Deadly Medicines and Organised Crime

by Peter Gotzsche

  Drug regulation builds on trust

  Economic theory predicts that firms will invest in corruption of the evidence base wherever its benefits exceed its costs. If detection is costly for regulators, corruption of the evidence base can be expected to be extensive.

  Alan Maynard, unpublished manuscript

  Drug regulators have told me that the regulatory system builds on trust, which they think is fine, as it would have too serious consequences for the companies if they cheated and it was detected. As Maynard explains, this argument doesn’t hold. Furthermore, as we have seen, big pharma means big crime, and where else in society would we trust what criminals tell us? Rats in toxicology studies may never have existed; they may have died more than once; they may be dead, although being described as being in good health in toxicology reports; tissues may be missing; data may have been fabricated; and the animals may have died too early before they developed drug-induced cancers.8,16

  Drug firms don’t trust each other, but drug agencies are supposed to trust the entire industry.16 The authorities know perfectly well that they cannot trust the industry and the reason they say the opposite is pragmatic. They cannot review more than a tiny fraction of the mountains of documents they receive. As an extreme example, one study report for a Tamiflu trial consisted of 8545 pages, which is a 1000 times greater than its published version.71 Understandably, most regulators only read summaries most of the time and, to my knowledge, it is only the FDA that routinely does its own statistical analyses on the submitted data, but the EMA now intends to do the same (see Chapter 11).

  Many of the thousands of pages are pretty useless, and I have no doubt that the industry deliberately drowns the regulators in data, which gives the industry two advantages. First, they reduce the risk that the regulators detect anything that might prevent the drug from being approved, or might hamper sales because of warnings on the label. Second, if problems arise, the industry can claim they didn’t conceal anything and that the regulators are therefore to blame. Although this isn’t entirely true, it might work out in court.

  The regulators are apparently so overloaded that they don’t even check that everything is there, which they should. We have found many examples that important appendices have been left out or that pages in the middle of a report were missing. Whole trials can also be missing, e.g. two out of seven negative studies of SSRIs in children,72 although this is against the law.

  It is not surprising that serious harms of new drugs may pass unnoticed, as they may be hidden so well in registration applications and other submissions that it would require time-consuming detective work to unravel them.1,73,74 An example of this is long-acting beta-agonists for treatment of asthma. In the 1990s, there were concerns that these drugs might increase asthma-related deaths rather than decrease them, and the FDA asked GlaxoSmithKline to carry out a large trial of salmeterol, the SMART trial.73 Glaxo’s handling of the trial was a bit too SMART, however, as the company manipulated the results it sent to the FDA.

  In 2003, the findings were presented at a meeting for chest physicians where Glaxo claimed that the results were inconclusive, but that was misleading. The Data and Safety Monitoring Board for the study had recommended its termination after 26 000 of the planned 60 000 patients had been enrolled, as there were more asthma-related deaths in the salmeterol arm than in the placebo arm, or alternatively, that 10 000 more patients were recruited.73

  The trial period was 28 weeks, but the investigators could – if they wanted – report serious adverse events that occurred in an additional 6-month period. The FDA assumed of course that the data they reviewed stemmed from the rigorously controlled randomised double-blind period. Only when the agency specifically queried the company as to which dataset had been provided, did Glaxo reveal that it had included the 6-months follow-up data. That made a huge difference. There was no statistically significant increase in asthma-related deaths when the follow-up data were included, whereas the risk was four times higher when only the trial data were considered, which was statistically significant. Independent researchers concluded that in the absence of the transparency associated with the advisory committee meetings at the FDA, these deceptions would never have come to public attention.73 Glaxo responded to the revelations by saying it had ‘acted responsibly and transparently’.74

  That wasn’t even all. Almost 3 years after the trial was finished, it still hadn’t been published. The SMART results confirmed the results of a large trial Glaxo had run and published already in 1993.75 Glaxo had compared salmeterol with its short-acting drug, salbutamol, and three times as many patients died from asthma when they received the long-acting drug (P = 0.11 for the difference). In 2006, a meta-analysis including the SMART study confirmed that long-acting beta-agonists increase asthma-related deaths.76 At a superficial glance, the absolute risk of dying seems small, only one per 1000 patients per year of use. However, salmeterol was one of the most prescribed drugs in the world and the increased risk translates into 4000–5000 extra asthma-related deaths every year in the United States alone.76

  In July 2005, the FDA considered whether long-acting beta-agonists should be removed from the market, but, instead, the agency opted for strong warnings and a recommendation that the drugs should only be used after other asthma drugs had failed.76 In 2010, the FDA warned again, this time about the increased risk of severe exacerbation of asthma symptoms, leading to hospitalisation and death, and said that these drugs must never be taken alone but should be combined with an inhaled corticosteroid.77 However, it doesn’t solve the problem to add inhaled corticosteroids, e.g. the risk of admission to hospital is still increased two-fold. The FDA also required the manufacturers to conduct additional clinical trials to further evaluate the safety of these drugs when used in combination with inhaled corticosteroids. I find this odd. FDA requirements of additional studies are usually ignored by the companies, and the FDA doesn’t enforce them. These drugs are dangerous – likely also when combined with steroids – and we don’t need them, so why not take them off the market?

  When Glaxo finally published the SMART trial in Chest, they mentioned the increase in asthma-related deaths, but the last two sentences in the abstract were interesting:78

  ‘Subgroup analyses suggest the risk may be greater in African Americans compared with Caucasian subjects. Whether this risk is due to factors including but not limited to a physiologic treatment effect, genetic factors, or patient behaviours leading to poor outcomes remains unkown.’

  Smoke and mirrors and the paper stinks: ‘Subpopulations were based on baseline characteristics such as [my emphasis] inhaled corticosteroid (ICS) use and study phase. Additionally, outcome events were analyzed separately for white and African-American subjects.’

  Such as? Glaxo doesn’t even tell us how many times they massaged the data before they found a subgroup result they could use to fool the readers into believing that the drug was only harmful for African Americans. Even the data massage itself was misleading. There wasn’t a test of interaction, which is what one needs to do before one can say there is a difference between the results in two subgroups. And, in fact, the relative risk for asthma-related deaths was very similar for Caucasians and African Americans. The Discussion section of the paper tells us about only one of the subgroups, which is misleading: ‘post hoc analyses showed no significant differences between treatments … in the Caucasian population’. Glaxo converted a clear harm into no harm. Words fail me, but it says a lot about why we cannot trust industry-sponsored trials. Two of the five authors were Glaxo employees and the other three were on Glaxo payroll.

  It seems that Glaxo did what it could to protect its drug rather than the patients.79 In a scathing editorial in the New England Journal of Medicine, the editors explained that Glaxo refused to provide a placebo inhaler for an NIH trial of salmeterol. The investigators had to spend $900 000 of taxpayers’ money to repackage the active drug and to create a visually identical placebo for use in the trial. The edito
rs furthermore wrote:

  Glaxo’s stated goal is ‘to improve the quality of human life’ but companies are able to develop and sell their treatments only because they can tap into a community resource: Patients who are willing to put themselves at risk as they participate in clinical trials. Companies, for their part, must therefore be willing to put their products at risk by providing them to legitimate third parties for study. Failure to do so is an unacceptable double standard.

  Drug companies may not only cheat the authorities in their submissions; they may also lie when questioned directly. In documents prepared for a 2005 FDA hearing, Pfizer denied that its NSAID celecoxib causes heart attacks, based on an analysis of 44 000 patients.80 But big numbers offered by the industry when it is on the defensive are often deceptive. Pfizer had unpublished evidence to the contrary,80,81 e.g. a 1999 trial in Alzheimer’s disease, and a Pfizer official admitted in an interview that its analysis didn’t include outside studies that indicated its drug causes heart problems. One such study, which Pfizer knew about,82 was conducted by the NIH and had been terminated after finding that high doses of celecoxib more than tripled the incidence of heart attacks and strokes.

  Other companies have also deceived the FDA by hiding studies and results showing that their drugs cause lethal harms.1,8,16,73,83,84,85

  There is one other reason why we know too little about the harms of drugs. Clinicians are supposed to report serious adverse events to the authorities but a common estimate is that only about 1% of such events get reported.86 Doctors are busy and may tend to think that an event isn’t drug-related and dismiss it, as this is convenient for them. If they report an event, they may learn never to do it again, as they might get harassed by a drug representative who keeps coming back with all sorts of questions about the patient, other drugs the patient was taking, etc. No one is really interested in harms it seems, apart from the victim. When I worked at a department of infectious diseases, I learned why many serious events in industry-funded AIDS trials didn’t get reported. The record forms were long and complicated and we didn’t have time for endless discussions with the drug company.

  Inadequate testing of new drugs

  When I lecture doctors in training to become clinical pharmacologists and explain why the regulatory demands for new drugs are inadequate and cannot ensure effective and safe drugs, and how the drug industry often manipulates its research, I’m met with mixed reactions. Some agree heartily and others are quite hostile, as if I had explained to a child that Santa Claus doesn’t exist. This worries me, as these are the doctors who are most likely to gets jobs in drug agencies and in the drug industry. I sometimes get the feeling that it’s already too late to talk sense to them.

  We could easily do far better than we currently do in protecting public health and avoiding wasting our money, and I shall give some examples.

  Only two placebo-controlled trials showing an effect isn’t enough

  Drug agencies consider efficacy to have been demonstrated if two placebo-controlled trials have shown an effect. As explained in Chapter 4, this is fairly easy to do for almost any drug for any condition because drugs have side effects, which will be expected to bias the assessment of a subjective outcome. If the sample size is large enough, any effect will become statistically significant, and the drug will be approved, if not too toxic.

  If the company didn’t succeed in its first two tries, it can perform more trials until two of them confess. On this background, it’s amusing that the Danish Minister of Health, after having consulted with the drug agency, replied to a politician that there is no requirement that a drug needs to be better than an existing drug to become approved, but it must be at least equally good, and under no circumstances worse than existing drugs. However, when only placebo-controlled trials are needed, we have no way of knowing whether new drugs are worse than existing ones.

  Companies are obliged by law to submit all trials they have carried out when they ask for drug approval, but the problem with this is that we cannot trust the drug companies. Trials may be missing and if they were conducted in countries with little public oversight, it might be impossible to know they existed.

  Cough medicines don’t work,87,88 but the drug industry has nevertheless succeeded getting hordes of medicines approved for cough and the sales are high.89 No less than 20% of all children up to 4 years are treated with asthma drugs such as terbutaline, which shows that the shady marketing I participated in when I worked for Astra was highly effective (see Chapter 2).

  In the United States, over-the-counter cough and cold medications were used by 39% of households during 3 years.90 Many of the drugs came on the market before 1972 when there was little control with medicines, but poison control centres had reported more than 750 000 calls of concern in 7 years related to such products, and the FDA had identified 123 deaths in children under six in its database. Adverse effects of the drugs include cardiac arrhythmias, hallucinations, depressed consciousness and encephalopathy. Manufacturers’ advertisements describe the drugs as safe and effective, both of which are untrue.

  A petition required the FDA to review the drugs, but the manufacturers claimed that the injuries could be prevented through parent education, which is a horrendous lie. In 2011, the FDA announced that the products shouldn’t be used in children below 2 years of age and that the ‘FDA strongly supports the actions taken by many pharmaceutical manufacturers to voluntarily withdraw cough and cold medicines that were being sold for use in this age group.’91 Why didn’t the FDA withdraw these useless and potentially dangerous products from the market? And why, after 4 years, was the FDA still reviewing the safety and expected to communicate its recommendations in the near future, as they said? Not even when useless drugs kill our children do the regulators act, whereas they have withdrawn many effective products, even though they caused fewer deaths. Drug regulation is not a consistent enterprise.

  I once discussed cough remedies with a drug regulator and he alerted me to studies included in a registration application that purported to have shown that the drugs worked. It is one of the weirdest papers I have ever seen (and I have seen a lot). The studies had been carried out in India. A sensitive miniature microphone developed by Procter & Gamble attached to the patient’s nose registered every little sound that perhaps was, or could develop into, a cough.92 All three drugs tested (guaiphenesin, bromhexine and dextromethorphan) had an effect. Surprise, surprise. These recordings were completely irrelevant for the patients. Two of the drugs also increased sputum volume. What are we to make out of that? If they increased sputum production, they would also increase ‘expectorant effects’ measured as sputum volume, but that would not be a beneficial but a harmful effect. The studies were published in Pulmonary Pharmacology, an obscure journal I’d never heard about. It’s not the regulators’ fault that they have to accept such nonsense; it’s the politicians’ fault that they have not required outcomes that matter to patients.

  Drug trials in countries with widespread corruption

  Nowadays, drug trials are outsourced more and more to countries with little oversight and widespread corruption. How are we to know whether the results have been made up when we have no possibility of controlling the trials? Despite considerable opposition from scientists, ethicists and consumer groups, the FDA decided in 2008 that clinical trials performed outside the United States no longer had to conform to the Declaration of Helsinki if used to support applications for registration of products in the US.93 Pardon me, but have they gone completely mad at the FDA? Has the FDA leadership never heard about the Nürnberg processes? Or about medical experiments on US prisoners where the Declaration of Helsinki wasn’t an issue? Or about the Tuskegee affair where researchers in Alabama followed 399 black men infected with syphilis without treating them for 40 years to study the natural course of the disease while preventing them from accessing treatment programmes available to others, and while many died of syphilis, wives contracted the disease and children were born with co
ngenital syphilis?94 Or that drug companies do research in poor countries for particularly dangerous drugs because peasants don’t sue big corporations for injury and because informed consent regulations either don’t exist or are weakly enforced?8 The most well-known example of the use of third world guinea pigs is oral contraceptives, which were first tested in Puerto Rico, later in Haiti and Mexico, and when tested in the United States, poor people were chosen, 90% of whom were either of Mexican or African origin.8

  In contrast to this indefensible move, the US Court of Appeals ruled shortly afterwards that the Declaration of Helsinki constituted a sufficient customary norm to be considered binding in Pfizer’s meningitis trial in Nigeria where the parents didn’t know that their children participated in a trial. The court reversed a dismissal by a lower court of a lawsuit by families of children who died or were injured while they received Pfizer’s experimental antibiotic, trovafloxacin, although a better drug was freely available through Médecin sans Frontières.95 Pfizer hired investigators to look for evidence of corruption against the Nigerian attorney general in an effort to persuade him to drop the legal action.96 It didn’t work out and Pfizer had to pay compensation to families whose children died. The drug was never intended for Africa. Pfizer planned to sell it in the United States and Europe, but its licence was withdrawn in Europe because of concerns over liver toxicity.

  An effect on a surrogate outcome isn’t enough

  One of the most harmful practices in drug regulation is to approve drugs based on their effects on surrogate outcomes. As this mistake has cost the lives of hundreds of thousands, or perhaps even millions, of patients (see below), it’s difficult to understand that the regulators don’t require proven effects on relevant outcomes.