Deadly Medicines and Organised Crime

by Peter Gotzsche

  Off-label promotion exposes patients to harms with no assurance of benefit. This criminal activity has increased and its victims have died, suffered heart attacks and strokes, had permanent nerve damage or lost their eyesight.14 In 2010, a jury found that Pfizer violated the federal Racketeer Influenced and Corrupt Organizations Act (RICO) and the company was to pay $142 million in damages.15 The jury found Pfizer engaged in a racketeering conspiracy over a 10-year period. Pfizer never told doctors or patients that its studies had shown that Neurontin was no more effective than a placebo for some of its off-label uses.

  References

  1 Tansey B. Huge penalty in drug fraud: Pfizer settles felony case in Neurontin off-label promotion. San Francisco Chronicle. 2004 May 14.

  2 Harris G. Pfizer to pay $430 million over promoting drug to doctors. New York Times. 2004 May 14.

  3 Lenzer J. Pfizer pleads guilty, but drug sales continue to soar. BMJ. 2004; 328: 1217.

  4 Angell M. The Truth about the Drug Companies: how they deceive us and what to do about it. New York: Random House; 2004.

  5 Petersen M. Our Daily Meds. New York: Sarah Crichton Books; 2008.

  6 Petersen M. Suit says company promoted drug in exam rooms. New York Times. 2002 May 15.

  7 Landefeld CS, Steinman MA. The Neurontin legacy – marketing through misinformation and manipulation. N Engl J Med. 2009; 360: 103–6.

  8 Petersen M. Court papers suggest scale of drug’s use. New York Times. 2003 May 30.

  9 Dickersin K. Reporting and other biases in studies of Neurontin for migraine, psychiatric/bipolar disorders, nociceptive pain, and neuropathic pain. Available online at: www.pharmalot.com/wp-content/uploads/2008/10/neurontin-dickersin-2.pdf (accessed 10 December 2008).

  10 Saul S. Experts conclude Pfizer manipulated studies. New York Times. 2008 Oct 8.

  11 Johansen HK, Gøtzsche PC. Problems in the design and reporting of trials of antifungal agents encountered during meta-analysis. JAMA. 1999; 282: 1752–9.

  12 Johansen HK, Gøtzsche PC. Amphotericin B vs fluconazole for controlling fungal infections in neutropenic cancer patients (Cochrane Review). In: The Cochrane Library, Issue 1. Oxford: Update Software; 2000.

  13 Krumholz SD, Egilman DS, Ross JS. Study of Neurontin: titrate to effect, profile of safety (STEPS) trial: a narrative account of a gabapentin seeding trial. Arch Intern Med. 2011; 171: 1100–7.

  14 Adams C, Young A. Off-label prescription case reflects federal concern over unsafe uses. Knight Ridder Newspapers. 2004 May 14.

  15 Voris B, Lawrence J. Pfizer Told to Pay $142.1 million for Neurontin Fraud. Bloomberg. 2010 March 25.

  13

  Merck, where the patients die first

  On 30 September 2004, Merck withdrew its COX-2 inhibitor, the anti-arthritis drug Vioxx (rofecoxib) from the market. I was in Canada and browsed the TV stations to induce natural sleep when I learned about it on Fox News. What was more surprising to me than the withdrawal of the drug was that the president of the US Arthritis Foundation lamented for about 10 minutes about what a great loss it was for the patients that Vioxx was no longer available. If I hadn’t known who was speaking, I would have guessed it was the CEO of Merck. Company talk all over. For a full 10 minutes. I usually get 30 seconds when I’m on the news.

  This speaks volumes about the extent to which patient organisations collude with big pharma. I checked the website for the Arthritis Foundation, and it had Pfizer’s logo on its opening page. In contrast to the Foundation’s hype about the drug, the jury in a court case stated that Merck showed ‘malicious, oppressive, and outrageous’ conduct and found it guilty of four counts of fraud in marketing rofecoxib.1

  It was known right from the start that COX-2 inhibitors, via their mechanism of action, must increase the risk of thrombosis. In 1996, Merck scientists discussed the heart attack risk,2 and investigators sponsored by Merck found that Vioxx reduced urinary metabolites of prostacyclin in healthy volunteers by about half,3 which indicates that Vioxx causes thrombosis. However, Merck convinced the authors to change what they had written into a meaningless sentence: ‘Cox-2 may play a role in the systemic biosynthesis of prostacyclin.’ Also in 1997, a Merck scientist said that if they didn’t allow patients to use aspirin in their trials (which decreases the risk of a heart attack), patients on Vioxx might have more heart attacks and that would ‘kill the drug’.4 Merck surely concealed how dangerous Vioxx was. A senior Merck scientist proposed to leave out people with a high risk of cardiovascular problems in the company’s planned VIGOR study so that the difference in heart complications between Vioxx and other NSAIDs ‘would not be evident’.5 None of the trials in the FDA submission were designed to evaluate the cardiovascular risk.3

  As mentioned in Chapter 10, FDA also had serious concerns about the drug. When the FDA approved rofecoxib for marketing in May 19994 despite disconcerting evidence in the application, it stated that it lacked ‘complete certainty’ that the drug increased cardiovascular risk.4,7

  I find this extraordinary. Imagine how absurd it would be if a doctor said to a patient: ‘I’m not completely sure that this drug might kill you, so please take it.’ If there had been patient representatives in the advisory committee, they would probably have rejected the application and demanded of Merck that it tested its drug more carefully, as it was clear that the drug must cause thrombosis. Further, as there were many other NSAIDs on the market, the drug wasn’t needed.

  The scandal of the COX-2 inhibitors is really monumental. The drugs were approved based on small, short-term trials that didn’t look for cardiovascular harms, in patients with a low risk for such events, although nearly half of real world patients with arthritis have coexisting cardiovascular disease.8,9 Merck did conduct two trials, however, trial 09010,11,12 and VIGOR13 that both showed that rofecoxib increased cardiovascular events. Trial 090 ended in 1999 but wasn’t published until 2006,12 2 years after Vioxx had been withdrawn when the publication couldn’t harm the sales.

  The other trial, with the catchy name VIGOR, was published in the New England Journal of Medicine in 2000.13 It compared Vioxx with naproxen. A year later, pharmacist Jennifer Hrachovec called a radio show on which the journal’s editor, Jeffrey Drazen, appeared and begged him to correct the paper because there were three more heart attacks on rofecoxib on the FDA’s website than in the journal article, but Drazen responded evasively.14 Two months earlier, Hrachovec had sent a letter to the journal, but it was rejected, officially because of ‘lack of space’, which is an excuse respectable journals cannot hide behind when scientific misconduct relevant for patient safety is suspected.

  The VIGOR trial would have looked very different if the three extra heart attacks had not been deliberately omitted from the trial report. Their inclusion would also have undermined the assertion in the article that only those who were already at high risk of a heart attack showed an increased risk after taking Vioxx, as the omitted heart attacks were all in the low-risk group.14

  There were other editorial blunders. The editors didn’t ensure that thromboses were appropriately described and discussed. There were two full tables of gastrointestinal adverse effects in the article, but no table of thromboses; they were only mentioned in a few lines in the text, and only as percentages, which made it impossible to calculate the true number of events, as not all of them were included! Based on the percentages, I calculated 32 versus 17 thrombotic events on Vioxx and naproxen, respectively, but there were actually another 15 versus 3 events.15 That wasn’t even all. The FDA reviewer found a death from a heart attack on Vioxx that was coded as something else and, conversely, two deaths too many on naproxen.11 Thus, the coding of the events favoured Vioxx and many more events disappeared on Vioxx than on naproxen in the published report. This looks like fraud to me.

  The editors allowed Merck to say that the reason Vioxx caused more thromboses than naproxen was that naproxen was protective rather than Vioxx being harmful. This interpretation was wholly speculative and later refuted, and it
was irrelevant for the patients. As there were more serious events overall with Vioxx, there could be no doubt that naproxen was the better drug.11

  The editors noted that forensic IT work on the submitted disc revealed that the three cases of myocardial infarction had been omitted from the manuscript 2 days before it was submitted to the journal.16 They also found out that Merck had selected an earlier cut-off date shortly before the trial ended for the thrombotic events than the cut-off date for the gastrointestinal events, which they were not informed about and which is deceitful.15 They blamed Merck and the clinical investigators but forgot to mention their own role in allowing the obviously flawed paper to appear in print. After 5 years of silence, when the drug had been withdrawn and the journal ran a risk of getting accused in court cases, the editors finally reacted by publishing an ‘expression of concern’.16 If they had acted earlier, it might have killed the sales of Vioxx instead of killing the patients, as the journal is so influential, and it would also have blunted the impact of the reprint sales.14 The New England Journal of Medicine sold 929 400 reprints of the article – more than one for every doctor in the country – and they brought in between $697 000 and $836 000.14 The journal won’t disclose its revenue, but its owner, Massachusetts Medical Society, listed $88 million in total publishing revenue for the year ending 31 May 2005.

  In 2001, independent researchers using FDA data documented that Vioxx doubled the risk of serious cardiovascular events significantly in the VIGOR trial (8076 patients),17 and in 2004, a meta-analysis performed by independent researchers showed that a clear relationship between Vioxx and increased risk of myocardial infarction existed already by the end of 2000.6 When this meta-analysis was published, the French drug agency felt it could be interpreted as an accusation of their own incompetence.18 They therefore wrote a letter to the editor to defend themselves, which, ironically, demonstrated their incompetence. They claimed there was no evidence of an increased risk before 2005 and put forward Merck’s false explanation that the reason Vioxx caused more thromboses than naproxen was that naproxen was protective rather than Vioxx being harmful. Sometimes it’s better to keep quiet with one’s ignorance. The rest of the world, including the FDA, had known since 1999 that Vioxx could cause thrombosis.7,13,17

  Two other meta-analyses, from 2001 and 2002, one with 28 465 patients and the other with 5435 patients, didn’t find an increase in cardiovascular risk with Vioxx compared with placebo, which is highly surprising given the huge number of patients, but not given that all the authors were employees or paid consultants for Merck.19,20 It is telling that the two meta-analyses performed by independent researchers were published in JAMA and the Lancet, whereas those performed by Merck were published in specialist journals, Circulation and the American Journal of Cardiology.19,20 Circulation is owned by the American Heart Association, which accepts drug company sponsorship;21 over a 10-year period, the association funded more than one billion dollars in research grants.22 This amount is surreal for a non-American. The website of the American Journal of Cardiology advertises many free CME programmes and, like Circulation, it also publishes supplements to the journal. The first supplement I came across was a paper that under Acknowledgments said that ‘Funding for publication and medical writing assistance were provided by Novo Nordisk Inc.’.23 Even in 2012, we are told about ‘medical writing assistance’, which means that the paper wasn’t written by its eight authors but by a ghost.

  Internal company documents24 showed that Merck in 2003 got away with publishing a huge seeding trial, the ADVANTAGE trial, which involved 600 sites and 5557 patients, in a prestigious journal, Annals of Internal Medicine.25 It compared Vioxx with naproxen and, as in the VIGOR trial, scientific misconduct was involved.26 Eight patients suffered heart attacks or sudden cardiac death on Vioxx compared with only one on naproxen, but in the publication, three of the Vioxx cases had disappeared so that the difference was no longer statistically significant. As an example, one of Merck’s scientists who had judged that a woman died from a heart attack was overruled by his boss, ‘so that we don’t raise concerns’. The cause of death was now called unknown, also in Merck’s report to the FDA. Merck’s top scientist, Edward Scolnick, noted in emails that he would personally pressure senior officials at the FDA if it took action against Vioxx.26

  The first author on the trial report said that Merck came to him after the study was completed and asked him to help with the editing. He was paid, which is highly unusual for a first author of a trial report, and the report was already written up by Merck; a Merck employee was thanked for ‘assistance with manuscript preparation’.25

  It confirms that we cannot trust drug companies that an independent meta-analysis of Vioxx studies found that those with an external endpoint committee reported four times more heart attacks with Vioxx than with the comparator, whereas trials without an external endpoint committee reported fewer heart attacks with Vioxx.6 Although the members of data and safety-monitoring boards in drug trials are supposed to be independent, even according to Merck’s own policy, the head of the VIGOR board was awarded a 2-year consulting contract with Merck 2 weeks before the VIGOR trial ended, and he disclosed family ownership of Merck shares worth $70 000.3 Before VIGOR was published in 2000, chief scientist Edward Scolnick admitted internally that Vioxx causes thromboses.

  Internal company documents show that Merck used guest and ghost authors for many of its papers.27 To investigate to which extent the medical literature is flawed and misleads the clinicians, we studied 397 abstracts on Vioxx.28 It was expected from the beginning that the drug would be a double-edged sword compared to older NSAIDs, i.e. causing less gastrointestinal bleeding and more thrombosis. From the patients’ point of view, both effects are important and should be investigated, emphasised and reported similarly. However, before the withdrawal of Vioxx, 3.4 times as many abstracts commented on gastrointestinal bleeding as those that commented on thrombotic effects, whereas after withdrawal, 1.8 times as many abstracts commented on thrombotic effects. Thus, the harms of Vioxx came into focus too late when the drug had been withdrawn.

  Merck also misled the readers by publishing a fake journal, the Australasian Journal of Bone and Joint Medicine, which looked like a peer-reviewed medical journal but was a marketing tool.29 Most of its articles presented data favourable to Merck products, including Vioxx, without disclosing the sponsorship.29

  Like Merck, the FDA failed badly in its duty towards the patients. A five times increase in heart attacks in the millions of people taking the drug wasn’t a public health emergency in the FDA’s eyes.7,30,31 Life-saving revisions of the Vioxx label took nearly 2 years to complete, as ‘We were trying to work out exactly what was acceptable to both sides.’7,30 I wonder what the thousands of grief-stricken spouses who lost their loved ones during these 2 years will say about this tempo in drug regulation. Many of the tens of thousands of patients who were killed by Vioxx4 shouldn’t have been treated with an NSAID, as paracetamol (acetaminophen) would have given the same effect, or as they could have done well without treatment.

  In February 2001, the FDA discussed the VIGOR study with Merck because of the five-fold increase in myocardial infarction with rofecoxib in comparison with naproxen, and the FDA asked Merck to make the doctors aware of these results.4,32 However, the next day, Merck instructed its sales force of more than 3000 people:

  ‘DO NOT INITIATE DISCUSSIONS ON THE FDA ARTHRITIS ADVISORY COMMITTEE … OR THE RESULTS OF THE … VIGOR STUDY.’

  If a physician inquired about VIGOR, the salesperson should indicate that the study showed a gastrointestinal benefit and then say, ‘I cannot discuss the study with you.’

  Merck also produced a pamphlet to its sales force indicating that rofecoxib was associated with one-eighth the mortality from cardiovascular causes of that found with other NSAIDs.32 The pamphlet presented a misleading analysis of short-term studies and didn’t include any data from the large VIGOR study. The card’s two references included ‘data on
file’ at Merck and a brief research abstract.33

  The corruption of the truth was total. In May 2001, Merck produced the press release ‘Merck reconfirms favorable cardiovascular safety of Vioxx’.4 Drug salespeople were only allowed to discuss approved results with the doctors, which were studies that provided ‘solid evidence as to why [doctors] should prescribe Merck products’. Distributing studies that raised safety questions about Merck’s drugs was ‘a clear violation of Company Policy’.32

  A Merck spokesperson, Kenneth C Frazier, lied when presented with Senator Henry A Waxman’s unequivocal account of all these issues.32 He said that ‘Our representatives were instructed to present a balanced description of the risks and benefits of Vioxx’, and that the data from the randomised trials (involving more than 28 000 patients) didn’t show an increased risk with Vioxx.34 Waxman replied that it was telling that the company relied on its 28 000 patients meta-analysis, as the FDA already in 2001 found it to have serious methodological limitations.35

  Eric Topol from Cleveland wrote about the issues 3 weeks after the withdrawal of Vioxx,4 and two Merck employees misinformed the readers in response.36 They claimed that the increase in cardiovascular risk began after 18 months of therapy. This marketing trick was widely believed at the time, even by clinical pharmacologists who should have known better. I told them that when you take the first dose of a drug that is thrombogenic, you might get a thrombosis. Merck’s misleading claim came from a trial in colorectal adenomas, and they propagated it – surprise, surprise – in the abstract in the New England Journal of Medicine.37 Merck had not used a correct statistical test, and they had excluded all events that occurred more than 2 weeks after stopping treatment, although some of these patients would be expected to have, and actually had,38 thrombotic events. It took 15 months before Merck was forced to retract its claim from the journal.39 Topol wrote that the harms were visible early on,40 and he also showed that two deaths, four heart attacks and three strokes with Vioxx were missing in the VIGOR publication compared with the data the FDA had access to, whereas the total number of such events was the same for the comparator drug, naproxen, in the two datasets. More fraud, it seems.