In January 1935, Sir Gilbert Barling, chairman of the Birmingham branch of the British Empire Cancer Campaign (now the Cancer Research Campaign), having secured £1,000 for the purpose, wrote to the Medical Research Council asking who, so endowed, might investigate ‘the asserted marked increase in cancer of the lung’ by looking into the statistics, both in the UK and abroad, conducting experimental work on atmospheric deposits, examining the effects of traffic, surgery, better diagnosis and other relevant factors. Professor Sir Edward Mellanby, Secretary of the MRC, replied that £1,000 was surely too little for so wide a survey – and the survey too wide for any one person. Sir Gilbert’s discouraged response, the last item on this thin file at the Public Record Office, dwells on his own age and infirmity, and the project seems to have been dropped.[1]
Nearly seven years later, a letter reached the MRC from a Wallasey general practitioner, Lennox Johnston, who had ‘for about twelve years . . . been engaged on an investigation into the effects of smoking on health . . . This work has cost me over £10,000 and nearly my life with nicotine’ (which he had experimented with, injecting himself and others with up to 1/8th grain). With the nation at war and in the conviction that his findings were of importance to the war effort, he had prepared a script for a broadcast talk, but the BBC had rejected it. Not, as the file repeatedly reveals, his own best advocate, Johnston had written of smoking as an addiction protected by a taboo analogous to that preventing discussion of sex and had pointed out that, while ‘Adolf Hitler neither smokes nor drinks’, Winston Churchill’s ‘cigar represents, I suggest, a symbol for our nation. It is a symbol of luxury and drugs’. If the nation quit smoking and drinking, he foresaw a ‘gigantic National renaissance’ leading to ‘increased factory output and . . . increased fighting efficiency’: ‘by conquering our weakness we should become unconquerable’.
The thesis he put to the MRC, however, was percipient. Smokers smoked for the effect of the nicotine, but their health suffered through the effect of the smoke on their lungs. Men smoked ten times as much as women, and their death rates after the age of 40 from ‘practically all the respiratory diseases, including phthisis and respiratory cancer, were more than double the female . . . There is no doubt that this drug [viz, nicotine] could be administered by a means less damaging to the respiratory system and unpleasant to non-smokers and yet with equal satisfaction to smokers, e.g., by atomizer, losenge etc.’
He sought support for a comparative study of time lost at work by matched groups of smokers and non-smokers. Apparently ignored by the MRC (despite two follow-up letters, one not for the last time denying that he was a crank), he wrote again in October 1942 proposing comparative studies of smokers and non-smokers in the Army and saying that the Director-General, Army Medical Services had agreed to cooperate if the MRC approved. He added, doubtless not helping his case, that he had had a paper on ‘Tobacco Smoking and Respiratory Disease’ rejected by the editor of the British Medical Journal which he ‘resented because the editor of the B.M.J. is a cigarette inhaler and one of the most important points in my paper is that on issues relating to this drug, the judgment of the addict is unconsciously disturbed.’
The MRC, in the person of Dr Arthur Landsborough Thomson, its Principal Assistant Secretary, sought an explanation of the Army’s backing from the Director of Medical Research at the War Office, Brigadier F A E Crew, commenting: ‘I may say frankly that we are not impressed by Dr Johnston’s proposals’. Brig. Crew replied: ‘A.M.D. [Army Medical Department] would be glad to cooperate if the M.R.C. approves the investigation. I should be very surprised indeed to learn, however, that such approval is given.’ It was apparently their policy always to agree to proposed enquiries in principle and leave others – like the MRC – to turn them down. This, after a meeting in January 1943 with Mellanby at which Johnston became excited and angry, they duly did; and despite several self-exculpatory letters from him, thereafter he received nothing but formal replies until 1948, by which time his renewed proposals were rejected since they had (he was told in confidence) been overtaken by work the MRC had commissioned from Professor Austin Bradford Hill and Dr Richard Doll. In this letter of rejection, Dr F H K Green, the MRC assistant secretary, wrote that there was already plenty of evidence of the deleterious effects of smoking and added prophetically ‘Even if it were possible to prove a correlation between smoking and coronary thrombosis – or, in your own investigation, smoking and cancer of the lung – I do not think the smoking public would pay much heed to it’.
Johnston replied (ad hominem as always that he inferred from this comment that Green was a smoker, to which Green replied lamely: ‘I hardly smoke at all’.Johnston had a short piece published in The Lancet[2], picked a quarrel with Bradford Hill, falsely accusing him of keeping him off a television programme[3] on smoking in January 1953, obtained application forms for a research grant from the MRC in August that year, but then disappears from the record.[4]
When Johnston was told in October 1948 of the work by Austin Bradford Hill and Richard Doll it was already almost two-and-a-half years since the first moves that led to that momentous collaboration. In May 1946 R J R Farrow, Deputy Registrar-General, wrote to J E Pater, an assistant secretary at the Ministry of Health, about moves within the British Empire Cancer Campaign to set up an investigation of cancer of the lung. These originated with Dr Percy Stocks, who was Chief Medical Statistician at the General Register Office from 1933 to 1950 but in this instance seems to have been working through the BECC group in Chester and North Wales. He had won the backing of Malcolm Donaldson, chairman of the statistical committee of the Radium Commission. Farrow wrote: ‘In the view of Stocks the continual increase in deaths attributed to cancer of the lung, about 8% every year, is a serious matter which ought to be studied with a view to deciding the vexed question whether it can really be attributed to more X-ray examinations, which is the facile explanation, or, if not, what is causing it.’
He suggested that an investigation sponsored by the MRC or the Minister’s Medical Advisory Committee would be preferable to one by ‘a more or less private committee’ as proposed.The Ministry took the views of Sir Ernest Rock Carling, its highly distinguished adviser on all matters related to cancer and radiotherapy[5], and Lord Amulree, one of its medical officers, and then asked the MRC to take on the work. In June, Sir Edward Mellanby wrote ‘I am in favour of a statistical investigation . . . Would it not be possible for Stocks to get in touch with Bradford Hill and arrange for an investigation of this kind to be made by the Statistical Committee of the Medical Research Council? It might be interesting to know not only the relative incidence of cancer of the lung in different trades, but also whether there is any difference between town and country dwellers, and whether smoking, especially cigarette smoking, is of any importance, etc . .’, although, as Dr Green of the MRC wrote to Bradford Hill two weeks later, ‘Mellanby . . . is doubtful whether the information obtained would throw much light on the question of aetiology.’ Agreeing to Green’s suggestion of a conference, Bradford Hill pointed out: ‘The much greater rise in the death rate of men compared with women suggests that X-ray examinations are not the sole answer though no doubt contributory.’
The informal MRC conference on the lung cancer problem was held on 6 February 1947. Among the thirteen people present were Stocks (who took the chair), Amulree, Rock Carling, Donaldson, Bradford Hill, Mellanby, Professor Alexander Haddow and Professor and Mrs E L Kennaway. In a ten-page brief for the meeting, Stocks wrote of the increase in lung cancer being greater in towns than rural areas, greater in the north than the south, and greater in areas of less rather than more sunshine. ‘The only adequate explanations of these facts which can be suggested are that either smokiness or pollution of the atmosphere in certain towns is an important causative factor for lung cancer or else that sunshine is an important preventive factor. The former explanation receives support from the finding that occupations connected with coal gas production, certain dusty trades and tobacco, have lung cancer mortalities above the average.’
A table showing no correlation between tobacco consumption and incidence of lung cancer in a number of countries was annotated that the data were not reliable.Professor Kennaway observed that ‘the absence of class differences in lung cancer rates suggested that the environmental causative factor was to be found in the atmosphere or in some other influence to which all sections of the population were subject’, adding (ignoring the problem of delayed reaction that was so often to confuse the question) that if coal smoke pollution, which had been on the decrease, were postulated, the increase in incidence of lung cancer could not be explained. He ‘thought it unlikely that sunlight had a direct protective action’ as coal miners had low rates of lung cancer, although sunlight might ‘act indirectly by destroying a carcinogenic substance – e.g., benzpyrene – in the atmosphere’ but he ‘suggested that smoking, particularly of cigarettes, might be responsible’.[6]
Other theories put forward included arsenic in cigarette smoke, radioactive dusts, and drugs The outcome of the conference was that Professors Haddow and Kennaway should continue to investigate benzpyrene in cigarette smoke and arsenic, radioactivity and benzpyrene in atmospheric pollution samples collected by the Department of Scientific and Industrial Research, while a further meeting would be held to plan ‘a large scale statistical study of the past smoking habits of those with cancer of the lung, and of two control groups, one consisting of patients with cancer of the stomach and the other of diabetics.’
The conference minute adds that ‘Sir Edward Mellanby said that the Council would probably be prepared to pay the salary of a whole-time worker for this study.’[7] By summer, however, Mellanby was suggesting that, partly for reasons of economy, the work should be assigned to the MRC’s new Social Medicine Research Unit rather than to Bradford Hill’s statistical unit. Hill replied vigorously rejecting the suggestion and writing of his and Stocks’ being ‘intensely interested’ and of there being no economies by doing the work elsewhere. As to ‘medical supervision . . . I shall have a Rockefeller Scholar in my department who might be used on it at little cost. A much better alternative would be Richard Doll who has been employed by the Council in the survey of peptic ulcer (sic) in industry and which is coming to an end I believe. I do know he is interested in cancer of the lung and I regard him as a very good worker to whom it is well worth while giving a wider experience in medical statistical work with an eye to the future. As you know, the number of medical persons who take at all kindly to careful statistical work is still small . . .’
Mellanby gave grudging consent to the Statistical Unit doing the work: ‘It seems to me that one of the main reasons against your Department carrying out the work is the extra medical cost involved, but if you can get your Rockefeller scholar to do this, that point is no longer worth consideration.’
[8] Bradford Hill had won his main point and played for time on the question of a medical appointment.Bradford Hill, Kennaway[9] and Stocks now prepared a draft protocol[10] for the planned retrospective investigation. This opened by identifying ‘the main problem at issue’ as the ‘possible association’ between lung cancer and tobacco smoking but said ‘it would clearly be unwise to limit attention entirely to smoking’. It proposed administering a questionnaire to lung cancer patients notified by cooperating hospitals. To ensure uniformity and completeness questionnaires would need to be filled in by specially appointed ‘well qualified social workers’ who would interview the patients, and to save travel costs the Greater London area was proposed. Control groups of patients with stomach cancer and patients with diseases other than cancer would be included in the study – the latter being picked by the social workers to match their lung cancer patients for sex and ‘approximately, if possible, . . . age – subject, of course to the patient being in a fit state to be interviewed’. Estimates on the basis of 1942 statistics (the latest available) suggested that two staff would be sufficient, each conducting seven 30-minute interviews per day.
The draft questionnaire sought not only a medical history and detailed history of the patients’ smoking but also their occupational history and places of residence since birth, together with the forms of heating in the workroom and living room respectively and the proximity of the nearest gas works to their homes. There were questions also about how often before the war they ate fried fish ‘cooked at home’ or ‘from shop’, or ‘fried bacon, sausage, ham, cooked at home’, and ‘Women patients only: fat generally used for frying pre-war’.
These proposals were discussed at a second MRC conference[11] on 29 September 1947 where Bradford Hill took the chair and the Ministry of Health was represented (but Amulree, Rock Carling, Haddow and Stocks were absent). The conference suggested extending the control group to include patients with cancer of the colon and rectum as well as the stomach and duodenum ‘to prevent any bias due to products of tobacco being swallowed with the saliva’ and made other proposals, including extra questions (‘the amount of time spent out of doors during work and leisure’ is annotated in manuscript ‘Impossible’).
They also, doubtless at Bradford Hill’s instigation, recorded their view that ‘the investigation would necessitate at least part-time direction by a medical man’, and formally recommended the proposals to the MRC.
At this stage Bradford Hill wrote to Dr Martin Ware, the MRC administrative officer, suggesting that Richard Doll be shown the previous papers so that he could comment on the paper to be put to the Council with the final proposal: ‘I have already talked to him tentatively and I know he is interested. If we are going to put his name forward as a whole or part time worker on it we ought to know what he thinks.’ Ware agreed, and the formal paper[12] to the Council on the study included ‘the appointment from January 1st, 1948 and, in the first place, for one year of W R S Doll, MD MRCP, to organise and direct the investigation at a part-time salary of £600 per annum’, along with two full-time social workers at £350 each to interview the patients, and a grant towards expenses of £300 per annum.
The study now commenced. It was not happening in a vacuum. The economy was struggling: the winter of 1947 had been one of the bitterest on record, rationing was becoming tighter, and dollar imports – such as 88% of the tobacco consumed in Britain – were tightly restricted. The April 1947 budget increased the price of cigarettes by a shilling – about 40-45% – leading to protests in Parliament about the hardship caused to pensioners. A Treasury official recorded the next day: ‘The Chancellor [Hugh Dalton] has been thinking over the question of a possible concession on tobacco to old age pensioners in the light of yesterday’s debate, and he has come to the pretty firm conclusion that a special arrangement of some kind must be devised . . .’ After much debate tokens worth 2s 0d a week (10 cigarettes cost about 1s 8d at the time) were issued to pensioners who declared themselves to be habitual users of tobacco or snuff[13] – a concession terminated only in 1958[14].
All purchases of US tobacco were stopped on 22 October. In December 1947, Sir Edward Bridges, Permanent Secretary at the Treasury, wrote to Sir John Woods, his opposite number at the Board of Trade: ‘The Chancellor [Sir Stafford Cripps] has given instructions that we should consider what action should be taken in the April Budget in the light of the probability that we shall not be able to afford any purchases of tobacco at all next year.’ Woods’ reply included a manuscript postscript: ‘The probability that we shall not be able to afford tobacco is one thing . . . the probability of our having to take dollar tobacco on a Marshall [Aid] string is another!’ (The point was borne out by a later minute: ‘The UK is the only really important market for the Virginia type of tobacco that we smoke and the speed with which the American Government stepped in to purchase the 1948 crop once we had withdrawn from the market is an indication of their nervousness about the internal effects of a cessation of British demand.’)
A working party began considering the alternatives of rationing or ‘a further steep increase in the tobacco duty’. Its report in January 1948, marked Top Secret, concluded that rationing would be impractical as it would lead to a black market in coupons as non-smokers and light smokers sold to heavy smokers, with the risk of undermining the rationing of food which was already needing stricter enforcement because of the ‘weakening of the public conscience’ in the face of the general decrease in food supplies. A joint ration for tobacco and sweets would reduce the black market because non-smokers would use the ration for sweets, but sweet coupons could be used for tobacco, substituting adults’ tobacco for children’s sweets, which was undesirable as ‘Ministry of Food experts attach importance to the nutritional value of sweets’; and a joint ration would create difficulties with retailers.
The report remarks: ‘As to the number of non-smokers, the only recent estimates we have seen suggest that one-fifth of the men and one-half of the women do not smoke at all. Thesefigures appear to be somewhat on the high side and we have assumed that 15% of the men and 40% of the women do not smoke.’
It adds: ‘We ought also to call attention to a possible long-term effect on the revenue of either rationing or a heavy increase of duty. It is probable that many people will permanently adopt the lower level of consumption enforced by rationing or an increase of duty and that others will be induced by a heavy increase of duty to give up smoking altogether. Such results would mean a permanent, and in the long run substantial, loss of revenue at any given level of duty.’ This worry about the revenue effects of reduced smoking recurs throughout the years.
The working party ended with a tentative recommendation of another 50% increase in tax: the increase in 1947 had produced a reduction in demand variously estimated at the time as 12-15% or 20% but put at 17% in a 1964 review; a further 50% increase (lifting the price of 20 Gold Flake from 3s 4d to 4s 8d) would perhaps cut consumption by about a third with neutral effect to the Revenue. In the event the tax increase in 1948 was restricted to 2d.[15]
None of Whitehall’s concern was known to the public, of course, nor to the MRC’s investigators nor, of course, could the Treasury know of the eventual outcome of Bradford Hill and Doll’s study, though whether they would have linked such knowledge with worries about the dollar crisis to adopt policies based on public health must be strongly doubted.
In January 1948, the MRC wrote to a number of London hospitals seeking their cooperation in the planned study by notifying Dr Doll of all patients diagnosed with cancer of the lung, bronchus, stomach, rectum and colon and by allowing the social workers to interview these patients and a randomly chosen control group with non-cancer diseases. By 1 May 1948 Doll was reporting on the first 156 interviews, saying there was already some bias towards smoking in lung cancer patients – but also towards residence near gasworks (or maybe, he noted, merely gasholders: not everyone would know the difference; and anyway the link disappeared as the study continued). A third social worker was engaged by June and in September Doll’s ten-page interim report stated: ‘The results appear to show a definite association between carcinoma of the lung and smoking, an association which is less strong for pipe smokers than for cigarette smokers’, although the lack of correlation between inhaling and cancer was ‘surprising’. No correlation was evident for the other cancers. By today’s standards the most interesting aspect of the statistics is the extent of smoking among the control patients: of 143 men with lung cancer, all smoked, but so did 135 of the 143 controls.[16]
In October the extension of the one-year study through 1949 was agreed, and in a lengthy fourth interim report in September 1949 Doll concluded ‘It is, therefore, probable that there is a real association between ca. lung and smoking.’ Less guardedly the same month he wrote to Dr Green at the MRC: ‘Incidentally the investigation has gone much better than I expected and it looks as if smoking will be incriminated to a major extent!’[17]
The following month, Green recorded a meeting between himself, Doll and Dr Harold Himsworth, the new Secretary of the MRC: ‘After challenging Doll on a number of detailed points, to which he seemed to have satisfactory answers, we told him that the results of the work so far were so striking and would no doubt cause so big a sensation when published, that we felt that it might even be desirable to repeat the whole study in order to see whether the answers came out the same.’ Doll was not keen[18] although he recognised that doing the same work in a rural area could be ‘very interesting’. The sequel does not appear in the files at the PRO and I am indebted to Sir Richard himself for it. ‘Himsworth, who was impressed by our paper,’ he writes[19], ‘pointed out that the subjects were nearly all Londoners and that the findings were so important that we ought to check that they could be reproduced in other parts of the UK. This we agreed to do and made arrangements to interview patients in Bristol, Cambridge, Leeds, and Newcastle. Before this extension was completed, however, Wynder and Graham published their findings in the USA (June 1950)[20], so we felt justified in publishing ours, which we did a few months later.’[21]
Ernst Wynder and Evarts Graham had conducted a case-control study of similar size and equally conclusive results to Doll and Hill’s. Thus by the end of 1950 two substantial studies, one in Britain and one in America, had established beyond reasonable doubt that the alarming rise in lung cancer was due to smoking. Other papers published in America showed similar results.[22]
But publication, so far from causing the predicted sensation, passed almost unnoticed in the press and totally unnoticed on the files (as preserved) of the Medical Research Council and the Ministry of Health.
Notes
1. PRO file FD 1.3567 [back]
2. ‘Points of View’: Cure of Tobacco Smoking, The Lancet, 6 September [back]
3. Dr Charles Fletcher took part in this programme, broadcast by the BBC on 12 January 1953. The producer wrote a week earlier to F H K Green at the MRC observing that Imperial Tobacco had ‘declined the offer of a chance to put their viewpoint in the same programme – perhaps wisely’. The programme lodged in the political memory: it was referred to in the briefing for Lord Salisbury at the Home Affairs Committee in February 1954 – PRO file FD 1.2009. [back]
4. PRO file FD 1.4761 [back]
5. Sir Richard Doll describes Rock Carling in 1946 as ‘a central figure in everything to do with cancer and radiation’ (personal communication, 19 January 1998). He chaired several MRC committees concerned with cancer and radiotherapy and had been a member of the MRC during the war years. [back]
6. Sir Richard Doll records that Kennaway and Mellanby were attracted by the idea that smoking was to blame, while Stocks was ‘convinced that atmospheric pollution was responsible’ – Statistical Methods in Medical Research 1998; 7: 87-117 [back]
7. PRO file FD 1.1989 [back]
8. PRO file FD 1.1993 [back]
9. About this time Dr A H Gale, a Medical Officer in the Ministry of Health, noted that the Kennaways had published in the British Journal of Cancer (1947; 1: 260-298 – September 1947) an analysis of death certificates: ‘The authors suggest tentatively that their findings are consistent with the hypothesis that the inhalation of cigarette smoke may be an aetiological factor [for lung cancer]’ – PRO file MH 55.1011. [back]
10. MRC 47/366, PRO file FD 1.1990 [back]
11 MRC 47/460, PRO file FD 1.1989 [back]
12. MRC 47/502, PRO file FD 1.1989 [back]
13. PRO file T 161.1302 [back]
14. PRO file T 171.792 [back]
15. PRO files T 233.221, T 171.792. The need to reduce dollar imports and the need not to risk revenue from tobacco again came into conflict in 1952, when the Cabinet, with Winston Churchill as Prime Minister in the chair, decided to postpone purchases of 20 million pounds of tobacco from 1952 to 1953. The Cabinet minutes for 6 March 1952 record: ‘This would save £4¾ million in dollars for the time being. While there was much to be said on social grounds for a general reduction in tobacco consumption, overseas expenditure on tobacco was small by comparison with the very heavy yield of internal revenue which it produced. If this were not maintained it would be necessary for the Exchequer to absorb purchasing power by other means.’ [back]
16. PRO file FD 1.1989 [back]
17. PRO file FD 1.1992 [back]
18. PRO file FD 1.1992 [back]
19. Personal communication, 23 October 1996 [back]
20. JAMA 1950; 143: 329-336 [back]
21. BMJ 1950; ii: 739-748. ‘These published results were essentially those we had shown Himsworth the year before. By the time of publication we had of course preliminary confirmative findings from other parts of the UK and Wynder and Graham’s work.’ (Sir Richard Doll, personal communication, 23 October 1996).[back]
22. Schrek et al, Cancer Research 1950; 10: 49-58; Levin et al, JAMA 1950; 143: 336-338; Mills et al, Cancer Research 1950; 10: 539-542. For a summary of the results of this research and for the general history of the research on smoking, see Richard Doll’s review in Statistical Methods in Medical Research 1998; 7: 87-117. [back]