Who was Philip Burch?
He is nowhere mentioned in Smoking Kills, the official biography of Sir Richard Doll. Yet of the select band of scientists who have doubted that cigarette smoking causes lung cancer, Philip Burch was among the most eminent. He was a Cambridge-educated professor of medical physics whose work on radiation and the human body earned him a personal chair: he built state-of-the-art radiation detectors and his papers on cancer were published in Nature. Around 1970, aged 50 and at the height of his powers, he investigated lung cancer in the light of his theories and concluded to his own surprise that smoking is a minor, possibly negligible, factor in its origins. In face of the silent treatment from a hostile medical establishment he published a long, technical book on cancer, ten or so major scientific papers on smoking and health, and numerous polemics against the low scientific standards of modern epidemiology. He maintained his heretical views until his death in 1987.
About this site
As a result of litigation in the 1990s, the cigarette companies were ordered to open their corporate archives and digitised versions were placed online. Since the industry avidly collected Burch’s papers, an unintended consequence is that most of his publications are available as free downloads from the Truth Tobacco Industry Documents. A substantial part of his scientific correspondence also found its way into the online archive, and those letters make it possible to tell a remarkable story of a remarkable man. This site consists of three main sections, A Half-life of Burch, The Essence of Burch and Sources, covering the story, the scientific issues and the primary documents. The Half-life is compiled entirely from publicly available sources and makes no pretence of being a complete biography – hence its title.
The inevitable questions
Burch was a lifelong non-smoker who said “I regard compulsive cigarette smoking as a disease”. He did not accept funding from the tobacco industry. He died of colorectal cancer, a non-smoking related disease, as he ironically noted. His surviving scientific correspondence shows him to have been sincere, scientifically rigorous, public-spirited and highly intelligent. Politically he was a classical liberal, with no religious beliefs so far as is known. He was also an English academic of the old school, donnishly understated and drily witty; a loving husband and father with many interests outside his work; and admirably free of the paranoia which too often affects those who know they are being deliberately ignored.
What is the Burch curve?
The Burch curve is the mathematical distillation of an unexpected pattern which Burch detected in statistics of lung cancer deaths broken down by age, covering many decades and countries. The uptake of cigarette smoking was supposedly to blame for a huge rise in premature death from lung cancer, and there was indeed an inexact correlation between the two over time (though less so from country to country). The problem was that the excess deaths were not premature. The peak of the curve, as fitted to the different data sets, rose to greatly differing heights, but scarcely wavered to left or right. Wherever you looked, lung cancer was predominantly what it always had been, a disease of late middle age. To Burch, this pattern was damning evidence which rendered Doll’s model of cigarette smoke as a carcinogen untenable. He identified various quantifiable tests of Doll’s theory and showed that it failed them.
Is there anything in it?
There are two sides to Burch’s work on smoking: a demolition of Richard Doll’s account of its role in lung cancer and Burch’s own theory of cancer. He could, clearly, have been right about Doll and wrong about cancer in general. It was never his aim to refute Doll. Rather, like a general wargaming his battles from the enemy perspective, he sought to test his own hypotheses by attempting to refute them. Burch came to cancer from physics, and regarded this falsificationist approach as the hallmark of hard science. Burch’s own view was that cancer is almost entirely a genetic disease and that carcinogens only affect its development at the margins. His ideas were inspired by the immune surveillance theory of the Australian immunologist and Nobel laureate Sir Macfarlane Burnet. They involve somatic mutation and the immune system, which are now believed to be important parts of the mechanism of cancer formation, though not in exactly the way Burch thought.
Many people know that the great statistician and geneticist R. A. Fisher took issue with the findings of Doll and Hill in the 1950s. He did not dispute that lung cancer is mainly a disease of smokers, but they had ignored an alternative explanation: it might be that genetic factors cause both the cancer and the smoking. Then, so most people think they know, new evidence, publicised in a thud, thud, thud of heavyweight reports, showed that Fisher was wrong and Doll and Hill were right. Burch was one of a minority who saw that this was not so. The new evidence was more of the same and the reports were an exercise in confirmation bias. Indeed, they relied explicitly on a novel scientific methodology, codified by Hill, which made confirmation, not falsification, the hallmark of proof and relegated contrary evidence to the status of ‘paradoxes’.
Medical, government and public opinion on smoking at that time largely reflected the views of one man, Dr Charles Fletcher, the principal author of the reports Smoking and Health (1962) and Smoking and Health Now (1971). When Burch’s book appeared early in 1976, Fletcher organised a seminar at which Burch and Richard Peto lectured and debated before an audience of senior medical specialists. That seminar, hostile reviews of the book and a discussion by letter with Burch persuaded Fletcher that Burch had been refuted and Doll vindicated. An editorial in the British Medical Journal informed its readers that the Burch affair was closed, the report Smoking or Health (1977) appeared as planned, and the official campaign against smoking continued unabated.
Nobody wanted to know. Burch’s claims were a threat to reputations, careers and the prestige of the medical profession, and if he was right, nothing could be done about cancer in the foreseeable future. The dangers of smoking were regarded as axiomatic and, after a brief period of notoriety, educated public opinion forgot about him. Burch was not for one moment deterred by the response he got, nor was he even particularly surprised by it. In plain language, he regarded Fletcher as a lightweight who did not know good science from bad, Doll as a pseudoscientist and Peto as a statistical charlatan. He continued to research and publish because for him the real issue transcended smoking. The corruption of public policy by bad science threatened scientific standards and freedom itself.
Fifteen crucial years
Burch’s career as a heretic spanned a crucial period in the campaign against smoking. In 1972, when he went public, the habit of smoking was still entrenched. Behind the scenes, however, campaigners had discovered the garlic which repels vampires, passive smoking, and by the early 1980s they were on the march. A new generation of anti-smoking activists was coming to prominence. They expanded the causes of lung cancer to include cigarette advertising, low taxation, corporate greed, cultural attitudes and renegades like Philip Burch. Also during these years, the first randomised prospective studies of smoking and health appeared, and the reception they got illuminates the new climate of opinion. Activists denounced them as poorly designed, which they were not, and tried to deprive them of publicity and funding. The reason was that the new studies failed to show a connection between smoking and disease and indicated that Fisher had been at least partly right.
A Half-life of Burch
The first 50 years of Burch’s life: the career of a highly regarded scientist with a distinguished record of publication at the boundary between physics and medicine.
Burch made a completely unexpected discovery and decided to pursue its heretical implications regardless of the professional consequences. With difficulty, he published in the academic journals and came to scientific attention.
In 1976, Burch’s book appeared and the general public became aware of him. After debates with Richard Doll, Richard Peto, Charles Fletcher and others the medical profession dismissed him.
Burch held that neither smoking, dietary fat nor other risk factors caused coronary heart disease. He published on the question in the years around 1980, gaining some support for his views.
The story then becomes the story of the anti-smoking movement. Burch continued to publish on smoking, and his criticisms of conventional epidemiology emphasised the need for randomisation and falsification, neglected by the school of Doll.
Burch was provoked by increasingly intrusive public health measures and much of his later work consists of polemics against the soft science they were based on. He sympathised with the tobacco industry’s attempts to defend itself, though he was careful not to accept tobacco funding.
From 1980, a flood of studies attempted to prove the dangers of passive smoking and events in 1986 made it orthodoxy. Burch saw the development as the next step in the advance of politicised pseudoscience and renewed his attacks on current epidemiology.
8. Last words
Burch died in middle age without recanting his views. He had made virtually no impact on cancer research, medical opinion and health policy – or had he?
The Essence of Burch
Burch is not easy reading. He wrote traditional academic English, and wrote it well, but he assumed an audience of his equals and was often subject to constraints of space. Above all, he took a physicist’s command of mathematics for granted, particularly the all-important exponential function and its relationship to powers and logarithms. This is not advanced mathematics – it is pre-calculus, it is in the high school curriculum – and a man of Burch’s calibre would not have thought to expound it in peer-reviewed publications. However, few people could grasp the implications of his formulas and diagrams as immediately as he could. This section of the site seeks to give his scientific reasoning with explanations of the steps he found too elementary to spell out.
Burch’s general theory of disease as applied to cancer, and the importance of the age at which it strikes.
Burch’s explanation of lung cancer as a largely genetic disease.
Burch’s work on coronary heart disease.
Burch’s arguments against the Doll-Hill theory of smoking as a cause of lung cancer.
Some background material.
Doll’s rise to fame as a protégé of Bradford Hill, his involvement with communism, his conflict with R. A. Fisher and his relationship with his statistical associates.
The great evolutionary biologist William Hamilton was a not-quite smoking sceptic.
Burch’s main sources of data and how to reproduce his results in the statistics programming language R.
A listing of the material used in this site with brief summaries. Not a complete bibliography of Burch but contains everything of relevance and also covers most of the literature of smoking scepticism.
Burch in Controversy
Some parts of the story are heavy on mathematics, for instance the long-heralded encounter with Richard Peto which persuaded the medical profession that Burch had been answered and convinced Burch that Doll and his associates were pseudoscientists. Short accounts of these episodes are included in the Half-life, and longer versions discussing the mathematics are designed to stand on their own.
Did Richard Peto win a decisive debate with Burch by statistical trickery?
Did Julian Peto demonstrate that 90% of lung cancer is caused by smoking, or 10%, or both?
How stringent were Dr Charles Fletcher’s scientific standards?
An audience of professional statisticians gave Burch a surprisingly sympathetic hearing.
Did Doll and Peto twice modify their dose-response relationship to suit the occasion?
Two views on the scientific logic of the smoking issue.
Did the health economist Marvin Kristein know what he was talking about?
Does active smoking cancel out the effects of passive smoking?
Did Repace and Lowrey have good reason to revise low estimates of the effects of passive smoking upwards?