by Burch, Druin
. . . my face became noticeably redder & the skin had a tendency to flake off, & a sort of rash appeared all over my body, especially down my back . . . It was very painful to swallow & I had to have a special diet for some weeks. There was now ulceration with blisters in my throat & in the insides of my cheeks, & the blood kept coming up into little blisters on my lips. At night these burst & bled considerably, so that in the morning my lips were always stuck together with blood & I had to bathe them before I could open my mouth. (Bastian, 2004)
Orwell’s hair and nails fell out. Like the bleeding and the rash, this seemed to be a reaction to the streptomycin. Orwell wrote to a friend, coining his own image to express the delicate balance his doctors were trying to find between streptomycin’s benefits and harms: ‘I suppose with all these drugs,’ he said, ‘it’s rather a case of sinking the ship to get rid of the rats.’
As his reaction grew worse, streptomycin was withdrawn. Most of the side effects soon faded, although Orwell’s nails never properly grew back. He was kept in hospital, prevented from spending time with his son and even – such were the demands of the prescribed bed-rest – prevented from writing. Desperate to finish his novel 1984, he tore his way through it, closing his life with the feeling that the rush and the sickness caused him to ruin it. ‘I ballsed it up rather,’ he concluded. The misery of feeling he had botched his book did not get rid of his urge to try and do something better: ‘I must try and stay alive for a while because apart from other considerations I have a good idea for a novel.’
Orwell died in January 1950. The final part of his life was destroyed partly by the side effects of streptomycin, which the MRC trial made clear was the flip-side of a generally beneficial drug, but also by the bed-rest – which was, in contrast, an entirely avoidable and pointless harm. Truth is corruptible – that was one of the messages of Orwell’s 1984. It is vulnerable, but so long as people cling to numbers, so long as they insist that two plus two make four, it can be protected. Opinions and propaganda are capable of distorting the world but numbers are not so easily swayed.
The Medical Research Council’s work, the culmination of a long historical rise in statistical thinking, showed something similar. The randomised controlled trial could sweep away the propaganda of drug companies wanting to sell expensive products, of eminent physicians blinded by their own self-belief, of doctors misled by good intentions and the play of chance. For the first time, doctors had a way of reliably discovering the truth about the world.
It gave rise to new problems. In his Principles of Medical Statistics, Austin Bradford Hill raised two questions that troubled him. ‘It is impossible to ignore the fact that in the random allocation of patients to the treated and untreated categories,’ he noted, ‘a difficult moral issue is often raised. The treatment is usually based on a priori evidence which suggests that it should have some curative effect. Can it, then, be justifiably withheld from any patient? And if it is withheld how extensive a trial is justifiable?’ Hill might have added a third question, too: ‘What sort of people should be randomly allocated to experimental trials, and what sort immediately given the newest and most promising treatment available?’
It was clear that the role of statistics was fundamental to medical research. If you wanted to know if something cured or killed, you had to be in the business of counting. If everyone who took it immediately dropped dead, coming to a conclusion was easy. If the difference was smaller then you needed to be more sophisticated. That meant thinking about statistics from the very beginning. You could not blunder your way thoughtlessly to begin with, then use some sort of statistical analysis to sort yourself out afterwards. Statistics needed to be part of your plan from the start.
Karl Pearson’s argument for randomisation was actually quite different from Hill’s. Pearson distrusted other techniques, fearing that they would not sufficiently spread traits across the groups being compared – Hill, on the other hand, just distrusted the doctors. His feeling was that they were simply unable to let go of their belief that they knew, in advance of a trial, which treatment was best. The experience of the MRC serum study, like the weight of thousands of years of medical history before it, appeared to prove him right.
It was also true, as Hill was clearly aware, that these statistical techniques offered problems as well as solutions. They had emotional costs. Going along with a randomised controlled trial meant summoning up the humility to put your ideas to the test, and designing experiments that were capable – despite how much you believed something was going to turn out true – of proving you wrong. Hill summed up what he expected doctors to screw themselves up into doing: ‘The more anxious we are to prove that a difference between groups is the result of some particular action we have taken or observed, the more exhaustive should be our search for an alternative and equally reasonable explanation of how that difference has arisen.’ That was the essence of the new medicine’s hard-won scepticism.
Beyond it, there were the ethical problems, ones that Hill had more difficulty in resolving. Unleashing a medicine without understanding its effects was clearly irresponsible. What about only giving it to half the people asking for it, though, because you are insisting on running a trial? Was that not also wrong? What about those people in the control group? The answers were not clear, but the alternatives were. Small and badly designed trials could too easily give the wrong answers. Life-saving treatments got abandoned as a result, and others that caused suffering and death were mistakenly adopted in their place. There were real moral difficulties about withholding a new treatment from half of those who might benefit, but was not giving an untested drug outside of a trial morally worse? It certainly held out the prospect of harming far greater numbers of people. As relying on therapeutic hunches had failed for millennia, there was no reason to think it was suddenly going to start becoming a successful strategy now.
Giving a drug outside of a trial not only meant that lost benefits and added harms were potentially vaster, it also meant they had no hope of coming to an end. Without a trial, the answer might never be known. Was it not at least better to give the wrong treatment to half of a small group of people, for a limited time, in the expectation of learning from the experience? Rather than just guessing and risking giving the wrong treatment to everyone for evermore?
John Crofton was a junior doctor involved in running the MRC trial of streptomycin. Almost sixty years later, he wrote an account of his experiences. His feeling for the importance of the study comes across. Streptomycin’s role in tuberculosis became rapidly understood in a way that was impossible through any other method. And the precedent set by the trial was superb. Crofton noted that ‘when Archie Cochrane was pondering which of the specialities within medicine had made most determined efforts to base policies and practices on the results of reliable research he had no hesitation in awarding the “gold medal” to the tuberculosis specialists’. Having seen the power of trials to illuminate their own field, these doctors were more persuaded than many of their colleagues. Crofton himself went on to be part of a group that found there were no benefits to TB patients from enforced bed-rest. Across the world, the sanatoria were closed.
Despite all this, Crofton made two revealing and surprising remarks. ‘Randomised trials like these’, he said of the streptomycin one, ‘were of great practical importance in developing effective treatment strategies, but they were not intellectually challenging.’ He knew he was making history with the streptomycin trial, and he still found something about it boring.
Then, right at the end of his reminiscences, Crofton recalls the many doctors from Britain and abroad who later came to visit his hospital. The great majority, he remembered, were interested in what his trials had found – not the methods by which they found them. You could set the best precedent in the world, and most of the medical profession was not going to be interested. Getting doctors to pay attention to results was straightforward; making them think about methods was torture.
Runn
ing a trial involved a lot of forms, a lot of protocol, and a lot of time. It was dull, it was methodical. The actual performance of it gave doctors no opportunity to enjoy the decision-making and intellectual effort that they normally relied on for job satisfaction. It couldn’t: the whole point of the trial was to temporarily remove the ability of individual doctors to make decisions. That was what you needed to do in order to see what happened when people were treated according to a strict protocol.
Methodical tests remain something that most doctors do not enjoy: an ongoing problem when it comes to doing them, and a clue as to why they took so long to develop in the first place. Their effects have been unmistakable all the same. ‘The Medical Research Council’s trials were designed according to statistical analyses made by Bradford Hill and, later, by Ian Sutherland,’ concluded the scientist Max Perutz:
They have been crucial to the near-eradication of tuberculosis in developed and many underdeveloped countries throughout the world. They were the first to evaluate the efficacy of treatments free from human bias and according to rigorous mathematical criteria, and they have helped to transform clinical practice from an art into a science.
Archie Cochrane played no part in the MRC streptomycin trial, but it affected him profoundly, as he recorded in his memoir:
It was at this time that I began to wonder and discuss with my colleagues how other forms of medical treatment would stand up to the test of the randomised controlled trial. Looking back, this is undoubtedly the point at which the immense potential of [it] . . . began to dawn on me. It offered clinical medicine, and health services generally, an experimental approach to questions of effectiveness and efficiency, and a massive step forward from ‘validation’ by clinical opinion and essentially subjective observations. I think it was the simplicity of the idea in relation to the magnitude of the advance it represented that captured my imagination.
Enthralled, Cochrane was one of the minority able to take intellectual delight in the opportunities this approach offered, and the doubts it cast on the benefits of expert opinion – something he was habitually sceptical about. Most of his work was on the lung diseases of Welsh miners, their causes and treatments, and the ways in which doctors could most usefully gather and think about the relevant information. He made a few forays into other areas. Cardiologists become a favourite target. Electrocardiograms are recordings of the heart’s electrical activity, the jagged lines so beloved of TV and film directors. Cardiologists claim skills in reading them that are beyond the measure of other doctors. Cochrane took randomly selected ECGs and sent copies to four different senior cardiologists, asking them what the tracings showed. He compared their opinions and found that these experts agreed only 3 per cent of the time. Their confidence in being able to look at the tracings and see the ‘truth’ did not seem justified. At least ninety-seven times out of a hundred, someone was getting something wrong. When Cochrane performed a similar test with professors of dentistry, asking them to evaluate the same mouths, he found that there was only a single thing that their diagnostic skills consistently agreed on: the number of teeth.
Convinced of the uselessness of most untested medical opinions, Cochrane was willing to take what appeared, to other doctors, to be risks. During a routine screening as part of his research, he found one man, ‘a very likeable, cheerful, tough young miner, married with two children’, whose chest X-ray showed lymphoma, a type of cancer. The man felt completely well. The accepted treatment was immediate radiotherapy. The side effects were known to be crippling, and no tests had ever been done to prove the widespread medical prejudgement that they were worthwhile. Cochrane decided not to tell the man, to ‘protect’ him from what the cancer specialists believed, without evidence, to be life-prolonging treatment. Instead he arranged for the man’s family doctor to follow him up surreptitiously. ‘He lived happily for another 10 years. He paid off his mortgage and had another child. I felt justified. He died rapidly after developing symptoms.’
In 1956 a surgeon removed a swollen gland from Cochrane’s armpit. It was meant to be a routine procedure, but when he woke from the operation Cochrane found his chest wrapped in vast swathes of bandages. ‘I must tell you the truth,’ the surgeon said, coming into the room soon after and looking grave. ‘Your axilla [armpit] is full of cancerous tissue.’ He explained to Cochrane that it seemed the cancer was advanced and inoperable. There was probably little time left.
The surgeon made his decision about the nature of Cochrane’s disease on the basis of what he saw on opening him up. The normal procedure, if there was any doubt about the appearance of a lump during an operation, was to have a piece of it examined immediately under a microscope by a pathologist. The surgeon was certain enough of his judgement to make that unnecessary. Trusting his own opinion, he had not stopped at cutting away the original gland, but had gone on to remove a large part of Cochrane’s chest wall and the attached muscles.
Cochrane lay in his bed planning how he should spent his last days, and thinking of where it was that he wanted to be when he died. For all his scepticism, when he was on the receiving end of such a serious medical opinion, he trusted the professional skills of his colleagues. The tissue cut out of him had not yet been fully examined, but he accepted that the diagnosis was certain. ‘I had been told that the pathologist had not yet reported, but I never doubted the surgeon’s words.’
The gland turned out not to contain any cancer at all; the surgeon’s impression of his own infallibility was a fiction. It was yet another lesson for Cochrane of the dangers of doctors who think too much of their own opinions. Restored to vigorous scepticism, and to equally good health, Cochrane continued his personal campaign to persuade people that whatever could be tested, should be tested. As a Cambridge medical student he had laboured to learn the detailed anatomy required of him; his efforts placed him first in his year. They were probably, he thought in retrospect, completely useless. What was the point of cramming incredibly intricate anatomical knowledge into medical students at the beginning of their courses? He suggested they should be tested, a few years later, to see if any of them remembered enough of it to justify the high opinion that their teachers placed on their efforts. The Professor of Anatomy, convinced without any such evidence that these early labours were good for the students, refused to co-operate.
When two different ways of running the local medical school were suggested, Cochrane suggested that, rather than trusting to their mutual ability to recognise the best one, they perform a trial. He suggested adopting one method at Cardiff and the other at nearby Bristol. Then they could allocate students, most of whom applied to both, randomly to one of the two schools. In a few years’ time they would be able to objectively see which system produced the best results. His colleagues laughed.
Having been beaten as a schoolboy, Cochrane wondered whether different forms of punishment in prisons and schools might be tested. ‘Corporal punishment had made an impression on me . . . Since then I had often wondered about the value of such “correction”, which so many extolled and yet so few had sought to validate.’ He tried and failed to find any schools willing to co-operate. He had no success at getting the government to test prison policies, reporting that ‘civil servants have consistently exhibited hysterical reactions to the mention of randomised controlled trials’. The Conservative Party’s 1979 suggestion that criminals needed a ‘short, sharp, shock’, Cochrane pointed out, could be tested.
I discussed the problem privately with a number of magistrates I knew. I suggested to them the value of controlled trials, but was horrified when they reacted like elderly physicians and headmasters. They too suffered from the God complex. They knew what to do without the help of any trials. It was depressing.
Then it was back to trying to irritate cardiologists out of their settled complacency. Cochrane was concerned at the introduction of coronary care units (CCUs) in hospitals. Miniature intensive care wards, designed to look after patients brought into hospital with heart attac
ks, they were new and expensive and theoretically excellent. Cochrane wondered if they actually delivered anything of value. Could they conceivably make things worse? He thought it was worth checking. Perhaps the noise and machinery in them, the awareness of being monitored so intensively, made people anxious – and perhaps that anxiety was an unhealthy thing for those whose hearts were unstable. Even if the units did not harm their patients, they took a lot of NHS money away from other areas, so merely assuming that they delivered sufficient benefits to make them worthwhile was not, argued Cochrane, a good enough way to proceed.
To his amazement, the government and the Medical Research Council agreed to set up a committee to look into whether a trial might be needed. The cardiologists said it was not. They presented evidence showing that in-patient death rates from heart attacks promptly declined whenever a hospital set up one of these new CCUs. Cochrane objected that whenever a unit was opened many of the healthiest patients, who would previously have been sent straight home, ended up being admitted. That resulted in the overall inpatient population becoming very much fitter. Was it then any surprise to find that a smaller percentage of them were dying? As with Galton’s investigation of prayer, an observational study could not distinguish what was cause and what was effect.
The committee was convinced, and approved a trial. The cardiologists responded by refusing to be involved in it, or let their patients near it. ‘Later I became fascinated by the psychology underlying the decision,’ wrote Cochrane. ‘The consultants whom I knew personally were ordinary, reasonable, intelligent people. They knew that the Platt committee had completely outflanked them intellectually and ethically, but they still felt a sacred right to treat patients as they wished. I was horrified.’
Cochrane eventually found some cardiologists willing to help, although he was amused to note their reasons. They were not going along with the trial because they felt their beliefs needed testing, but out of the conviction that the results would confirm their views and strengthen their arguments for more of the new units. Six months later, the committee sat again. This time the aim was to look at the data that the trial had gathered up to that point, comparing the death rates of patients randomly allocated to CCUs with those randomly selected to be sent home. Cochrane, presenting the data, knew that it showed an advantage for those patients who were sent home. It was too small, however, to be reliable – given the small numbers of patients then in the trial, it could too easily be down to luck.
