A Half-life of Burch
3. In the spotlight
Publicity for The Biology of Cancer
By the start of 1975 Burch had completed his book The Biology of Cancer, and in August it was accepted for publication by the Medical and Technical Publishing Company of Lancaster. The Information Officer of the University of Leeds arranged the customary press release, and Burch briefly found himself at the centre of a media circus. “You are on the wrong side of the Atlantic!” he wrote to Seltzer.
We have been having great fun and games over here… I spent virtually the whole of Monday either talking to correspondents or new agencies on the phone, or in the BBC radio studios or in the ITV television studios. Two broadcast interviews and one T/V report went out. Then on Tuesday (yesterday) I was slightly less busy on the telephone, but did one live broadcast (BBC) and one live T/V interview (BBC). Stephen taped the sound from both programmes and I was particularly pleased with the broadcast. On top of all that, I made the front-page headlines in The Sun (copy enclosed) and inside headlines, with photograph, in The Daily Mirror (copy enclosed): I am hoping that ASH will write to The Times so that I can reply to them…
This morning the BBC asked me whether I would consider doing an ‘in-depth’ T/V programme, in a new series on smoking and lung cancer. I showed some sympathy but stressed the difficulties.
The Sun 19 August 1975:
‘SMOKING MAY NOT KILL’ STORM
Professor rapped by medical men
The Daily Express 20 August 1975:
Smoking: Does anyone really know the truth?
Non-smoking Professor Philip Burch lit up a king-size controversy yesterday…
The Daily Telegraph 20 August 1975:
Smoking link to cancer queried
The Daily Mirror 21 August:
CANCER ‘JURY’ TO DECIDE
Controversy in The Times
British newspapers fall into something of a hierarchy; the mass-circulation tabloids such as the Sun and the Daily Mirror, and the ‘quality press’ aimed at a smaller, more educated readership. Burch knew that his ideas were too complex to be simplified for a mass audience though, like many before and since, he discovered that the slanted, sensational tabloids report definite fact more accurately than their high-minded competitors. However, he took educated public opinion seriously, and took great pains in communicating his views in an interview with Neville Hodgkinson of The Times, in those days far the most intelligent of the national dailies. (“The final product is still, in my view a poor piece of writing – not as good as that in the ‘vulgar’ press.”) Hodgkinson’s report on 18 August occupied some 21 column-inches, and could hardly be ignored away by the anti-smoking establishment. A series of letters to The Times ensued.
The participants were heavyweights: Professor Michael Stoker FRS (noted for his work on viruses and cancer), Charles Fletcher and his colleague Keith Ball, Herbert Dingle (Professor Emeritus of the History and Philosophy of Science, University College, London), Sir Richard Doll and Richard Peto, J. H. Edwards (Professor of Human Genetics, University of Birmingham) and Dr Roger James. Four of the letters concerned scientific reasoning. Fletcher and Ball accused Burch of emphasising negative evidence over positive evidence, though there was far more of the latter: he was alone in his views and irresponsible. Dingle (in support of Burch) retorted that scientific questions are not decided by majority vote. Edwards wrote that “The evidence now takes the form of a massive and untidy net, some of whose strands are weak, while others are obscured by encrustations of misleading statistical argument.” James, interestingly, claimed that the “Fisher-Burch” theory did not meet Karl Popper’s criterion of falsifiability. Two letters concerned fact. Stoker asserted that radiation is carcinogenic because it is mutagenic, thus cigarette smoke, which is mutagenic, must be carcinogenic. Doll and Peto set out five main reasons for rejecting Burch’s hypotheses.
Burch replied to each criticism. To Stoker, he outlined his reasons for thinking that cigarette smoke is not mutagenic. To Fletcher and Ball he asserted that he had examined far more of the evidence than any of the RCP reports. To Edwards he retorted that the ‘net’ theory of inference was nonsense. James had identified the real issue, falsification, but Burch, not James, was the Popperian. (James later published a book, Return to Reason, on the application of Popper to public policy, making many good points about the planning disasters of the 1970s, and also repeating his case against Burch.)
Doll and Peto’s letter took pride of place on the correspondence page, occupying three columns. Burch’s reply occupied two. Neither party said anything they had not said before, including Doll and Peto’s claim that Burch’s ideas had “already been examined and found wanting”, though it was probably new to most readers of The Times. Doll and Peto invoked evidence from animal experiments, the non-smoking Mormon and Adventist sects, industrial workers exposed to carcinogens, the alleged presumption of causality when a correlation is sufficiently large, the benefits of stopping smoking, and cohort effects. Burch’s reply stressed self-selection, dose-response relationships, and the properties of the age distribution. He finished with the issue of scientific reasoning:
Doll, Peto and others make the fundamental error of dwelling on what they believe to be corroborative evidence. Sir Karl Popper has taught some of us that, when testing theories, we must attend mainly to those crucial observations that threaten to refute them.
An editorial in New Scientist took note of the clash. It informed its readers that “Burch is no crank” and that “Historical precedent favours the case of the informed and intelligent rebel”. However, “Meanwhile, it would seem sensible to continue to regard tobacco as the devil’s own weed.”
Exchange with William Rees-Mogg
Burch’s third letter, addressing Edwards and James, nearly failed to appear. Space in the letters page of The Times is coveted – far more so than an appearance on television – and the letters editor wished to close the correspondence. Burch, who had little use for the tabloids and television, wanted to be heard by the kind of people who read The Times, and took the matter to the top. In a letter to Seltzer he wrote “Despite everything I remain buoyant but somewhat disturbed by recent events.” In the course of the summer he had become a public figure and to some a public enemy, the pages of Nature were now closed to him, and new battles loomed on the horizon.
Letters from Edwards and James (enclosed) were published in The Times on 19 September and I replied (see enclosed) on 22 September. This reply never appeared and so I wrote a personal letter to the Editor, Mr. Rees-Mogg. So far, no reply from him.
The point Burch made to Rees-Mogg (29 September) was this.
As you will be aware, editors of medical and scientific journals allow the author of an accepted paper the privilege of replying to criticisms published in the correspondence columns. I recognize that Mr. Hodgkinson’s report was not an article by me although I was quoted very extensively and under the circumstances I feel – and I hope you will agree with me – that I should be permitted to reply to the criticism of Professor Edwards and Dr. James.
Rees-Mogg, a thoughtful and fair-minded man, saw the justice of this. “We do not follow the practice of invariably giving the last word to the initiator of a controversy,” he wrote to Burch personally (9 October), “but we felt that your third letter did provide substantial new material to the argument and were very glad to publish it.” “I did Rees-Mogg an injustice,” Burch reported to Seltzer.
Hodgkinson also wrote to Burch (8 October) to congratulate him on his tenacity and to inform him of a forthcoming interview with Mehrishi in Cambridge. Hodgkinson wrote a number of reports about the smoking controversy during this period. He covered Seltzer on 17 November 1975, Hans Eysenck on 30 April 1976 and the anti-smoking campaign ASH on 16 December 1976.
Correspondence with the Minister of State for Health
One immediate result of the burst of publicity was that the government referred the matter to the Independent Scientific Committee on Smoking and Health, known as the Hunter Committee after its chairman, Dr R. B. Hunter, established by the Department of Health in 1973 to create guidelines concerning the New Smoking Material then under development. The Minister of State (the junior of the two health ministers) at the time was David Owen, later Lord Owen, but then Dr Owen. (He insisted on the title as he was a qualified physician, although it was the usual courtesy title accorded to medics who are not actually MDs.) He held conventional anti-smoking views. Owen wrote to Burch on 22 August to inform him that government policy on smoking was based on reports by the Royal College of Physicians and others, but that the committee would be considering his theory.
Owen was the kind of intelligent, influential figure who Burch wanted to persuade, and he replied at some length from his holiday home in Wales on 1 September. He had no very high expectations of the Hunter Committee. “I have sufficient experience of committees to be slightly sceptical of its capacity to cope rigorously with the difficult and emotionally laden subject in dispute… but I realise that no practical alternative to a committee is available to you.” He restated his objections to the 1971 report Smoking and Health Now: “the restricted nature of the quoted evidence (uncomfortable observations are omitted), the propagandist line and … the spurious logic and incompetent science”.
He also had two positive suggestions for state-funded research projects. One was to set up a national twin registry on the same lines as those in Sweden and Finland, a project which had encountered resistance in Britain. The other was an investigation of the causative role of smoking in chronic bronchitis, which Burch did not dispute.
Volunteer addicted smokers might be persuaded to smoke different types of cigarette, on a double-blind basis, for specified periods of time chosen to obviate seasonal and other complicating factors. Subjective and objective measures of coughing and bronchitis could be used to assess the effects of different forms of smoking. Although such an investigation would call for careful planning it would not be too expensive and could make a genuine contribution to health.
Burch arranged for the committee to be supplied with page proofs of The Biology of Cancer, then with the printers. The Hunter Committee returned a negative assessment of Burch’s views to Owen early in 1976. Burch, never one to let a rebuttal go unrebutted, drew up a point by point response and sent it to Owen.
Approaches from the Royal College of Physicians
At this time, two important publications were in the pipeline. One was the analysis and publication of the results from the Doctors’ Study 20 years after its inception, and the other was a third report on smoking and health from the Royal College of Physicians. Both were scheduled to appear in 1976. From the perspective of those at work on them, Burch was a problem to be disposed of. The sequence of events was set in motion in the autumn of 1975 when, as Burch related to Seltzer:
Ball came along to see me as planned and we spent about 2 ½ hours in fascinating conversation. It is fairly clear that he was acting as the extremely diplomatic ambassador from the Royal College of Physicians. He did not presume to be able to challenge me on scientific grounds – for obvious reasons. He tried to exert some moral pressure, but his primary objective was to get me to go along to the Royal College for an informal discussion with a few interested parties. The President and Sir Richard were mentioned. I suggested that we should first attempt to deal with the scientific issues through correspondence. They can write to tell me where I am wrong: I can then respond. If at the end of a suitable exchange both sides desire a meeting then that can be arranged. (A meeting with the President alone might be worth contemplating.)
(“Ball” was probably Dr Peter Ball, one of the governing officers of the RCP.)
The president of the RCP, Sir Cyril Clarke, duly wrote to Burch with proposals for a meeting and discussion with representatives of the College. Burch replied on 13 November that he thought a discussion should follow and not precede a written exchange of views. “I feel that it is most important that the 1976 report should reflect more scientific credit on the Royal College than did the last one,” he observed with barbed courtesy. Further letters were exchanged on 18, 20, 25 and 28 November in which Burch offered Clarke advance copies of The Biology of Cancer in return for a draft of the new report. Clarke arranged for Burch to be sent the draft of chapter 4 of the report, the chapter on lung cancer, and Burch provided Clarke with a copy of his book.
The draft of chapter 4, written by Dr Charles Fletcher, came with an appendix “The Statistical Controversy on the Relation Between Smoking and Lung Cancer” by Harvey Goldstein and Keith Horsfield, specifically directed at Burch’s version of the constitutional hypothesis. Burch immediately drew up a detailed criticism of the appendix and dispatched it to Clarke on 23 December in the expectation of a response from Goldstein and Horsfield. Clarke replied on the 31st confirming that it had been forwarded.
Christmas at Cambridge
A glimpse of Burch’s Christmas is too good to leave out. “We had a very enjoyable Christmas in Cambridge. Eight to dinner on Christmas evening and 18 to lunch on Boxing day. We saw Stephen’s new rooms at Trinity (palatial) and the radioastronomy and new Cavendish labs. All most impressive.”
Clarke passes the buck
Burch passed the correspondence with Clarke on to Seltzer. “The tone, you will notice, is very friendly. Perhaps he would like me to get him off the hook? Maybe he is not keen to have the Royal College exploited in this way?” Burch was an astute man and, sure enough, it transpired that Clarke had handed over responsibility for the Burch problem to Fletcher, who wrote to Burch on 14 January with a new proposal.
In the first week of May, we have what we call a Specialist Course here which is for registrars and consultants in respiratory diseases. On 5th May, 1976, at 11.45 a.m. we are having a seminar for the course which will also be attended by members of the staff here on lung cancer and smoking. Richard Peto will be presenting some of the latest results from the doctors’ study and I would be very pleased if you could come and present your views which would be followed by a discussion. The whole session lasts an hour and a quarter… I have asked Richard Peto to speak for 25 minutes; if you could do the same we would then have adequate time for discussion.
Burch replied on the 16th:
Provided I can have an advanced copy of Richard Peto’s paper in time to allow me to make informed comment on it, I shall be very pleased to join in. In the meantime, I look forward to receiving your response to my criticisms of the draft Appendix to the Report of the Royal College.
Geoffrey Todd, who was monitoring the draft report on behalf of the Tobacco Research Council, had the inside story. The Royal College had received Burch’s rebuttal of the appendix but “Horsfield could not understand it”. The tables of data from the Doctors’ Study had now been digitised, but “Peto had barely begun to analyse the data.” On 5 February he disclosed that the meeting between Burch and the RCP appeared to be off. “They wanted to be able to say that they had listened to all that Burch had to say… Perhaps it is being replaced by the Peto/Burch confrontation on 5th May.” Peto expected to complete his analysis by the summer and “first results” would be presented at Hammersmith.
Neither Peto’s text nor Goldstein and Horsfield’s reply materialised. Burch wrote to Clarke on 10 February about it. “Surely he [Goldstein] cannot have seen the light. It would be so out of character.” Clarke responded with the news that Goldstein considered that he had nothing to say which he had not said already. On 26 March Burch wrote to Clarke and, separately, to Fletcher, making his cold contempt for Goldstein clear. Though Burch did not know it at the time, the idea of an appendix had been dropped and the final version of the report would appear without it.
Ireland, January 1976
Burch visited Dublin and Belfast for a day each in January to lecture on his theories. “Pray that I escape the bullets and bombings!” he wrote to Seltzer beforehand (Northern Ireland was then nightly in the news) but in the event all went more than well. Characteristically, he found himself more at home in supposedly puritan Belfast than in the south.
My talk in Dublin was well attended and the question session went on for well over half an hour. However, the audience was very mixed, not well informed (no medical statisticians) and not very clever; it included a smattering of fanatical anti-smokers. The audience in Belfast was totally different. Dr. MacRae, who invited me, is a medical statistician and obviously a disciple; there were also four representatives from the tobacco industry which is active in Ireland. I could see that I had a sympathetic audience and I rose to the occasion. If any fanatical anti-smokers were present, they remained silent.
Socially, both cities were a great success. I was put up in splendid accommodation at Trinity College, Dublin and in the Staff House at Queen’s University, Belfast. Two sherries and three large glasses of wine during an early dinner prior to my Belfast talk might have contributed to its success.
The purpose of the visit was to lecture to regional branches of the Royal Statistical Society (there was still a branch in the Irish Republic at this time). Burch undertook other tours of this kind before he lectured at the Society’s London headquarters in 1978.
Nicholas Wald and his smoking machine
A minor controversy took place in the pages of the Lancet early in 1976. On 17 January there appeared a paper by Dr Nicholas Wald of the Department of the Regius Professor of Medicine at the Radcliffe Infirmary, Oxford. Its aim was to show that the introduction of filter cigarettes had modified the dangers of smoking. A novel feature of this study was the use of a smoking machine to measure the products delivered by different kinds of cigarettes.
Wald compared smoking rates in men and women to 1970 (static in men, rising in women) with rates of lung cancer and heart disease in the periods 1956-60 as against 1969-73. He explained lower rates of lung cancer and rising rates of heart disease by blaming the former on tar and the latter on carbon monoxide (the new cigarettes delivered less tar at the cost of more carbon monoxide). The matter is complicated by a major change in diagnostic criteria for coronary heart disease, as defined in the International Classification of Disease, which took place with the transition from ICD7 to ICD8 in 1968.
A correspondence ensued. Geoffrey Myddleton observed that Wald’s data on lung cancer and smoking in women was taken from the Hammond study, which actually showed a very weak correlation between the two. Niri Orlien complained that the experimental design was poor, partly because he relied on puff counts from his smoking machine. Carl Seltzer objected that Wald had underplayed the “irreparable break” involved in the new ICD definitions. A comparison of male and female rates in 1950-4 against 1963-67 (ICD7) and 1968-69 against 1972-73 (ICD8) showed lower rates of heart disease in women. Wald replied to Myddleton that Hammond’s data was unrepresentative of US women (true, but Wald had cited it himself). To Seltzer he replied that the alternative time periods suggested were misleading because they were not of equal duration.
A further letter from Philip Burch defended Seltzer against Wald. Wald’s hypothesis predicted a higher increase of CHD in women than men. This did not happen over the ICD7 period, and in the ICD8 period for three ages groups only, no doubt by coincidence those to which Wald appealed.
Publication and reception of The Biology of Cancer
The Biology of Cancer was finally published in January 1976. It is perhaps unfortunate that the Leeds University Information Officer did not hold his fire until this point. As it was, Burch had his week of media exposure when scientifically literate opinion only had the New Scientist exchange to go on, and for those with access to the Lancet, a complicated correspondence to be tracked through dozens of weekly issues. A second press release was issued to mark publication, but it was the same news twice, and the actual appearance of the book was something of an anti-climax. Reviews, ranging from tepid to hostile, appeared in the journals, none of them engaging with its fundamental theses, the theory of central growth control and the concept of autoaggressive disease.
The Lancet objected that five parameters will fit any data, and that Burch’s refutation of Doll assumed the reality of hypothetical quantities such as his parameter S. New Scientist thought that he divided cancers into early and late onset forms ad hoc and ignored obvious environmental factors in cancer. The British Medical Journal complained that he supplied no experimental evidence and that studies of age-dependence were insufficient to make his case. Burch’s comment on this was: “The current BMJ carries a review of my book, fairly hostile as expected… There are few signs that the reviewer has read the book, or that he is in his right mind.” (Letter to Seltzer, 18 May 1976.)
In March, the British Journal of Cancer published an extremely hostile review of the book by Julian Peto, younger brother of Richard Peto, which led to a clash between him and Burch in the journal’s correspondence pages. Since this exchange took place after the public war of words between Burch and Richard Peto at Hammersmith, and refers to points raised there, that event needs to be considered first. The British Journal of Cancer was the house journal of the cancer specialists forming the bulk of the audience, and they will have arrived at Hammersmith with little doubt as to the official line on Burch.
Israel, April 1976
Burch’s frequent letters to Seltzer during this period often include personal news, not least his enjoyable accounts of trips to conferences overseas, often accompanied by Jane Burch, and of the banqueting, intelligent discussion and sight-seeing which generally followed them.
He and Jane visited Israel in April for three engagements; a workshop on Huntington’s chorea and two lectures on radiobiology and lung cancer.
We returned from Israel on Saturday after a splendid, busy and fascinating two weeks. The Huntington’s chorea workshop was, on the whole, unexciting being too near the Leuven (Louvain) meeting of 1 1/2 years ago. The session which I chaired and in which I also presented a paper, went very well but the members of the workshop do not really understand what I am getting at, or why. But they are nice chaps and we all (or nearly all of us) get on very well together.
Jane went along to Be’er Sheva with me and heard my lecture on biological effects at low doses and low dose-rates of radiation. Jane thought it went very well and there were lots of questions. My paper on lung cancer and smoking was given at the Israel nuclear centre at Soreq. There was much sympathetic nodding in the rather small audience and some intelligent questions. But the most satisfying session was at the Cancer Registry in Jerusalem with a chat to an audience of four, one of whom was a dentist (now administrator) who was trained at Birmingham and knows Douglas Jackson well. He was terribly bright and, with a woman in the cancer registry, asked all the right (highly intelligent) questions.
Luckily we found some time for pure tourism and from Jerusalem we visited the old city, the Israel Museum, the Rockefeller museum, the model at the outlying hotel, Tiberias (by public ‘bus) etc. etc. Expensive everywhere but enjoyable.
(Douglas Jackson worked with Burch on a series of studies of tooth decay not covered in this website.)
Switzerland, April 1976
Burch visited Zurich at the end of April for a symposium, Smoking in Scientific Controversy, organised by the cigarette manufacturers of Switzerland and Germany (he was generally willing to speak at industry events for expenses, but did not accept industry funding of his research). “Can you call in at Leeds before the Symposium?” he wrote to Seltzer in advance. “I am wondering whether I ought to arrive in Zurich on the 27th April, allowing one day for sightseeing, museum visiting, and the like.”
It was here that Burch delivered his talk Smoking and Lung Cancer: a Causal Connexion?, possibly his best exposition of his theory for a non-mathematical audience. There were also addresses from Seltzer on smoking and coronary heart disease and Hans Eysenck on smoking and personality. He wrote afterwards:
It was very pleasant meeting you and Ruth in Zurich, and I found the whole meeting – Kinli included – most enjoyable and entertaining. I was very pleased to meet Eysenck for the first time and I was impressed by his manner as much as by his excellent brain.
(“Kinli” is probably the handsome old Hotel Kindli.)
Confrontation at Hammersmith
The public exchange between Philip Burch and Richard Peto went ahead on 6 May 1976 at Hammersmith Hospital Postgraduate Medical School in west London before an audience of senior doctors specialising in respiratory disease. Richard Doll should also have been there but was prevented by a misunderstanding about dates. We do not have the text of either man’s lecture, but we have a handout which Peto distributed to the audience and two very different accounts of the meeting by Burch and Peter Lee, who was also present. Lee thought that the confrontation was a clear victory for Peto, while Burch came away feeling that he had been subjected to a show trial and convinced of Peto’s intellectual fraudulence.
According to Lee:
All in all, I got the clear impression that PB did not come out of the discussion very well. RP showed that the causation hypothesis was consistent with all the observed facts… However, on the basis of Occam’s Razor, PB’s arguments did not come off very well, needing to invoke more and more assumptions to fit the data.
Peto spoke first, followed by Burch.
PB’s reply was for most of the time off the point. He went to painstaking lengths to churn out all his old graphs to prove the point that cigarette smoking could not explain the whole of the observed secular changes in lung cancer rates. As RP had already accepted that diagnostic changes were a very large determinant of those secular changes, PB’s effort was not really worthwhile.
Burch gave his version to Seltzer in a letter dated 10 May:
The Richard Nixon ‘dirty tricks’ department could have learned something from a study of the Richard Doll organization.
Peto had agreed to let me have a draft of his paper by mid-April on the understanding that my task would be to comment on it. Needless to say, nothing came along. That did not surprise me. They wanted to disadvantage me as much as possible. I arrived at Fletcher’s office at about 10.45 and we greeted one another cordially. He gave me Peto’s handout ‘Smoking and Lung Cancer’ (copy enclosed) and apologised on behalf of Peto who shortly turned up and joined our discussion. He is a young, short, shrill, angry man with abundant hair who looks at you sideways. I straightway accused him of abominable behaviour and needless to say he was full of insincere apologies. After a fairly rancorous discussion we went into the lecture theatre. During this discussion I asked him why ‘Trends in Mortality among British Doctors in Relation to Their Smoking Habits’ had been published by Doll and Pike without confidence limits. He thought this was done for ‘propaganda purposes’! Fletcher could hardly believe his ears!
Peto duly delivered his bit, angrily and with various misrepresentations. I delivered mine, correcting various errors in Peto’s presentation and pointing out conflicts between his model and the data. Fortunately, I kept cool and refrained from insults. Quite a number of faces in the audience looked sympathetic. We then had a slanging match for about 1/2 hour and Fletcher finally said a few words. He admitted that I had corrected some of his false beliefs…
Although I find Peto an execrable character (he is nevertheless clever in a slick, smart alec kind of way) I must confess I rather enjoyed the whole business. Their tricks were not wholly successful.
Burch and Lee cannot both have been right. To understand why Burch saw it as he did, the logic of his case and Peto’s, and also the structure of the meeting need discussion: see the longer version of this section here.
In later years, Burch refused on principle to address any gathering also addressed by Richard Peto. There can be no doubt that Peto withheld an advance copy of his paper, that he circulated a poorly formatted handout to his audience, and demanded that Burch answer nine tricky questions in 25 minutes at the end of a morning session. Whether his actions were merely thoughtless or a deliberate attempt to discredit Burch must be a matter of opinion.
The Burch-Fletcher correspondence
Burch wrote to Fletcher on 18 May setting out his considered, critical judgement on the Peto lecture. They proceeded to discuss the draft of chapter 4 of the new RCP report by exchange of letter:
|Burch 29 June||Fletcher 6 July|
|Burch 19 July||Fletcher 30 July|
|Burch 5 August||Fletcher 15 September|
|Burch 29 September|
Fletcher’s letters are characterised by increasing impatience and incredulity that Burch does not find his arguments convincing. He simply cannot see that Burch has the superior intelligence and grasp of scientific method. In the final exchange, the two men put their cards on the table in a revealing discussion of their different outlooks: Fletcher the practical man with humanitarian motives and limited tolerance for what he regards as Burch’s sophisms and obsessiveness, Burch the visionary who wishes to make medicine a science to compare with physics and contemptuous of the soft science of epidemiology and its sloppy intellectual standards.
Fletcher on Burch:
When you came to lecture here, I got the impression that you were open to logical persuasion. As our correspondence has continued, I have become less and less confident of this and more convinced that you are the victim of an “idee fixe”. Another difficulty I find is that you seem to look on the whole question as if it were a problem in theoretical physics in which the outcome might be of practical importance in half a century from now. Doctors have to decide what to do about smoking and health now and the causative hypothesis is consistent with so many more and inconsistent with so many fewer facts than the pure genetic hypothesis that from the practical point of view, we have to persuade people not to smoke and to encourage the manufacturers to introduce less harmful forms of smoking.
Burch on Fletcher:
You are justified in contrasting our attitudes. Doctors do have to decide what advice they should give to their patients and I imagine that the majority would tend to err (if they err at all) on the side of caution. That is right and proper. But I am bound to say that humanitarian considerations, however noble, are almost certain to induce bias where the investigator is concerned. Those of us who have been trained in a more rigorous science than medicine ought to feel free (or even obliged) to challenge conclusions that appear to us to be unjustified. We frequently read claims in the medical literature that some environmental factor causes a certain disease. These claims often rest on the flimsiest of evidence and a wholly unjustified interpretation. Watch dogs are essential.
It is much more difficult to practice this method in medicine than in physics, and it is no coincidence that Popper is unpopular among certain epidemiologists.
A longer version of this section quoting the two men in their own words as they consider the report section by section is here.
The BMJ announces the new line
An editorial in the British Medical Journal for 21 August entitled Lung cancer and smoking: is there proof? dealt with the issue of Philip Burch by name. “Fletcher obviously had a large hand in it,” he wrote to Seltzer. It rehearses all Peto’s arguments at Hammersmith, presumably to promulgate them to the medical public: the alleged implausibility of the constitutional hypothesis, the new reliance on sex-ratios in the generations of 1850, 1890 and 1930, animal studies and trends in ex-smokers.
Burch’s reply (11 September) noted that the well-known words of R. A. Fisher, “a catastrophic and conspicuous howler,” had been wrongly attributed to Burch. The argument from sex-ratios “is based on unpublished, superficial and incomplete examination of the evidence” (i.e. Peto’s lecture and handout). “The problems of self-selection and misdiagnosis have not been adequately addressed, though there has been some improvement on the RCP report of 1971.”
Finland, August 1976
A conference on CHD took Burch to Helsinki in August, with an excursion to St Petersburg (then called Leningrad).
I had a most enjoyable congress in Helsinki and short trip to Leningrad. Helsinki is an attractive city in its modest way whereas Leningrad is a city of extreme contrasts. We have the gilded imperial pomp of the Czars on the one hand and great stretches of unrestored proletarian squalor on the other.
In the CHD sessions all the speakers agreed on the evils of smoking. A question from me about the problems of self-selection produced an admission that the phenomenon was present, but it did not shake their basic faith. My own paper about genetic factors in connective tissue diseases was well received and various people I didn’t know came up and said how much they had enjoyed it. Which surprised me.
Exchange with Julian Peto
Earlier that year, the British Journal of Cancer published its review of The Biology of Cancer, or rather, as Burch acerbically noted, “one part of one chapter of a 12-chapter book”. That section is Burch’s calculation of lung cancer rates predicted by Doll’s model and data on smoking, rates far lower than those actually observed. The reviewer was Julian Peto, younger brother of Richard Peto, who worked on Doll’s studies of exposure to asbestos. He had no particular expertise on the subject of cancer (indeed he missed the association of asbestos with mesothelioma). The review consists mainly of rhetoric (“a bewildering array”, “enthusiastic extrapolation”, “Procrustean model”) obfuscating the burden of proof. A controversy by letter ensued, in which Burch responded to the review in the sarcastic tones it had initiated, barely pretending to treat Julian Peto as other than his brother’s mouthpiece.
Peto proposed to account for a 100 fold increase in lung cancer rates by taking early 20th century levels, multiplying by a factor of 10 reflecting the real increase and again by a factor of 10 representing improved diagnosis. “The hundred-fold increase in recorded rates may thus be crudely summarized as the effect of a 10-fold increase due to smoking and a 10-fold increase due to diagnosis. (These 2 factors must of course be multiplied. Burch’s “best estimate from published data” is obtained by assuming that they are additive.)”
In reply, Burch contrasted his “Procrustean” and therefore falsifiable hypothesis with Peto’s resort to “fudge factor”. As for the two factors of 10%, they are each 50% of Peto’s explanation of lung cancer rates.
Peto got the final word, asserting that not 10% of the rise in lung cancer but 90% is caused by smoking. “By the same token, of course, 90% could be attributed to improved diagnosis.”
A fuller version of the arguments analysing the mathematics in more detail is here.
The Doctors’ Study at 20 years
Richard Peto’s analysis of the Doctors’ Study was finally published in the British Medical Journal for 25 December. It consists of
- A table of data on the survey, with discussion.
- 10 tables of data on smoking and causes of death, with discussion.
- 2 figures derived from 2 of the tables, with discussion.
- 2 tables summarising the conclusion.
The primary data on causes of death in various categories of smokers and non-smokers contained no surprises. The strong correlation of cigarette smoking with lung cancer, the negative correlation with Parkinson’s disease and the puzzling anomaly about inhalation were all confirmed, as was the association of cigarette smoking with heart disease (strong in younger age groups, weaker in the old). The details were set out in eight tables comparing male doctors with male doctors.
The authors were careful to emphasise that association is not causation and that the results were only new “insofar as they provide detailed comparisons between mortality rates in the general population and a subgroup that reduced its consumption of cigarettes more than average.” This referred to two figures and the two tables (II and XI) they are based on which compare male doctors with the general male population. Both figures showed regression lines running through a cloud of data points plotted left to right from past to present. Doll and Peto explained: “If any slope is negative, the mortality among doctors is progressively improving compared with that in the general population.” And indeed the regression lines for both smoking rates and death rates in doctors were negative, while those for the general population were flat.
This, Doll and Peto claimed, warranted the conclusion that “much of the excess mortality in cigarette smokers can be attributed with certainty to the habit, but whether the greater part is attributable depends on the interpretation of the evidence regarding ischaemic heart disease.”
The footnotes and references issued a couple of academic IOUs. A footnote promised that the full data sets which the study rested on were held “in the Radcliffe Science Library, Oxford, and in the library of the National Institute of Medical Research, Mill Hill. Two statements, that “the results of the Swedish twin study do not support Fisher’s genetic explanation” and that a question relating to dose-response relationships for lung cancer “will be discussed elsewhere” were supported by references to “Cederlöf, R, Friberg, L, and Lundman, T, in press” and “Peto, R and Doll, R, in preparation”.
That lethal weed
The same issue of the British Medical Journal featured an editorial, That lethal weed presenting the report’s conclusions as incontrovertible fact, and calling for new restrictions on smoking. “The time has come to make smoking unattractive and socially unacceptable.”
Burch had his say on this in a letter to the editor in January 1977. Parts of it alluded to the Hammersmith exchange, which centred on sex ratios, misdiagnosis and lung cancer rates in old age. The Doll-Peto paper related to men only, more or less ignored misdiagnosis, and as Burch pointed out, showed no evidence for ever higher rates with advancing age. But his main criticism of the paper related to figure 2 and Table XI, which provided comparative data on doctors as against all men but no absolute figure for lung cancer mortality. Had absolute rates fallen? “Professor Doll and Mr Peto could readily enlighten us on this important issue.” Burch had reason to think that absolute rates had not fallen. He pursued the matter both by seeking Peto’s complete dataset and by calculating absolute rates from available evidence.
The Challenge of Cancer
Possibly in consequence of his recent notoriety, on 31 January 1977 Burch delivered a lecture at Leeds called The Challenge of Cancer, though he alluded to his heresy only in parentheses: “(A few may even doubt that lung cancer is ‘almost entirely caused by cigarette smoking’.)” He began by emphasising that he will only be discussing causation – not treatment or cure, “because I am not competent to do so”. According to the prevailing view of oncologists at that time, cancer was mainly of environmental origin, with some holding that this was true of 99.9% of cases. Burch noted that an advantage of this view, “and one of the reasons for holding it,” was the prospect of prevention, a point which he illustrated from a recent pronouncement of Sir Richard Doll as reported in The Times.
People will either have to stop smoking and drinking or find a way to eliminate the dangerous chemicals in tobacco and alcoholic liquor; sexual promiscuity must also be reduced or a way found of preventing it from causing cervical cancer. It may be necessary to reduce the content of meat and animal fat in our diet…
The immediate priority, Sir Richard suggests, is for doctors to press the Government to ensure that neither commercial interests nor the pattern of taxation should be allowed to encourage people to undermine their health.
You will notice that the Government is being asked to assume responsibility for those of our everyday habits that are supposed to affect our health.
He proceeded to identify four major fallacies of current epidemiology
- Equating association with cause
- Comparing disease rates in groups which differ in numerous other ways
- Ignoring self-selection
- Supposing that emigrants are a random sample of the populations they left behind
Burch then illustrated the autoaggressive theory with new results on cancer of the lip, pancreas and cervix. He concluded
I can appreciate that many well-meaning persons welcome the justness of the notion that cancers are caused by that appalling trio of vices: smoking, alcohol and sex. To these, many would add the modern diet. But… I am not alone in fearing that the reputation of medical science – and of medicine in general – is bound to suffer from this relaxation of logical rigour. Moreover, to inflict on the cancer patient the additional, and very probably unwarranted burden of guilt, for what some are pleased to call a self-inflicted illness, is not only presumptuous, it is also bad medicine.
New health warnings
Burch’s name briefly returned to the daily newspapers in March 1977. A press cutting enclosed in a letter to Seltzer begins:
After last week’s renewed attack on smoking in which Mr Ennals, Social Services Secretary, called on the Health Education Council to outsmart the persuasiveness of tobacco advertising, an expert in medical physics has denounced the fallacious interpretation of anti-smoking statistics.
Smoking has been much in the news here as you can see from the enclosed excerpt from The Daily Telegraph. ASH, I gather, wanted cigarette packets to carry the warning ‘Smoking causes lung cancer, bronchitis and heart disease’ – or words to that effect. I can imagine the tobacco manufacturers being sued for libel if they made such declarations!
Rejection by the Lancet: exchange with Sharp
On 5 April 1977 Burch sent a letter to the Lancet criticising a recent editorial Are 90% of cancers preventable?
The editorial begins “There is a rumour going round that some 90% of cancers are due to environmental factors and thus could be prevented.” It ends “If cancer is to be prevented, far-reaching social changes will be required. Changes in behaviour, diet and alcohol consumption as well as smoking will probably be necessary. In the words of Sir Richard Doll, ‘the problems that we are now having to face in preventing tobacco-induced cancer will seem childishly simple’.” That anything up to 90% of cancers are of environmental origin was the thesis of Doll’s forthcoming book, The Causes of Cancer: Quantitative Estimates of Avoidable Risks of Cancer in the United States Today, written with Richard Peto and published in 1981.
Burch’s letter was rejected in the following terms.
Thank you for sending us that letter arising out of our editorial on cancer prevention. As is our custom I sent this letter to the Oxford group to give them an opportunity to comment… As you know, we have kept more or less “open house” in our correspondence columns on the arguments over smoking and disease, but on this occasion I think we had better say No.
I understand that the Oxford workers would be willing to provide more detailed comments, and if you would like to have them I will be happy to arrange it.
“It looks as if their policy has changed under the new editor, Ian Monro,” Burch observed to Seltzer.
The rejection came from the deputy editor of the Lancet, D. W. Sharp. Letters to journals are not peer-reviewed but accepted or rejected according to the judgement of the editors, often taking outside advice, as was done in this case from “the Oxford workers” (i.e. Richard Peto). This jarring phrase is indicative of the way Sharp regarded Burch, as somewhat like a schoolboy writing to NASA in need of gentle enlightenment. He helpfully explained the latest orthodoxy in the remainder of his letter. But Sharp was the kind of intelligent, knowledgeable man whose opinions Burch hoped to sway, and Burch wrote back making some important points.
According to Sharp, the risk of smoking rises as a proportion of years of smoking raised to a power. Successive cohorts of British doctors have had more years of smoking, and this is the reason why Doll and Peto did not present lung cancer rates in doctors as absolute rates, standardised only for age, but standardised against rates in the general population as well.
Burch, in reply, made three related points.
The first is one, originally made by Peter Armitage to Doll’s face when he addressed the Royal Statistical Society: that Doll was trying to validate two incompatible theories of carcinogenesis. Burch was to raise it again in his own appearance before that body, in discussion with Armitage: it is considered below.
The second is that the current orthodoxy on lung cancer had been revised: Doll and his then statistical adviser Malcolm Pike had seen no need to standardise their results from the Doctors’ Study as recently as 1972, and nor had the 1971 RCP report, which was based on their work.
The third is that Doll and Peto’s actual data was supposed to have been deposited in the Radcliffe Science Library at Oxford, and at another library, but was not available from either. Could Sharp try to obtain it?
In search of Doll’s data
A footnote in Mortality in relation to smoking: 20 years’ observations on male British doctors reads:
Tables showing the numbers of deaths observed and expected for each group of causes and each smoking category from which the rates shown in tables III-IX have been calculated are held in the Radcliffe Science Library, Oxford, and in the library of the National Institute of Medical Research, Mill Hill.
Burch wrote to those institutions requesting a copy of the data. The Radcliffe library replied on 2 April that the tables were not held there or at Mill Hill. “What Doll and Peto stated in the footnote to their paper was evidently a black lie,” he wrote to Seltzer. “Perhaps they are not too keen on the idea that some of us might get hold of their details?”
Peter Lee had noted Burch’s BMJ letter of 15 January and gone in pursuit of the data too. A direct request to Peto on 9 February received a reply the next day:
The tables aren’t, yet, held at the Radcliffe Science Library, although they should be sometime over the next few months, and when they are I’ll send you copies of them. Meanwhile, you’d be welcome in Oxford to examine what we do have already available.
Lee continued to pursue the matter and obtained the tables from Peto: they are the basis for his 1979 paper Has the mortality of male doctors improved with the reductions in their cigarette smoking? To this question his answer was that the benefit of stopping smoking was “not so great as has commonly been assumed”. That paper gives the missing tables of raw data but they relate only to male doctors aged under 65, omitting the higher age groups which interested Burch. Lee circulated the absolute numbers in a TRC memo on 4 December 1980.
When is significant not significant?
A short exchange in the British Medical Journal shed further light on the quality of Doll and Peto’s statistical work. The data from the Doctors’ Study seemed to show that physicians specialising in different areas of medicine had a tendency to die of different diseases. In a new paper, Doll and Peto went through all the combinations of specialism with disease looking for significant associations (a practice known as data dredging). Some letters to the editor followed. Professor Hugh Dudley complained that they had arbitrarily dismissed some associations which came out as statistically significant. Doll and Peto defended themselves, and Dr G. Silverman rebuked Dudley for not seeing that decision makers should not be slaves to statistical methods. A letter from Burch began “Professor Hugh Dudley raises… and Mr R Peto, Sir Richard Doll and Dr G Silverman discuss…. a statistical problem of severe and evidently unresolved difficulty.” In fact the problem was well known and the solution given in textbooks. Doll and Peto had greatly underestimated the likelihood of apparently significant results occurring by chance: a p value (the probability of a freak result) they gave as 0.001 ought in fact to be 0.33. In any case, hypotheses ought to be formed at the beginning of a study and not at the end.
Interlude: Burch in Italy
The letters to Seltzer provide us with another delightful glimpse of Burch combining conference business with sightseeing pleasure, this time in Italy. “Whether or not we shall get there seems to be in the hands of maintenance engineers at Heathrow,” he grumbled (5 April – in Britain those years were punctuated by one strike after another). But (27 April)
Jane and I duly got back from Italy late on Sunday evening having had a splendid time in Rome and Florence. My paper in Rome went well – although I had literally to read it – and three people congratulated me afterwards. I managed to mix Congress with sightseeing in Rome in a reasonably satisfactory ratio although I had to forego some favourite paintings. In Florence we met up with Belinda who was in excellent fettle. We took her to Pisa one glorious day and on another, we took her and her friend to San Gimigniano. These students are hard pressed for money and we were able to give them meals at a level beyond their normal financial means. We revelled in the Italian renaissance, especially the quattrocento, and appreciated the real stature of Ghirlandaio for the first time.
(By “read” is meant “read out word for word”.)
Belinda was 22 in 1977. Burch at that age had been in uniform protecting his country on a serviceman’s pay. Interludes like this, one imagines, were what it had all been for.
Mortality associated with the pill
Burch submitted a letter on this topic to the Lancet partly to establish whether he was subject to a general blacklisting and to publicise his forthcoming paper on coronary heart disease in Gerontology. “Somewhat to my surprise The Lancet published my letter,” he wrote to Seltzer. “At least there is no clampdown on letters critical of official dogma.” The background to this consists of three papers in the Lancet on the health effects of the contraceptive pill: a retrospective study by Valerie Beral and two prospective studies inspired by it, one conducted by Beral for the Royal College of General Practitioners and another by a group from Oxford University. All three established a correlation between use of the pill and cardiovascular disease. The prospective studies both appeared on 8 October 1977, together with an editorial noting the result and that the number of excess deaths was low. “Caution, more or less, should prevail,” was its conclusion.
Letters concerning the issue from P. B. S. Fowler and from Burch were published on 22 October. Fowler referred to a scathing attack on the reports printed in the Daily Express – in his eyes justified in view of the low number of deaths blamed on the pill.
Burch noted with approval the emphasis the RCGP report placed on the problem of self-selection, so often ignored in the case of smoking-related disease. He also had praise for Beral’s retrospective study, which analysed change in death rates from cardiovascular disease in 21 countries over a period of ten years, endorsing her conclusion that excess mortality in “takers” had a causal explanation. His main point is that this causal factor was independent of smoking, which is thereby exonerated of causing cardiovascular disease.
Effects of Cigarette-Smoking on Fetus and Child
This policy statement by the Committee On Environmental Hazards of the American Academy of Pediatrics, in its journal Pediatrics, begins and ends with the slogan “For the Health of Our Children, Please Don’t Smoke.” Figure 1 (there is no Figure 2) rams home the message. It illustrates a placard distributed to members of the academy for them to display in their waiting-rooms, which read
FOR THE HEALTH OF OUR CHILDREN
PLEASE DON’T SMOKE
The statement contains no new observations and appeals (in footnotes) to numerous correlations between smoking and hazards to child development, beginning with low birth weight and ending with teenage smoking. Yerushalmy, Burch and Goldstein get a brief discussion, as do animal studies by a Dane named Astrup of the effects of carbon monoxide which seemingly uphold a causal explanation of these results: a rebuttal of the Danish study by Richard Hickey (citing genetic factors) is also given consideration. The report dismisses non-causal explanations inspired by Yerushalmy as “strained”.
Burch wrote a letter of protest, stressing that Yerushalmy had always regarded his hypotheses as tentative for the very reason that he adhered to the scientific method of hypothesis testing and falsification. Astrup’s study of rabbits, if extrapolated to human populations, predicted that babies developing in the womb should lose 1% of their expected weight when in fact, at 7%, the discrepancy is larger.
Dose-response effect of maternal smoking
This letter from S. M. Garn, H. A. Shaw and K. D. McCabe appeared in Pediatrics in 1979 but may as well be discussed here. It reported on a study of smoking during pregnancy involving 19,000 normal ‘singlings’ (healthy non-multiple births) and identified two significant patterns in their data (black and white subjects were analysed separately). One was a linear dose-response relationship between the rates at which the mothers smoked and the birth-weight of the baby. They also found a similar relationship between smoking rates and haemoglobin levels in the baby’s blood (a new discovery). On this evidence, they pronounced that “with such truly large-sample data now available, the constitutional explanation must now be stretched to the uttermost.”
Since the letter discussed Burch by name he was given the opportunity to reply. He wrote that the constitutional hypothesis also predicted an apparent dose-response relationship of some kind, because heavy smoking mothers were more likely, and light smoking mothers less likely, to be constitutional smokers. The same thing was seen in connection with coronary heart disease, and was very likely of genetic origin. The studies by Donovan and Silverman suggested that the same thing was true of low birth-weight.
Doll at the Royal Statistical Society
Burch published no new research on lung cancer until 1980, but in 1978 he gave a lecture to a meeting of the Royal Statistical Society – an honour, since he was not a professional statistician, and a long-sought opportunity to present his ideas to an audience equipped to understand them. Doll had delivered a lecture to the Society in 1970 on Age Distributions of Cancer: Implications for Carcinogenesis which needs discussion because Burch referred to it when addressing the Society eight years later. Doll’s lecture concerns all forms of cancer, and in remarks on lung cancer he argued for a linear dose-response relationship between smoking and lung cancer. This gave rise to an important question posed by Professor Peter Armitage.
Professor Doll has referred to some of the earlier work which he and I did together 10 or 15 years ago… In this connection I have always been somewhat puzzled about the effect of cigarette smoke as a carcinogen. The dose-response relationship seems to be linear, which suggests that the carcinogen affects the rate of occurrence of critical events at one stage, and one stage only, in the induction period. (If it affected two stages, one might have expected a quadratic relationship and so on.) Does this crucial event happen early or late in the induction period?
The evidence here seems to conflict.
The supposed 30 year lag between rising smoking rates in the 20th century and the rise in lung cancer rates suggests an early stage in the process, but the fall or levelling off of lung cancer rates in ex-smokers suggests a late one.
I should be interested to know whether Professor Doll has considered this anomaly and can resolve it.
Doll replied in writing, having had several weeks to consider the question.
The anomaly to which Professor Armitage refers is, in my opinion, crucial; if we could resolve it we should be much closer to understanding the mechanism of cancer induction than we are now. The suggestion that cigarette smoke might affect two stages in a multi-stage process and that the effect on each stage might vary with the square root of the intensity of exposure is ingenious. I am not aware, however, that it is supported by any set of experimental data with other agents. The only explanation of the anomaly that I can offer is the rather unsatisfactory one that cigarette smoking acts principally on an intermediate stage in the process of cancer induction…
The point is this. Lung cancer rates in non-smokers at different ages of life fall roughly on a curve generated by a function involving age in years raised to the power of an integer, possibly age4 or age5. Plotted on a logarithmic scale, the curve is a straight line rising with age. Smokers are more prone to lung cancer than non-smokers, and their rates of the disease seemed to fit a straight line parallel to the first but higher up. Doll and Peto explained this in terms of what is called a ‘hit’ theory: cancer results when a cell undergoes so many successive damaging changes, possibly mutations, possibly not. Smoking, they held, increases the risk of one of those changes taking place, and with it the risk of lung cancer. Peto, at the Hammersmith seminar, suggested a formula involving
c · d · d · d · d (= cd4)
where c is cigarette consumption and d is decades of smoking, c multiplies the risk of the first hit occurring and hits occur about a decade apart.
The dose-response relationship of smoking rates to lung cancer rates would be linear on the ordinary scale. Armitage’s point is that this formula fits the data, but not the Doll-Hill theory of smoking and lung cancer. If smoking affects the first hit and the last, you would expect the formula to look like this:
c · d · d · d · c · d (= c2d4).
This has two consequences.
1. Plotted logarithmically, the new formula would still be a straight line, but with a steeper slope.
2. The dose-response relationship would be quadratic – a rising curve when plotted on the arithmetic scale and linear when plotted on the logarithmic scale.
Doll and Armitage went back a long way. Both were protégés of Hill, Armitage having joined Hill’s staff first and thus, if anything, senior to Doll professionally, though 12 years his junior by age. Their connection began when Armitage was assigned to subject a study by Doll of peptic ulcer to statistical scrutiny. In papers published in 1954, 1957 and 1961 they proposed a general explanation of cancer (all cancer, not just lung cancer) based on the ‘hit’ theory, which they did much to popularise, though they were not its originators.
Doll acquired his knowledge of statistics when he undertook his first epidemiological study, of peptic ulcer, on a three month course taught by Hill, based on Hill’s elementary textbook Principles of Medical Statistics. When Hill took early retirement in 1961, there was a general shake-up of his unit, in which Armitage succeeded Hill as Professor of Medical Statistics. As the official biography tactfully puts it:
Doll was a metonym for epidemiology not statistics and he liked to have a big-brained statistician close at hand. The role was filled by Malcolm Pike, who came out of a job interview with Armitage, to find Doll waiting in the corridor. “You didn’t get the job. I think you should come and work with me instead.”
He published similar work with Pike, and when Pike moved on in 1972, Richard Peto took on the role. The statistical work on the ‘hit’ theory was undoubtedly performed by Armitage.
These facts are taken from the official biography of Doll, Smoking Kills by Conrad Keating. It is mainly based on Doll’s reminiscences as an old man, which were not always accurate. Information about Hill unmediated by Doll is hard to come by. There is no biography and the Royal Society obituary of Hill was written by Doll.
In asking his tricky question, Armitage was fairly obviously showing his teeth, but he was a personal friend of Doll and admired his approach to medical statistics. Doll had, after all, more or less created the ‘risk factor’ epidemiology in which Armitage had made his career.
Burch at the Royal Statistical Society
Burch’s opportunity to address the society came on 17 May 1978. His lecture was entitled Smoking and Lung Cancer: the Problem of Inferring Cause. The story can be partly told in his own words, as he wrote his own account of it for the non-technical readership of Tabak-Journal called Smoking, lung cancer and scientific debate.
The Royal Statistical Society convenes some of its ordinary meetings in London to a fairly elaborate protocol. First, the Council of the Society chooses a potential speaker and subject, and the Meetings Secretary invites the person so honoured to “submit a paper”. Submission of an invited paper does not, however, guarantee acceptance: the paper undergoes “the normal referring process” by two referees, and “acceptability depends on the reports received”. A paper that survives these hurdles is set up by the printer in galley proof form and despatched to persons known by the Society, or the author, to be interested in the subject matter.
At the Meeting, galley proofs are placed at the entrance to the lecture theatre for the benefit of those attending who have not already availed themselves of copies. The author is not expected to read the paper literally but to follow it “fairly closely” in an oral and more informal version. The speaker’s presentation is followed by contributions from the proposer and seconder. “Their function is not merely to be polite. If they disagree with the paper in any particulars, it is the tradition of the Society that they, like all other speakers, are expected to be uninhibited in saying so.” The seconder is followed by contributions from the floor, including written contributions submitted by persons unable to be present at the meeting. At the end of the discussion the author is expected to make a brief verbal reply. After the meeting remarks of discussants, together with written contributions, are sent to the author, who is expected to prepare a “considered reply” for publication with the paper and the discussion. The procedure is complicated, but provided all deeply interested persons are alerted and can be persuaded to participate, it could be thorough.
The audience included Peter Armitage, who proposed the vote of thanks, as he had done for Doll’s paper, Peter Lee observing for the Tobacco Research Council, and Carl Seltzer, who had come to London for the occasion and made a contribution from the floor. Not present were Doll, Peto or any of their associates. “I am disappointed, as I have been in the past, by the low profile maintained by the staff of the Regius Professor of Medicine at Oxford,” Burch sardonically noted.
As to content, the paper made Burch’s now standard case against the causal theory of smoking and lung cancer. He emphasised statistical inference, dose-response relationships and the statistical distributions associated with them in great detail, while saying little about his biological theories. Some of the material was new, for instance an analysis of Doll and Peto’s dose-response relationship, using data from their 1976 paper on the Doctors’ Study. Also new was a demonstration that the falling rates of lung cancer in British male doctors, which they claimed to demonstrate in that paper, were illusory. Doll and Peto had quoted relative rates with respect to the general population, and when these were converted back into absolute rates, no fall in incidence was seen. Burch also called attention to the latest results from the Swedish twins study, which continued to confirm Burch’s hypothesis (despite Doll and Peto’s insinuation to the contrary in the 1976 paper) and to Armitage’s comments on Doll’s lecture considered above. For readers with the mathematical and statistical grasp of a Burch, this paper is the best exposition of his views, while for most it will not be.
The really interesting material comes in the audience discussion and Burch’s considered responses to it. Some participants were confused and uncomprehending, and Burch easily disposed of them. Most were penetrating and raised points that go to the heart of the matter: Burch considered these at length and answered them in detail. In response to several contributions questioning Burch’s use of the parameter S as a measure of lung cancer incidence, he repeated certain crucial calculations using the more usual standardised death rate (SDR) and showed that his conclusions still stood: an important result.
The words of Dr P. D. Oldham illustrate the scepticism of many statisticians about risk factor epidemiology: “too often the authors of such studies succeed in implying that a tenable hypothesis, developed from a careful epidemiological survey, is a proven one.” He also had some interesting observations on the way the case against cigarettes took hold in the 1950s.
Most marked of all was the immediate adoption by health authorities (but not revenue authorities) of the view that all that was needed in response to the evidence of an association was to utter loud and clear warnings to the general public, which would then immediately cease to smoke and so cease to acquire lung cancer. Indeed, this turned out to be quite effective with the medical profession, who after a brief period of chain-smoking miniature cigars (under the illusion that these were the cigars whose association with lung cancer was not significant in the first studies, and that not significant means unassociated) were able in large numbers to give up smoking completely, I would guess with some relief. Suggestions that studies of how and why cigarettes came to be associated with lung cancer should be carried out, and that the nature of the desire of people to smoke should be investigated, with the idea that a safe form of smoking might be found, an alternative means of satisfying the desire, or at least that scientific knowledge might benefit, were received with scorn and hostility; I can remember an exceedingly uncomfortable luncheon with some professors of epidemiology and social medicine which ended in my being virtually ostracized from their company for being unwise enough to say something of the sort.
A fuller account of the discussion is given here.
Peter Lee was in the audience and submitted a report to the TRC which did not get into the Tobacco Documents, but his report on the subsequent discussion survives, as does one on Smoking, Lung Cancer and Scientific Debate. He reported that Burch had successfully made his case that the parameter S was not essential to the argument, and that absolute rates of lung cancer in doctors had not fallen. However his sympathies were with Armitage and Townsend. He unfavourably contrasted Burch’s reliance on five parameters and neglect of curve-fitting with Doll’s simpler power laws and provision of p-values which fitted numerous data sets very well. A paragraph of observations on the discussion headed “General” is probably fair comment:
Burch in general got quite a lot of support, although a number of the speakers had little special knowledge of smoking and health. That no-one from Oxford bothered to turn up was understandable in some ways but was probably ill-judged. Certainly it would have added entertainment had Richard Peto arrived with guns at the ready! In his absence Burch really came out pretty well.
The Doll curve
Lee, reporting on Burch’s lecture, also complained that Burch did not take into account Doll’s current views on the dose-response relationship between smoking and lung cancer. This was unfair. It is true that Doll, or rather Doll and Peto, rethought the relationship, but their paper Cigarette smoking and bronchial carcinoma: dose and time relationships among regular smokers and lifelong non-smokers appeared early in 1978, too late for Burch to consider it in a lecture delivered in April from a text submitted in draft months before.
The Doll-Peto paper was a long overdue attempt to answer the tricky question posed by Armitage to Doll at the Royal Statistical Society in 1970. As they mildly put it:
Epidemiological data on smoking and bronchial carcinoma are more extensive that for any other cause of human carcinomas, so it may be profitable to ask which stage or process smoking affects most strongly. This has already been done (Doll, 1971), but the epidemiological evidence was difficult to fit together plausibly (Armitage, 1971).
The aim of the paper was to present arguments using data from the Doctors’ Study which “may allow circumvention of the difficulty discussed by Armitage.” Doll and Peto managed to fit a quadratic curve to eight data points extracted from the study. They excluded a ninth, representing the heaviest smokers, who had strikingly low rates of lung cancer compared with predicted values. They had three possible explanations of this inconvenient anomaly.
1 It is possible that heavy smokers may have a different constitution (essentially, different genes) from other groups.
2 It is possible that the men in this group are exaggerating or lying.
3 It is possible that heavy smokers are predominantly inhalers, and inhalation has the ‘paradoxical effect’ of protecting against lung cancer.
These things are indeed possible. A longer version of this section giving more detail about Doll and Peto’s procedure is here.