by Burch, Druin
Over a year later, in December 1984, the FDA heard Sterling’s case again. This time Peto dressed both himself and his talk up with all the paraphernalia and style that the committee expected. Rather than scribbling out explanations as he went, he had slides. Rather than corduroy, he wore a suit.
You will hear some doctor saying, well, if it doesn’t show up in a trial with a couple of hundred patients then it can’t be worth bothering with. This is not medical wisdom but statistical unwisdom.
Interventions that exert moderate effects on common diseases, Peto explained, can save a lot of people’s lives.
Some of these people will be old, some will be horrible people who would be better dead anyway, but a fair number of these are going to be in middle age with a reasonable chance of enjoying life. So this kind of thing is worth doing.
This time the committee was convinced. The FDA approved labelling aspirin as a life-saving treatment for those with heart disease. A year later the US Health Secretary was explaining the drug’s benefits to the press.
Jeffreys’ account is a vivid one, but it makes no mention of the cost of the delay. In between the FDA’s rejecting Sterling’s first presentation and accepting the second, heart disease carried on its normal business of killing. In America alone, 20,000 people died that year whom aspirin would have kept alive. The FDA’s jurisdiction only applies to America, but its influence is worldwide. The impact of the FDA’s delay was huge. So was the delay of governments worldwide in advertising aspirin’s benefits. Why was there a year’s gap between FDA approval and the US Health Secretary holding up a packet of aspirin at a press conference? Even when science provided firm answers, bureaucrats were slow to respond.
Aspirin was gradually taken up, in the wake of academic papers and conference presentations, advertisements and press conferences. Dr Khan’s encyclopedia suggests that by 1983, cardiologists had enthusiastically adopted it as a treatment for heart disease. The British Heart Foundation looked, in 1987, to see if that was true. They counted how often coronary care units – the pinnacle of cardiological practice – gave aspirin to their patients. Of those for whom it was potentially life-saving, it was being given to one in ten.
The original meaning of the term ‘antibiotic’ was not a substance that helped snuff out an infecting bug, but something that one organism’s physiology manufactured in order to wage war with another. Some of these bugs, having evolved to attack us, produce molecules tailored by natural selection to do so. And in the same way that all beneficial drugs have the capacity to harm, some harmful ones are also able to help.
The streptococcus that causes puerperal fever no longer preys so much upon mothers. There are occasional cases, but the last epidemic outbreak was in 1965 in Boston, Massachusetts. The reason for the decline is not understood, but may represent the bacteria’s having evolved to focus on other things than causing disease, or at least this disease. Nevertheless, the streptococcus is still around, and still retains much of its basic molecular equipment. One of the ways in which it was deadly was through its production of a drug acting on human blood. It is a molecule that stops clots from forming, or breaks them up if they are already there. From the point of view of the streptococcus, the drug helps it avoid being walled off by a host’s defences, entombed within a blood clot.
From the 1950s, doctors were experimenting with a drug made out of the molecule. It was called streptokinase and it dissolved blood clots regardless of whether bacteria were present. During the following decades, trials were undertaken to see if it could usefully attack the clots that form in coronary arteries. By the time the FDA approved the change in labelling for aspirin, streptokinase had been generally rejected. It was clearly dangerous – causing far more bleeding even than the anticoagulant drugs – and the trials of it were disappointing. True, they were all too small to actually give a reliable answer, but the cardiologists were not going to have their decision-making slowed down by such quibbles. They had carried out a few trials and now they were ready to move on to trying something else.
Combining the results of lots of small trials – putting similar experiments together to increase the validity of their conclusions – was, however, starting to become popular. Richard Peto was involved in an overview combining the data from thirty-three separate trials of streptokinase. Individually, each was so small as to be meaningless. Taken together, they showed something surprising: streptokinase was dangerous, but it worked. It caused more fatal bleeding than aspirin, but it also stopped more fatal heart attacks. On balance, it did good and saved lives. Together with two colleagues, the cardiologist Salim Yusuf and the trials specialist Rory Collins, Richard Peto published this analysis in a specialist cardiology journal. Most cardiologists who read their paper did not believe it.
Faced with a group of specialists who understood so little about the nature of medical knowledge, the researchers came up with another strategy. They had uncovered the truth about aspirin and streptokinase, and most cardiologists had ignored them. Now they looked for ways to get the cardiologists to do better. Doctors could clearly not be relied on to understand scientific method; attention had to be paid to educating them.
A large trial was therefore designed with a dual purpose. It was an attempt to alert cardiologists to the nature of reality, but also designed to uncover a bit of new knowledge along the way. What happened, for example, if you gave streptokinase and aspirin? Given that streptokinase carried such a risk of terrific blood loss, was there a way of figuring out exactly those people it was either likely to kill or to save, so as to more profitably choose whom to give it to?
The new trial was so large that, rather than being listed under the names of those organising it, it had its own name. There were too many people involved (and perhaps too many medical egos) to name it any other way. ISIS-2 was actually the second in a series. The first International Study of Infarct Survival had looked at drugs called beta blockers, and their ability to help people live through their heart attacks. This second one now took 17,000 heart attack patients and divided them into four groups. One group got aspirin, another streptokinase, a third both; the fourth group got nothing at all. Given the fact that both aspirin and streptokinase were already proven to save lives, a control group that got neither was an ethically curious decision.
The trial published its results in 1988. Of the patients who got nothing at all, 87 per cent survived to the end of the month after their heart attack. Aspirin improved things, and so did streptokinase. In fact, the two drugs worked out pretty much the same, improving people’s chances by similar amounts (just over a 2 per cent absolute increase in survival rates for each) despite that fact that one had dramatic results – both good and bad – while the other was milder. Given together, they provided the best of all outcomes. Among that group, 92 per cent survived at the end of the month. Taken together, these two drugs reduced someone’s chance of dying from a heart attack, their relative risk, by about 40 per cent.
This time, the trial was so big and so conclusive that people took notice. The British Heart Foundation had found that 10 per cent of heart attack patients were being given aspirin the year before ISIS-2 came out. The year after it was published, in 1989, they repeated their survey. This time the figure was 90 per cent.
The research unit that Richard Peto helped to set up continues to operate, to undertake large trials and report on their effects. They estimate that aspirin prevents around 4,000 deaths a year in Britain alone. If everyone who was meant to be on it actually was, and all those people took their tablets, that number would be 7,000. Medicine’s main benefits for human health are in these accretions of moderate effects.
‘What needs to be overcome’, said William Silverman, an American paediatrician energised by the battle to do his patients more good than harm, ‘is a naïve “all-and-none concept”.’ A patient within a double-blind, randomised controlled trial is taking a risk. One of the treatments is likely to be better than another, and patients have no control
over which one they get. In the case of ISIS-2, some of the harm could have been avoided by getting rid of the placebo arm – in other words, if cardiologists had been a touch more modest and an ounce more numerate, an estimated seventeen deaths would have been avoided among the trial patients alone. (To a significant extent, cardiology is now a speciality that has learnt its lesson. That is to say, many treatments within cardiology are now based on good evidence. Many others are not.)
What is needed is a culture, regulatory and intellectual, where every attempt is made to ensure new medical interventions are used solely in randomised trials. Only when their effects have been determined should they become available for use outside a trial setting. Until then there is a moral obligation on doctors to use unknown drugs and treatments only in such a way as to come to an understanding of them, and a moral obligation on patients to demand treatments that are either supported by sound evidence, or only given as part of a trial which will uncover some.
No regulatory system can encourage all innovation and stifle all errors. No amount of trials can make sure that all medical decisions are based on the firmest conceivable evidence. That is no reason not to keep trying.
22 The Battle for Hearts and Minds
AROUND A QUARTER of a century ago doctors began using a new type of drug to treat their patients.
Although some of those with heart disease are killed by clots in their coronary arteries, others die in different ways. The electrical co-ordination of the heart, particularly in past sufferers of heart attacks, can go wrong. Rather than beating, the heart just quivers. This is precisely the condition for which people are given electric shocks. The paddles on their chest deliver a charge that can reset the heart’s conducting system. As a medical approach, it has more than a little in common with switching a computer off and on again. Sometime it works, sometimes not.
On the reasonable grounds that prevention was better than cure, cardiologists tried to think of ways of protecting the heart from these sorts of problems. They noticed that there were certain patterns of electrical activity that predicted future disaster. Everyone’s heart skips out of its normal rhythm, at least a few times each day. It throws in extra beats and misses the odd one or two as well. After a heart attack, though, people with lots of extra beats seemed to be at higher risk of dropping down dead from a sudden and total electrical failure of cardiac co-ordination.
A class of drugs were developed – antiarrhythmics – that prevented these extra beats. This was where doctors let themselves down. They reasoned that if you suppressed the extra beats, people would live longer. As a theory it was perfectly respectable, but theories are hypotheses that you test. Whereas cardiologists had once been suspicious of aspirin because of their inability to explain the mechanism by which it thinned the blood, with antiarrhythmics they seemed content to behave in a very different way. Perhaps because these drugs were newer, and linked in their minds with all the allure of modern technology, they were far more confident about them. Whatever the reasons, they were aware of the gap in their knowledge but believed it did not matter. Why go to the trouble of testing drugs that were so obviously going to work? Why delay the introduction of potentially life-saving medication?
This line of thought was set out explicitly by a doctor called Bernard Lown, speaking at the 1978 convention of the American College of Cardiology. Four hundred thousand people a year were dying suddenly when their hearts stopped, said Lown. The way to help save them was to stop the extra heartbeats that came before their deaths. He admitted that there was no direct proof that preventing the beats prevented the deaths, but when people were dying at such a rate, what purpose could be served by waiting to confirm the theory? ‘In medicine,’ he explained, ‘great rewards have flowed from partial answers and usually have preceded complete solutions. This is the case with sudden cardiac death.’
A few doctors disagreed, and said they thought this partial knowledge was not enough. They suggested that experiments needed to be done in order to show that the drugs suppressing the extra heartbeats were actually life-saving. They got very little attention. ‘In American medicine,’ said Thomas Moore, who wrote a book about the whole shambolic episode, ‘words of caution rarely slow the rush to treatment.’ The thought was in tune with Oliver Wendell Holmes’s ideas about the American longing for heroic cures, but allocating blame to Americans in particular seems unreasonable. The drugs also began to be used eagerly in Europe and elsewhere. The Soviet Union developed a novel one of their own.
The idea that suppressing extra beats was life-saving took hold in the 1970s. At the end of the decade, in 1979, American doctors were writing out 12 million prescriptions a year for these drugs, trying to keep the hearts of their nation beating along in an orderly fashion. By 1981, an editorial in the New England Journal of Medicine praised their utility. One such drug was described as ‘an important addition to the current antiarrhythmic armamentarium’. If that last word sounds pre-modern, almost medieval, then so was the untested thought behind it.
For the best part of a hundred years, America had already possessed regulation claiming to make sure drugs were safe and effective before they were used. Each time a drug disaster showed the regulation to be a mockery, it was improved a little bit. Never enough, but a little. In the 1970s and 1980s the requirement to prove effectiveness still did not specify how ‘effectiveness’ should be measured. In this case the drugs were effective in the eyes of the cardiologists; that was enough for the FDA, and enough for the cardiologists themselves. Today there is still no absolute requirement for proving hard end points like mortality. A soft proxy, in this case the suppression of extra beats, is often still acceptable.
To begin with, the main avenue for prescribing these drugs was through ‘compassionate use exemptions’ allowing the use of experimental drugs on people too sick to wait for tests to be completed. The whole idea of the exemptions is based on the presumption that they are likely, once the testing has been finished off, to actually work. This is itself based on the belief that new drugs are more likely to help than harm. Compassionate use allows the drugs to be prescribed outside a trial. It often supplies doctors with their earliest experiences of a treatment. Thus the medical profession frequently forms its views on the basis of anecdotes, of observations without any of the trappings even of second-rate methodology. The drugs, thought the doctors, worked well. The extra heartbeats were effectively suppressed. Some of the patients died, but then they were people with damaged hearts. Doubtless some were always going to die. The cardiologists reckoned that there would have been more deaths if not for their new drugs.
A major attempt to test any of these drugs came with one called flecainide in 1983. Flecainide interferes with the way heart cells use sodium ions to control their contractions, altering the overall way in which the heart beats.
As so often, when doctors are persuaded to trial something, the aim was not to see if their theories were correct. They were certain about that. Their aim was to persuade other doctors, whose intuitions were different. This first study was in fifty-five patients. That turned out, as could have been easily predicted, to be too small a number to determine the drug’s actual effects. The next study, this time of 630 people and of a similar drug called mexiletine, also ended up being too small to give an answer. That was partly because, when more people started dying in the treatment group than the placebo one, the study was brought to an early halt. It was not heading towards proving the benefit that the doctors knew was there, so there was no point in carrying on with it. The drug, noted the trial summary, appeared to be good at controlling abnormal heartbeats. That was enough for now.
At a cardiology meeting in Bermuda in the same year, 1983, paid for by the makers of flecainide, opinions about the drug were still positive. The doctors remained aware of the trial data they were missing, and remained happy that these were not essential. ‘We do not need to hold up marketing of the drug until we have all of this information,’ said one. Although ente
ring any of their patients into a trial required an intimidating amount of paperwork, and a formal ritual of informed consent, simply giving it to them without any reliable trials needed none.
By 1984, the issue was still undecided. Some experts thought the drugs worked, a few did not. In the absence of any reliable trials, there was nothing better to go on than opinion. The FDA approved flecainide on the basis that the cardiologists believed in it. They noted that a minority of doctors were worried that flecainide was dangerous. Therefore, suggested the FDA, its use ‘should be reserved for patients in whom, in the opinion of the physician, the benefits of treatment outweigh the risks’. Given that the physicians had nothing to base that decision on besides their own optimism, what happened next was not a surprise. Prescriptions of the drugs kept rising.
At the heart of the FDA’s thinking, and it was betrayed in their words, was the old idea that benefits are certain but harms are not. And there was no conception that opinions, regardless of a doctor’s personal attributes, are no way to assess risks.
Thomas Moore’s history of these antiarrhythmics, Deadly Medicine, notes another problem. The amount of money that companies spent advertising their drugs was greater than their expenditure on research. And their research cost billions of dollars a year. In the views of the doctors, those advertising budgets did not matter. Only a tiny minority of doctors believed their own views were ever swayed by something so crude a thing as advertising. They were sure their decisions were based on an ability to weigh up rationally what was best for their patients. Good evidence, however, said that doctors were mistaken. In one study, when they were shown identical drugs advertised in different ways, half of all doctors thought that one of them was very much better than the other. Time and again research showed that doctors were influenced by advertising; this remained something that the doctors themselves could not quite believe.
