Deadly Medicines and Organised Crime

by Peter Gotzsche

  In November 2010, Nancy and Shaun McCartney’s 18-year-old son, Brennan, went to their family doctor with a chest cold.11 The extroverted high school student mentioned feeling sad over breaking up with a girl he’d been seeing for 3 months. He left with a script for an antibiotic and a sample pack of Cipralex. Nancy expressed concern, as Brennan had no history of depression, but he assured her the doctor had said it would help. On the fourth day, Brennan seemed agitated when he left the house and he failed to come home. The next day his body was found. He had hung himself in a local park. Nancy wanted to warn other Canadians about Cipralex and submitted an adverse reaction report, and when she noticed a typo on her entry, she called the Vigilance Branch requesting a correction. She also asked for an updated copy but was told she’d have to file an access to information request. Seven months later, anyone searching Cipralex on MedEffect would find 317 reports, including five suicides, 12 suicide attempts and many references to suicidal ideation, but not Nancy’s submission. When the journalist writing about the tragedy asked Health Canada why, its spokesperson responded weeks later saying the entry was in the database and provided a screen grab. However, subsequent searches using the same terms failed to find it. It’s unbelievable. Not even suicides reported to the authorities may be traceable in their records.

  Here is an example that the advertising of prescription drugs to the public, which is legal in the United States, can kill healthy people who don’t need them:12

  Ten years ago my irrepressible teenage daughter Caitlin returned from holiday with relatives in the US, where prescription drugs are widely advertised; she saw an ad for an antidepressant drug called Prozac and wanted to try it. She went to our local GP and it took her 8 minutes to get the prescription. Sixty-three days later, during which time she descended into unprecedented chaos, including neural twitches, violent nightmares and self-harm, she hanged herself.

  Concealing suicides and suicide attempts in clinical trials

  I shall explore here what the true risks of suicide and suicidality with SSRIs are. They are certainly much larger than what the drug companies have told us. David Healy performed a study in 20 healthy volunteers – all with no history of depression or other mental illness – and to his big surprise two of them became suicidal when they received sertraline.13 One of them was on her way out the door to kill herself in front of a train or a car when a phone call saved her. Both volunteers remained disturbed several months later and seriously questioned the stability of their personalities. Pfizer’s own studies in healthy volunteers had shown similar deleterious effects, but most of these data are hidden in company files.13

  FDA reviewers and independent researchers found that the big companies had concealed cases of suicidal thoughts and acts by labelling them ‘emotional lability’.13,14,15 However, the FDA bosses suppressed this information. When safety officer Andrew Mosholder concluded that SSRIs cause increased suicidality among teenagers, the FDA prevented him from presenting his findings at an advisory meeting and suppressed his report. When the report was leaked, the FDA’s reaction was to do a criminal investigation into the leak.16,17

  There were other problems. In data submitted by GlaxoSmithKline to the FDA in the late 1980s and early 1990s, the company had included suicide attempts from the washout period before the patients were randomised in the results for the placebo arms of trials, but not from the paroxetine arms. A Harvard psychiatrist, Joseph Glenmullen, who studied the released papers for the lawyers, said that it’s virtually impossible that Glaxo simply misunderstood the data. Martin Brecher, the FDA scientist who reviewed paroxetine’s safety, said that this use of the washout data was scientifically illegitimate.18 Indeed. I believe it’s fraud.

  David Healy wrote in 200219 that, based on data he had obtained from the FDA, three of five suicide attempts on placebo in a sertraline trial20 had occurred during washout rather than while on placebo and that two suicides and three of six attempts on placebo in a paroxetine trial20 had also occurred in the washout period. Healy’s observations weren’t denied by Pfizer and Glaxo,21,22 but Glaxo again provided a glaring example that their lies are not of this world:22

  The ‘drug’ v. ‘true placebo’ analysis Dr Healy describes is not only scientifically invalid, but also misleading. Major depressive disorder is a potentially very serious illness associated with substantial morbidity, mortality, suicidal ideation, suicide attempts and completed suicide. Unwarranted conclusions about the use and risk of antidepressants, including paroxetine, do a disservice to patients and physicians.

  So, should we trust people who deliberately hide suicidal harms of their drug and hide trials that showed no effect and make billions out of their frauds, who are only responsible to their shareholders, and who nonetheless wants us to believe that patient welfare is their primary concern? Or should we trust an academic like Healy whose job it is to take care of the patients?

  At least three companies, Glaxo, Lilly and Pfizer, added cases of suicide and suicide attempts in patients to the placebo arm of their trials, although they didn’t occur while the patients were randomised to placebo.13,19,23,24,25 These omissions can be important for the companies in court cases. For example, a man on paroxetine had murdered his wife, daughter and granddaughter and committed suicide, but in its defence, Glaxo said that its trials didn’t show an increased risk of suicide on paroxetine.26

  The pervasive scientific misconduct has distorted seriously our perception of the benefits and harms of SSRIs. As an example, a 2004 systematic review showed that, when unpublished trials were included, a favourable risk–benefit profile changed to an unfavourable one for several of the SSRIs.27 Also in 2004, a researcher used the full reports of Glaxo’s trials that were made available on the internet as a result of litigation, and he found in his meta-analysis that paroxetine increased significantly suicidal tendencies, odds ratio 2.77 (95% confidence interval 1.03 to 7.41).14 He included three trials, among them the unpublished study 377, which didn’t show that paroxetine was better than placebo (Glaxo had stated in an internal document that ‘There are no plans to publish data from Study 377.’)28 He also included the infamous study 329. He described that an 11-year-old boy who threatened to harm himself and was hospitalised was coded as a case of exacerbated depression, and that a 14-year-old boy who had harmed himself and expressed hopelessness and possible suicide thoughts and was hospitalised was coded as a case of aggression.

  It is widely believed that SSRIs only increase suicidal behaviour in people below 25 years of age, but this is not correct. A 2006 FDA analysis of 372 placebo-controlled trials of SSRIs and similar drugs involving 100 000 patients found that up to about 40 years of age, the drugs increased suicidal behaviour, and in older patients, they decreased it (see Figure 18.1).29 However, as explained below, it is much worse than this. A major weakness of the FDA study is that the agency asked the companies to adjudicate possibly suicide-related adverse events and send them to the FDA, which didn’t verify whether they were correct or whether some had been left out. We already know that the companies have cheated shamelessly when publishing suicidal events. Why should they not continue cheating when they know that the FDA doesn’t check what they are doing? Furthermore, collection of adverse events was limited to within one day of stopping randomised treatment, although stopping an SSRI increases the risk of suicidality for several days or weeks. This rule therefore also seriously underestimated the harms of SSRIs.

  Figure 18.1 FDA meta-analysis of 372 placebo-controlled trials of SSRIs and similar drugs involving nearly 100 000 patients. Odds ratios for suicidal behaviour for active drug relative to placebo by age

  Other data show that the huge FDA analysis cannot be reliable. An internal Lilly memo from 1984 reported that the German drug agency described two suicides and 16 suicide attempts among only 1427 patients on fluoxetine in clinical trials even though patients at risk of suicide were excluded from the trials.30 A memo from Lilly Germany listed nine suicides in 6993 patients on fluoxetine in
the trials.31 In contrast, there were only five suicides in total in FDA’s analysis of 52 960 patients on SSRI drugs, or one per 10 000 patients, although one would have expected 74 and 68, respectively, based on the two Lilly reports, or 13 per 10 000 patients.

  Many suicides are missing in the FDA analysis. In a 1995 meta-analysis, there were five suicides on paroxetine in 2963 patients,32 which is 17 per 10 000 patients. This meta-analysis wrongly reported two suicides on placebo, which had occurred in the washout period. The UK drug regulator was much more careful than the FDA and did not only search for suicide terms in the documents but also read text in case report forms and narratives.33 They showed that paroxetine was harmful in adults with major depressive disorder. There were 11 suicide attempts on paroxetine (3455 patients) and only one on placebo (1978 patients), P = 0.058 for the difference. I wonder why no suicides were reported, as we would have expected six on paroxetine.

  A 2005 meta-analysis that built on data in a report the UK drug regulator had made found nine suicides in 23 804 patients,34 or four per 10 000. This was an unusually low rate, and it has been shown that the companies underreported the suicide risk.35 There were other oddities; the researchers found that non-fatal self-harm and suicidality were seriously underreported compared to the reported suicides.

  A 2005 meta-analysis of published trials including 87 650 patients conducted by independent researchers included all ages and found double as many suicide attempts on drug than on placebo.36 Even so, they found that many suicide attempts must have been missing, e.g. by asking the investigators, some of whom responded that there were suicide attempts they had not reported, while others replied that they didn’t even look for them in their trials. There were other issues related to trial design that likely led to underestimation of suicide attempts, e.g. events occurring shortly after active treatment is stopped might very well be caused by the drug but were not counted.

  It is abundantly clear that suicides, suicidality and violence caused by SSRIs are grossly underestimated,37 and we also know the reasons. First, there is outright fraud. Second, many suicidal events have been coded as something else. Third, the drug industry has taken great care to bias its trials by only recruiting people at very low risk of committing suicide. Fourth, the companies have urged the investigators to use benzodiazepines in addition to the trial drugs, which blunt some of the violent reactions that would otherwise have occurred. Fifth, some trials have run-in periods on active drug, and patients who don’t tolerate it aren’t randomised, which comes close to scientific misconduct, as it artificially minimises the occurrence of suicidality. Sixth, and perhaps the worst of all the biases, events occurring shortly after active treatment is stopped, e.g. because the patients feel very badly, might very well be suicidal events caused by the drug but are often not registered. Seventh, many trials are buried in company archives and these are not the most positive ones.

  Given what I have just described above, and earlier, e.g. that middle-aged women who use duloxetine for urinary incontinence have a suicide attempt rate that is more than double the rate among other women of a similar age, my take on all this is:

  SSRIs likely increase the risk of suicide at all ages. These drugs are immensely harmful.

  Lundbeck’s evergreening of citalopram

  Lundbeck launched citalopram (Cipramil or Celexa) in 1989. It became one of the most widely used SSRIs and provided the company with most of its income. That was a risky situation to be in but Lundbeck was lucky. Citalopram is a stereoisomer and consists of two halves, which are mirror images of each other, but only one of them is active.

  Lundbeck patented the active half before the old patent ran out and called the rejuvenated me-again drug escitalopram (Cipralex or Lexapro), which it launched in 2002. When the patent for citalopram expired, generics of Cipramil entered the market at much lower prices, but the price of Cipralex continued to be very high. When I checked the Danish prices in 2009, Cipralex cost 19 times as much for a daily dose as Cipramil. This enormous price difference should have deterred the doctors from using Cipralex, but it didn’t. The sales of Cipralex were six times higher in monetary terms than the sales of citalopram both at hospitals and in primary care. I calculated that if all patients had received the cheapest citalopram instead of Cipralex or other SSRIs, Danish taxpayers could have saved around €30 million a year, or 87% of the total amount spent on SSRIs.

  How is it possible for doctors to have such a blatant disregard for the public purse to which we all contribute and why can it continue year after year? The old recipe with a blend of money and hyped research seems infallible. A psychiatrist described vividly that when Lundbeck launched Cipralex in 2002, most of the Danish psychiatrists (she did say most, although there are more than a thousand psychiatrists in Denmark) were invited to a meeting in Paris. That meeting seems to have been enjoyable, ‘with expensive lecturers – of course from Lundbeck’s own “stable” – luxurious hotel and gourmet food. A so-called whore trip. Under influence? No, of course not, a doctor doesn’t get influenced, right?’38

  When the patent of Cipramil was expiring, Jack M Gorman published an article in a special supplement of CNS Spectrums, a neuropsychiatric journal he edits.39 The article concluded that escitalopram may have a faster onset of action and greater overall effect than citalopram’40 Gorman was a paid consultant to Forest that marketed both drugs in North America, and Forest paid Medworks Media, the publisher of CNS Spectrums, to print the article. At the same time, Medical Letter, an independent drug bulletin with no advertising, also reviewed the two drugs and found no difference between them.41

  On one of the occasions where I was invited to give a lecture for Danish psychiatrists, I expressed my doubts that a drug could be better than itself to a person sitting close to me at the lunch table. She was a chemist working at Lundbeck and didn’t agree. She sent me a copy of Gorman’s paper, which on page 2 says: ‘Brought to you by an unrestricted educational grant from Forest Pharmaceuticals, Inc.’ Oh no, I thought I would never accept ‘an unrestricted educational grant’ from a drug company, not even in the form of a reprint, but here it was. All three authors worked for Forest, Gorman as a consultant and the others in the company. The paper was a meta-analysis of three trials that compared the two drugs with placebo.

  What am I supposed to make out of a paper published in a bought supplement to a journal edited by a person who is also bought by the company? Nothing, I would say. We cannot trust the drug industry, and a paper published this way is nothing but an advertisement. There are so many ways a trial can be manipulated, and in SSRI trials it’s particularly crucial how the statistician deals with dropped out patients and other missing values.42 On top of this, Lundbeck was in a pretty desperate situation. I therefore wouldn’t believe anything unless I got access to the raw data and analysed them myself.

  But it isn’t necessary to go to such lengths. What Forest published was small differences between the two drugs and between active drugs and placebo (see Figure 18.2). After 8 weeks, the difference between the two drugs was 1, on a scale that goes from 0 to 60, and the difference between active drugs and placebo was 3. Obviously, a difference of 1 on a 60-point scale has no importance for the patients. Furthermore, as explained in Chapter 4, it doesn’t take much unblinding before we find a difference of 3 between active drugs and placebo, even if the drugs have no effect on depression. There is therefore no good reason to use a drug that is 19 times more expensive than itself.

  Figure 18.2 Change from baseline in MADRS score throughout 8 weeks; the scale goes from 0 to 60. Redrawn

  The official task of the government-funded Danish Institute for Rational Drug Therapy is to inform Danish doctors about drugs in an evidence-based fashion. In 2002, the institute reviewed the clinical documentation for Lundbeck’s me-again drug, escitalopram, and informed Danish doctors that it didn’t have clear advantages over the old drug, which contained the same active substance.43 Lundbeck complained loudly about this in the press and said
it was beyond the institute’s competence to give statements that could affect the international competition and damage Danish drug exports.44

  Although it wasn’t beyond the institute’s competence to give recommendations about new drugs, whatever the consequences for drug exports, the institute was reprimanded by the minister of health and it declined to comment when asked by a journalist, for pretty obvious reasons. The Danish drug industry has tried for years to get political backing for closing down the institute, which is a thorn in its flesh, as it reduces sales of expensive drugs, but it hasn’t succeeded.

  It seems that our highly praised governmental institute is only allowed to tell the truth about imported drugs, not about drugs we export. An untenable position that shows that principles are only valid as long as they don’t cost too much.

  Two years after these events, the institute announced that escitalopram was better than citalopram and might be tried if the effect of citalopram hadn’t been satisfactory.45 The institute must have stepped on its toes to find a politically correct way to express themselves.46 Its information to doctors now stated that they should usually choose the cheapest SSRI, as there are no major differences between the drugs. About escitalopram it said that ‘Two studies have shown that the effect of escitalopram comes somewhat faster than that of venlafaxine and citalopram, but with about the same maximum effect’, and ‘In a single study it was made likely in a subgroup analysis that escitalopram is a little better in severe depression than venlafaxine and citalopram.’