A Half-life of Burch
6. Preventive medicine
A new friendship
In the last weeks of 1983 a new figure entered the story. Burch received a letter from Ray Thornton of BAT Industries (British American Tobacco) expressing an interest in Burch’s work and requesting a meeting. Thornton was a senior executive in the Research and Development department at Southampton who monitored and funded scientific research. The chairman of BAT Industries, Mr (later Sir) Patrick Sheehy, appointed in 1982, took a keen interest in cigarette scepticism and Thornton had his backing. Burch was, of course, delighted to discuss his views with somebody who wanted to understand them. The two met at Leeds on 11 January 1984 and entered into a correspondence.
Thornton genuinely wanted to understand Burch’s theory and they exchanged letters on the subject for much of 1984. Thornton encountered the usual stumbling blocks, mainly the meaning of the parameters n, r and above all S, hugely variable but apparently supposed to be the proportion of the population with lung cancer genes. Burch answered him patiently and fully, for instance explaining the difference between his equation and a superficially similar one devised by the Danish researcher Knud Juel (who Burch regarded as inconsequential).
The Surgeon General on CHD
Thornton provided Burch with an early copy of the Surgeon General’s report for 1983, which concentrated on heart disease, perhaps hoping for a demolition job to match Burch’s critique of the 1982 report.
Burch responded in January
Its bulk is somewhat intimidating and I hope that the essence of the argument can be extracted without wading through the whole tome.
My initial reaction is one of outrage. We read through page after page of case-control and prospective studies of self-selected smokers, non-smokers and ex-smokers. In the CHD section I waited, with the proverbial bated breath, to see how the SG would deal with the awkward (from his viewpoint) evidence on secular trends, twins studies, randomized trials and the sex- and age-patterns of smoking and CHD. Of course, I should have guessed. He ignores it; he does not attempt to deal with awkward evidence or analysis in any shape or form; anything unfavourable to his dogmatic and absurd conclusions is simply suppressed.
But in July,
I have to admit that I have not given the SG’s report the attention it doubtless deserves. Or, on second thoughts, perhaps I have ?
Visit to BAT at Southampton
On 30 March, Thornton invited Burch to give a talk at the BAT Research Centre in Southampton to an audience of scientists and statisticians, for such expenses as Burch felt able to claim. “In respect of your wish to be independent, could I mention that Mr. Richard Peto visited us at Southampton last June and gave us a talk on various aspects of cancer.” Burch agreed to speak: “I hope that my immortal soul will not be imperilled through acceptance. The most convenient day of the week for me to travel would be a Friday…” The title of the talk, A Unified Theory of Growth and Disease was settled on and Burch delivered it on 1 June. On his return, he wrote a note of thanks which was, perhaps, also a reminder to Thornton that he had a fulfilling life of his own in Leeds:
The journey back was uneventful and the garden party we attended on my return was most entertaining with early music (live in a fascinating garden), an extraordinary display of cameras extending from the first-ever onwards and a couple of videos on the flight of pterodactyls and models of them (our host was – is – a zoologist).
But the hotel and dinner in the New Forest were excellent, especially the claret.
More on Burch’s theory
One last pair of letters in the Burch-Thornton exchange concerned the biological basis of Burch’s theory.
Thornton asked Burch in a letter of 18 July 1984
How do the extraordinary developments in molecular biology that have occurred in recent years fit in with chapters 1-5 of your book, which was written in 1976?
Burch replied on 25 July.
Your serious question is simple enough but, as you clearly appreciate, not easily answered. Once upon a time, between writing the book and starting another, I kept a loose-leaf classified record of papers that impinged on my theory. Although I have compendious and numerous files that I can raid when working on a particular topic I cannot easily select from the many to respond to your question in its most general form.
Burch then discussed some new developments in physiology, emphasising the distinction between mutations to germ cells and somatic cells: his stress on the latter had received support from more recent work and his general theory had not been disproved. Burch was particularly interested in Byers et al. Predilection of lung cancer for the upper lobes which revealed an age-related pattern to the site of lung cancer: such patterns were a prediction of the auto-aggressive theory. “The authors’ discussion is very perceptive; they appreciate the threat their observations pose for orthodox views on causation.”
Thereafter, Thornton became another of Burch’s friends and intellectual confidants with whom he shared personal and scientific news, though Burch occasionally had to keep Thornton’s largesse at arm’s length (“I fear that I shall have to decline the invitation to Wimbledon but look forward to seeing you here on Thursday…”)
(“Wimbledon”: i.e. the tennis championship.)
Philip Morris and tobacco funding
In 1983 and early 1984, Burch was in negotiations with Nicholas Martin of the Department of Human Genetics at Virginia Commonwealth University which illustrate his policy on tobacco funding. Martin wrote to Burch on 21 July 1983.
I have spoken to Walter Nance and Lindon Eaves about the idea that you might spend some time here, probably next summer. They are both keen on the proposal, particularly if we can persuade you to give us a series of informal seminars for our group. Inevitably, however, the ugly subject of money must be raised. Walter thinks we can raise about $1,000 from departmental sources but if you need more than that, we shall have to apply to the University and perhaps also to the tobacco companies. (Richmond is a centre of the tobacco industry.)
Burch apparently knew Nance and Eaves, possibly from their research into twins. Eaves also collaborated with Eysenck on The Causes and Effects of Smoking. He replied on 2 August:
Splendid! I shall be very pleased to spend some time with you all (3-4 weeks?) at the Medical College next summer. Would September be suitable? My wife and I will have to be back in Leeds for the beginning of term in October.
To the ugly subject and what I regard as desirable. I should like my return air fare, living accommodation for two, and a stipend of around $3,000 if that is at all possible.
Martin wrote on 8 February 1984:
I have now heard from Mr. Robert Moore of Philip Morris, Incorporated that they will not grant us the money for your visit, on the pleasantly surprising (and I think genuine) grounds that they would not wish to compromise your scientific integrity in any way.
Mr. Moore also indicated to me that they might be interested in supporting our research into the genetics of smoking. I shall be writing today asking for such support and if we are successful in our request this may free some departmental funds for your visit which are currently otherwise committed. I shall let you know how things progress.
Burch replied on 14 February:
I was pleased to get your letter because I had felt somewhat uneasy about getting financial support from Philip Morris. Evidently Mr. Moore must know something about my published work.
It looks as though I might, with luck, be assisted ‘by other means’; anyway, I hope so.
The Burchs visited the US in March to April 1984, whether at Virginia Commonwealth University or elsewhere is not known.
Philip Morris undoubtedly took an interest in Burch’s work, not least because Seltzer kept them informed about Burch and could have procured industry funding for him. But as he wrote to a contact at Philip Morris in 1983, forwarding work by Burch on cervical cancer:
Please do not contact Burch directly, since we have agreed to maintain a certain relationship which would not affect his credibility.
Normal growth and neoplastic growth
Burch’s last publication on his general theory of autoaggressive disease concerns retinoblastoma, an often hereditary childhood cancer of the eye. (The condition is not smoking-related and the paper is not in the tobacco archive.) Retinoblastoma cannot be discussed without introducing one of the most important figures in modern cancer research, Alfred Knudson, whose publications on the subject in 1971, 1976 and 1986 led to a revolution in the understanding of the molecular biology of cancer. Burch knew of Knudson’s work, naming him in a letter to Seltzer as one of a very few scientists in the USA investigating somatic mutation as a cause of cancer, and referred to him in the present paper.
Retinoblastoma comes in two forms distinguished by three criteria which overlap but do not entirely coincide.
Firstly, most sufferers have close relatives who are also affected by the disease: this is referred to as the familial form. A minority have the “sporadic” form in which this is not so, though they sometimes have more distant relatives who were also affected.
Secondly, the age distribution for retinoblastoma suggests that there are two forms with onset in early and later childhood.
Thirdly, retinoblastoma is either unilateral, affecting a single eye, or bilateral, affecting both eyes.
Knudson deduced from the age distribution that retinoblastoma is caused by two mutations to the same gene in paired chromosomes. In the familial form of the disease one mutation is supposed to occur in a germ-line cell and the other in a somatic cell. What this means is that the first mutation was present in the sperm or the ovum, inherited from a parent. As the embryo forms by division and differentiation of the original fertilised ovum, that mutation is present in every cell of the body. A familial sufferer is therefore born one mutation away from the cancer, which develops when the second mutation occurs in a cell of the eye. Sporadic sufferers are thought to be conceived free of the germ-line mutation, and both mutations occur in a single cell of the eye after it has formed, in a much rarer stroke of ill luck.
Burch held that the familial form of retinoblastoma occurs in a population of predisposed individuals after a single mutation to a single cell (n=1, r=1 in terms of his theory). This accords well with Knudson’s account: the predisposed population have the germ-line mutation and the disease develops after a single somatic mutation. Knudson discovered the specific mutation in tumour cells in 1976 (a deletion on chromosome 13), and Burch mentions the development. The gene was sequenced in 1988 and, under the name Rb, has since been implicated in many other cancers.
Burch believed that he could show that the sporadic form of the disease involved three mutations to a single somatic cell (n=1, r=3), as distinct from Knudson’s two. There were other differences between the two researchers involving the increasing number of somatic cells in the body as childhood progresses and Knudson’s reliance on a ‘hit’ theory in his analysis of the age distribution.
There is no point in adjudicating the matter. Knudson’s views won the day: his name is in all modern textbooks of oncology and Burch’s in none of them. Any re-examination of Burch’s intriguing statistical patterns and his audacious claim to explain cancer in terms of particle decay would have to take modern knowledge as the starting point.
Long term effect of stopping on angina
Seltzer and Burch both contributed letters to the British Medical Journal criticising this study of heart disease by Dr Leslie Daly and others including Risteard Mulcahy, the pioneer of preventive medicine in Ireland. The subjects of the study were 498 men retained from an earlier study in which 555 men who had survived a heart attack by two weeks were traced for two years. In the present study, the survivors, now numbering 498, were tracked for another five years. The results showed that the risk of death during this second period for three classes of subject was:
The researchers laid great emphasis on the fact that the never-smokers were at high risk of CHD for other reasons such as lack of exercise and hypertension. The researchers therefore concluded that stopping smoking lowers the risk of death after surviving a heart attack.
Seltzer’s main point was this.
Antismoking orthodoxy demands that the order of decreasing mortality should be: smokers, ex-smokers, non-smokers, or at least that ex-smokers and non-smokers should be alike. Clearly, the order found in this article – that is, smokers, non-smokers, ex-smokers – does not fit the traditional causation hypothesis.
By what logic does one claim cause and effect for one finding and reject it out of hand for another finding in the same context? The “anomalous position of non-smokers” should have been a red light to the investigators for a reassessment of all the findings within the strictures of the study design.
He also pointed out that the researchers calculated risk afresh from the end of a two year period in which 13% of non-smokers and only 8% of smokers died, and had nothing to say about the severity of the initial attack which all 555 men survived by two weeks, although this is a very important determinant of survival rates.
Daly and his colleagues replied that the non-smokers must have had an abnormally high risk of a heart attack by virtue of the very fact that they had suffered one. It was true that survival rates from a heart attack depend on its severity, but after two years the two survival curves (survival beyond the two-year mark and survival as predicted by initial severity) run in parallel so that only the one curve need be studied.
Burch wrote that the three groups were not comparable at entry into the study. They differed in their initial risk factors and, as Seltzer had pointed out, the severity of the initial attack. It was statistically illegitimate to treat them as a single population and draw conclusions about the single factor of cigarette smoking. The best way was to control for all other variables, and the Whitehall study, which did so, showed negative results. Burch had shown the same thing using a different approach in Ischaemic Heart Disease.
Daly and colleagues replied that they had only been concerned to compare continuing smokers with ex-smokers, and non-smokers were “not a major part of our paper”. The relative risk of 2.4 between the two groups corresponded to no other observed risk factor, and “a strong genetic influence on giving up smoking” was “an unlikely hypothesis”.
A letter to the British Medical Journal concerns a paper of this name by Jay H. Lubin and others. A case control study of 718 smokers or ex-smokers with lung cancer in western Europe and 11, 000 matched controls, also smokers or ex-smokers, showed that those who had not smoked for at least 10 years had half the lung cancer risk of continuing smokers, and the fewer years as a smoker the lower the risk.
Neither the design of the study, the results, nor the conclusion were novel and neither was Burch’s rebuttal. Case control studies ignored the problem of recall bias and the fact that ex-smokers are self-selected. Randomised prospective studies such as Whitehall and MRFIT avoided these biases, and the pooled data from both studies showed no significant difference in lung cancer rates in the intervention and control group.
Diet and ischaemic heart disease
A letter from Burch in the Lancet appeared on 24 September 1983 but was part of a controversy which had been running for years and continued into 1984.
As long ago as 1976, the Royal College of Physicians had issued a report Prevention of Coronary Heart Disease blaming fatty foods for high rates of CHD in Britain and recommending a change in the national diet. However, medical opinion was divided on the matter and government advice to the public about diet had not changed in decades, despite the best efforts of the Coronary Prevention Group, a lobby modelled on Action on Smoking and Health whose members included Dr Keith Ball and Geoffrey Rose (now an adviser to the WHO).
An unsigned Lancet editorial (6 August) acknowledged that the medical profession was divided but took the side of Ball and Rose in calling for change. “Too much should not be read” into the failure of MRFIT, for which there might be many explanations. Those, such as Professor Michael Oliver, who favoured inaction were “nihilists”, there was a presumption in favour of action, and the allegedly suppressed NACNE report should be published. A letter in the same issue from Dr Keith Ball concurred: the gravity of the situation required the publication of the report.
NACNE was the National Advisory Committee on Nutrition Education, created in 1979. Its report, favouring a low-fat diet, had been in preparation since 1981.
Burch’s contribution questioned the claim that a changed diet was the reason for the currently falling rates of CHD in the United States. An analysis of data on whites broken down by age and sex suggested that other factors were at work.
End of static decade?
The NACNE report was also the subject of a guest editorial in the British Medical Journal by Professor A. Stewart Truswell and a letter by Burch disputing its premises and conclusions. Truswell began with a lament: “Mortality from coronary heart disease has been static in Britain since 1974, when the Department of Health published its first report on diet and coronary heart disease.” Alas that document had been “conservative and confusing”. After a more promising report in 1977, “for the first time in Britain, newspaper writers took up the topic of diet and coronary heart disease enthusiastically with the inevitable oversimplifications (like “Eating animal fats may be dangerous to your health”). Unfortunately, “the butter industry reacted with the Butter information Council, and Sir John McMichael started a correspondence opposing the dietary fat hypothesis.” More recently the NACNE report had met a similar fate. However, now at last the Department of Health and Social Security had spoken with one voice.
Doctors should be able to accept the conclusions of the 1984 report into the body of received medical knowledge – “reduced saturated fat helps to prevent CHD” like “not smoking reduces lung diseases.” If dissension and doubt…now diminish then journalists, dietitians and schoolteachers can get on with their work of informing and educating people who have been confused. Food manufacturers and politicians will respond to public consensus and demand and Britain can at last expect a decline in premature deaths from coronary disease.
(Sir John McMichael FRS was an eminent cardiologist, and also the father of A. J. McMichael whose study of oesophageal cancer was discussed above.)
Burch’s letter, backed up by a table of changing death rates by age group, disputed the facts. Mortality from CHD in Britain had indeed been static from 1974 to 1977, but from 1978 to 1983 there were clear signs of falling rates like those seen in the United States ten years earlier. As for saturated fats and other risk factors, Burch’s work in Ischaemic Heart Disease showed that they were not causes of heart disease.
The food scandal
Truswell’s story of avarice in the form of the Butter Information Council at war with inconvenient truth in the form of the Department of Health and Social Security was partly derived from the investigative journalism of Geoffrey Cannon which appeared in New Scientist, the Sunday Times and a book The Food Scandal (1984), co-written with Caroline Walker. A sympathetic review of that book in the Listener magazine by Derek Cooper welcomed its appearance, noting that The Food Programme on BBC radio had covered the story the previous year. This was not coincidental, as the Listener, a cultural and current affairs weekly, was published by the BBC to furnish background to its broadcast output. Cooper also reviewed the “sensible” Diet 2000, a popularisation of the NACNE report with a recommended scheme to help the public eat a healthier diet.
Burch wrote to the Listener on 12 July 1984 protesting that Cooper called the case against saturated fat “irrefutable” when it was anything but. Yet another epidemiological study (of Adventists in California) had revealed “that women, but not men, who eat eggs, and men, but not women, who eat meat have a significantly higher mortality than those who refrain”- and yet again correlation was simply assumed to be causation.
A letter from one Colin W. Brown on 26 July reminded readers that Burch was a smoking sceptic and not to be taken seriously.
Indeed, that ‘association implies causation’ is a logical fallacy, but the standard of proof required for action to be taken on this scientific issue, acknowledged to be one of life and death, is lower than simple theories of science suppose.
Burch replied to this on 9 August. “Dr Brown… contends that we have to adopt a relaxed ‘standard of proof’ where matters of ‘life and death’ are concerned. This deliberate abandonment of scientific method might be justified on the grounds of prudence if we had no alternative.” However, Burch’s has subjected the causal theory to rigorous tests and found it wanting.
Anyone who troubles to work through all this (and there is plenty more like it) will surely be struck how little original thought there is in the world and how much depends on who can get organisations like the BBC to whistle their tune. One of the secrets is to persuade people like Geoffrey Cannon, undoubtedly an honest man and, as befits a journalist, properly suspicious of vested interests, not so much that truth is on your side as that deception and corruption are on that of your opponents.
Controversy with Kristein
Marvin M. Kristein was an economist specialising in health care costs. In policy papers such as Wanted: smoking policies in the workplace he variously estimated the costs of smoking to the US economy as $36 billion per year and “$16 plus zillion a year” (this latter in testimony to Congress). His paper 40 Years of US Cigarette Smoking purported to refute the conclusion of Burch’s Smoking and Mortality in England and Wales by a computer-assisted calculation. An Acknowledgement, in very small print at the end of Kristein’s paper, informs us that the computer calculations were performed by L. Mihalko. A controversy followed, published in three issues of the Journal of Chronic Disease.
|Kristein:||40 years of US cigarette smoking|
|Burch:||Cigarette smoking and mortality rates|
|Hickey:||Cigarette smoking, cancer, statistics and judgement|
By the end of Burch’s second contribution it is clear that Mihalko had done all the mathematical work and that Kristein had no idea what calculation he had performed. Kristein refused to admit defeat and threw an intellectual tantrum, accusing Burch of “exonerating smoking” by “a series of sophisticated abuses of scholarship” including “ignoring logic and evidence”. Burch left it at that, but Richard Hickey submitted a letter rebuking Kristein for his resort to personal attack and rehearsing the case against the Surgeon General’s methodology. He contrasted that methodology with Berkson’s invocation of Bertrand Russell: “Morally, a philosopher who uses his professional competence for anything except a disinterested search for truth is guilty of a kind of treachery.”
In a Response to Hickey, Kristein claimed to be motivated by the same lofty spirit, “viz. using professional competence except for finding the truth is a kind of treachery. Also, the Talmud tells us that consciously teaching an error is a sin.” This is interesting. Judaism is indeed a scholarly religion. It also attaches huge importance to the preservation of life, and gives considerable weight to the judgement of a doctor on medical matters. Can that be the reason why Kristein regards the pronouncements of the Surgeon General as unchallengeable and unarguable? There were, of course, Jews on both sides of the debate on smoking and health.
A fuller version of this episode discussing the mathematics is here.
Lee, though no more sympathetic to Burch’s views than before, was under no illusion as to the quality of Kristein’s work. In a report to the TRC he wrote
I described the Burch paper as ‘silly’. In many ways this is even sillier.
This is incredibly facile and betrays that he has just not read Burch’s papers.
The rest of Kristein’s reply is not really of interest. It is already clear that he has neither the knowledge nor ability to refute Burch’s position, although he no doubt thinks he has done so.
Early in 1985, two minor letters making familiar points about self-selection, Whitehall and MRFIT appeared in New Scientist (7 February) and the British Medical Journal (2 March).
Fag End referred to a New Scientist feature article, The Control of Lung Cancer by Richard Peto and Richard Doll, a popularised version of a Keynote Address of the same name delivered by Peto (“in collaboration with Sir Richard Doll”) to the International Lung Cancer Update Conference at New Orleans in March 1983. Peto and Doll rehearsed their standard arguments for the causal theory and called for lung cancer to be reduced by increased taxation, advertising restrictions, tar limits and propaganda aimed at governments (this was now to be a separate matter from propaganda aimed at the general public). The most interesting sentence in the Keynote Address (not found in the New Scientist version) is this (p. 8): “Partly because of difficulties of dosimetry, it is not really known whether, as Doll and Peto have tentatively suggested, a doubling of the true dose rate produces an approximately fourfold increase in the age-specific effect, or whether, as is suggested by much other data, it merely produces a twofold increase.” The quadratic dose-response relationship was on the way out.
Burch wrote to Thornton (14 February): “Do you see the New Scientist? In case you don’t a photocopy of recent letters is enclosed. When despatching mine I was quite confident that it would not be published because recent editorial policy has been fanatically anti-smoking. Imagine my surprise on finding two letters critical of Doll and Peto in the issue of 7 February.” And on 6 March: “Doll and Peto have declined my New Scientist challenge. They must detest these randomized intervention trials and their wretched results.”
The World Cancer Burden
The World Cancer Burden refers to a guest editorial of that name in the British Medical Journal by C. S. Muir and D. M. Parkin. These two had done useful work assembling data on the global incidence of various cancers from national reporting. On the basis of The Causes of Cancer by Doll and Peto, they pronounced that “as much as 80-90% of human cancer is determined environmentally” and that “tobacco is the most important single aetiological factor in cancer”. Burch wrote a brief letter about self-selection, Whitehall and MRFIT.
Burch wrote to Thornton (6 March):
Having rather abandoned hope that the BMJ would publish my letter, ‘The world cancer burden’, I was delighted to see it yesterday (copy enclosed). The editors probably sent it to Muir and Parkin with an invitation to respond. It is not wholly surprising that no response has come. Perhaps ‘other experts’ will be braver? Perhaps not.
Non-antipodean smoking view
A long-running war of words in the Medical Journal of Australia, to which Burch made a late contribution, illustrated the increasingly politicised nature of the anti-smoking movement as its leadership passed to a new generation of professional activists. The context was the narrow failure of a bill to ban tobacco advertising in Western Australia, with similar legislation pending elsewhere in the country. The issue for 14 April 1984 published two papers relating to this setback for the campaign against smoking, and a leading article commenting on the papers.
Not biting the hand that feeds you, by Simon Chapman BA, later Professor Simon Chapman AO, sociologist and tobacco control advocate, analysed press coverage of the proposal in an attempt to discover whether the newspapers had given equal space to both sides in the debate. It seems that they did not: 47% of items published were negative, 36% positive and 17% neutral. Letter writing was dominated by ‘health and medical professionals’ on the positive side and “five main writers from the tobacco and advertising industries” on the other. The press was apparently biased by advertising revenues.
Learning from legislative disasters by Allan Peachment discussed what went wrong and what was to be done. The anti-smoking campaign was hampered by its “lack of control and influence over the main thrust of the debate,” a problem for which Peachment had three remedies; quasi-experimentation, social audits and science courts. (Details of these mechanisms for manufacturing opinion are given in the paper.)
The editor of the Medical Journal of Australia, Alister Brass, did not share these views. His piece, written in a jovial, knockabout style, defended smoking as a fact of life against “crusaders,” “authority figures,” “muesli eaters,” a “jihad” and “self-righteous wowserism”. “After all, this is 1984!”
The issue for 26 May found Brass printing a two minutes hate in the letters department, led by Simon Chapman and the director of the National Heart Foundation of Australia, R. L. Hodge. The National Heart Foundation had arranged for the editorial to be given national newspaper coverage.
“Contemptible,” “embarrassingly trite,” “British,” “feather-brained,” “Quo vadis, Dr Brass?” “Rejection of almost everything that medicine stands for,” “intellectual poverty,” “superficiality and insensitivity.” “Disappointed.” “Glib and philosophical.” “Implacably against it.” “From which cave shambled,” “blinkered,” “dinosaurian.” “Et tu Brute!” “Thurberesque,” “well-worn aphorisms.” “Infamous.” “Disgusted.”
The correspondence continued for weeks.
Brass was up against what, a few years later, would come to be known as political correctness, and had mistakenly supposed that it was “a middle- and upper-class phenomenon, another fashion”. Not a bit of it. The group Billboard Using Graffitists Against Unhealthy Products (BUGAUP), who he had dismissed as “merry pranksters getting a kick out of vandalising billboards” were in fact “serious-minded professionals, angry and frustrated at the immorality of the cigarette message”, or so the president, chairman and secretary of the Australian Council on Smoking and Health informed him in a joint letter. He was lucky to keep his job: there were reports of a “confrontation” between him and the Journal’s board of directors, and of at least two calls for his resignation. He claimed to have received substantial private support from doctors who feared to speak publicly.
On 4 February 1984, Burch poked the hornet’s nest. He had not published in Australia before, and may have been alerted to the row by Thornton, who was there in early 1984. He sent Brass what he described to Thornton as “a letter that, I hope, might provoke some response among irate readers”. His point was, again, that activists took the dangers of smoking for granted but Whitehall, MRFIT and Burch’s own investigations had discovered no evidence for this.
Letters challenging Burch appeared from Konrad Jamrozik (1 April) and J. P. Pierce (13 May). Jamrozik restated the conventional case against smoking with reference to the latest reports. His tone was civil, that of Pierce a lot less so. Pierce’s degrees were in Physiology and Pharmacology, Clinical Epidemiology and Biostatistics, Public Administration, Psychology and Communication Research (1971, 1975, 1977, 1981, 1981). To a man of his generation, smoking always had been dangerous and epidemiology always had been risk factor epidemiology. His published papers applied the risk factor methodology to health education, demonstrating that things like “health beliefs” and “social influences” were factors in giving up smoking, to be manipulated in the name of public health.
Pierce revived Lilienfeld’s charges of scientific malpractice by Burch and cited one fact from the Whitehall study and one from MRFIT which he claimed to be supportive of the causal hypothesis. On 19 August the Journal printed letters from Carl Seltzer and J. R. Johnstone (author of The Scientific Scandal of Antismoking) in support of Burch. Seltzer noted in connection with heart disease, and Johnstone for lung cancer and general mortality, that neither of Pierce’s facts was statistically significant. Then Burch and Pierce had it out.
On 30 September, Burch restated his approach to causation and, in answer to Pierce, drew attention to these numbers
|−43% to +18%||the confidence limits for Pierce’s facts about Whitehall (unimpressively wide)|
|−46% to +61%|
|40.2 per 1000||the death rates in the two MRFIT groups (astonishingly close)|
|40.4 per 1000|
Pierce’s answer to this is crucial. After repeating Lilienfeld’s charges he returned to the randomised studies and the use to be made of them. The studies furnished positive evidence against smoking if you compare smokers and non-smokers (“subgroups” other than the randomised groups), whereas the apparent negative results ought to be rejected because the trials lacked “power” to detect positive results.
The last two letters in the exchange appeared on 14 April 1986. Burch emphasised that he practised orthodox scientific method, that Pierce had put words into his mouth and that Pierce’s talk of “subgroup analysis” defeated the whole purpose of randomisation. As for Lilienfeld, Burch had answered him at length. Pierce closed the correspondence, saying that he sympathised with Kristein’s view of Burch. He condemned the use of an alleged $360 million on randomised trials “that predictably showed no result”.
This is revealing. Pierce was an angry man who seriously believed that Burch’s motive was “to exonerate smoking”. To him, as to Lilienfeld, a negative result was not a result, and Burch’s falsificationist approach to science an obstacle to the activism which for him was the only conceivable purpose of medical research.
Burch’s work on oesophageal cancer reached print as a major paper in the Journal of Chronic Disease in late 1984: he sent Thornton an offprint at the start of 1985 – “good news (in part) to start your New Year”.
Oesophageal cancer was known to be correlated with smoking, drinking and urbanisation, but these three factors are also correlated with each other, and after controlling for the presence of one in the presence of another, no clear pattern emerged. Burch’s approach was to examine death rates over the twentieth century broken down by age groups, looking for long term trends. He established that there was falling incidence of the disease in younger age groups to the 1950s followed by a rise in most age groups from 1960.
This pattern bears no resemblance to trends in smoking but some to drinking. Rates of oesophageal cancer tracked falling alcohol consumption on a 30 year time lag to the 1950s and rise with rising consumption from 1960 but, puzzlingly, with a zero time lag. One possible explanation is that alcohol precipitates the cancer on top of an unidentified infection, perhaps more prevalent in cities, which became less common during the first half of the century. Burch therefore set up a precipitator hypothesis and proceeded to test it. (A precipitator acts to produce disease at the end of a long process: it is the proverbial last straw which breaks the camel’s back.) As a matter of scientific logic, it is very similar to the hypothesis concerning smoking and lung cancer which he had been testing: an agent (cigarettes/alcohol) precipitates (lung/oesophageal) cancer on a short time lag in a subset of the population with certain genes which mutate in accordance with the Weibull distribution, the whole pattern being obscured by other trends (misdiagnosis/the mystery infection).
Burch first estimated alcohol consumption in each age group from scanty official data with dry asides about problems like the discrepancy between self-reported drinking and Customs and Excise records. He then tested how closely changes in consumption fitted changing levels of disease. His hypothesis that alcohol precipitates the cancer withstood his attempt to falsify it and he provisionally accepted it. He stated his tentative conclusion and finished thus:
For a lifelong non-smoker who happens to be a light to moderate drinker these conclusions are not entirely welcome. Nevertheless, in choosing between the pleasures of many forms of alcohol and the (hypothetical) extra risk of esophageal cancer, I have no hesitation in opting for the former.
Oesophageal cancer: Burch’s reasoning
The mathematics of Burch’s reasoning is relatively simple and may as well be discussed here. He constructed an equation
RA = R0 (1 + αA)
relating the risk of cancer to grams of alcohol A multiplied by a constant α, where R0 is the risk in teetotallers and RA the risk from A grams of alcohol. Changes in αA would then show up as changes in the proportion of people affected by the disease (changes in the apparent size of the constant S). For a certain initial estimate of α he had a good fit to his British data and similar data on US whites. This disconfirmed the null hypothesis that alcohol made no difference, and on the evidence then available to him Burch accepted the alternate hypothesis that it does.
In a report to the TRC, Peter Lee identified four main weaknesses of the paper: the data on alcohol consumption, a neglect of pipe and cigar smoking, the strategy of analysing recent (as opposed to lifetime) exposure to alcohol and tobacco, and the fact that Burch treated the two separately rather than taking “a combined alcohol and tobacco index”. Burch’s conclusions were therefore “premature”.
Burch returned to the Listener in 1985. The occasion was a column by the author, journalist and communist Mat Coward in the Langham Diary slot, offering jocular observations on life’s oddities, including the recent ban on the sale of cigarettes in his local hospital and the relegation of smoking to special areas.
Cigarettes may or may not be bad for you: all I’m saying is that, after all the errors medical theorists have made over the centuries, you’d think they’d be a little more cautious… they have always believed, at any given moment, that the sum of the present knowledge equals the sum of all knowledge available.
Professor M. P. Haggard objected to this levity on 9 May, but Burch (30 May) concurred. Coward’s jibe was true: medical scientists proceed as if everything is already known. Whitehall and MRFIT had not been widely publicised, but they cast serious doubt on Haggard’s certainties. Burch had made the same point in numerous outlets, but this is his most accessible statement of it. He cannot not often have found himself in agreement with Coward, who wrote for the Morning Star, but he commented to Thornton (15 June):
BAT should feel grateful to Mat Coward. His words will be more influential than yours or mine.
Heart of the matter
An article in The Times (1 October 1985) by Dr James Lefanu discussed the introduction of the NACNE guidelines on diet, finally published the previous day. Lefanu noted the role of Ancel Keys in the campaign against dietary fat and a recent radio interview in which Keys said that that “the critics either had ‘no knowledge whatever’ of the issues or had ‘other fish to fry’”. British medical opinion was actually divided on the value of the new diet and that much depended on the value of epidemiological studies as against the negative experimental evidence furnished by studies like MRFIT. The guidelines represented a new departure in other ways. “Armed with the new proposals, medical authority can stretch beyond its legitimate confines to touch the essential culture”.
A short letter from Burch agreed with Lefanu, noting that Japan had recently overtakn Sweden as the country with the longest lifespan, having done nothing in the way of modifying lifestyle to deserve it.
A study by Samuel M. Lasko and others compared women with endometrial cancer and a control group of otherwise similar women with other forms of cancer. The results showed a negative association of endometrial cancer with smoking, progressively more negative in ex-smokers, current smokers and postmenopausal current smokers. However, the finding did not have “direct public health importance since cigarettes, overall, have serious deleterious effects”. Similar results had previously been attained by other researchers including Noel S. Weiss, who contributed an editorial to the same issue. He alleged that smoking by postmenopausal women would cause 30 times the number of deaths it prevented.
A letter from Gunnar R. Nordenstam questioned whether the two groups of women studied were adequately matched. One from Burch noted that both Lasko and Weiss simply assumed that negative correlation is causation of a negative and had ignored the possibility of a third factor affecting both rates of the cancer and smoking levels.
Meeting with Sheehy
Thornton was cultivating Burch as part of a brief to investigate scepticism about the health effects of smoking and how BAT might put it to use. He reported to Eric Bruell, the chairman of British American Tobacco, which was the main holding of the conglomerate BAT Industries under the chairmanship of Patrick Sheehy. Two of his projects at this time were a compendium of alternative views on epidemiology and a projected academic conference on the subject. Discussion on the matter took him to the US, Australia and many other destinations during this period.
Burch, now in active retirement, also loved travel and would update Thornton and compare notes with him. “We leave tomorrow for the U.S. by the slow route. I trust you enjoyed Concord[e].” (21 March). “I am off to Italy after lunch.” (25 June). “I have popped into the office this morning but leave for Scotland before lunch. We are spending a few days there with my daughter.” (9 September).
Bruell and Sheehy both took a keen interest in Thornton’s work and Thornton set up a meeting between them and Burch for 24 April, sending Burch a draft of his epidemiological compendium as the proposed topic of discussion. Thornton briefed Bruell and Sheehy on Burch, and accompanied them to Leeds in Sheehy’s private aircraft for two and a half hours of discussion over lunch in a private room at Queens Hotel.
Burch seems to have liked the idea of a conference more than the compendium, which he subjected to a critique in a letter of 9 May. What would be its readership and its intellectual level? His main point was the obvious one that few people would be receptive to smoking scepticism from any quarter, let alone an executive in the tobacco industry. In any case,
The bulk of epidemiology is rubbish. How can you persuade people to believe you when Fisher, Yerushalmy, etc. all failed?
He and Thornton met again on 21 May, and then Burch seems to have forgotten all about the compendium. It was a confidential document, and Thornton had to request its return via the secure delivery company Securicor.
Burch knew perfectly well how useful he could be to BAT, but he wanted to be heard and saw that BAT could also be useful to him. About this time, an American official called Karl Kronebusch had drafted a report on the economic costs of smoking and circulated it to various parties for comment including, bizarrely, Philip Burch, who received it on 2o May, the day before his meeting with Thornton. Thornton reported that he “had evidently formed a very poor view of the document, not to say contemptuous, and was planning various forms of action in relation to it”.
Kronebusch worked for the Office of Technology Assessment, a body formed to advise the Congress of the United States on scientific matters. It was preparing the report at the request of the Health Subcommittee of the House Committee on Ways and Means. The report reflected the conclusions of a workshop on 18 April (attended by, among others, Jay H. Lubin) and followed a simple methodology pioneered by, among others, Marvin Kristein. For every disease commoner in smokers than non-smokers, the total illness and death in smokers, minus the total in non-smokers, was labelled “attributable risk” and multiplied by the alleged cost to the economy in dollars of one case of that disease.
Although Kronebusch’s request reached Leeds after the deadline for submissions, Burch replied (29 May) challenging the assumptions behind the calculation, both medical and economic such as “the costs to society of an increased lifespan”. He ended “When I was a young man… selective citation was generally regarded as a crime. Does the end now justify the means?” Coming from Burch, language like this was a sign of outrage, and he had decided on his own initiative to circulate his reply to sympathetic members of Congress. He sought Thornton’s advice on who to approach and Thornton, after taking soundings, suggested Senator Jesse Helms. However, on reflection, Burch decided to stay his hand until the final version of the Kronebusch report.
This appeared in September and was widely publicised. Hickey sent Burch a clipping from the New York Times (“Smokers’ Ills Cost Billions, U.S. Says”, 16 September). Kronebusch sent Burch a copy on 30 October with the same form letter requesting comment, though he had not replied to Burch’s letter 29 May, and Burch obliged with what he described to Hickey as “my fairly testy reply”. He focussed on the increasing tendency to conflate attempts to influence social policy and the procedures whereby fact is established, with a concomitant tendency to ignore negative results. Burch quotes a paper by P. C. Elwood and J. E. Gallacher on leaded petrol which is worth reading in full for its insights into this trend. He ended “I feel bound to issue a protest in the hope that someone, somewhere, might take notice and reflect on these matters.” Finally, in November, Burch was sent yet another copy of the report by Judith Wagner, Project Director. To Hickey he wrote “I must say that the OTA is persistent. Yesterday, I received another copy of their fictional exercise”. To Wagner he sent a letter covering the same ground again and ending “How do you, as Project Director, justify such practices?” No reply ever came.