Last Words

A Half-life of Burch

8. Last Words

Tribute to Eysenck

Smoking and Health was Burch’s contribution to Hans Eysenck: Consensus and Controversy, a symposium on all aspects of Eysenck’s work. “It is a glossy publication of 421 pages – and therefore expensive – but it should not be beyond the resources of BAT. (Purchases will not benefit me; I have already received a princely £50.)” Burch told Thornton.

Eysenck, like Burch, had come to the smoking issue from outside the field of medicine. What they had in common was exceptional intelligence, an independent cast of mind, a grounding in a rigorous science, a mastery of statistics and a commitment to scientific method. Under the heading Heresy and its Foundations, Burch identified the smoking heretics – Fisher, Berkson, Yerushalmy, Eysenck and more recently Feinstein – as upholding the falsificationist principles of Popper against the soft science of risk-factor epidemiology.

In a section Eysenck’s Perspective, Burch identified a difference between that perspective and his own. Eysenck dismissed the “large-scale epidemiological studies” as against proposed small-scale experimental studies of small samples of smokers and non-smokers of different kinds. Burch thought that epidemiology had been corrupted and politicised but might yet become a hard science and in the rest of the paper he explained how this might be achieved. Much of the material overlaps with his critique of the Surgeon General’s reports, though some of it is new. A long section on twins presented the latest Swedish data and contested the widely held view that monozygotic twins are genetically identical. Burch had evidence that sometimes they are not, a fact which he explained in terms of his autoaggresssive theory – with implications for the origin of breast cancer and schizophrenia.

A Coda reads, in part

Finally, I should like to pay tribute to Professor Hans Eysenck for his long and unswerving devotion to the scientific investigation of human behaviour. A deep distrust of genetic explanations for behaviour and disease permeates many supposedly scientific disciplines as well as the laity; to maintain a rigorous approach in defiance of widespread hostility calls for courage and perseverance, qualities that Hans possesses in abundance.


Resolution of the American Society for Cancer Research

Smoking and lung cancer: an overview was a “position paper” presented to a meeting of the American Association for Cancer Research in May 1984: that meeting unanimously adopted a “position statement” identifying smoking as the major preventable cause of human lung cancer and calling for six actions to be taken against it including advertising bans and restrictions on smoking in public places. Its five authors were responsible for its four sections.

Carcinogenesis Lawrence A. Loeb, John Abbots
Epidemiology Virginia L. Ernster
Ecomomics of smoking and lung cancer Kenneth E. Warner
Clinical aspects of lung cancer John Laszlo

The report contained nothing very new, though the space accorded to the economics and treatment of the disease reflect the increasing tendency for social science to dominate the anti-smoking movement (Warner attended the workshop which drafted the Kronebusch report). Burch remarked to Hickey:

[I]f the members of the audience at the annual meeting in Houston unanimously accepted the content of the Loeb et al. paper then it must be correct. It is a democratic society is it not? Suppose there had been one dissenter. Do you think he would have been torn limb from limb?

Anyway, it will be interesting to see how Loeb et al. reply to my criticism in Cancer Research.

He had submitted a letter making familiar points about the direction of causality, twentieth century trends in smoking and cancer, and about MRFIT and Whitehall (here he discusses the anomalous rates of cancer other than lung cancer in the intervention and control groups). The issue in which it appeared contained two separate replies to him by Loeb and Ernster.

Concerning rates of change, Loeb wrote “Contrary to Burch’s interpretation, a replot of his data on a linear scale shows more definitely that from 1930 to 1950, lung cancer rates among men rose more rapidly than rates among women.” And on MRFIT: “Whether those who quit smoking in these somewhat inconclusive studies were “self-selected” or not, the decreased mortality rates among quitters lend further support for the idea that smoking cessation is beneficial to health.” Ernster made these identical points in different words.

Both replies are angry in tone. Loeb concludes:

The scientific debate on smoking or health was over long ago, and the tobacco industry lost. Discussion should now turn to the best methods of reducing the preventable epidemic of smoking-related diseases.

Ernster’s last words are:

In the face of the evidence available in the mid-1980s, Burch’s stance is more than academic folly. It is a disservice to the public health.

Burch commented to Hickey:

You will be interested, I am sure, to see what Cancer Research has finally decided to publish about the effort by Loeb et al. (They received my first letter in January, 1985 and publication is now scheduled for May 1986. I trust that readers of the journal will survive the shock.)

I find the replies fascinating. The authors do not appreciate the distinction between absolute and proportionate changes. When a randomized trial fails to produce the “right” result what do you do? You break the randomization – a procedure that was initiated by the authors of MRFIT, poor souls. It is in a way comforting to discover that these anti-smoking fanatics are also incompetent as scientists.


The Ghost of Yerushalmy

Burch returned to the question of smoking during pregnancy as part of a correspondence in the Journal of the American Medical Association about a paper from 1984 by Mary Sexton and others concerning a randomised, prospective study in which some pregnant women were strongly advised to give up smoking and others were not. It was the second such study to be conducted: a similar one from Britain by J. W. Donovan had appeared in 1977. There was an important difference between the two. The Donovan study estimated maternal smoking rates by self-reporting while the Sexton study used a saliva test. The Donovan study detected no significant difference in the birth weight of the babies born to the two groups of women, whereas the Sexton study found that women who gave up smoking had heavier babies and that the difference was significant. Since the Donovan study challenged orthodoxy, it was inevitably balanced by a paper by Harvey Goldstein, in which he claimed once more to have answered Yerushalmy in 1973.

Sexton accordingly claimed to have falsified the conclusions of Donovan and Yerushalmy. An editorial by Richard H. Aubry concurred. “Surely this article, in context with the remaining literature on the subject, leads even the most careful scientist to conclude that smoking while pregnant is unhealthy…”

A letter by Stuart M. Berman and Carol J. R. Hogue (8 March 1985) lamented “how difficult it is, as observed by Goldstein, to nail the coffin closed on the ghost of Yerushalmy”. Berman and Hogue questioned various aspects of the Sexton trial. In particular, the saliva test used (thiocyanate) was not an accurate measure of smoking rates. Sexton’s own data showed lower levels of the marker in both groups at 8 months, though the controls reported higher smoking levels. Sexton replied that the conclusion was “strengthened, surely not weakened” by the “compatibility” of the assay and self-reporting.

A letter from Burch (28 February 1986) noted some differences between the Donovan and Sexton studies. He held that the ideal study would randomise before conception, and pointed out that if the samples were combined, the negative conclusion of Donovan still stood. Was the saliva test conducted blind? Sexton replied that the differences between the two studies were discussed in her paper, and that the weighing of the babies was performed blind.


Period and cohort trends in mortality from cancers of the uterus

A short but important paper concerned a new development in Burch’s theory of cancer.

Uterine cancer falls into two diagnostic categories, cancer of the corpus and cancer of the cervix, and Burch identified an early and a late onset form of the latter from its age distribution. He fitted all three types of uterine cancer to Weibull curves having the same values of his parameters n and r, but different values of S and k.

The main part of the paper distinguishes between two sorts of trend over time: those affecting women born around the same year (cohort trends) and those apparent in all age groups during a particular period of time (period trends). There is a clear period trend in Burch’s data, falling rates of the disease from 1921 to 1982. As Burch noted, this trend bears no relation to smoking rates in women, which rose for most of this period, although uterine cancer is commoner in smokers. On top of this, Burch identified two cohort trends: relatively low rates of the disease, at all ages, in the generations born around 1911 and 1936.

Burch explained the period trend as evidence that a precipitating agent, probably an infection, had become less common over the decades. However, the cohort trends could not be explained in that way. He formed a hypothesis that they were the result of an autoaggressive disease which struck in the womb and altered the genetic makeup of the zygote or early embryo, so that different generations really did have different values of the parameter S. He had seen this pattern in no other cancer and formed a prediction by which his hypothesis could be tested (though he never assembled the data to do so): rates of early onset cervical cancer should be correlated with mother’s age at birth.


National trends in mortality from ischaemic heart disease: implications for prevention

Burch returned to the subject of heart disease in a letter to the Lancet in May 1986. Professor A. G. Shaper had started a correspondence on a paper by Thomas J. Thom and others in the International Journal of Epidemiology. That paper, Shaper claimed, showed that countries with high rates of ischaemic heart disease, such as the USA and Britain, must do far more to reduce the consumption of saturated fats and cholesterol. A figure from Thom illustrated the magnitude of the task, and another from a paper by H. Malmros allegedly proved the role of diet in heart disease. Malmros had found rising levels of IHD in the USA during World War II as against a temporary fall in Sweden, Finland and Norway. The reason could only be the wartime diet, though England and Wales (not considered by Malmros) had only seen a levelling off, not a fall, “despite the rationing acutely remembered as deprivation by the older generation”. Shaper concluded: “The implication of these observations is that the prevention of IHD is a very difficult task, requiring considerable effort and dedication from everyone concerned, including governments. It cannot be left to “take its natural course” in the hope that market forces (i.e. public opinion) will produce sufficient change at an acceptable rate. It is clear from the wartime data and the 1950-78 trends that a positive effort of considerable magnitude is necessary.”

Thom’s paper was an essay in descriptive statistics illustrated by a figure. His data consisted of male and female rates of heart disease in 26 countries at six points in time from 1950 to 1978. The most glaring feature of the figure reproduced was lower rates of heart disease in women than men in every country studied. The second most glaring was that the relative ranking of countries was almost unchanged over the 28-year period.

Alexander McNair and M. F. Oliver (26 April) both pointed out that Shaper had simply ignored these patterns in Thom’s data. McNair noted that Shaper simply assumed that diet was the causal factor, but the overall pattern made that very unlikely.

Burch (17 May) criticised Shaper from a different angle. Shaper assumed that Thom’s data referred to coronary (ischaemic) heart disease, but in fact it referred to a broader category of heart disease chosen to avoid problems arising from the change in classification between ICD7 and ICD8. Burch had used a different approach to this problem in his two main papers on heart disease and his results suggested that the cause was a precipitating infection, not diet.


The Big Kill

A new exercise in government anti-smoking propaganda appeared late in 1985. The Big Kill was a joint publication of the Health Education Council and the British Medical Association consisting of 15 volumes of about 100 pages each concerning the alleged death toll from smoking in England and Wales. A grand total of 77, 774 such deaths (55, 107 men and 22, 667 women) is broken down into regional and local totals with absurd precision: for instance, 286 of the yearly toll are alleged to live in Cheltenham. The document is obviously computer-generated – then a technological novelty – from actual reported deaths multiplied by percentage estimates of the proportion attributable to smoking.

Burch was, of course, incensed. Can Epidemiology Become a Rigorous Science: How Big is the Big Kill? is a critique of the report and a restatement of Burch’s view of the correct approach to epidemiology. “The biologically ignorant but numerate reader might be forgiven for concluding that epidemiology is not only a rigorous science but an incredibly accurate one, with an implied error in mortality estimates of less than 1 part in 77 774. The scientifically trained person can be expected to ask: How were these impressive figures calculated?”

Burch summarised the methodology of The Big Kill. The proportion of deaths attributable to smoking for each disease was extrapolated from the Doctors’ Study of doctors aged 35 to 64, except that attributable risk for lung cancer in women was raised from 40% to 80% “on the authority of Sir Richard Doll”. He then proceeded to say much that he had said before, for instance in his critique of the Surgeon General’s reports, this time more concisely and punchily (he described the paper to Lee as “my excursus into ridicule”) with some shrewd observations on the capacity of the medical profession to decide the matter. Some quotations.

The process of reaching sound conclusions about causation is, however, more of a scientific than a medical task. Medical skills are required, of course, to reach an accurate diagnosis of the cause of death and a proper appreciation of limitations in the evidence, but analysis of the resulting statistics calls for familiarity and dexterity with scientific logic. The two skills are not incompatible but they are not always combined in the same person.

By transforming a rather difficult scientific problem into a simple and essentially moral issue… the crusaders have disarmed many a potential critic. What bright but ambitious young epidemiologist would dare to challenge the establishment? It will be, and has been, contended that in matters of life and death some relaxation of scientific rigour is justifiable; opposition to orthodoxy, with the taint of sophistication, can be represented as obscurantism.

I met a director of a medical research unit (not an epidemiologist, I hasten to add) who seemed incapable of understanding that association does not necessarily imply causation.

Finally, do critical essays like the present one server any useful purpose? Past experience indicates that they are largely ineffectual and change few opinions; but if the condition of freedom is eternal vigilance then the delineated standards of science surely call out for occasional recapitulation.

Lee’s verdict

Lee submitted a report on How Big is the Big Kill to the TRC. He began “I agree wholeheartedly that claims that 77,774 people are killed annually in England and Wales from heart disease, lung cancer and bronchitis/emphysema due to smoking imply an order of accuracy that it is totally impossible to defend on a scientific basis.” He found much to agree with in the paper, but also had some specific points of disagreement. Burch had no reason except extrapolation to suppose that smokers aged over 80 should be less prone to heart disease than non-smokers; that Burch relied exclusively on Passey and Herrold in his argument about the age of onset of lung cancer in light and heavy smokers; and that his discussion of the relation of smoking to all causes of death concentrated on current, rather than lifetime, exposure to smoke. As for the constitutional hypothesis, “Smokers differ from non-smokers in many respects, but not to the extent required.”


Approach to Nigel Lawson

Burch sent a copy of How Big is the Big Kill? to the Chancellor of the Exchequer, Nigel Lawson, later Lord Lawson.

Opportunities to cut Government expenditure are doubtless welcome to you. Mr. Digby Anderson in The Times of November 4 makes reference to a book, “Smoking and Society: A More Balanced Assessment”, which questions “…whether smoking causes lung cancer, coronary heart disease and other medical problems…” The enclosed offprint also deals with those issues and, in particular, directs criticism to the antics of the Health Education Council.

Unexpectedly, the Health Education Council was axed (only to be replaced by the Health Education Authority). Among its last acts were a publicity campaign (“Now it’s official”) announcing that passive smoking causes lung cancer and the publication of a report The Health Divide, blaming the Thatcher government for an increasing gap between the health of rich and poor. Burch wrote to Thornton:

My Big Kill paper has had an obviously lethal effect on the Health Education Council – I don’t think. Government ministers cannot be that wise! Nevertheless, the punishment has fitted the crime even though both might have complex origins.

The Treasury passed Burch’s letter on to the Department of Health and Social Security. A woman signing her name as K. James wrote an official reply. She conceded the spurious accuracy of the figures in The Big Kill, explaining that the compilers wanted subtotals to add up and to “emphasise in some way the substantial magnitude of the effects of smoking”. The general case for this was that the relationship between smoking and disease satisfied Bradford Hill’s criteria for identifying a correlation as a cause.

Naturally, Burch replied to her.

Thank you for your letter in which you make a brave attempt to defend the scientifically indefensible. I appreciate that you have been asked to present an “official” apologia for “The Big Kill” and that the views you put forward do not necessarily represent your personal convictions.

His theme was that smoking-related illness did not satisfy the Hill criteria, which were in any case not Popperian science. He also sent James’s letter and his reply to Lawson.

I fear that your DHSS colleagues seem unable to provide a convincing defence of the Health Education Council’s fiction, “The Big Kill”. This comes as no surprise to me; I have not yet met anyone who can achieve that miracle.

Burch was a mischief maker, but he was also willing to take pains with people like James and Lawson who knew enough to be worth trying to persuade.

Burch was corresponding with Lee at this time, and they discussed The Big Kill. Lee told him, “I agree with you that no reliable estimate can be made, although I suspect my estimate would be quite a bit higher than yours. I enjoyed reading it and agree with a lot of the points you made, though not all.” Burch replied “I am pleased that you enjoyed my ridicule of The Big Kill – an easy target. It is more important, however, that I should ponder disagreements (I wish to “get it right”) and if you can spare the time to jot down your criticisms of my arguments I should be so much obliged.” They exchanged a couple of letters, both making sound points on specific issues. However, Lee was never going to be won over by Burch: he simply found it too implausible that a third factor could cause both smoking and disease.



Whither preventive medicine?

An opinion piece in Alvan Feinstein’s Journal of Chronic Disease brought insight and foresight to bear on questions about the prevention of disease, the ageing population, the end and beginning of life, and the relationship of science to ethics. It is as relevant today as it was in the 1980s. Burch observed the increasing tendency for medicine to ally itself with moral crusades.

If people would only refrain from self-indulgence over smoking, abandon diets rich in eggs, salt, butter, cream and animal fats and rouse themselves from a slovenly indolence to jogging, then most of those 153, 250 deaths could be avoided – or so we are often led to believe. The story appeals to that strand of puritanism, however slender, in each one of us…

Burch preferred to be undisturbed in his enjoyment of burgundy, claret and butter (he had “a passion for… spreading the delicious stuff good and thick”). But larger issues were also at stake, for instance that the elimination of one disease would increase mortality from others perhaps more unpleasant, or result in a senility not worth having. The fundamental issue was the relationship of science to policy making. He quoted E. H. Ahrens, who had written a paper on heart disease which Burch admired.

I believe that as scientists we are expected by the public to render scientifically sound advice. Policy-makers must come to their own conclusions, and will do so for a complex of reasons – political, social and economic. That is their affair; ours is to be sound, as sound as current evidence permits, stating clearly where the gaps in knowledge exist.

This contrasted with the relationship of the Surgeon General’s scientific methodology and the campaign against smoking. According to Burch, “We must hope that the policy makers will be wise enough to pay as much attention to the failures of scientists as to the self-proclaimed “successes” of the pseudo-scientists.”


Period and cohort trends for mortality and cigarette consumption

The last of the major papers on smoking and health continues the approach of Smoking and Mortality in England and Wales. It considers trends in smoking and mortality from all causes in the population as a whole regardless of smoking status. The difference is that the earlier paper analysed trends in men and women separately whereas the later one concerned trends in the difference between men and women. The main data set is laid out in two tables.

Table 1 presents death rates in men and women aged 35 to 84 in the years 1946 to 1980 broken down into seven five-year periods and ten five-year age groups: that is, the male and female death rates at age 35-39 in 1956-50 and on up to rates at age 80-84 in 1976-80, displayed in a seven-by-ten table. The ratios between males and female rates in each cell of the table give 70 data points relating to mortality.

Table 2 presents smoking rates in men and women in those age groups in the years 1941 to 1975 broken down in the same way: thus, smoking rates five years before death for each of the 70 subgroups, again expressed as male-female ratios giving 70 data points relating to smoking.

Burch’s aim was a further test of the hypothesis that smoking causes disease on a linear dose-response relationship, precipitating death on a five-year time lag (he also investigated a 10 and 20 year time lag). If this was true then the data points in the first table ought to be strongly correlated with the corresponding data points in the second, low values in one table corresponding to low values in the other and high values to high. Plotted against each other as a graph the points ought to rise in a straight line from bottom left to top right.

Burch plotted the data in figure 2 (four panels represent four variants of the hypothesis). The lines linking the data points all started at top left and wobbled down towards bottom right. There was no correlation to be seen between differential smoking rates and death rates in men and women. Male and female smoking rates converged over the period, while the life span increased in both sexes. Women tended to live longer than men, and the tendency was most pronounced in the middle of the period studied. Burch had already shown that rising life expectancy in men and women could be explained almost entirely by the near-eradication of tuberculosis. This further anomaly, on Burch’s analysis, could be explained by rates of bronchitis in the first decade of the twentieth century.

(Peter Lee, in a report to the Tobacco Research Council, was in broad agreement with this. “The conclusion must be, as Burch says, that smoking is not as major a determinant of overall mortality trends as some would have us believe.”)

Some quotations from the final section, Conclusions, follow.

Either the fatal consequences of smoking are fortuitously concealed by counter factors or few if any consequences exist to be hidden.

A null effect is notoriously difficult (if not impossible) to establish in the empirical domain.

From the outset this general field of inquiry has suffered polarization. Certain statisticians (including Fisher, Berkson, Brownlee and Yerushalmy) have stressed the potential importance of constitutional factors, problems of investigative bias, and the phenomenon of self-selection, but they found the evidence inadequate (as I do) for definitive conclusions about the lethal effects of smoking. In contrast, certain epidemiologists and their many evangelistic followers have had little or even no hesitation, it appears, in identifying positive and more or less reproducible associations with causation. In my judgement, the statisticians have sought to follow rigorous procedures but the complexity of the problem and the limitations in the evidence have prevented them from defining the relative contributions of intrinsic and extrinsic factors to observed associations. Anti-smoking evangelists, on the other hand, have converted a difficult scientific problem into a simple moral issue. I can only regret and deplore their current triumph.

These were his last words on the subject. The paper was published posthumously.


Illness and death

The end came suddenly and far too soon. On 25 February 1987 Burch wrote optimistically to Thornton:

Now I must mention a slight cloud on my personal horizon. This afternoon I learned from the Professor of Surgery that I have carcinoma of the rectum. Some surveys, as you know, show a negative association between smoking and colorectal cancers and so I am paying the consequences of not smoking! As, no doubt, the HEC would be eager to tell me. But to paraphrase my hero, R. A. Fisher, those of us who refuse the jump from positive association to causation will not be tempted to infer prophylaxis from negative association. Needless to say, I have looked up the pertinent survival statistics and I find that two distinctive groups are involved, one with a very poor prognosis (half-life for relative survival of 1 year) and the other with a very good prognosis (average survival equal to that of members of the population without colorectal cancer). I become a patient on Monday (2 March) for operation on Thursday; by the end of March I should be fit again because I have reason to believe that I am a member of the good prognosis group.

Given reasonable luck, Hamburg should be welcome and enjoyable. Jane is very keen on the trip and we would hope to spend the week-end in Hamburg before the meeting.

But on 25 March, it was Jane Burch who wrote to Thornton:

I am afraid that I must write you a very sad letter. Philip simply failed to recover from his operation, and died last Sunday*.

He was found to have metastases in the liver, and the surgeon decided to operate on the liver as well as the rectum. It was evidently too ambitious: after good initial progress there were progressively worse setbacks, and Philip became weaker and weaker.

Finally he haemorrhaged, and died peacefully after saying goodbye to us.

I am only thankful that he had no long illness, no pain apart from post-operative discomfort, and remained absolutely himself all through. He was optimistic and forward-looking until the day before he died: we thought that he was recovering, albeit slowly and with difficulty. But finally he knew he would not recover, and characteristically refused to allow an emergency operation. He felt that surgeons were driven to keep on intervening however remote the chances of success, and he wanted to die peacefully and in command of himself, rather than while being rushed about on a stretcher or on the operating table. So he had his wish, and we had a calm and loving parting before he slipped away.

Perhaps I have written too much – but I thought you would want to know what had happened to him.

*22 March 1987



Jane Burch

Jane Burch was the mainstay of Burch’s life, and he of hers. When he died, her life fell apart and she worked hard to rebuild it. She renewed old friendships and threw herself into her work at the Leeds Biochemistry Department. She continued the country walks they had both loved, now with the University Walking Club, and travelled to increasingly adventurous destinations including Nepal, Malawi and Cambodia, sometimes with her daughter Belinda. She wrote to a friend, “as time goes along one becomes accustomed to one’s new circumstances and ceases to expect all the things one found it so hard to be without”. The first of four grandchildren was born in 1988. When she retired from university teaching in 1991 she embarked on a second career volunteering at the Leeds Citizens Advice Bureau. She was active up to the last six months of her life, when her health failed. She died on 31 July 2010.


The selected papers

Part of Jane Burch’s coming to terms with widowhood was her work on Philip’s selected scientific papers. The editor was to have been Geoffrey Burwell, whose collaboration with Burch in the 1960s engendered the theory of autoaggressive disease. (A bid for the editorship by Theodor Sterling was swiftly quashed.) However, Burwell dropped out because of other commitments and the major editorial burden fell on Jane and Dr Susan Chesters of the Department of Medical Physics. They had the support of the Department and of BAT. Thornton was involved in the project throughout and procured them a subsidy from the company. (On a visit to Leeds, he was pleased and surprised to hear the acting Head of Department say that he and his colleagues thought Burch was “on to something”.) The book appeared from Leeds University Press in 1989 as The Biological Basis of Disease, with introductory material by Burwell, Alvan Feinstein and F. W. Spiers. (Frederick Spiers, Emeritus Professor of Medical Physics, was a departmental colleague of Burch who understood his work on radiation.)



Litigation against the tobacco industry began as early as 1954 with the unsuccessful suit of Eva Cooper against R. J. Reynolds: however, the industry only suffered its first defeat in 1988 when the Liggett Group was ordered to pay $400,000 to the widower of Rose Cipollone, a smoker who died of lung cancer in 1984. In the 1990s, the anti-smoking movement had increasing success with a strategy of litigation for punitive damages against the tobacco industry in the USA, and the industry was forced to accept the “Master Settlement Agreement” with state attorneys in 1998. One result of the settlement was the abolition of the industry-funded Tobacco Institute and Council for Tobacco Research, and another was the creation of the online Truth Tobacco Archive used in these pages. Sir Richard Doll, who had stood apart from the publicity campaigns against smoking, appeared as an expert witness for the prosecution in several of these trials.


Kenneth MacRae

Jane Burch, writing to Ray Thornton about her editorial labours, lamented “the zero impact that Philip’s work has apparently had… I suppose I should be used to it – he certainly was.” As an estimate of Burch’s influence on oncology this is undoubtedly accurate, but there were those who kept alive the approach to epidemiology which he learned from Fisher and Yerushalmy.

One of them was Kenneth MacRae (1942-2002), who Burch met in Belfast in 1976 and identified as “obviously a disciple”. Ken MacRae, a bon viveur and keen foe of political correctness, was, successively, a lecturer, reader and professor of medical statistics at Queen’s University, Belfast, Charing Cross Hospital Medical School and the University of Surrey. He testified as an expert witness for the defence in tobacco litigation in the United States, and in the first European anti-tobacco suit, the protracted Aho case in Finland. (That case concerned laryngeal cancer, but MacRae was also a sceptic about smoking and lung cancer: he cited The Biology of Cancer in at least one of his papers.) He should have appeared for the defence in the first British lawsuit, McTear – much of the defence case in that trial bears his stamp and, indirectly, that of Burch – but for his sudden death, aged 60, in 2002.



Alfred ‘Jock’ McTear, an unimpressive character with a criminal record and a patchy employment history, was a cigarette smoker for all his adult life. Shortly after he was diagnosed with lung cancer in 1992, he heard that ASH Scotland (Action on Smoking and Health) were looking for test cases in which lung cancer victims would sue the tobacco companies and brought a suit against Imperial Tobacco Ltd which his widow, Margaret, continued after his death in 1993. At one point in the process, a spokesman for ASH was alleged to have said “we just need one breakthrough, we just need one victory. […] We just have to win one case to win everything”. After many delays, the case came to trial in 2005 at the Court of Session, the highest civil court in Scotland, and was heard by Lord Nimmo Smith, sitting without a jury, as is the rule in Scottish civil cases.

The prosecution brought five expert witnesses: Dr Sheila McCarroll, Professor James Friend, Professor Gerard Hastings, Dr Keith Kerr and Sir Richard Doll. McCarroll was McTear’s doctor, Hastings an expert on tobacco marketing and Kerr a pathologist who examined McTear’s biopsy. Friend was an oncologist who presented the orthodox case that smoking causes lung cancer, and Sir Richard Doll was – Sir Richard Doll.

There were six expert witnesses for the defence, all smoking sceptics. They were Professor Jeffrey Gray, Dr Derek James, Professor Jeffrey Idle, Dr Arnold Cohen, Dr Michael Lewis and Professor Charles Platz. Gray cast doubt on the addictiveness of nicotine, James brought a mass of evidence to bear on the historical unreliability of lung cancer diagnosis, Idle, a biochemist, described competing theories of the causes of cancer, Cohen discussed the problems of animal experimentation and Lewis questioned the scientific basis of risk factor epidemiology. Platz, a pathologist, had also examined the biopsy: Platz and Kerr agreed on a diagnosis of squamous cell carcinoma of the bronchi.

The written judgment makes heavy reading but the sections reporting the testimony of the expert witnesses contain a wealth of interesting, documented facts. The form in which it reports what the witnesses said needs to be decoded slightly. A third person sentence like “Sir Richard was honestly not sure whether Feinstein was a sincere critic” represents a question and answer:

Counsel: Was he (Feinstein) a sincere critic?

Sir Richard: I’m honestly not sure.

Most interesting of all is Lord Nimmo Smith’s account of Sir Richard Doll and the quality of his testimony.


Sir Richard Doll and the close of his system

Some extracts from the judgment follow.

Professor Sir Richard Doll, aged 91, had not provided a CV and reference was made instead to his entry in Who’s Who.

“Do you know who I am?” might have worked on a jury, but it did not impress a senior judge, who also noted that Doll was the only witness “not to avail himself of my invitation to be seated while giving evidence”.

The prosecution evidence largely relied on a lecture given by Doll at Green College, Oxford in 1997, which recounted the story of Doll’s main studies of smoking and lung cancer and how medical opinion came round to his views. The prosecution took him through that lecture and Doll testified that he agreed with quotations from his own words. He made little attempt to explain how he arrived at his conclusions or what his facts and reasoning were.

The defence then took him through a long list of scientists who had disputed his findings and Doll gave his opinion of them. At his advanced age, he understandably had trouble remembering names unless prompted.

Sir Ronald Fisher?

He [Fisher] had done some genetic work but no epidemiological work.

He described Fisher as an “ignorant geneticist” in expressing this extreme view which other geneticists would not have taken.

In relation to smoking, Fisher was an ignorant geneticist.

Joseph Berkson?

He described Berkson’s views as “silly nonsense”.

Hans Eysenck?

He said that he was familiar with the name. He [Eysenck] was the psychiatrist Sir Richard had referred to earlier. He died in 1977. He did a lot of research for the tobacco industry. He was on Sir Richard’s list of eccentrics; indeed he would go further and come straight out, he was a liar.

(In fact Eysenck died in 1997.)

Philip Burch?

Sir Richard agreed that Professor Philip Burch was one of the people he described as “eccentrics”.

What Burch wrote should be regarded with some care. He wrote “a terrible lot”, and some of it was “nonsense”.

Burch’s exchange with Doll in New Scientist?

Sir Richard said that by this time he had not responded to Burch because there was no controversy with him, he just made wild statements which people ignored. Of course it was quite untrue that Fisher’s position had not been disproved, it had been by 1974. Sir Richard did not enter into debates with Burch. He had ignored him intentionally.

Carl Seltzer?

He said that Seltzer was one of “a rather peculiar group of people who took some odd views”. He did not accept the causal hypothesis, and a lot of other things as well. Sir Richard was unable to recall what had led him to describe Seltzer as one of a rather peculiar group of people, but he confirmed that he was one of the people he had previously described as “eccentrics”… There was still a Flat Earthers Society at that time.

K. A. Brownlee and his review of the Surgeon General’s report?

Sir Richard accepted that Brownlee was someone who, in 1965, was not prepared just to accept the causal hypothesis, and who turned to Fisher and apparently took his views seriously.

Counsel then turned to Burch’s lecture to the Royal Statistical Society in 1978 and the subsequent discussion.

G. A. Stern?

Sir Richard said that this was quite untrue… Sir Richard said that he did not agree with any of this.

H. Gwynne Jones?

Sir Richard said that he rejected this, it was just nonsense.

Alvan Feinstein?

Sir Richard described [him] as an eminent but peculiar scientist who tended to have a different view to anybody else on any subject. He was well known for this.

He [Doll] knew him quite well and had spoken in his department, but had difficulty making up his mind about him. “He liked to do things because it was naughty, I think.”

Richard Passey?

Passey challenged the causal hypothesis, but Sir Richard regarded the challenge as extremely fallacious.

Doll dismissed the views of other sceptics in similar terms.

Lord Nimmo Smith was not impressed.

No attempt was made by him to explain why it was that they were wrong to disagree with him. They were all, on the face of it, people of some degree of professional standing who had put up reasoned objections to his conclusions. A reasoned response would have served to show why they were wrong. Yet Sir Richard relied before me principally on ad hominem arguments of a kind which is surely unacceptable in rational academic debate… I can find little in these passages beyond assertions that those who disagreed with Sir Richard were wrong, coupled from time to time with epithets which I quite frankly found it unbecoming for a man of his stature to have chosen to use.

The judge had similar things to say about the evidence of Professor Friend and, to a lesser extent Professor Hastings. The evidence of Professor Idle, on the other hand, was the kind of thing he had been expecting.

Professor Idle’s research did not appear to me to have omitted reference to any publication (apart from the epidemiological literature and the reports and other publications based on it) which might have served to undermine his conclusions, nor did it appear to me that he misrepresented the substance of any publication to which he did refer. His own conclusions seem to me to have been fairly and objectively based on the literature he had considered… What his evidence demonstrates is not simply that there is no basis in the literature relating to any branch of scientific inquiry other than epidemiology for holding that cigarette smoking can cause lung cancer, but that an enormous amount of effort, resources and ingenuity have been devoted to the pursuit of scientific inquiry on this issue, with essentially a negative result.

For these reasons, and others concerning the legal requirements for proof, Lord Nimmo Smith dismissed Mrs McTear’s suit.

The Scottish Minister for Health, Andy Kerr, announced that the judgment would not halt the impending ban on smoking in public places and explained why. “Intuitively, we all know smoking’s bad for you.”


The End