Doll in context

Appendix A

Sir Richard Doll in context

Who was Richard Doll?

In 1927, aged 15, he was an outstanding schoolboy at Westminster, wavering between evangelical Christianity and the secular gospel of communism. In 1937, recently qualified from Guys Hospital, he was a young physician still seeking a role and a livelihood while the clouds of war gathered. In 1947, home from service in the Mediterranean theatre, he had found his niche in epidemiology under Dr Francis Avery Jones at the Central Middlesex Hospital and Austin Bradford Hill at the London School of Hygiene and Tropical Medicine, working on a study of peptic ulcer. The next ten years were the making of him. As of 1957 he was Richard Doll, the great scientist who discovered the cause of lung cancer. A lifetime of honours ensued: a fellowship of the Royal Society in 1966, the Regius Chair of Medicine at Oxford in 1969, a knighthood in 1971 and the wardenship of newly founded Green College, Oxford in 1979. He died in 2005.

Doll’s annus mirabilis

1957 was the year in which the United Kingdom tested its first hydrogen bomb. It was also the year in which Doll became Richard-Doll-the-man-who-proved-in-1956-that-smoking-causes-lung-cancer. On the strength of five years of his and Hill’s study of British doctors, the Medical Research Council, the Department of Health and the British Medical Journal issued announcements that the case was closed. Though the eminent statistician Sir Ronald Fisher dissented in the very next issue of the BMJ, medical opinion was increasingly convinced. Important figures, notably Drs Charles Fletcher and George Godber, threw their weight behind him and began to plan a campaign of action against smoking. Doll’s star was rising. He published the second of a series of papers with his colleague Peter Armitage on carcinogenic mechanisms and became a fellow of the Royal College of Physicians, the academy of his profession. And it was in 1957 that Doll resigned from the Communist Party.

Doll and communism

Doll joined the Communist Party of Great Britain in reaction to the 1920s depression and left it in response to Lysenkoism and the Soviet invasion of Hungary. These were respectable reasons, though Fisher had got the measure of Lysenko years before: “The reward he is so eagerly grasping is Power, power for himself, power to threaten, power to torture, power to kill.” Had Doll come to see that the revolution would not leave people like him in charge? Smoking was never an issue for the communists, and the main significance of the Party in his life was social networking, since it was how he met Richard and Julian Peto (through their communist parents), as well as the senior medic at the Central Middlesex, Horace Joules, and (together with her first husband) his wife Joan. There can also be no doubt that, had he stayed in the party, he would never have been awarded a chair at Oxford or a knighthood.

Allies in high places

In the mid-1950s, the conservative pre-war leadership of the medical profession was giving way to a younger generation. Typical of the new men were Charles Fletcher and George Godber, both fellows of the Royal College of Physicians and both socialists who had campaigned for a National Health Service of the kind established in 1947. Fletcher was a prominent chest specialist at Hammersmith Hospital, well-known for his television documentaries on medicine. His work on the miners’ disease pneumoconiosis in south Wales made him a champion of preventive medicine, and Doll convinced him as early as 1950 that smoking causes cancer. The RCP reports were his brainchild.

He had a powerful ally in George Godber, then Deputy Chief and, from 1960, Chief Medical Officer to the government. During Godber’s tenure, advertising restrictions were imposed, health warnings introduced and the Health Education Council created. Later in his career he formulated the anti-smoking policies of the World Health Organisation. Another significant figure was Geoffrey Todd, an employee of Imperial Tobacco and a statistician at the Tobacco Research Council who later became its director. After initial scepticism, he came to agree with Doll and Hill around 1957 and under his influence the council conducted research into smoking and health advised by Doll and, later, Richard Peto.

Transatlantic developments

A parallel series of events unfolded in the United States. The first post-war study of smoking and lung cancer was published by Ernst Wynder and Evarts Graham in the Journal of the American Medical Association in 1950. There followed the American Cancer Society study by E. Cuyler Hammond and Daniel Horn. On the strength of early results, Hammond, in an address to the American Medical Association in 1954, went so far as to blame two thirds of all deaths on smoking. In 1961, the ACS and three other medical charities called for a commission on smoking and health: President Kennedy appointed one, and the first Report of the Surgeon General was issued in 1965. (A searching critique of it by the statistician K. A. Brownlee fell on deaf ears.) The then Surgeon General was Luther Terry but further reports, most of them merely updates, appeared annually, even when the office was vacant. The most important are the 1965 and 1979 reports and the neglected Smoking and Health Characteristics (a sort of minority report from 1965). Unlike Britain, the USA was a tobacco-producing nation, and the industry refused to accept the alleged dangers of smoking. The major cigarette manufacturers founded the Tobacco Institute in 1958 to promote their point of view and channelled funding to smoking sceptics including Carl Seltzer.

Breach with Fisher

Fisher continued to contest Doll and Hill’s findings. One remarkable feature of the 1950 Hospitals study was that smokers who inhaled had a lower rate of lung cancer than those who did not, and Fisher wrote to Hill requesting the full dataset. This was held on worn-out punch cards and Hill tried offering Fisher the 1952 data instead. After some pointed remarks by Fisher and a testy exchange of letters, Hill came up with the 1950 data, but relations between the two men, who had once been on friendly terms, deteriorated. When both of them addressed a Cambridge symposium on the issue, Fisher said his piece and then walked out before Hill spoke. On another occasion, he publicly stated that Hill should never have been made a fellow of the Royal Society. Also around this time, Fisher assembled data on twins suggestive of a genetic influence on smoking, which he included in his book Smoking: the Cancer Controversy (1959). He then retired to Australia, where he died in 1962.

The apostolic succession

Doll liked to claim that he stood in an “apostolic succession” from the pioneer of medical statistics in Britain, Karl Pearson: Pearson, Greenwood, Hill, Doll, Peto. The first three of these names occupied chairs at neighbouring colleges of the University of London.

The Galton chair of Eugenics The chair of Epidemiology
University College, London London School of Hygiene and Tropical Medicine
1912-1933 Karl Pearson 1928-1947 Major Greenwood
1934-1943 Ronald Fisher 1947-1961 Austin Bradford Hill
1945-1965 Lionel Penrose 1961-1976 Peter Armitage

The Galton chair of Eugenics was founded and endowed by that fascinating man Francis Galton, among other things a eugenist and a cousin of Charles Darwin. Karl Pearson, his personal choice for professor, was also a eugenist, a Germanophile and the leading British statistician of his day. Pearson and his successor kept Galton’s memory green, but under Penrose, in the altered climate of opinion after the defeat of Nazism, the chair became the chair of Genetics.

Greenwood (given name Major) and Hill (given names Austin Bradford) were less important figures intellectually. Both were medical statisticians whose life work was the practical application to their chosen field of theoretical advances which others had made. Hill was an admirable man and generally liked. He was a former World War I air ace who had been debarred by serious illness from the medical career he had hoped for. Greenwood, a family friend, steered him into epidemiology and in due course the heir apparent succeeded to the throne.

Doll and Armitage were, in turn, protégés of Hill. Their first collaboration was the peptic ulcer study, Armitage being the statistician while Doll collected data under the ulcer specialist Avery Jones (given name Francis). This division of labour continued into the investigation of lung cancer. Doll, directed by Hill, amassed data at two London hospitals, then four more in the provinces. In the early years of the Doctors’ Study, Doll processed the data while Armitage, with Doll as second author, published papers on the statistics of cancer.

When Hill retired, Armitage was the obvious choice to succeed him. Doll now needed a new statistician and Malcolm Pike was appointed in 1963 (tellingly, it was Armitage, not Doll, who interviewed him). When Doll accepted the Regius chair he brought many of his London staff, including Pike and Richard Peto, with him to Oxford, and when Pike moved on in 1972 the role was given to Peto.

Chi-square

For obvious reasons Fisher is not in the apostolic succession, yet he was the actual successor of Pearson both as Galton professor and in the history of science. In 1912 statisticians used the methods of Pearson and in 1933 they used those of Fisher. Fisher had remade the foundations of statistics, using multi-dimensional geometry, in a series of brilliant papers. The first, concerning the t-distribution of William (‘Student’) Gosset, made it possible to use far smaller samples than those required by Pearson’s system. The last one concerned the chi-square distribution.

When Doll divided his sample of London hospital patients two ways, into those with lung cancer and those without and into those who smoked and those who did not, he proceeded to perform a two-by-two chi-square test with one degree of freedom. According to his biographer, “Fisher’s book, Statistical Methods for Research Workers, led Doll to Pearson’s X2 (Chi2) test.” He had first read it in 1937: “The book was extremely complex and the only part Doll could understand was the use of the Chi-squared test.”

The greatest statistician of the twentieth century

Fisher succeeded Pearson because he had gone beyond him. In particular, it was Fisher who showed that a two-by-two chi-square test should be performed with one degree of freedom (Pearson specified three). Pearson never really accepted this, but his colleagues, including Greenwood and Pearson’s own son Egon, also a statistician of repute at UCL, knew that Fisher was right. Hill’s textbook Principles of Medical Statistics (1937), originally a series of articles in the Lancet and essentially a popularisation of Fisher’s system, gives Fisher’s version of the chi-square test, in simplified language which Doll admired.

Hill was a good man with scientific achievements, such as the blind testing of drugs, to his name, and the rift with Fisher hurt and puzzled him. Doll was less forgiving (at one point in the dispute over the Hospitals Study, he considered suing Fisher) and consistently sought to minimise his importance. Doll’s claim that Hill was “the greatest statistician of the twentieth century” is not to be taken seriously. Fisher created the tools that Hill used, and Doll relied on the statistical work of others. Neither was in the same league as him, and they not only did not refute Fisher’s views on smoking, they did not so much as try.