Deadly Medicines and Organised Crime

by Peter Gotzsche

  Here is an example. When I had been a doctor for only 2 years, I diagnosed mild type 2 diabetes in an old man that had been admitted for something else to the department of hepatology where I worked. I wrote in his files that it was common practice to start treatment with tolbutamide, but since the only large trial of tolbutamide ever performed was stopped prematurely because of an excess of cardiovascular deaths, and since those patients who took most of the their daily doses were also those that had the highest event rate, I decided not to institute treatment with tolbutamide.

  My superior in the hierarchy blew me up when he saw my notes. ‘How dare you not start tolbutamide in violation of the guidelines the endocrinologists have written?’ I explained calmly but firmly that I knew more about this drug than the endocrinologists because I had read the trial report carefully, plus the many articles and letters that followed, and also a book that discussed the issues. The study – the University Group Diabetes Project (UGDP) – had been carried out independently of the drug industry, and it had been heavily debated and reanalysed by several other groups than those who conducted the study. I had no doubt about who were right.

  Tolbutamide lowers blood glucose but this is a surrogate outcome. We don’t treat patients to lower their blood glucose; we treat them to prevent complications to diabetes, in particular cardiovascular ones. I therefore considered it absurd, and still do, that people used this drug when the only trial studying cardiovascular complications was stopped because the drug killed the patients. It was particularly convincing that good compliers with tolbutamide had a greater mortality rate than poor compliers,97 because patients who do what they are told are generally more healthy than others and therefore have better survival even when the drug is placebo. A trial of a lipid-lowering agent, clofibrate, demonstrated this.98 There was no difference in mortality between drug and placebo, but among those who took more than 80% of the drug, only 15% died, compared to 25% among the rest (P = 0.0001). This doesn’t prove that the drug works of course, and the same difference was seen in the group that received placebo, 15% versus 28% (P = 5 ∙ 10–16).

  Upjohn, the maker of tolbutamide, launched an aggressive campaign to discredit the UGDP study findings by using leading and well-remunerated academics, and the arguments became increasingly ad hominem.99 Lawsuits were brought by the company to prevent the FDA from mentioning the study’s results in the package inserts, and the FDA was even forced to carry out an investigation that concluded that the data in the study hadn’t been falsified!97

  The use of tolbutamide should have been stopped by withdrawing the drug from the market, at least temporarily, while those who were sceptical towards the trial’s result conducted another trial. But the FDA never required Upjohn to do this and it was never done.

  No one seems willing to learn anything – or at least not much – from history when it comes to drug regulation. History repeats itself all the time. For the next 40 years after the UGDP trial, industry simply stopped performing trials that might have revealed that their diabetes drugs increased cardiovascular events, and our drug regulators let them get away with this,99 which is pretty scandalous. Rosiglitazone is a recent example of a diabetes drug that was approved based on its effect on blood glucose, but as this drug also increased the cardiovascular complications it was supposed to prevent, it was taken off the market in Europe in 2010 after having killed thousands of patients (see Chapter 16).

  Similar stories can be told from other therapeutic areas.100 A cardiac arrhythmia suppression trial (CAST) was stopped prematurely because the two active drugs, encainide and flecainide, killed the patients. This trial was originally designed as one-sided, which means that the drug can only be beneficial or neutral, since the cardiologists couldn’t imagine that the treatments could be harmful.101 At the peak of their use in the late 1980s, anti-arrhythmic drugs were likely causing about 50 000 deaths every year in the United States alone, which is of the same order of magnitude as the total number of Americans who died in the Vietnam War.102 The drugs were widely used because they had an effect on a surrogate outcome, the ECG, and although the FDA had serious safety concerns, they gave in to the pressures of the companies, which – quite predictably – led to the drugs being used in many completely healthy people with benign rhythm disturbances that many of us have.

  Tumour shrinkage is another popular but misleading surrogate outcome. Cancer patients’ primary interest is to stay alive, but some treatments that reduce the size of the cancer increase mortality, e.g. radiotherapy in women who had their breast cancer diagnosed at screening.103 This can be said about many, if not most, cancer drugs. High doses may have a better effect on the cancer but may also kill more patients. If the dose is high enough, all cancers will die but so will all patients. This shows how absurd this surrogate outcome is.

  In 2008, FDA granted bevacizumab (Avastin) accelerated approval for treatment of metastatic breast cancer, although it didn’t increase survival, only progression-free survival.104 This is not only a surrogate outcome but also one that is prone to bias, as it is pretty subjective to decide whether progression has occurred. The FDA obligated the company to conduct more trials and these didn’t show an effect on progression-free survival whereas they showed serious harms, including deaths. Three years later, the drug, which cost the same every year as several new cars, about $88 000, was revoked for breast cancer.105

  Lack of adequate safety data isn’t acceptable

  It’s a gross failure in drug regulation when drugs with known harms are approved without adequate safety data. The COX-2 inhibitors are a perfect example, as their mechanism of action predicted an increased risk of cardiovascular mortality. When I discussed this with a drug regulator, he replied that if they were to demand such data, it would delay the introduction of valuable drugs for years.

  I don’t buy the argument. A drug company could easily perform a large trial of its COX-2 inhibitor that could tell us what the risks are and it’s the industry’s own fault when it thinks it can get away with cutting corners. If rofecoxib (Vioxx) had been studied in relevant patient populations, its harms would have been detected very quickly, as the number needed to treat for 1 year to cause one extra myocardial infarction is only 70 patients.19 There is also an overriding ethical issue, which cannot be trumped by petty claims about practicalities and potential loss of income. Unfortunately, the drug agencies give in to the drug companies’ unsustainable arguments.

  Vioxx was withdrawn in 2004 and valdecoxib (Bextra) in 2005. Before Bextra was pulled off the market, nine of the 10 FDA advisers with industry ties voted to keep it on the market!106

  In 2008, the FDA considered whether, in future, it should require post-marketing studies with relevant outcomes such as cardiovascular morbidity and mortality.107 However, only one-third of such studies are ever performed,46 and the FDA is known not to enforce them because it lacks the authority to do so.22 From 2007, failure to perform a post-marketing study or to make a needed label change can result in fines, but only up to $10 million.54 As this is peanuts for big pharma, it’s window-dressing, or a fake fix. Even when studies have been carried out, they might show that a drug has killed thousands of patients, which we could have avoided by requiring relevant trials before drug agencies decide whether a drug should be approved. Post-marketing studies are therefore a very bad idea compared to rejecting the application for marketing authorisation. We need relevant data for every new drug in a therapeutic class, as a new drug might kill people even though 10 similar drugs didn’t.

  An additional problem is that required post-marketing studies are not necessarily randomised trials but may merely be observational studies, which are very poor in detecting signals of harm. Those who are being treated differ in many ways from a control group that is not being treated, and a doubling in the rate of heart attacks in elderly people may simply be because these patients are more prone to get a heart attack than other patients. Patients with rheumatoid arthritis, for example, are more prone to get h
eart attacks than other people of the same age, which makes it difficult to detect that COX-2 inhibitors kill them.

  The spontaneous reporting of serious adverse events to the regulators for marketed drugs is also a weak method for harm detection. In 2010, the FDA warned Pfizer in a 12-page letter for failing to quickly report serious and unexpected potential side effects from its drugs after having conducted a 6-week inspection of Pfizer’s headquarters.108 Pfizer had misclassified or downgraded reports to non-serious without reasonable justification and had failed to submit reports on blindness caused by Viagra (sildenafil) and similar medications within the agency’s 15-day deadline. Pfizer was also warned in 2009, but the FDA noted that the company’s delays in telling the agency about harms had only grown. Pfizer was told that failure to fix the problems could result in legal action without notice and delays in approving the company’s pending drugs.

  In 2012, Roche was reprimanded by the EMA for failing to report up to 80 000 possible adverse reactions from its drugs, including 15 161 deaths in the United States.109 Regulators identified additional deficiencies related to the evaluation and reporting to national drug agencies of suspected adverse reactions in 23 000 other patients and 600 participants in clinical trials.

  Too many warnings and too many drugs

  All drugs come with a long list of warnings, contraindications and precautions, for example explaining types of patients, conditions or other drugs the patients take that make it risky to use the drug. Have a look at an advertisement in a medical journal and you’ll see how overwhelming it is; there can be more than 20 warnings for a single drug. Here is an example.

  Statins

  Some of my colleagues are obsessed with cholesterol and believe that everyone over 50 should take a statin, no matter what their cholesterol level is, as it will reduce their risk of dying. They also say that statins have no side effects worth mentioning, or even that they have no side effects.110 Let’s have a look at an advertisement that appeared on the first pages of JAMA on 19 September 2012. It said, ‘Try LIVALO® to lower LDL-C and improve other lipid parameters’.

  That’s not why you might consider taking a statin, is it? You would want it to reduce your risk of dying, not to improve some laboratory values. Can you be sure that a particular statin reduces your risk of dying? No, you cannot, as statins are approved based on their effect on plasma lipids. LIVALO might reduce your risk of dying from heart disease but it might also increase your risk of dying from other causes, so you cannot know what your chances are with and without LIVALO.

  Just by reading the first two lines in the ad, I would say no thanks. We shouldn’t take ‘life-saving drugs’ without knowing whether they will decrease or increase our risk of dying.

  But let’s go on. Page 2 of the ad says, ‘Drug therapy should be one component of multiple-risk factor intervention in individuals who require modifications of their lipid profile. Lipid-altering agents should be used in addition to a diet restricted in saturated fat and cholesterol only when the response to diet and other nonpharmacological measures has been inadequate.’

  Aha. This is not what my well-meaning colleagues say when they are close to proposing we should get statins with our drinking water. I am not on a diet or on some ‘other nonpharmacological measures’ (what on earth is that?) and how can anyone decide whether I require a modification of my lipid profile? Can you see how subjective all this is and how woolly the regulatory language is?

  Further ahead comes what I wanted to know, but curiously under a subheading called ‘Limitations of use’:

  Doses of LIVALO greater than 4 mg once daily were associated with an increased risk for severe myopathy in premarketing clinical studies. Do not exceed 4-mg, once-daily dosing of LIVALO.

  The effect of LIVALO on cardiovascular morbidity and mortality has not been determined.

  I knew it! We don’t have a clue whether LIVALO does what we want it to do. And I would run a risk of severe muscle damage. People absorb and metabolise drugs differently, and some will undoubtedly get severe muscle damage even if they don’t exceed 4 mg a day. It could be me. At this point, my free interpretation of the drug’s name is LEAVE ME ALONE!

  Page 1 of the ad doesn’t tell us about the possible benefit of the drug, apart from the headline about the lipids, which isn’t useful. The rest of the page is about harms, which are called ‘Important safety information’. My scepticism increases:

  ‘Cases of myopathy and rhabdomyolysis with acute renal failure secondary to myoglobinuria have been reported with HMG-CoA reductase inhibitors, including LIVALO.’ Such effects increase with dose, with advanced age (>65 years), renal impairment, inadequately treated hypothyroidism, and in combination with fibrates or lipid-modifying doses of niacin (≥1 g/day).

  And then it becomes really difficult. ‘LIVALO therapy should be discontinued if markedly elevated CK levels occur or myopathy is diagnosed or suspected’, and ‘Advise patients to promptly report unexplained muscle pain, tenderness, or weakness, particularly if accompanied by malaise or fever, and to discontinue LIVALO if these signs or symptoms appear.’

  Good Heavens. CK means creatine kinase, a muscle enzyme. Patients treated with statins often have such symptoms111 (although the ad wrongly says they are rare), so how would the patients know when to discontinue LIVALO?

  We are also told about liver injuries. Liver enzyme tests should be performed before treatment starts and if signs or symptoms of liver injury occur. It seems a bit late to measure liver enzymes if the liver is already injured. ‘There have been rare post-marketing reports of fatal and non-fatal hepatic failure in patients taking statins, including pitavastatin.’ The drug might kill me.

  LIVALO may also increase blood glucose, which will increase my risk of dying from cardiovascular problems, which LIVALO was supposed to protect me against.

  I’ll stop here, but it’s important to realise that drugs are never safe. Life jackets on boats are good to have, as they may save your life. They won’t kill you. Drugs are not like that. Taking a statin may reduce your risk of dying from heart disease, but it will also increase your risk of dying from some other causes. Not much, but one of the statins, cerivastatin (Baycol), was taken off the market after patients had died because of muscle damage and renal failure.

  Anyone of us will need to consider the pros and cons of taking a drug, and our doctor isn’t always the best person to ask, as most doctors have been brainwashed and many have been bribed by the drug industry. What we would like to know is this: how much longer will we live, on average, if we take this drug? The older we are, the smaller the benefit. If we don’t die from heart disease, we’ll surely die from something else. A male 65-year-old non-smoker with a systolic blood pressure of 140 mm Hg and a cholesterol of 5 mmol/L will be expected to live 3 months longer if he takes a statin for the rest of his life.112 That’s not much, particularly not if the bonus comes when he sits demented and incontinent in a nursing home and would rather have preferred a drug that shortened the length of this misery. We should also ask the patients what their experiences are. A survey of over 10 000 people found that muscular side effects were reported in 60% of former users and 25% of current users.110

  Other lipid-altering drugs are also interesting. It was expected to be beneficial to increase high-density lipoprotein, but a drug that does this had no effect on the progression of coronary atherosclerosis in trials of about 1000 patients.107 The chemical name of the drug is torcetrapib. Can you pronounce and remember this? One reason why the chemical names, which are invented by the drug companies, are so foolish is that doctors are then forced to use the trade name and therefore less likely to prescribe a cheaper generic when the drug comes off patent. Luckily, the company did a large trial in 15 000 patients, and since it showed that the drug kills people, the manufacturer halted the development of the drug.

  Another lipid-altering drug, ezetimibe, was approved by the FDA in 2002 because it had lowered low-density cholesterol in the blood by 15%.107 In 200
7, sales of the drug reached $5 billion in the United States, although no one knows whether it’s beneficial or harmful.

  Warnings are fake fixes

  It’s impossible for clinicians to know what they need to know about drugs to prescribe them safely, and it’s therefore not surprising that doctors make many medical errors. The fundamental problem is that regulators think about drugs one by one and don’t care that doctors cannot possibly know all the warnings about the drugs they use. What matters to regulators is: not our fault, we did warn you, didn’t we?

  Every doctor knows that the anticoagulant warfarin can interact dangerously with other drugs and some food items, but doctors cannot even use this drug safely. In one study, 65% of the patients were given at least one other drug that could increase the risk of bleeding with warfarin, and in another study, about a third of the patients received such drugs.113

  Cisapride (Propulsid from Johnson & Johnson) was supposed to promote gastric emptying, but it is no longer on the market as it causes cardiac arrhythmias that kill people. In 1998, the FDA warned about the contraindications for the drug through additions to the black box label, and practitioners were furthermore warned through a dear doctor letter sent by the manufacturer. These warnings had barely any effect.114 In the year prior to regulatory action, cisapride use was contraindicated for 26%, 30% and 60% of users in three study sites, and in the year after regulatory action use was contraindicated for 24%, 28% and 58% of users. Johnson & Johnson sold the drug for more than a billion dollars each year, although it should never have been approved. When the FDA called for a public meeting in 2000, a company executive admitted that they had not even been able to show that the drug worked.85 Again, regulatory insufficiency resulted in tragedies for real people:115