the reviews; otherwise, your reviews will be garbage.”

This was a dire forewarning. The NIHR ceased their funding of the Cochrane review groups based in the UK in March 2023, leaving Cochrane in big disarray and considerable confusion.²⁹ Now, little Denmark, my home country, was the biggest contributor to Cochrane.³⁰

I informed Cochrane on 24 June 2025 that the information on its website was untruthful. It said that “NIHR is the largest single funder of Cochrane, and currently supports 21 Cochrane Review Groups based in England.” I asked Cochrane to change the false information about who the major funders are and encouraged Cochrane to put a date on its official announcements, as it is unprofessional not to do so.

Cochrane’s response was to remove all information about who their funders are. This is not appropriate for a charity considered by outsiders to be too close to industry.¹⁴ I assume it was too embarrassing for Cochrane that the person they kicked out in 2018 after one of the worst show trials ever in academia²⁹ – me – was also the one who ensured that his country became the biggest funder of Cochrane. The Danish Government is now the only institution in the world that contributes over £1 million annually to Cochrane, the upper limit that was used on the Cochrane website.³⁰ In fact, Denmark contributes about £2 million annually to Cochrane.

The Cochrane review of antidepressants for generalised anxiety disorder is scandalously poor, misleading and harmful for the patients. I have described many other scandals in Cochrane in my two recent papers, “Cochrane on a suicide mission“ and “Requiem for the Cochrane Collaboration.¹⁴

It is a shame that this once magnificent organisation, which I co-founded in 1993, is now moribund.

References

1 Kopcalic K, Arcaro J, Pinto A, et al. Antidepressants versus placebo for generalised anxiety disorder (GAD). Cochrane Database Syst Rev 2025;1:CD012942.

2 Wise J. Antidepressants are effective for managing anxiety, finds Cochrane review. BMJ 2025;388:r209.

3 Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol 1959;32:50-5.

4 Cipriani A, Zhou X, Del Giovane C, et al. Comparative efficacy and tolerability of antidepressants for major depressive disorder in children and adolescents: a network meta-analysis. Lancet 2016;388:881-90 and Jakobsen JC, Katakam KK, Schou A, et al. Selective serotonin reuptake inhibitors versus placebo in patients with major depressive disorder. A systematic review with meta-analysis and Trial Sequential Analysis. BMC Psychiatry 2017;17:58 (In the clinical study reports of depression pills I obtained from the European Medicines Agency, the median standard deviation on the Hamilton scale after treatment was 7.5. Thus, an effect size of 0.25 corresponds to 2 on the Hamilton scale).

5 Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960;23:56-62.

6 Leucht S, Fennema H, Engel R, et al. What does the HAMD mean? J Affect Disord 2013;148:243-8.

7 Gøtzsche PC, Healy D. Restoring the two pivotal fluoxetine trials in children and adolescents with depression. Int J Risk Saf Med 2022;33:385-408.

8 Gøtzsche PC. Response rates in psychiatric drug trials are statistical nonsense. Brownstone Institute 2025;July 11.

9 Gøtzsche PC. Number needed to treat with a psychiatric drug to benefit one patient is an illusion. Mad in America 2022;Dec 13.

10 Montejo A, Llorca G, Izquierdo J, et al. Incidence of sexual dysfunction associated with antidepressant agents: a prospective multicenter study of 1022 outpatients. Spanish Working Group for the study of psychotropic-related sexual dysfunction. J Clin Psychiatry 2001;62 (suppl 3):10-21.

11 Paludan-Müller AS, Sharma T, Rasmussen K, et al. Extensive selective reporting of quality of life in clinical study reports and publications of placebo-controlled trials of antidepressants. Int J Risk Saf Med 2021;32:87-99.

12 Sharma T, Guski LS, Freund N, et al. Drop-out rates in placebo-controlled trials of antidepressant drugs: A systematic review and meta-analysis based on clinical study reports. Int J Risk Saf Med 2019;30:217-32.

13 Sharma T, Guski LS, Freund N, et al. Suicidality and aggression during antidepressant treatment: systematic review and meta-analyses based on clinical study reports. BMJ 2016;352:i65; Melander H, Ahlqvist-Rastad J, Meijer G, et al. Evidence b(i)ased medicine – selective reporting from studies sponsored by pharmaceutical industry: review of studies in new drug applications. BMJ 2003;326:1171–3; Turner EH, Matthews AM, Linardatos E, et al. Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med 2008;358:252-60; Maund E, Tendal B, Hróbjartsson A, et al. Benefits and harms in clinical trials of duloxetine for treatment of major depressive disorder: comparison of clinical study reports, trial registries, and publications. BMJ 2014;348:g3510; Le Noury J, Nardo JM, Healy D, et al. Restoring Study 329: efficacy and harms of paroxetine and imipramine in treatment of major depression in adolescence. BMJ 2015;351:h4320; Gøtzsche PC. Deadly psychiatry and organised denial. Copenhagen: People’s Press; 2015; Maund E, Guski LS, Gøtzsche PC. Considering benefits and harms of duloxetine for treatment of stress urinary incontinence: a meta-analysis of clinical study reports. CMAJ 2017;189:E194-203.

14 Godlee F. Reinvigorating Cochrane. BMJ 2018;362:k3966; Gøtzsche PC. The decline and fall of the Cochrane empire. Copenhagen: Institute for Scientific Freedom; 2022 (freely available); Gøtzsche PC. Cochrane on a suicide mission. Brownstone Institute 2025;June 20; Gøtzsche PC. Requiem for the Cochrane Collaboration. Brownstone Institute 2025;July 18.

15 Moncrieff J, Cooper RE, Stockmann T, et al. The serotonin theory of depression: a systematic umbrella review of the evidence. Mol Psychiatry 2023;28:3243-56.

16 Gøtzsche PC. Is psychiatry a crime against humanity? Copenhagen: Institute for Scientific Freedom; 2024 (freely available).

17 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.

18 Sharma T, Guski LS, Freund N, et al. Suicidality and aggression during antidepressant treatment: systematic review and meta-analyses based on clinical study reports. BMJ 2016;352:i65; Le Noury J, Nardo JM, Healy D, et al. Restoring Study 329: efficacy and harms of paroxetine and imipramine in treatment of major depression in adolescence. BMJ 2015;351:h4320; Gøtzsche PC. Deadly psychiatry and organised denial. Copenhagen: People’s Press; 2015; Bielefeldt AØ, Danborg PB, Gøtzsche PC. Precursors to suicidality and violence on antidepressants: systematic review of trials in adult healthy volunteers. J R Soc Med 2016;109:381-92.

19 Hengartner MP, Plöderl M. Newer-generation antidepressants and suicide risk in randomized controlled trials: a re-analysis of the FDA database. Psychother Psychosom 2019;88:247-8; Hengartner MP, Plöderl M. Reply to the Letter to the Editor: “Newer-Generation Antidepressants and Suicide Risk: Thoughts on Hengartner and Plöderl’s ReAnalysis.” Psychother Psychosom 2019;88:373-4; Plöderl M, Hengartner MP, Bschor T, et al. Commentary to ”antidepressants and suicidality: A re-analysis of the re-analysis“. J Affect Dis 2020;273:252-3; Hengartner MP, Amendola S, Kaminski JA, et al. Suicide risk with selective serotonin reuptake inhibitors and other new-generation antidepressants in adults: a systematic review and meta-analysis of observational studies. J Epidemiol Community Health 2021;75:523–30; Gøtzsche PC. Observational studies confirm trial results that antidepressants double suicides. Mad in America 2025;Feb 8.

20 Nielsen M, Gøtzsche P. An analysis of psychotropic drug sales. Increasing sales of selective serotonin reuptake inhibitors are closely related to number of products. Int J Risk Saf Med 2011;23:125-32.

21 Nielsen M, Hansen EH, Gøtzsche PC. What is the difference between dependence and withdrawal reactions? A comparison of benzodiazepines and selective serotonin re-uptake inhibitors. Addiction 2012;107:900–8.

22 Davies J, Read J. A systematic review into the incidence, severity and duration of antidepressant withdrawal effects: Are guidelines evidence-based? Addict Behav 2019;97:111-21.

23 Moncrieff J, Read J, Horowitz MA. The nature and impact of antidepressant withdrawal symptoms and proposal of the Discriminatory Antidepressant Withdrawal Symptoms Scale (DAWSS). J Affect Dis Rep 2024;16:100765.

24 Gøtzsche PC. Critical psychiatry textbook. Copenhagen: Institute for Scientific Freedom; 2022, page 125 (freely available).

25 Gøtzsche PC, Gøtzsche PK. Cognitive behavioural therapy halves the risk of repeated suicide attempts: systematic review. J R Soc Med 2017;110:404-10.

26 Hetrick SE, McKenzie JE, Bailey AP, et al. New generation antidepressants for depression in children and adolescents: a network meta-analysis. Cochrane Database Syst Rev 2021;5:CD013674.

27 Gøtzsche PC. Garbage in, garbage out: the newest Cochrane meta-analysis of depression pills in children. Mad in America 2021;Aug 19.

28 Gøtzsche PC. Cochrane: a sinking supertanker? Funding of UK Cochrane groups in jeopardy. Institute for Scientific Freedom 2021;May 15.

29 Gøtzsche PC. Whistleblower in healthcare (autobiography). Copenhagen: Institute for Scientific Freedom 2025;April 8 (freely available).

30 Our funders and partners. Cochrane website, undated.