A Half-Life of Burch
2. Emergence of a Heretic
Burch at 50
In 1970, Burch was awarded a personal chair (a professorship created for him personally). He was 50, at the height of his powers, and had what he wanted out of life: fulfilling work in a place he loved, international recognition, a supportive wife and colleagues and three bright children. The only further advancement in his profession would be a fellowship of the Royal Society awarded for some major achievement – such as Sir Richard Doll’s discovery that smoking causes lung cancer. A demonstration that Doll had simply failed to make his case could do Burch no good at all, and he knew it. He could simply have omitted lung cancer from his projected book The Biology of Cancer, pleading the complexity of the issue and the need for more research. At this time, researchers into intelligence were learning to tiptoe round the question of racial differences, leaving informed readers to draw their own conclusions from silence. Unlike William Shockley, another physicist who saw further into a soft science than many of its practitioners, Burch was not a man who sought headlines and issued angry challenges. But, “I felt, contrary to my original intentions, that I could not, really avoid discussing lung cancer.” He went public slowly, accepting the prospect of a long struggle to publish his heresy in a scientific journal.
Burch delivered his inaugural lecture as a professor on 17 January 1972: it was later published as Can We Have Absolutely True Scientific Laws? An inaugural lecture is supposed to set out the new professor’s vision for his subject, and Burch ranged widely over physics and biology with occasional provocative asides. “Was Shakespeare an evolutionist? Did he anticipate Darwin,” he mused after quoting Hamlet.
Burch began by declaring allegiance to Popper’s principle of falsification, according to which, scientific laws are always subject to revision in the light of new observations (and Heisenberg had shown that observations are inherently imprecise). Even the laws of physics are not quite absolutely true. He continued:
I wish now to turn to biology, a sector of science that, because of its paucity of mathematically formulated laws, is customarily regarded as being somewhat imprecise and second grade. A subject for girls but not worthy of the attention of scientifically ambitious boys.
Yet the amazing perfection of living organisms, the structure of the body, of the cell and of the central nervous system went to indicate that there must exist exact laws in biology. Burch pointed to the age distribution of disease and its strongly mathematical character. A quantitative study of gout in men and women enabled him to state precisely an observation first made by Hippocrates.
Then Burch presented his theory of autoaggressive disease and announced his recent extension of it to cancer, illustrated with an analysis of data on ovarian cancer. He proceeded to expound at length the evolutionary considerations informing his speculations about a biochemical feedback loop between a central growth control system and in target tissues. He stressed the Popperian falsifiablity of this hypothesis and finished off with a bold challenge.
If we ever discover little green men on Mars, I am prepared to wager a very large sum on the prediction that their central and cognate target recognition macromolecules will contain identical sequences. Any takers?
The Lancet correspondence
Burch’s first published work on smoking was a contribution to a controversy in the correspondence columns of the Lancet. The episode began with a paper by the smoking sceptic Carl Seltzer, Critical Appraisal of the Royal College of Physicians’ Report on Smoking and Health. The 1971 report claimed to show the benefits of giving up smoking by a comparison of British male doctors with the general male population of England and Wales. Seltzer showed that it made extremely selective use of the Doctors’ Study and that a fuller data set including the “missing time period,” the “missing ages” and the “missing geography” showed no clear cut trend.
Inevitably, the same issue of the Lancet disavowed Seltzer in an editorial, arguing that the selective citation of data to enhance the case against smoking is good practice. To rub the point home, the next week’s issue and the one after that published letters from Sir Richard Doll and Dr Charles Fletcher, the originator and main author of Smoking and Health (1962) and Smoking and Health Now (1971). Both disputed minor points and asserted that Seltzer had “already been answered”. Doll finished “I have reached many silly conclusions in my life, though none (I think) in the company of Sir Austin Bradford Hill and none (I hope) as silly as Dr Seltzer implies.” Among Doll’s hallmarks were the false modesty of a man who regards himself as great and the burnishing of Hill’s reputation the better to enhance his own.
Seltzer replied in March, reinforcing his main points with new data and calculations. In April and June, two smoking sceptics weighed in to support him. One was Theodore “Ted” Sterling, whose views were well known. The other was Philip Burch. “Sir, Are we to judge from their silence that Dr. Seltzer’s critics have conceded his case? Is cigarette-smoking non-lethal?” (He knew perfectly well that Doll’s idea of debate was a single, apparently crushing, response followed by lofty silence.) Burch observed that R. A. Fisher’s critique of Doll and Hill remained valid, and that recent results from a study of Swedish twins seemed to bear it out.
The Tobacco Research Council and Burch
Burch’s declaration that he had joined the ranks of the smoking sceptics was not headline news, but it aroused interest in various quarters including the Tobacco Research Council. This organisation, originally an industry body, now performed anti-smoking research coordinated with Doll and his associates at industry expense.
Burch’s first letter to the Lancet, on 10 June 1972, called for a world-wide study of twins: “And why should the cigarette manufacturers not foot the bill?” The director of the TRC, Geoffrey Todd, immediately sought a meeting with Burch, and travelled to Leeds with his colleague Francis Roe for discussions on 30 June. The two men both wrote internal reports for the TRC. Roe was uncomprehending and hostile, wrongly dismissing Burch as an “armchair scientist”. Todd was fairer but there was much he did not understand.
The course of the discussions can be reconstructed from the reports: they began by talking about twin studies.
Burch: I am surprised that the cigarette manufacturers have not done any work with twins.
Todd: On the contrary, it is one of our oldest lines of research. We have supported work in Germany and Italy, and we have a small twin panel of our own.
Burch: Why not get general practitioners to recruit a larger panel?
Todd: That seems like a good suggestion.
Burch: On Fisher’s hypothesis, monozygotic and dizygotic twins should have differing patterns of lung cancer.
Todd: I thought that Fisher expected smokers and non-smokers to have different rates of lung cancer? And why do you get monozygotic twin pairs where one twin is a smoker and the other not?
Burch: In such cases the smoker is probably a social smoker. The prediction is not that monozygotic twins have different rates of lung cancer from dizygotics, but that monozygotics have the same rate as each other regardless of smoking.
Burch then expounded his theory of disease, which will have accounted for most of the time spent, and Todd took a note of his stochastic formula. Roe, in his report, admitted to not understanding it.
Roe: Have you any evidence for this central control system?
Burch: Liver regeneration requires integrity of the bone marrow, and kidney regeneration in mice requires integrity of the spleen.
Roe: How do you explain skin carcinogenesis by actinic radiation?
Roe reported that Burch had ‘no understandable explanation’.
Roe: Don’t migrants have different rates of cancer from rates in their countries of origin?
Burch: Migrants are self-selected.
Roe: What about lower rates of lung cancer in ex-smokers?
Burch: Selection again. “Ex-smokers are smokers by mistake.”
Todd: Will you be publishing your work on lung cancer?
Burch: It would require a long paper and it is so heretical that with one possible exception I don’t see any journal accepting it.
It was agreed that Burch would draw up a statement of his theory for consideration by the TRC: this is The Analysis of Cancer Statistics. He also sent them a draft journal paper, probably Lung Cancer and Cigarette Smoking: Secular Trends.
Report by Sir Charles Ellis
Todd sought a second opinion from the eminent physicist Sir Charles Ellis FRS, at that time a consultant to BAT (British American Tobacco). Ellis, whose experiments at the Cavendish led to the discovery of the neutrino, was excellently qualified to appraise a man of Burch’s background, but when sent some material to read he replied (13 July) that he had not been able to understand it. He added, however, that a man of Burch’s distinction could not be the crank Roe took him for, and in failing to see what Burch meant “I felt that it was I who was at fault, not him”.
The Lancet correspondence and new friendships
Burch’s contribution to the Lancet was merely the start of a long, involved and often tedious correspondence on smoking which continued for the next year and more. Its significance is that Burch used his contributions to get his main arguments in print, and that they were the only statement of his views that educated opinion had to go on as he became nationally known. Fletcher and Doll again rebuked Seltzer, Burch and Sterling before dropping out. Twins, identical and otherwise, then became the topic for the rest of 1972.
Over this period, Burch entered into personal correspondence with several participants. He tried to interest the Swedish twin researchers Lars Friberg and Rune Cederlöf in his ideas, exchanging several letters and a draft paper with Friberg. Richard Hickey, an authority on environmental carcinogens, became an on-off correspondent of Burch’s for many years. Hickey was disheartened by the fact that the sceptics seemed to make no headway against what he saw as the obviously inferior science of their opponents, and he provoked some of Burch’s most interesting thoughts on the matter. But it was Carl Seltzer, whose Critical Appraisal had sparked Burch off, who swiftly became a lifelong correspondent and personal friend. He and his wife Ruth were welcome guests at the Burchs’ house in Leeds and Burch’s letters to Seltzer often contain delightful glimpses of his family and travels. Many of them give the inside story of Burch’s publications and reveal his unvarnished opinions of the anti-smoking establishment: they are quoted at relevant points in the story.
Heart disease in the New England Journal of Medicine
Seltzer’s specialism was heart disease, and he continued to publish on the topic. His letter Smoking and Coronary Heart Disease in the New England Journal of Medicine (31 May 1973) attacked the report Smoking and Health Now from a new angle: if British doctors had reduced their rate of lung cancer between 1953 and 1965 by giving up smoking, why had their rates of heart disease risen? The medical establishment weighed in again. Nicholas Wald (an Oxford subordinate of Richard Doll) and others explained again that the case against smoking is more clear cut if you only look at the right selection of the evidence. George F. Grannis accused Seltzer of “inappropriate statistical manipulation” and Seltzer defended himself.
This was the context of a brief journal letter by Burch to say that Wald’s logic was wrong and Seltzer’s mathematics was right.
The Statistical Subcommittee
The Tobacco Research Council was still deciding what to do about Burch. It had a Statistical Subcommittee which met every few months and Burch was discussed at the next four of its meetings.
On 18 July 1972 Todd and Roe reported on their trip to Leeds, and it was agreed to wait for Burch’s statement and Ellis’s evaluation. Peter Lee, the Council’s professional statistician, was directed to study Burch’s existing work. By 11 October, Burch had submitted two papers to the Committee and it was agreed to defer consideration of them. On 18 January 1973 it was minuted that Peter Lee had analysed Burch’s papers and it was agreed that he would write up his note on them as a report. On 13 March, comments were invited on Burch’s papers and Lee’s report and it was decided to put the matter into abeyance.
It is not difficult to read between the dry lines. Only Lee came close to understanding Burch and he was assigned an unofficial brief as the TRC’s Burch-watcher.
On 9 March 1973, Burch wrote to Todd enquiring what progress the Council had made, and also enquiring about smoking rates among Native Americans, who appeared to have one third the lung cancer rates of whites. “Don’t American Indians smoke heavily? Or are we misled by all those peace pipes?”
The Lancet correspondence turns to lung cancer
The Lancet correspondence came to life again in the spring of 1973 when J. N. Mehrishi of Cambridge University wrote proposing the formation of a group to study mechanisms whereby cigarette smoke causes cancer. The response cannot have been what he expected. In a letter of 24 April, Burch claimed to have disconfirmed the existence of such mechanisms. The 1972 correspondence had concerned smoking and health in general but this was Burch’s first public challenge to the dogma that smoking causes lung cancer. On the strength of it, he wrote to Peter Lee at the Tobacco Research Council in slightly jaunty tones, seeking Lee’s reaction.
The editors of the Lancet were taking a risk publishing Burch and were remarkably generous with space, allocating the best part of a page to scientific heresy. On Burch’s part it was a masterpiece of concision to reduce a complex argument to 1400 words and two diagrams. Every sentence is pared to the bone, at the price of making no concessions to the reader. Few will have grasped the meaning of a bald statement such as “k is a kinetic constant proportional to the nth power of the average rate of gene mutation”. Lee was closer to Burch’s intellectual level than most people, and was in possession of the material supplied to the TRC the previous year, but even he initially misunderstood important points of the Lancet letter.
Burch’s theory of lung cancer is set out in the appendices and need not be repeated here.
Richard Peto and John Mathews took issue with Burch on 19 May, contrasting his method of vertical analysis unfavourably with their preferred method of cohort analysis. Professor Alwyn Smith (23 June) and T. J. Cole (7 July) discussed the crucial age distribution of lung cancer in some detail and Burch replied (23 June, 14 July) expounding his reasoning in more detail. The Lee correspondence proceeded in tandem with this. After initial misunderstandings on both sides, Lee came to see what Burch’s argument was, but was never persuaded. He had ideas of his own about cancer and about statistical method and the two men were largely talking past each other.
Lee was also exchanging letters with Richard Peto during this period. At the time the two were on friendly terms, though it was Peto who issued the instructions (“P.S. Don’t forget the pocket chess set!”). Sent Lee’s Burch material, Peto responded with a studied lack of enthusiasm. To say the least, he was far less categorical in private than in public: “John Mathews wants to reply to Burch but I don’t; I think that the data as they stand will not prove or disprove any hypothesis, and that most of the time spent on them is therefore time wasted.”
The Lancet correspondence more or less ended in July 1973, although a brief exchange about stress as a possible cause of lung cancer in 1974 should be mentioned for completeness.
Sir Richard Doll in search of funding
Sir Richard Doll took no further part in the Lancet correspondence, but he continued to monitor Burch, mentioning him in a letter to Geoffrey Todd (16 July 1973) on a more important subject, money.
Some time ago I asked if there might be a chance of approaching the tobacco industry for an endowment of an academic position in Oxford. When I wrote you thought that the time was not opportune, and I am wondering whether it might be now…
I should very much like to obtain an endowment in medical statistics which might make a suitable post for Richard Peto. At the present moment he has a lectureship in my department, but the appointment is only for three years at a time, and the maximum salary is much less than he deserves. University policy is, I am afraid, to invest money with a view to its long term value, so that a readership costs something between £100,000 and £120,000…
Yours, as ever
Presumably the time was still not opportune, as Richard Peto was appointed Imperial Cancer Research Fund reader in cancer studies in 1974.
In a postscript, Doll mused on Burch and his wrongheadedness.
P.S. I was delighted to hear that you were hoping to publish some estimates for the distribution of smokers by amount at different ages at different periods, and that Peter Lee and Richard Peto had been using these to make some more precise calculations than our friend Burch. It fascinates me that he can write so much to demonstrate that changes in diagnostic standards and in amount smoked cannot individually account for the increase of lung cancer without ever apparently asking himself whether the increase can be accounted for by a combination of better diagnosis and increase of smoking.
Rejection by Nature
Lung cancer and cigarette smoking: secular trends was submitted to Nature in late 1972 and rejected in March 1973. “You will not be surprised to learn that Nature turned down my paper on lung cancer and smoking,” he wrote to Geoffrey Todd. “Anti-establishment views cannot easily get an airing.” But Burch was a patient and tenacious man and proceeded to publish papers in journals lower down in the scientific hierarchy.
Yerushalmy, Goldstein and smoking in pregnancy
During this period, Burch contributed to overlapping correspondences in two journals concerning the relationship between low birth weight in babies and the smoking rates of their mothers. A lot of context is required to make sense of a few short letters by Burch, but the affair is interesting firstly as a case study in the manufacture of public opinion and secondly because of the important, neglected work of Jacob Yerushalmy.
By the mid-1960s a number of writers including Fisher, Neyman, Berkson and Brownlee had dissected the poor statistical reasoning underlying the campaign against smoking and pointed out that a third factor might bring about an apparent causal relationship between two others. Yerushalmy, in his studies of low birth weight, was the first investigator to identify a case where this seemed to be so. His results suggested that a particular kind of woman tended both to smoke and to have underweight babies. Yerushalmy’s papers (of which three are relevant here) are elegant, displaying a clarity of language which reflects clarity of thought. He set up hypotheses like bowling pins only to demolish them with well-aimed evidence, and Burch regarded him as a model of how to conduct epidemiology.
In 1972, Mr Harvey Goldstein, a British anti-smoking researcher who specialised in this area, took issue with Yerushalmy’s results in two different journals. Yerushalmy answered him with arguments which satisfied not only sceptics such as Burch, but causationists such as Geoffrey Todd. Goldstein continued to publish studies of his own in which he spoke as if Yerushalmy had been refuted, and in a 1973 interview which was picked up by the national press, spoke as if the causal interpretation was beyond doubt. An editorial in the journal Nature cried foul and a correspondence ensued involving both Burch and Goldstein, in which Goldstein conceded that causation had by no mean been proved. However, in a paper published later that year, he carried on as before, and he was one of the moving spirits behind subsequent government campaigns in the media to persuade pregnant women to give up smoking. (These campaigns focused on premature birth, which is not strictly the same thing as low birth weight, although the two are often conflated.)
It is instructive to observe how defeat in controversy on the matter with Yerushalmy in a specialist publication softens into agnosticism in the general science publication Nature, becomes a probability in the British Medical Journal, and is presented to the general public as a certainty in a magazine interview and second-hand reports in the daily newspapers.
Yerushalmy died during the course of this controversy, but Burch returned to the issue of smoking and pregnancy from time to time in the years that followed.
Yerushalmy and self-selection
Yerushalmy was an important influence on Burch, particularly because of his work on the problem of self-selection. Everyone knows that you cannot predict the outcome of a general election by asking your friends how they will vote. They will be far too similar to you with respect to age, class, race and so on to be a representative sample of voters. In the early days of opinion polls, news magazines used to forecast elections by polling their readers, and did so very inaccurately for similar reasons. The readers were a sample of the population, but nothing like a random one: by deciding to buy a favourite magazine, they were selecting themselves for untypical reasons.
Yerushalmy saw that the same problem arose in epidemiological studies. Those who decided to smoke, or give up smoking (or drinking, or eating a certain diet) were self-selected, and therefore not an adequate sample of the general population. Burch identified this as a huge design flaw in all the major studies of smoking and health, and it was one of his main aims to find solutions to it.
Yerushalmy was also an early critic of Dr Ancel Keys and his Six Countries (later Seven Countries) study of dietary fat as a factor in coronary heart disease. Yerushalmy examined evidence from all countries for which it was available, and showed that an apparently impressive correlation of fat consumption with deaths from heart disease only obtained in the six countries included in Keys’s study (six other countries showed no correlation at all). Keys had, consciously or unconsciously, selected the data confirming his hypothesis and ignored that which falsified it.
The growing strength of the anti-smoking campaign was reflected by the introduction of health warnings on cigarette packets in 1971. By 1973 there was a new demand to change the wording from “Warning by H.M. Government: Smoking Can Damage Your Health” to “Cigarettes are our biggest killers”. A letter from Burch to the Lancet protested that the work of Friberg, Yerushalmy and Burch himself showed that no such thing had been proved. However, Swedish twin studies demonstrated that smoking can “help to cause” bronchitis and he considered the warning “Smoking may damage your health” to be “subtly appropriate”.
At the Eugenics Society
At the end of September 1973, Burch went to London to deliver a talk to the Eugenics Society. Was Burch a eugenist? As a brilliant man who had married into a Cambridge dynasty and fathered three promising children perhaps he should have been, but it seems that he was not. Membership of the society was a matter of public knowledge and there exists a book listing every known member. It is an illustrious roster, headed by the name of John Maynard Keynes, but Burch is not in it. Paradoxically, from today’s point of view, the eugenics movement, with its origins in schemes to improve the teeming industrial slums, may have been too optimistic for him. The message of his talk was that society had no right to expect that the medical advances of the past 100 years could be matched in the next.
Toward perfect man
Dr Roger Lewin reported on the meeting for New Scientist. An initial summary captures the prevailing mood. “Last week in London the Eugenics Society tried to grapple with the implications of ‘the new biology’. In spite of the contemporary nature of its topic the meeting still emanated the unmistakable 1930s ethos of the perfectibility of man.” Indeed. “Just what can be achieved in terms of individual happiness and social economics by promoting free access to contraceptives and abortion facilities has been demonstrated by the Aberdeen experiment,” exulted one speaker. “Those antisocial people who ignore a plea to reduce family size will breed preferentially, thus increasing the pool of antisocial genes,” lamented another. But Lewin devoted a lot of space to Burch’s talk, which is important as a clear, accessible exposition of his general theory of disease and the only one he wrote for a lay audience. He wrote to Seltzer: “Much to my surprise, the New Scientist has given me a disproportionate share of space. Editorial policy generally favours the anti-smoking lobby. Although Lewin hasn’t reported very accurately, he has got the general drift and has not attempted to suppress the inconvenient evidence.”
Burch and Doll: the New Scientist exchange
Possibly as a result of Lewin’s report, in February 1974, Burch was given the opportunity to present his work on lung cancer to a wider audience in the pages of New Scientist, then as now an important source of science news for the educated British public. His four page article Does smoking cause lung cancer? appeared on 25 February, inevitably accompanied by an editorial stressing that Burch was expressing personal opinion, and balanced by a restatement of orthodoxy by Sir Richard Doll entitled Smoking, lung cancer and Occam’s razor.
Burch’s article is a good place to start in understanding his arguments. It places lung cancer in the context of his general theory of disease, explains why the Burch curve is evidence that the cause is not smoking, emphasises the problems of self-selection and misdiagnosis, and finishes by highlighting the central place of anomalies such as the inhalation ‘paradox’ in scientific reasoning as championed by Fisher.
Doll’s reply disputed five points which he regarded as crucial to Burch’s argument, his theme being that “arguments” must be weighed in the balance against “a mass of evidence” so large that it creates a presumption that correlation is causation. A second article by Burch later that month discussed these five points, observing that Doll had acknowledged a role for misdiagnosis for the first time, and that previously unpublished data which Doll had included in his article did not prove what he wanted it to.
The editorial, significantly entitled Propaganda and Integrity, stressed the need to balance health education against open scientific debate. It alluded to the recent controversy in Nature about Goldstein on smoking in pregnancy and the double standards under which seemed the anti-smoking lobby seemed to operate.
An American correspondent of Burch’s felt that the editorial was a sign of weakness in face of “the propagandist party line”. Burch wrote in response:
I hope you will forgive me if I defend Dr. Bernard Dixon, the editor of New Scientist against your (marginal) criticism. The climate of opinion in this country is not conducive, at the moment, to rational discussion of the supposed health hazards of smoking. Several highly emotional campaigns are being mounted by anti-smoking groups and supported by influential medical scientists. The last thing the sponsors want is critical comment from a reputable magazine of fairly wide circulation. Very few editors (excepting those of the specialized journals Nature and the Lancet) would have the courage to publish my uncomfortable heresies. It is only fair to point out that Bernard Dixon’s tenure is much less secure than mine. I thought his Editorial presented a remarkably fair and balanced discussion of the moral and scientific issues.
You may be interested to hear that Dixon invited Sir Richard Doll to answer my reply to him, but the Regius Professor declined the offer. He was not altogether pleased, I gather, that Dixon published my reply. His touchiness over this issue can be readily understood.
Correspondence in New Scientist
A correspondence in the letters page of New Scientist ensued. The Health Education Council defended its work on smoking in pregnancy and Harvey Goldstein announced plans to study the different effects on the health of babies which he would cause by distributing different publicity to different groups of mothers-to-be. Charles Fletcher confirmed that Dixon had added “social irresponsibility to mere eccentricity”. The Edinburgh geneticist Charlotte Auerbach FRS engaged Burch longest with three obtuse letters. His final reply to her assured her that no offence was intended by his statement that “some people, even among the medically qualified, find it almost impossible to accept that association does not necessarily imply cause.”
Correspondence with Congressman Richardson Preyer
Burch’s American correspondent was Congressman Richardson Preyer (Democrat of North Carolina) who saw Burch’s articles in New Scientist and wrote to him on 10 April 1974. He wrote:
When the legislative hearings on cigarette advertising and public health began in 1969, I expected that I would have to defend the interests of my tobacco growing constituents against what Sir Richard Doll referred to as ‘a mass of evidence to support the belief that cigarette smoking is one of the principal causes’ of lung cancer. After thirteen days of testimony, which for the first time exposed both sides of the controversy to public scrutiny, I was frankly surprised by the many serious questions raised about the evidence and the conclusions that had been drawn from it by the antismoking fraternity, most of which had permeated the mass media as fact.
Preyer had proposed a law establishing a commission to examine the issue.
Burch was delighted by the letter. In his reply, he enlarged on the weakness of the case against smoking and the low scientific standards prevalent in epidemiology, but for that very reason he was pessimistic about the calibre of any commission likely to be appointed.
The Commission needs genuine experts (at least two) in scientific logic and statistics who are also familiar with the medical and biological field and the complications presented by the phenomena of self-selection and genetic heterogeneity. One of these should be the chairman. The Commission would also need at least one human geneticist, although that, I fear, is another discipline lacking in scientific rigour. A pathologist or two and an epidemiologist should make up the numbers.
Five such individuals with medical qualifications would be impossible to find, certainly in Britain.
This was the first of several forays which Burch made into the political process in Britain and the USA. He wrote to Preyer again suggesting a meeting during a trip to America in 1975 but it is not known whether it took place.
At the Society of Occupational Medicine
In July 1974, Burch addressed the Society of Occupational Medicine on the subject of cancer. He reported to Seltzer:
My paper… went very well indeed. The questions were sensible and non-hostile. At coffee, I was introduced to the editor of the Society’s journal who asked me whether I had despatched my paper. One member at his shoulder said that my paper had to be published all costs! Perhaps their readiness to accept my arguments followed from the prestige that still attaches to Professors in this country? Various members had corroboratory stories about the unreliability of diagnosis.
The paper, covering the same ground as the Lancet letters and the New Scientist articles but at more satisfactory length, was duly published in the Journal of Occupational Medicine early in 1975. Burch had at last succeeded in getting an entire paper into print in a medical journal.
Publication in Tabak-Journal
Smoking and disease: problems of interpretation is a minor piece published in Tabak-Journal International “which, apparently, is some form of trade journal,” Burch told Seltzer. “I have allowed myself somewhat sterner language than is customary in scientific journals. Tabak-Journal is printed in six languages and I do not envy the translators their task.”
Framingham and self-selection
Burch continued to write in the correspondence columns of the Lancet. A letter of December 1974 concerns the important Framingham study, in which the health and smoking habits of 5209 men and women in the town of Framingham, Massachusetts were monitored over many years (18 years of data had accumulated at this point). A paper by Tavia Gordon and others claimed to demonstrate that smoking causes coronary heart disease from a positive correlation between the two and lower rates of the disease in those who gave up smoking. Burch made the obvious point that correlation need not imply causation, and the persistently neglected one that those who give up smoking are self-selected, and therefore not a random sample of the population. The Framingham study was vulnerable to this problem because it was prospective but not randomised. The MRFIT study, which was both, was designed to overcome the difficulty, and took the Framingham results as a baseline but failed to corroborate its results.
Burch had similar criticisms of a study of the alleged effects of continuing to smoke after coronary thrombosis by Dr Risteard Mulcahy. Mulcahy, a reformed smoker who admitted to the zeal of the convert, was a pioneer of preventive medicine in Ireland and noted for his then unusual practice of jogging through the streets of Dublin.
Lee’s definitive assessment of Burch was submitted to the Tobacco Research Council in June 1974. One of his reports gives a glimpse of Burch in that year: Lee was in Leeds for discussions with a different researcher.
Shortly after we had started talking I noticed P.R.J.B. in another corner of the same room. He spotted me and came over to talk. He bore no personal grudge against me for arguing with him for so long by post a year or two ago and was in fact very friendly. He talked non-stop for half an hour and has clearly not changed his views. I did not attempt to argue with him, not that he gave me much chance to say anything at all! I did mention to him that the E.E.C epidemiologists had recommended a Common Market Twin Register, which was news to him.
(The EEC, or Common Market, is now the European Union.)
Lee deftly summarised most of what Burch had written in the Lancet and New Scientist, classifying the points as negative evidence against the causal theory or positive evidence for it, with the burden of proof on Burch to show that causationism is false. However, he saw a presumption in favour of causation and regarded the constitutional hypothesis as a far-fetched logical possibility. He did not cover the general theory of disease and discussed the mathematics of the curve only in the context of his own rival account of carcinogenesis.
Although Lee never took Burch entirely seriously, this and his occasional analyses of Burch’s later work, are important critiques of Burch by a rare reader who understood him and gave him a fair hearing.
Burch in Michigan
Burch visited the USA in May 1975 to deliver a course of lectures at Michigan State University. We know only of his plans for the visit and the fact that it took place. He was in contact with Seltzer, Ted Sterling and Congressman Richardson Preyer about possible meetings, and it is known that he met Sterling in Vancouver either on this or another occasion.
New smoking products
This item relates to a forgotten episode in the war on smoking: the introduction of New Smoking Material, a substitute for tobacco consisting of cellulose doped with nicotine. A Lancet editorial noted that animal experiments seemed to show that NSM was less carcinogenic than tobacco and that the government had established the Hunter committee to advise on it. A letter from Burch reaffirmed that the alleged dangers of tobacco were still at issue, although the Swedish twin study incriminated it as a cause of bronchitis.
Another rejection by Nature
At some point in 1975, Burch submitted a second paper to Nature entitled Cigarette-Associated Cancers Other Than Lung Cancer: Some Paradoxes. Burch expected it to be rejected and it was. The text seems not to have been preserved, but Burch discusses these cancers in chapter 10 of The Biology of Cancer, where he points out that cancer of the stomach, bladder and pharynx, though commoner in smokers than in non-smokers, had not increased in frequency with rising rates of smoking. The Biological Manuscripts Editor, Peter Newmark, wrote to Burch that the decision was based on “the opinion of expert reviewers” and cited two considerations. “The first half of your paper contains a narrow selection of data which support your case but cannot counter the many studies which indicate the contrary… The other half illustrates an apparent paradox but does not allow for the explanations which have been put forward for it in the past.”
The issue here is clearly scientific method: Burch’s Popperian approach in which evidence refuting a hypothesis (the so-called paradoxes) is crucial, as against risk factor epidemiology with its emphasis on large amounts of data confirming the hypothesis, and its willingness to be content with possible, rather than demonstrable, explanations of anomalies. “The forces of righteousness are preparing a counter-attack! And it won’t be on scientific grounds,” Burch commented to Seltzer.
Seltzer at Leeds
No doubt at Burch’s invitation, Seltzer visited the University of Leeds to lecture on smoking and heart disease in November 1975. “The Professor of Medicine, George McNicol,” Burch warned him, “ is a fanatical anti-smoking crusader, with whom I clashed at a committee meeting last week. His special field is CHD. You can expect fierce and implacable opposition from him… Anyway, we look forward to seeing you and Ruth and I suggest we have dinner at 48 Henconner Lane after your lecture.” The title and content of the talk is unknown, but when Seltzer’s departmental boss, Frederick Stare, wrote asking how Seltzer’s “controversial” views had gone down, Burch replied:
I had expected the audience to be critical or downright hostile. In this I was proved to be quite mistaken. The general reaction was best summed-up, perhaps, by Dr. S.H. Taylor, who is in charge of our Intensive Care Unit. Because he is an authority on heart disease and also open-minded on matters of aetiology and pathogenesis, I had asked him to propose the vote of thanks. I did not anticipate his eulogy. Not only was he persuaded by Carl’s well-presented arguments, he also pointed out that the type of analytical approach that Carl had used was conspicuously absent from most medical education – at least in this country. He very much regretted that medical undergraduates were not present at the lecture. Having myself reached similar conclusions about the deficiencies in our education of medical students, I was intrigued to find that a busy clinician was equally disturbed.
The question-session following Carl’s lecture was a long one and it included the inevitable queries from convinced ‘causationists’, but these proved to be no embarrassment to Carl. At the end of the formal session, and as always happens after a highly successful lecture, a small group gathered around the lecturer. The group included my colleague, Professor R.E. Ellis. I knew that he was very sceptical of the ‘constitutional’ case before the lecture, but the data presented and analysed by Carl made him change his mind. Subsequently, various other members of the audience told me how impressed they had been.
It is fair to say that the lecture was something of an event. Seldom do we hear an exposition on a fundamental topic that causes some of us to reformulate our thinking. Carl succeeded in doing just that.
The Need to Know
By now, the tobacco companies were taking a keen interest in Burch. In 1975, the Tobacco Institute commissioned a film. This body was funded by the US cigarette manufacturers and, unlike the British Tobacco Research Council, had not been captured by health campaigners. The purpose of the film was to challenge the belief that smoking is harmful, and its substantive content consists of clips from interviews with all the prominent cigarette sceptics of the day. It was aimed at the general public and was repackaged for distribution to different audiences under the names The Need to Know or The Answers we Seek.
One of the sceptics interviewed was Philip Burch. It was arguably an error of judgement for him to participate, although there is no reason to suppose that he did so for financial gain. The film is propaganda and, as such, no worse than the government funded anti-smoking propaganda films of that period. It consists largely of factoids and sound bites from scientists including Brownlee, Seltzer and Eysenck. Burch performed well to camera but he did not really come up with what was wanted because he was not able, and probably not concerned, to simplify to the required level. He was a man who thought, and spoke, in paragraphs.
Burch: Now when you think this thing through, and take into account various other studies on twins etc., this indicates that the genes that are predisposing to lung cancer, whether or not they predispose the particular individual to smoking, are associated with genes that predispose to smoking, because we find the first degree relatives of the non-smoking lung cancer cases are more common, more frequent than among the control, first degree relatives of the non-smoking…
Interviewer: I’d just like to amplify that bit again. What you are saying is that in their study, the close relatives of lung cancer victims who smoked –
Burch: Who didn’t smoke.
Interviewer: Oh, of lung cancer victims who didn’t smoke, their close relatives were also –
Burch: They smoked more frequently than the first degree relatives of the non-smoking controls.
Interviewer: Would you mind saying that one again? I don’t think it came out clearly. Would you mind launching into that one again.
Burch: Right ho.
One brief clip of him survives in a version of the film called The Answers we Seek (at 13.00):
We still live in an era of environmental determinism. That’s to say, we have an essentially optimistic philosophy. We want to believe that all our ills can be attributed to something in the environment… It means that all we have to do is eliminate smoking, eliminate soft water, and so on, and we will eliminate lung cancer, cardiovascular disease and so on.
(‘Hard’ water has limestone dissolved in it, ‘soft’ water does not.)