Bradford Hill

The two authors of the 1950 London Hospitals study and the British Doctors study were Bradford Hill and Richard Doll. Richard Doll remained a well-known public figure for the next 50 years. But Bradford Hill, the senior partner in the studies, seems to have more or less vanished.

Austin_Bradford_HillLast year I bought a copy of  Smoking Kills, the 2009 biography of Richard Doll by Conrad Keating. It’s a bit of a hagiography of Doll, but it sheds some light on his mentor and co-author, Austin Bradford Hill (right).

There don’t seem to be any biographies of Bradford Hill, who has been completely eclipsed by his junior partner, Richard Doll. But there are a couple of chapters in Smoking Kills which have quite a lot to say about him.

Hill was invalided out of the Royal Naval Air Service with tuberculosis in 1917 – with an index-linked full disability pension that he continued to draw until his death in 1991. He’d wanted to be a doctor, but his health did not permit it. He instead turned his attention to preventive medicine, for which no medical qualification was necessary.


No academic qualifications in either medicine or statistics! Richard Doll was at least a qualified medical doctor. It seems he was no great mathematician either:hill_mathematics

Hill’s route into epidemiology and statistics seems to have been provided by Major Greenwood, who had been a demonstrator for Hill’s father, the physiologist Leonard Hill. Greenwood became interested in statistics, studied under Karl Pearson, and eventually became President of the Royal Statistical Society in 1934.  From The origins of Austin Bradford Hill’s classic textbook of medical statistics:

In 1927 Hill joined Major Greenwood’s department at the London School of Hygiene and Tropical Medicine (LSHTM), having obtained an honours degree in economics at University College London in 1922, and then a grant from the Medical Research Council to examine the high mortality in young adults in rural areas of England.

While carrying out this study he attended Karl Pearson’s course on statistics at London University. Hill’s first four papers, for all of which he was the sole author, were published in 1925.  The following year he was awarded a PhD by the University of London with a thesis entitled ’A physiological and economic study of the diets of workers in rural areas as compared with those of workers resident in urban areas,’ which consisted of these four papers (we have not been able to trace the identity of Hill’s examiners). In 1930 Hill obtained a DSc from the University of London, with a thesis entitled ’An investigation of sickness in various industrial occupations.’ This comprised eight (single-authored) papers published between 1925 and 1929, including the four submitted for his PhD degree. His external DSc examiner was WP Elderton, author of Frequency curves and correlation (1906) and Primer of statistics (1906). Major Greenwood was the internal examiner.

By 1936 Hill had published thirty-nine book reviews, eight research reports, and sixteen papers, including nine in the British Medical Journal or The Lancet, and four in the Journal of the Royal Statistical Society. Many of the reviews were of books about population, poverty, industrial working conditions, migration, mortality, and the social conditions in London, subjects of obvious central interest at the LSHTM.

Perhaps Hill’s principal achievement was to have persuaded the medical profession to think quantitatively. He taught the elements of statistics to medical students, and was the author of The Principles of Medical Statistics.

The relevant sections of the book:

pages 62 + 63   Chapter 5 – No Hill No Doll
pages 64 + 65
pages 66 + 67
pages 68 + 69
pages 70 + 71
pages 72 + 73
pages 74 + 75
pages 76 + 77   Chapter 6 – The Paradigm Shift
pages 78 + 79
pages 80 + 81
pages 82 + 83
pages 84 + 85
pages 86 + 87
pages 88 + 89
pages 90 + 91
pages 92 + 93
pages 94 + 95
pages 96 + 97


About Frank Davis

This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

31 Responses to Bradford Hill

  1. Harleyrider1978 says:

    They had been using junk statistics back in 1894 when mark twain wrote the moral statistician. All Hill did was perpetuate the junk science to the rest of the world and now we find he had no mathematical background and they use him as the father of today’s junk science we put up with outlawing us and it has no value other than propaganda!

  2. legiron says:

    Sixteen papers! I have 84 and still nobody listens.

  3. Tony says:

    One quick comment:
    13,000 deaths from lung cancer in 1950 (p77), a 1,500% increase since the end of the first world war. Doll published his (supposedly) landmark paper in 1950. It is well established that smoking prevalence declined from then on. At a very fast rate with doctors and then, perhaps 10-20 years later, with the population as a whole. Male rate went from 80% in 1950 down to 30%(supposedly) by 1980 and 20% nowadays. The female rate fell from 40% down to 20% over the same period.

    Last year there were only 35,000 deaths from lung cancer. What wonderful preventive medicine!

    OK, I accept that the UK population has increased from 40 or so million to nearer 60 and life expectancy has also increased but then cancer survival has increased too, to a limited degree.

    • nisakiman says:

      I would imagine that a major reason for the 1500% increase in LC deaths after WW1 would be the exposure of troops to mustard gas during the fighting, which as they aged manifested itself in their succumbing to lung cancer.

      And because mustard gas damages the DNA in human cells, recovered victims are at greater risk for developing certain cancers.

      It would be interesting to know what the deaths from LC per 100,000 were in 1950 as compared to 2015. From the figures you give above, a rough mental calculation suggests that the rate has almost doubled in the 65 years interim.

      • Rose says:

        Deposition of Wilhem C Hueper
        Director of Environmental Cancer, National Cancer Institute


        “Soon after World War 1 when mustard gas had been used extensively for war purposes on both sides,the claim was advanced by some investigators that the corrosive trauma to the respiratory system resulting from a single exposure to mustard gas might have causal relation to this subsequent development of cancer of the larynx and of the lung among veterans of the various armies”.

        “During recent years somewhat more valid evidence has been advanced.

        Japanese investigators observed and reported during recent years that employees of war gas factories in Japan, exposed to mustard gas, died from chronic respiratory diseases which, for a number of years, had been clinically diagnosed as tuberculosis of the lung.

        When subsequently on some of these death cases autopsies were performed, it was found that every one of these workers …….dying from tuberculosis of the lung had a cancer of the lung and in more recent years the Japanese observer also clinically confirmed the diagnosis, confirmed by biopsies, of cancer of the larynx among those workers.”

        There is; moreover, a large scale English study on the relative frequency of cancer of the lung among English veterans of World War I on record in which the relative frequency of cancer of the lung among this population group in relation to the standard values was determined.

        “It was found that English veterans who had been exposed to mustard gas poisoning during World War I had an excessive frequency of cancer of the lung,”

        No idea which page

        • Rose says:

          Wilhelm C. Hueper was one of the non-believers to such a degree that he merited his own special trashing from Robert Proctor.

          I learnt a great deal reading that entire testimony.

        • nisakiman says:

          Thank you. Rose, for that link. I had a look at the first few pages, and it looks very interesting. I’ve saved it for further perusal when I have a bit more time.

    • Tony says:

      Late last night, from memory, I hurriedly gave a figure of 35,000 UK lung cancer deaths last year. A more accurate figure appears to be 31,000 (male+female) from here:

    • Tony says:

      Some more figures:
      In their 1950 hospital study, Doll and Hill stated that deaths from lung cancer increased from 612 in 1922 to 9,287 in 1947, or roughly fifteen fold.
      Rate per 100,000 1901-1920, males 1.1, females 0.7. Rates 1936-9 per 100,000, males 10.6, females 2.5.

      • Frank Davis says:

        So 612 lung cancer deaths in 1922 (M&F?) and 9,287 in 1947, and 31,000 in 2015 – after adult male smoking prevalence had fallen from 80% to 20% on being told by the likes of Richard Doll that 90% of lung cancers were caused by smoking.

        • nisakiman says:

          Those are the lines along which I was thinking, Frank. Something doesn’t add up here, and the errors all seem to be with the claims of the anti-smokers.

  4. Cecily Collingridge says:

    [bottom of page 81] “On 6 February 1947 a conference was called at the MRC to discuss the rapid increase in deaths in the United Kingdom attributed to the disease. Among the thirteen attending were Sir Ernest Rock Carling, Percy Stocks, Edward Mellanby, Ernest and Nina Kennaway, Alice Stewart and Austin Bradford Hill… Various hypotheses were advanced at the meeting. Stocks believed that urban atmospheric pollution was the probable carcinogen, while Ernest Kennaway believed that smoking could be a possible causative factor. Other theories put forward were arsenic in cigarette smoke, radioactive dusts and drugs. There was no consensus of opinion.”

    I noted the presence of Alice Stewart at this conference. You might be interested in something published in 2011 (Perspect Biol Med. 2011 Autumn;54(4):504-31. doi: 10.1353/pbm.2011.0042):

    Richard Doll and Alice Stewart: reputation and the shaping of scientific “truth”
    Greene G
    As the world watched the Fukushima reactors release radionuclides into the ocean and atmosphere, the warnings of Dr. Alice Stewart about radiation risk and the reassurances of Sir Richard Doll assumed renewed relevance. Doll and Stewart, pioneer cancer epidemiologists who made major contributions in the 1950s-he by demonstrating the link between lung cancer and smoking, she by discovering that fetal X-rays double the chance of a childhood cancer-were locked into opposition about low-dose radiation risk. When she went public with the discovery that radiation at a fraction of the dose “known” to be dangerous could kill a child, her reputation plummeted, whereas Doll, foremost among her detractors, was knighted and lauded as “the world’s most distinguished medical epidemiologist” for his work. Their lives and careers, so closely intertwined, took contrary courses, he becoming “more of the establishment” (as he said), while she became more oppositional. When it was discovered, after his death, that he’d been taking large sums of money from industries whose chemicals he was clearing of cancer risk, his reputation remained unscathed; it is now enshrined in the “Authorized Biography” (2009) commissioned by the Wellcome Institute, along with Doll’s denigration of Stewart as an “embittered” woman and biased scientist. Stewart lived long enough to see radiation science move her way, to see international committees affirm, in the 1990s, that there is no threshold beneath which radiation ceases to be dangerous; recent evidence from Chernobyl is bearing out her warnings. But a look at the making and breaking of these reputations reveals the power of status, position, and image to shape scientific “knowledge” and social policy.

  5. Rose says:

    How peculiar, Alice Stewart’s obituary from the Independent has gone entirely missing and try as I may, I just can’t find it.

    This is all I’ve got left.

    “The incidence of child leukaemias was increasing and no one knew why. She suspected that the mothers might remember something the doctors did not, so she interviewed them and rapidly saw the correlation with X-rays, which she demonstrated statistically.

    X-rays were medicine’s new toy and were being used for everything from examining the position of the foetus to treating acne; even shoe shops had X-ray machines where customers could see how their footwear fitted.

    This was at the height of the arms race, when the British and US governments were trying to build up public trust in the friendly atom and did not want people to get the idea that low-dose radiation could kill their children.”

    • Harleyrider1978 says:

      Smoking doesn’t give you but an extra 260 mrems per year thyroid obulation 18,000,000 mrems

    • Cecily Collingridge says:

      There was another abstract in PubMed that was very concise. Unfortunately, the article itself is in French and I can’t access it anyway, but it makes me curious about what was in the German studies mentioned.

      Richard Doll. A surprising story of conflicts of interest
      Sculier JP.
      Richard Doll is a very famous English physician epidemiologist. He is credited with discovering the link between smoking and lung cancer. His reputation was recently vitiated by two facts, ignorance of German studies prior to his work and the existence of major conflicts of interest with industry that led him to minimize the role of chemical products in carcinogenesis.

      • Frank Davis says:

        Richard Doll did know about some of the German studies: the Doll and Hill London Hospitals study paper references (Franz?) Muller 1939. There were a whole bunch of German researchers – e.g. Fritz Lickint – in the Nazi era using the sort of questionnaire science which Doll and Hill later adopted (and added statistics). Hitler was personally funding the work of Schairer and Schoenegger in 1944. Robert Proctor’s Nazi War on Cancer is informative in this respect, even if Proctor is a virulent antismoker who likens Nazi antismoking science to German rocket science.

    • mikef317 says:

      Rose, might these help? The wiki page says the Independent obit is a dead link.

      • Rose says:

        Thank you Mike, it’s that I liked the way the Independent one was written.
        Some rather gloss over things.

        But this one I just found on the Telegraph is rather good too.

        Edited highlights

        Alice Stewart

        “Alice Stewart became involved in this field in 1955 when, as head of the Oxford Social Medicine Unit, she noted a rapid increase in lymphatic leukaemia among young children. Her own godchild had recently died of the disease, which gave impetus to her interest. She was immediately surprised to find a higher incidence of childhood leukaemia in areas with better medical care and lower overall childhood death rates.

        Feeling that the solution might lie in the children’s pre-natal medical care, she proposed a project to interview the mothers of the children who had died.

        “In what later became known as the Oxford Childhood Cancer Survey, Alice Stewart and her team visited all 203 public health departments around the country to obtain details of every child who had died of leukaemia between 1953 and 1955. As part of the survey, a questionnaire was sent to the mothers of these children as well as to the mothers of children in two control groups – children who had died of other cancers, and children who were alive and well.

        Alice Stewart and her statistician, George Kneale, noticed an astonishingly strong pattern. Children who had died of any form of cancer before the age of 10 had been X-rayed twice as often in utero as the live children. A single X-ray, well within official safety limits, was enough to double the risk of early cancer, they concluded.

        When Alice Stewart published this finding in 1956, most of her fellow scientists were sceptical, even openly hostile. Nuclear technology was riding high and there were huge vested interests involved. The suggestion that very low doses of radiation might pose a risk to human health was not one the scientific establishment wanted to hear.”

        “The controversy stirred by the Oxford Survey was as nothing compared to the storm that broke out in the 1970s over Alice Stewart’s findings on the effects of low-level radiation on workers at the gargantuan Hanford plutonium weapons complex in Washington State. She had become involved in this research by chance in 1974 when, on a visit to America, she and George Kneale were invited by Thomas Mancuso, of the University of Pittsburgh Medical School, to become consultants on an extensive investigation he was directing for the American government.”

        “When the British pair looked at Mancuso’s data, they concluded that the risks of cancer from low-level radiation at Hanford were about 20 times that predicted by the A-bomb studies. When their results became known, the American authorities reacted with outrage. Mancuso was deprived of his directorship and forbidden to publish his findings; the use of outside consultants was banned. “Everyone in America who took our side lost their funding,” Alice Stewart recalled. “They don’t burn you at the stake any more, but they do the equivalent, in terms of cutting you off from your means to work.”

        “The epidemiologist Richard Doll (one of the first scientists to establish a link between smoking and cancer) was appointed Regius Professor in 1969; but he had little time for Alice Stewart, criticising her methodology and suggesting that she had “gone off the rails” with her later work on nuclear workers.”

        “When asked why recognition had eluded her in her own country, she replied: “Good people are seldom fully recognised during their lifetimes, and here there are serious problems of corruption. One day it will be realised that my findings should have been acknowledged.”

      • Rose says:

        The Bomb Test men

        A summary of mortality and incidence of cancer in men from the United Kingdom who participated in the United Kingdom’s atmospheric nuclear weapon tests and experimental programmes

        R. Doll

        “Most of the differences observed between the participants and controls were interpreted as due to chance, but some may be due to differences in smoking habits”

  6. Tony says:

    There is so much in those book extracts that deserves comment that it is hard to know where to start. But maybe this bit, concerning the 1950 hospital study is particularly noteworthy:

    From page 94 paragraph 2: “it was not until a month after the patients were discharged from hospital that Doll and Hill came to know the outcome… This eliminated the possibility of bias…

    But here’s Doll’s own recollection of the events: “And my job was to go round after the patient was discharged, look at the hospital notes and decide what the patient had actually got. page 330

    So it was Doll himself that determined the final outcome.

    • Tony says:

      The book extract (pages 84-85) makes a big deal of the “fearsome questionnaire” used in the hospital study. 50 questions, including where subjects lived, but only 11 on smoking. But there is no mention of any of these other questions in the published paper of 1950. Not even an acknowledgement that they were asked (unless I missed something). I don’t know if these results were published in separate papers but I strongly suspect not.

      • Frank Davis says:

        There was supposed to be a second paper discussing the other questions asked, but it was never written. In effect, these questions may as well never have been asked. It may as well have all just been questions about smoking. And in the subsequent British Doctors study, they were just that.

      • Some French bloke says:

        there is no mention of any of these other questions in the published paper of 1950

        All that remains of these other questions is in Table II of the study (Comparison Between Lung-carcinoma Patients and Non-cancer Patients Selected as Controls, With Regard to Sex, Age, Social Class, and Place of Residence).
        The social class categories are too gross to permit the identification of any occupational hazard, and going into more detail concerning places of residence (considering that fewer heavy smokers were residing in less urbanised areas) would have shed an unconvenient light upon the rural-urban divide. So, just as Frank says: It may as well have all just been questions about smoking, and that morphed into the Doctors’ Study’s ‘smoking only’ template.

        Two juicy quotes from p. 85 of Conrad Keating’s “Smoking Kills”

        The questionaire was deliberately wide-ranging because they wanted to eliminate any bias from the patient’s responses.

        Hill and Doll’s aim was to make the field observations mirror an experimental design as nearly as possible. In other words they sought, as in an experiment, to limit the variables.

        In other words, this “wide-ranging” stuff is far too taxing intellectually, so let’s narrow it down to just one factor! Limiting the variables surely makes a toxological study more rigorous; in the context of epidemiological research, however, it just translates to “ignoring confounders”, making the study more open to bias. In the first case, the material impact of other factors is effectively removed, when it comes to a ‘soft science’ like epidemiology, the impact of the researcher’s prejudices can increase exponentially! The worst case scenario being when such intellectual dishonesty is allowed to run amok the world over for five or six decades.

        • Some French bloke says:

          read: toxicological…

        • Tony says:

          Thanks for that. I hadn’t spotted that residency information in table 2, although it looks too vague to be of much use.

          I suspect you’re right about how limiting variables will have a very different effect on toxicological compared to epidemiological studies. But there is also a huge difference between limiting and ignoring.

        • Some French bloke says:

          “there is also a huge difference between limiting and ignoring”

          Indeed, that why I specified “effectively removed” concerning other factors in the context of toxicology. Anyway, thanks for the accolade, Tony!

  7. Clicky says:

  8. smokingscot says:


    It is becoming more than a little difficult for the established parties within the EU to keep a lid on things.

    It’s only an election for the Presidency of Austria and the far right party (the Freedom Party) candidate looks to have won the first round!

    Candidates from the two dominant parties in Austria were unable to even get to this point.

    There’s going to be a run-off between the Freedom chap and a guy from the Greens on the 22 May. Then there’s their General Election that must be held by 2018.

    So when’s their total smoking ban due?

    2018 is what this lot tells me.

    Oh, sorry, so sorry, just happens to be a minor coincidence. I shouldn’t have even mentioned it… NOT.

  9. Some French bloke says:

    @ Cecily. Here’s a link to a pdf of the original article: “Richard Doll. Une surprenante histoire de conflits d’intérêt” (Pr J.-P. Sculier):

    The references to those German “studies” cited by Sculier, with a few comments:
    9. Müller FH: Tabakmissbrauch und lungencarcinoma.
    Zeitschrift Krebsforschung 1939 ; 49 : 57-65
    – The raw facts as presented by Sculier: based on 96 autopsies (incl. 10 women) for each group, Müller found 3 non-smokers and 56 heavy smokers in cases, as against 14 and 31 in controls.

    10. Schairer E, Schöniger E: Lungenkrebs und Tabakverbrauch.
    Zeitschrift Krebsforschung 1943; 51 : 264-9
    – According to Sculier, the authors state that lung cancer represented 0.46 % of autopsies conducted between 1910 and 1914, and 2.77 % of those conducted between 1935 and 1939.
    From their questions, put to the families of 195 deceased LC victims about the latter’s smoking statuses (“habitudes tabagiques”), as compared to 700 living controls of the same average age (“de même âge moyen”), they found the proportion of heavy smokers to be 52% and 26.7%, respectively, and of non-smokers: 3% and 15.9%. Sculier adds: “at the time, statistical tests were not yet performed systematically. The p (value) of the chi-squared test was in fact < 0.000000114", citing:
    14. Smith GD, Egger M: The first reports on smoking and lung cancer – why are they consistently ignored?
    Bulletin of the World Health Organization 2005 ; 83 : 799-800

    Note: The "700 living controls" bit above is misleading, since a detailed article on that same study asks: "how did the authors identify the 700 men residing in Jena, specifically aged 53 or 54 years, including their addresses, and send them a questionnaire? What were the characteristics of the 270 respondents compared with those who did not?"
    More on that, including an English translation of the questionnaire, in: Lung cancer and tobacco consumption: technical evaluation of the 1943 paper by Schairer and Schoeniger published in Nazi Germany
    Alfredo Morabia

    Another note: The study was unbalanced anyway, considering the questions to the living controls were put to them directly, whereas the information concerning the deceased’s habits was obtained indirectly, leading to a form of recall bias “by proxy”.

    – This 1943 “study” was finally published in English in 2001, with a commentary by Doll, stating he hadn’t been aware of it until Proctor “talked to him about it” in 1997:
    11. Schairer E, Schoniger E: Lung cancer and tobacco consumption.
    Int J Epidemiol 2001 ; 30 : 24-7

    – Schöniger, born in 1917, supposedly died in the war, and a similar assumption goes for Müller (born 1914). Schairer was born in 1907, and died of dementia in 96.
    13. Schairer HS: Commentary : In memoriam of my father,
    Prof. Dr. Med. Dietrich Eberhard Schairer.
    Int J Epidemiol 2001 ; 30 : 28-9

    – A lesser-known case-control thingy, published two years before the Doctors Study’s preliminary report, is also mentioned:
    3. Doll R, Hill AB: A study of the aetiology of carcinoma of the lung. Br Med J 1952 ; 2 : 1271-86

    The article just says it consisted of 1,357 men and 108 women pitted against as many controls, and that the p values were < 0.000001 (men) and < 0.001 (women). Needless to say, their "conclusion" was identical to that of the better-known 1950 London Hospitals "Study".

    As for Sculier's conclusion, while calling for more awareness of "conflicts of interest's perverse effect in the field of medicine", and stressing their role in the denial ("véritable dénégation") of the "major impact" of chemical-industrial carcinogens, he also hails Doll as "a brilliant epidemiologist who conducted high quality studies that display a high level of proof. He was thus able to clearly establish the negative health impact of smoking ("tabagisme")."

No need to log in

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s