by Burch, Druin
There were also times when what was required from a doctor was an entirely intuitive therapeutic intervention. Trials were not the only way of discovering what worked. At the end of a long day in Elsterhorst, the Germans handed over to Cochrane a young Soviet prisoner, screaming in pain and dying. Examining him, Cochrane found that tuberculosis had eaten up the man’s lungs, and that the outer covering of one of them, the pleura, was inflamed. It rubbed agonisingly against the inside of the man’s ribs each time he took a breath.
Unwilling to let the man’s cries wake the ward, Cochrane brought him into his own room and treated him there. Aspirin had no effect; there was no morphine available. The man’s screams were awful. Cochrane knew a little Russian, but not enough, and there was no one who could translate. At last, unable to bear the man’s howls any longer, Cochrane sat down on the bed and held him. The screams faded away. A little while later, apparently peacefully, the man died. The inflamed coverings of his lungs were not what had been most painful to the dying man: that had been loneliness and terror. ‘It was a wonderful education about the care of the dying,’ wrote Cochrane. ‘I was ashamed of my [initial] misdiagnosis and kept the story secret.’
In January 1945 the Germans attempted to move some of the desperately sick tuberculous patients, loading them carelessly onto the open back of a truck while the snow came down. ‘I had looked after these patients for a long time and had done my best. I loved them dearly. I thought this murder. I lost my temper and jumped on the wagon and made a loud speech to the Germans.’ Their behaviour, he told them, was shameful: ‘Das ist doch ein Skandal im Lande von Robert Koch’: ‘This is a scandal in the land of Robert Koch.’ The truck was replaced with an ambulance but Cochrane was dismissed from his post in the camp and replaced with another doctor.
By the end, Cochrane felt he had done as well as he was able. He was troubled, though, by the thought that he had shortened some of his patients’ lives, not knowing the actual impacts of the medical choices he was making. The day-to-day demands of clinical work, he discovered, were also not enough for him. ‘I had found it emotionally satisfying and distressing at the same time. I knew my patients so well that I was miserable when they died. I also found ‘caring’ intellectually unsatisfactory.’ Had he been less of a man, had he been able to convince himself of his own power always to determine what was best, like when to deflate a lung, he would not have been left so unsatisfied. Cochrane clung successfully and painfully to an awareness of his own ignorance. It was something that only a minority of doctors before him seemed to have managed. It was something that came partly by virtue of his character, partly because the world in which he developed was beginning to change. Within the medical sciences, humility was spreading.
On 25 April 1945, some German civilians brought in a young girl with a badly damaged arm. In the filth and darkness of a cellar, Cochrane cut it off. When he emerged, he found that the guards were all gone. He was free.
Soon the Soviets arrived. Finding him dressing the German girl’s stump of an arm, a Russian soldier informed Cochrane that he was to be shot for aiding the enemy. The threat was sincere, and the soldier meant to carry out his sentence immediately, but he allowed himself to be distracted by Cochrane’s questions about his medals. Vanity got the upper hand. The American soldiers, when they arrived, seemed to Cochrane almost as bad. They were less violent, but they stole his belongings and showed a ready interest in rape. Being liberated was not such a pleasant experience.
When he got home, Cochrane found that the British army was not quite done with him. For some months after the war finished he worked in a military hospital in the small cathedral city of St Albans, running the tuberculosis ward. Then, funded by the same Rockefeller Foundation that enabled Florey’s production of penicillin, Cochrane turned his back on clinical work and chose research instead. Prior to a period of training in Philadelphia, the Rockefeller Foundation sent him on a preparatory course at the London School of Hygiene and Tropical Medicine. There, from 1946 to 1947, Cochrane studied medical statistics under Austin Bradford Hill. The two men were to be instrumental in bringing to medicine an entirely novel degree of methodological rigour – changing the way doctors viewed the nature of evidence and having a more profound effect on human health than any newly discovered drug. One of the diseases on which their ideas most quickly had an effect was tuberculosis.
* * *
1 ‘I had one interesting meeting in a bar in Barcelona with a tall Englishman with big feet,’ he remembered, but found George Orwell a difficult man to like. ‘I later enjoyed his books more than I had that conversation.’ Hemingway, whom he met around the same time, struck him as ‘an alcoholic bore’.
16 Captain of the Men of Death
BY 1947, TUBERCULOSIS in Britain was only an eighth as common as it had been at the start of the century. It was not because treatments had improved, or the climate grown healthier. Better food and housing, as well as sanitation and hygiene, made people more resistant to Mycobacterium tuberculosis. Sulfonamides and penicillin, although they had no effect on the germ itself, helped by treating the added bacterial infections that TB sufferers were vulnerable to. In 1935, around 70,000 people died of tuberculosis in England and Wales. By 1947 it was only 55,000.
For centuries, ideas about testing had popped up here and there, only to have no apparent impact and to disappear for the span of an age. That, finally, was changing. The idea of using controls, and sample sizes big enough to overcome the flukes of chance, were taking root in the medical mind. By the time physicists were building nuclear weapons, doctors were conscious that reliable evidence only came from comparing groups that were significantly bigger than one person who took a drug and one who did not.
From Francis Galton onwards, progress formed a palpable chain of events. Not only did the methods for figuring medicines out improve rapidly, they did so in a sort of order. Francis Galton was idolised, and his work continued, by Karl Pearson. They met around the start of the twentieth century, when Pearson was forty and Galton eighty. It would be Galton, Pearson proclaimed, whom history would recognise as the greatest grandson of the eighteenth-century poet-physician Erasmus Darwin. The other competitor, Charles Darwin, was comparatively second-rate. The theory of evolution, to Pearson’s mind, was infinitely less useful to humanity than a grasp of statistical method.
Pearson’s work was broad, but the thrust of it was epistemological. How could people order their experience of the world in order to better understand it? How could they use numbers to help them think about the nature of things, explore the relationships between events, distinguish coincidence from causation? Pearson felt people were on the verge of new ways of understanding and dealing with the world. His radicalism was widespread. Beyond statistics and the theories of knowledge that went with it, he was passionately attached to other causes. Revolutionary socialism was one, and he combined it freely with a totalitarian view of eugenics and of racial differences, as well as a belief in the value of warfare for keeping the superior races pure.
Unsurprisingly, it was Pearson’s brilliance that attracted people to him, not his human warmth. The young Einstein read him and recommended him to others. Pearson’s impact was to turn statistics from an interest of gentleman amateurs into an academic discipline at the centre of scientific method. The chairs he held – Applied Mathematics and Mechanics at University College London, Geometry at the Royal Society’s home of Gresham College – were ones whose names did not quite suit him. That was because statistics was not the sort of thing academia had ever taken seriously. In Pearson’s wake, that changed.
An important ally was Major Greenwood, whose curious first name was not actually a title. Greenwood qualified as a doctor in 1904 at the age of twenty-four and, like Archie Cochrane, left clinical work behind as soon as he possibly could. He went to work instead with a physiologist in London, Leonard Hill. Hill’s wife disapproved. ‘I do hope’, she declared to her husband, ‘you are not going to appoint that cynical
little man to your staff.’ ‘The boy has brains,’ replied Leonard Hill. ‘He’ll never be any use as a doctor, and I must give him a chance in science.’ Greenwood was soon writing articles for Biometrika, Karl Pearson’s statistical journal, and went on to study with Pearson himself. The two men were comparable in their love of numbers, and perhaps also in the reasons why they found the world of mathematics attractive. ‘Some have thought that he selected statistics as his life’s work because that branch of medicine was most remote from the emotional,’ explained a close friend of Greenwood’s. ‘Perhaps they were right.’
In 1928, now forty-eight, Greenwood was appointed to the first professorship in Epidemiology and Vital Statistics at the London School of Hygiene and Tropical Medicine. Like his election to the Royal Society, his professorial chair acknowledged the influence that statisticians were beginning to exert. Greenwood, declared his Royal Society nomination, ‘has done much to encourage and develop the use of modern statistical methods by medical laboratory investigators, and, as Chairman of the Medical Research Council’s Statistical Committee, to secure the adequate planning and execution of field investigations’.
Medical tests were becoming team efforts, just as pharmacology had been for almost a century. That meant there was a pressure for people to think about how they were organised, and an opportunity for those at the centre to control how the humble workers in the field collected and ordered their experiences and experiments. ‘In the future,’ read one of Greenwood’s obituaries, ‘it may well indeed seem that one of his greatest contributions, if not the greatest, lay merely in his outlook, in his statistical approach to medicine, then a new approach and one long regarded with suspicion. And he fought this fight continuously and honestly – for logic, for accuracy, for “little sums”.’ The obituary also noted that Greenwood’s feeling for numbers was mixed with a clumsiness with people.
To some he may have seemed distant and unapproachable, to some cynical and censorious. He was indeed not a person whom it was easy to get to know well . . . He was certainly censorious of pomposity and pretentiousness, and could be a devastating critic of the illogical and the stupid. In the medical journals his letters, epigrammatic and satirical, were widely read with enjoyment or pain – according to one’s position.
The obituary writer was Austin Bradford Hill, the son of the physiologist Leonard. Born in 1897, at the close of the nineteenth century, Austin Bradford Hill grew up with the advantage of a father who was educated, thoughtful and attentive. Leonard Hill’s publications included innovative physiological works, the autobiographical Philosophy of a Biologist and, more usefully for Austin’s childhood, The Monkey Moo Book and other collections of fairytales. The family home was a good place for a growing boy to learn that a feeling for numbers could fit perfectly well with human warmth.
When the First World War began Hill joined up as a pilot. The Royal Naval Air Service was officially formed in 1914, although the navy (offended at the army’s control of the Royal Flying Corps, founded in 1912) had been operating it without permission for some time. Posted to Gallipoli in 1917, two years after the catastrophic British and Commonwealth battle to secure it, Hill travelled overland through France and Italy. By the time he reached the Aegean an attack of tuberculosis, possibly acquired en route, was so bad that he was sent straight home.
His tuberculosis was pulmonary, of the lungs, and that made it unpredictable. For a time it seemed likely to kill him. After two years in hospital, however, he emerged, debilitated but alive. Too weak to follow his father and obtain a medical degree, Hill spent a further two years convalescing before he became strong enough to look for work. At that point Major Greenwood stepped in, returning the support that Leonard Hill had once given him, and offered Austin a start in medical statistics. Working for the Medical Research Council, Hill studied statistics under Karl Pearson. Then, when Greenwood was made Professor of Epidemiology and Vital Statistics, he appointed Austin as Reader.
One of Hill’s duties was teaching statistics to medics. It gave him an appreciation of just how difficult it was for doctors to tolerate the subject, let alone understand it. During the 1930s, he drew his lectures together into a series of articles, publishing them in the Lancet. Later they were collected and released as a book: Principles of Medical Statistics.
‘Is the application of the numerical method to the subject-matter of medicine a trivial and time-wasting ingenuity as some hold, or is it an important stage in the development of our art, as others proclaim it?’ the Lancet asked in 1921. Opinion in the medical profession was inclined to hope that it was the former. There was something much more comfortable about the days in which a doctor’s sage opinions were considered sufficient, and the nuisance of numbers could be largely avoided. Hill’s book noted this reluctance and did a fair job of patiently sympathising with it. At the same time it pointed out that doctors could not conscionably give in to it. Their tendency was to regard statistics as ‘obscure and even repellent’. The obscurity had to be wiped away through hard work and study, while the repellent nature of statistics was something they had either to overcome or learn to put up with.
Statistics was now installing itself at the centre of medical research. That did not mean that the clinical community, mainly made up of doctors who finished their training before statistics bit into the curriculum, were already converted. Although medical papers were increasingly smattered with numbers, few of the researchers yet understood them. Controls and case series, percentages and other numerical measures were used as a form of dressing. Just as doctors once packed their writing with references to Galen and Hippocrates, now they did so with mathematical jargon. These meaningless mockeries of statistics added to the distrust doctors felt: many of the numbers filling medical journals and textbooks were patently nonsense.
Against these feelings, Hill argued that practical experience and wisdom were not enough to make sense of the way diseases behaved and treatments acted. ‘Are simple methods of the interpretation of figures only a synonym for common sense or do they involve an art or knowledge which can be imparted?’ he asked. ‘Familiarity with medical statistics’, he tempted doctors to believe, ‘leads inevitably to the conclusion that common sense is not enough.’ Hill was sensitive to frightening off any incipient interest. He was almost apologetic in his book, pointing out that what he had to say was so obvious it was barely worth saying – except, he added in the politest possible way, for the fact that doctors seemed unable to notice it for themselves.
Statistics, explained Hill, was a method for turning the world into a laboratory. A chemist at a bench could control all the factors affecting his experiment, altering a single one at a time and thereby learning its effects. Doctors were faced with situations in which an infinite number of influences were always colliding. Statistics was simply a way of teasing them apart. Given the wild and wilful nature of the world, analysing numbers was the only way to understand great expanses of it.
Take control groups, for example. ‘The essence of the problem in a simple experiment is . . . to ensure beforehand that, as far as is possible, the control and treated groups are the same in all relevant respects.’ Hill’s emphasis drew attention to the difficulty. No matter what your suspicions, you could never know for certain all the important factors that were going to predict your patient’s response to a therapy. You might carefully match your controls with respect to age or height or weight, but could you ever be sure that they were identical in every way that mattered? Even if you were, there was always the possibility of differences between them that were inconceivable to you, that were unforeseen. That meant your control group might always be crucially ill-matched. For pulmonary tuberculosis, it might be that the people whose lungs you collapsed were already fated, in some hidden way, to do better than those you left alone. Cause and effect got too easily muddled up.
This was a problem that Francis Galton had touched upon. Half seriously, but with his characteristic ingenuity, he had once tried to
apply statistical methods to investigating the power of prayer. The liturgy of the Church of England included prayers for the long life of the nobility. The higher up the aristocrat, the more churchgoers would be praying for their continued survival. At the top of the tree, prayed for by more people than any of those beneath them, were the kings and queens. Yet they lived, Galton noted, shorter lives than any of the other aristocrats – shorter even than army officers, artists, writers and scientists, people for whom virtually no one thought to pray. It seemed as though prayer had no effect whatsoever. The possibility remained, however, that aristocratic prestige went along with grave and ascending ill-health, and that without those masses of prayers the royal family and others might have lived even shorter lives. Just from the association of two observations, there was no way of working out for certain if one caused the other.
So Hill was aware that a statistical association was not the same as a demonstration of causality, and that no control group could ever be matched with perfect precision for every possible factor. Doctors objected that these problems meant the utility of statistics was a foolish illusion; expert opinion might be flawed, but it was less flawed than the alternatives.
Not true, said Hill. Differentiating causation from association was possible if you intervened rather than simply observed. Or at least it was possible if you could solve the problem of matching two groups of people with enough confidence to be sure that they were effectively identical. That is, if you could find two groups who were exactly the same as one another, in every imaginable and unimaginable way, then you could do something to one of them and not to the other. Any resulting difference between the two would then be definitely due to your intervention. But how could two people ever be found who were identical? Even in the womb, twins that share exactly the same genes already begin to have different experiences. No two people were ever precisely the same.
