Deadly Medicines and Organised Crime

by Peter Gotzsche

  Independent testing of drugs would also mean that we could compare new drugs with old, cheap ones under fair circumstances where the old drug is not given in a dose that is too high (so that the manufacturer can falsely claim that their drug is better tolerated), or a dose that is too low (so that the manufacturer can falsely claim that their drug is more effective). There are also far too few comparisons with non-pharmacological interventions. When a study found that a programme of exercise and weight loss was better for preventing type 2 diabetes than metformin (56% and 31% effect, respectively),52 the only thing the Wall Street Journal wrote about was the effect of the drug!

  We also need much better and honest reporting of harms, which we’ll never get in industry trials. Although the first thing we know about any drug is that it may cause harm, a survey of 192 reports of trials of at least 100 patients showed that the space devoted to harms was 0.3 page, which was similar to the amount of space devoted to contributor names and their affiliations.53

  Finally, the rationale for a proposed clinical trial must be based on a rigorous, recent systematic review of all previous trials with similar drugs, with meta-analysis if possible.54 This will often tell us that what looked like ‘conflicting’ results in previous trials are not at all conflicting. If this isn’t done, many unethical trials will be approved, as the type of drug might already have been shown – or could have been shown – to be either life-saving or harmful. Such a requirement was introduced in Denmark in 1997,55 but for reasons unknown to me it was quietly taken out when the law about trials was revised. The research ethics committees have failed miserably to pay attention to what matters most to patients. They have done nothing to ensure that the trials they approve are ethical; that the informed consent forms accurately depicts the state of knowledge and uncertainty; and that the results of all trials become publicly available, although this was pointed out to them in a well-argued BMJ paper in 1996.54

  We won’t get rid of industry-conducted trials for a long time. But in the interim, we could decide to let the regulatory authorities select those clinicians who would be allowed to test drugs, instead of industry’s picking of the most willing doctors,41 who are also those that are easiest to corrupt and most likely to keep quiet about industry’s manipulations of the evidence. Patients should know everything about money involved in trials and investigators’ conflicts of interest. If doctors are uncomfortable about this, they have something to hide, which makes it even more pertinent to ensure transparency. Protocols and contracts with drug companies should be publicly available so that we can all see whether to trust our institutions. It’s a sad fact that many still accept gagging clauses,56 and also that when authors solemnly and routinely declare in journals that they had access to all the data, it is rarely true.57,58

  Drug companies should be required by law to deliver placebos for independent research at no more than the manufacturing cost as a condition for having a product on the market. It should also be obligatory for companies to deliver the pure drug, e.g. in powder form, for independent research. If drug companies want to be a part of society, they must be willing to further public health by letting other researchers do research on their products.

  Trials come to their full fruition when assembled in systematic reviews, and such reviews should always tell the readers how many trials, outcomes and results that might be missing, and also to which extent the reviewed trials were sponsored by the manufacturer. Here is a good example from a Cochrane review of hypertension, where the Plain Language Summary aimed at patients says:

  Most of the trials in this review were funded by companies that make ACE inhibitors and serious adverse effects were not reported by the authors of many of these trials. This could mean that the drug companies are withholding unfavorable findings related to their drugs … Prescribing the least expensive ACE inhibitor in lower doses will lead to substantial cost savings, and possibly a reduction in dose-related adverse events.59

  Finally, seeding trials should be made illegal, as they now are in the EU. The directive on pharmacovigilance (post-authorisation safety studies) was amended in 2010 to say ‘Studies shall not be performed where the act of conducting the study promotes the use of a medicinal product.’60

  Drug regulatory agencies

  Drug agencies have a major responsibility for the many drug deaths. They approve many dangerous drugs and use fake fixes by issuing an enormous number of warnings and precautions, although they know perfectly well that they won’t work.

  We would never accept it if airlines crashed several times a day the year round because their construction made them too difficult for the pilots to fly.

  We need a revolution in drug regulation. Drug agencies need to become evidence-based and realise that the current system isn’t working and cannot work. They should reject far more drugs and require sufficient safety data.

  Surrogate outcomes should not be accepted

  I have given many examples earlier61 and in this book how terribly misleading surrogates can be. The patients may be injured or die while their surrogate improves. Cancer drugs are such a disaster area. The regulatory requirements for cancer drugs are close to none. They are often approved based on single-arm studies that cannot say anything about whether the drug increases or lowers mortality.

  Of 27 different indications in Europe, of which 14 were new applications and 13 extensions of earlier approvals, the clinical documentation consisted entirely of small single-arm studies in eight cases.62 The total number of patients was small, a median of 238, and in half of the cases, only surrogate outcomes such as complete or partial tumour response were provided, although they are highly prone to biased assessment, particularly in single-arm studies. This is very worrying because most cancer drugs may cause both tumour shrinkage and increased mortality, e.g. if given in a too high dose. Those drug studies that did report on survival found a median difference of only 1 month.62 Another study, of 12 new cancer drugs approved in Europe from 1995 to 2000, showed that none of them offered any significant progress and yet one cost 350 times more than the competitor.63

  It’s even worse at the FDA, which approves most cancer drugs (68%) based on outcomes other than survival. In addition, 35% of the drugs were granted a licence even without a single randomised trial.62

  Based on published data, I have calculated that 33 years of randomised trials in solid tumours sponsored by the UK Medical Research Council hasn’t led to any progress against cancer, on average.64 It was a large material, 32 trials comparing one treatment with another, and 6500 deaths, and the mortality on the new treatment was the same as that on the control treatment. Other cancer surveys have confirmed this, e.g. for 57 trials of radiotherapy the relative risk was 1.01,65 and for 126 trials in childhood cancer, the odds ratio was 0.96.66

  Zero progress against cancer, on average, means that it is very difficult to find new treatments that are better than those we already have. Rarely, a new treatment is better, and rarely, it is worse. As long as our drug agencies don’t require mortality data from randomised trials, they will allow harmful drugs to be introduced on the market without anybody knowing.

  Relevant patient populations, comparators and outcomes

  Patients above 65 years of age are routinely excluded from industry-sponsored trials,67,68,69 e.g. only 2% of patients in NSAID trials were 65 years or older,68 although these are the patients most likely to take drugs and most likely to be harmed. Exclusion of these age groups also makes it difficult to detect harms caused by the combined effects of taking many drugs (polypharmacy). The EMA recently announced that they will from now on expect the age distribution of patients to be representative in studies presented for marketing authorisation.70

  This is good but still not good enough, as drug firms can still write in their protocols that a condition for participation is that the patients don’t receive other drugs than the trial drugs, and don’t have more than one disease. We need to ensure that drugs are tested in realistic settings,
which isn’t the case today. A survey showed that common medical conditions formed the basis for exclusion in 81% of the trials, and patients receiving commonly prescribed medicines were excluded in 54% of the trials.69 Such exclusions were significantly more common in industry-sponsored trials.

  According to the Helsinki Declaration, a new drug must be tested against those of the best current proven intervention, and placebo should only be used when no such intervention exists, or where there are compelling and scientifically sound methodological reasons for the use of placebo (e.g. when current treatments have doubtful effect).71 I therefore believe drug agencies should require relevant head-to-head comparisons against commonly used drugs, and that, when placebo is needed, they require that some of the trials have used active placebos to reduce the risk that they approve useless drugs (see Chapter 4).72

  Safety

  The current practice, where drugs are approved based on only 500–3000 patients73 in short-term trials, even when the drugs are to be used for decades, sets the stage for major drug disasters, which are very expensive because of legal fees and settlements. These costs are subsequently added to the price of other medicines.74

  Apart from identical twins, people are genetically different, e.g. in how quickly they metabolise a drug or how susceptible they are to its effects. It is therefore expected that for most drugs, some people will react very badly. These reactions can only be detected reliably if many patients are studied in randomised trials. If, for example, a drug causes fatal liver failure in one patient out of 2000, even a trial of 20 000 patients may not detect it (since we would only expect to see five liver failures in the 10 000 randomised to the new drug, we might have seen none at all). If that drug is an analgesic to be used in, say, 50 million people, 25 000 of these will die of liver failure although they didn’t need the drug, as there are so many other analgesics.

  This won’t happen in practice of course. The drug would be withdrawn long before the 25 000 liver failures, but if it had been an increase in heart attacks, we might never have found out, as so many will get a heart attack anyway.

  Drug agencies should require much larger numbers of treated patients before they make any decisions, and they should also require trials that run over several years, if the drug is to be taken for years, as harms may take time to develop, e.g. if the drug causes cancer. The standard excuse that it would take too long to get valuable new drugs on the market, if requirements were tightened, doesn’t hold. There are years between such an outstanding drug and the fact that drugs are the third most common cause of death says more than enough about the current system.

  If post-marketing studies are required because of remaining safety concerns, it’s essential that these be conducted by, and commissioned to, as part of the regulatory approval, independent investigators. Companies have every reason in the world to blindfold themselves by conducting poor studies, or fail to report them, or fail to even do them.

  All clinical data must be publicly available

  It’s a terrible misconception that a company can own clinical trial data. According to the European ombudsman, data and results belong to society, for obvious reasons. Patients don’t volunteer for trials and run a personal risk to benefit the shareholders of a particular company. They do so to contribute to science and help improve treatments for future patients. If we accept that companies can claim ownership to trial data, we also accept that it’s legitimate to exploit patients for commercial benefit. It obviously isn’t and would violate the Helsinki Declaration.71 We should therefore force companies to make all their trial data available, including raw anonymised data in statistical programmes, which is what the EMA intends to do for new drugs (see Chapter 11).

  We should also use our muscle, e.g. by deciding not to recommend drugs or not to buy them until all data have been made available. UK law makes it possible to take legal action without notice and to delay approving the company’s pending drugs, or even to withdraw a marketed drug if the company refuses to provide all the data. Confiscation of the company’s patents is another penalty that is under consideration. If a company abuses a patent by marketing a drug for a purpose it has not been approved or tested for, why should it then continue to benefit from the exclusivity?75

  Health Technology Assessment agencies should follow the lead of the German agency IQWIG and refuse to assess a drug unless all data are provided from all trials without any conditions about confidentiality, so that the public can also see the data.

  As suggested by the Danish drug agency,76 there should be full public access to all documents in drug agencies, including data from toxicology studies.49 Drug companies should be required to submit all documentation in easily searchable formats, e.g. as text-recognisable pdf files; the agencies should check that the files are complete and include all documents listed in indexes; and the files should be made publicly available. US legislation from 2007 ensured that the data bank at clinicaltrials.gov would be expanded to include all phase 2 and subsequent trials, and that results information would be added after the product has been approved for marketing.77 The restriction to marketed drugs needs to be removed, however, as unknown harms of a drug might lead to unnecessary experimentation and additional harms of similar drugs in future. The six healthy volunteers in the United Kingdom who nearly died in a phase 1 trial is a case worth remembering.78

  Above all, there should be no redactions, which will require changes of some national laws. We should not accept receiving documents from drug agencies that are so heavily censored that they look more like military intelligence documents than drug studies,79 and where all the harms of the drug have been wiped out, which we experienced when we received study reports for a slimming pill from the Danish drug agency and on SSRIs from the Dutch agency. An additional problem is that redaction is arbitrary. Sidney Wolfe, director of the Public Citizen’s Health Research Group, said, ‘I’ve never been able to get any kind of protocol for what [FDA staff] are instructed to redact, but in general they redact way more than they should.’79 He added, ‘Of course, it’s a catch-22, because if you don’t know what they are redacting you can’t argue that it should not have been redacted.’ Alastair Wood, the head of the FDA’s advisory committee on the safety of COX-2 inhibitors, insisted that there is no reason ever to redact clinical trial data. Absurdly, without knowing what information is being withheld or the rules guiding redaction, the interpretation of what constitutes a trade secret seems itself to be a trade secret.

  We also need laws requiring firms to disclose all knowledge about their drugs and research data,80 and that not only allow but require drug agencies to publish what they know. Currently, the companies may not disclose anything even when they know that their drugs are harmful.

  Conflicts of interest

  Drug agencies should be publicly funded, as user fees create competition between the agencies about becoming the fastest and therefore also the least critical agency. For example, it was a goal in the contract between the Danish drug agency and the Ministry of Health to be more attractive to the industry than other agencies.81 Drummond Rennie believes that user fees are profoundly corrupting and that ‘It is ludicrous to imagine that the FDA could truly work for the public interest if they continue to be paid not to.’13

  Divisions in drug agencies that deal with harms of drugs should be separated from the divisions that approve drugs and should have their own authority, enabling them to remove drugs from the market. According to laws of public administration, a person or a body must never come in a position where it evaluates itself. For this reason alone, it’s clear that the two functions must be separated. Alastair Wood, whose nomination as the new FDA commissioner was withdrawn in the last minute because he put too much emphasis on drug safety (see Chapter 10), has noted that ‘When a plane crashes, we don’t turn over the investigation to [the airline] and the air-traffic controllers. We get someone else to do it.’82 The Danish drug agency understands this and has separated the two functions,83 but
the FDA won’t.

  Since our drugs kill us, drug agencies should be evaluated by how well they handle safety issues. Currently, however, the emphasis is on the speed with which new drugs are being approved,1,84 with performance-based salaries to the top executives, e.g. in the Danish drug agency.81 This incentive is not only perverse; it is lethal.

  Labelling of drugs

  If the drug agencies’ clients were the people and not the industry, the label on drugs would look very different, somewhat like this (inspired by drug epidemiologist Jerry Avorn):67

  This new drug hasn’t been shown to be any better than currently available drugs, and we know much less about its harms, including the lethal ones, than we do for old drugs. There is no evidence that its higher price is accompanied by any therapeutic advantage. It’s generally safer to take an old drug, as many new drugs come off the market later because of safety problems.

  Patients should be informed about what the drug does, with numbers they can understand for both the benefits and the harms. Researchers from Dartmouth showed that if patients are told the facts, they are much better at choosing the better drug and in knowing what the effect is.85 If people knew that the effect of a sleeping pill is to make them fall asleep 15 minutes faster,86 and could make them dizzy and drowsy the next day, they might be less interested in taking one, and if they also knew that the effect would disappear within 2 weeks if they take it every night, there would be few long-term users. The Dartmouth researchers convinced the FDA’s Risk Communication Advisory Committee that the agency adopt their suggestions. However, after having thought about it for a year, the Department of Health and Human Services announced it needed at least three more years to come to a decision.87 Of course they do. An initiative that indisputably helps patients to choose much more rationally between drugs, or even to say no to drugs, is almost like an attack on the state, as it could lead to loss of income for the drug industry.