The doyen of the world's epidemiologists, Sir Richard Doll, who has died aged 92, proved the link between lung cancer and tobacco smoking.
He hit the headlines in 1950 when, with Sir Austin Bradford Hill, he shattered government and public health complacency with research among patients in 20 London hospitals. Over three decades the registrar general's records had shown a very rapid and unexplained increase in lung cancer deaths among men. The search, involving patients with and without lung cancer, had no obvious causes to target. The prime suspects were the smuts from coal fires, exhaust fumes from cars, and tarring of the roads in response to the expansion of car ownership.
Doll, working with Bradford Hill, recorded the lifestyle and habits of men admitted to London hospitals with suspected lung cancer. Later, after the patients had a diagnosis, he found that those whose suspected lung cancer was confirmed were the smokers, and those in whom lung cancer was ruled out were the non-smokers.
Within a few months, cigarette smoking, then believed to be generally harmless apart from "smoker's cough", had emerged unambiguously as the only dominant factor - so dominant as to seem causal. Yet publication of this crucial Medical Research Council (MRC) study, which was completed in 1949, was delayed for a year at the insistence of Sir Harold Himsworth, then secretary of the MRC. He felt that the findings were so important and unexpected that they should not be released until they had been confirmed by a second study embracing hospitals around the country.
Doll and Bradford Hill quickly confirmed that cigarette smoking was the single and overwhelmingly most powerful connection with lung cancer. A preliminary study along the same lines also appeared in the United States, so the MRC team could not claim the priority their work deserved.
As Doll later commented, this was profoundly disappointing. The MRC study was by far the more powerful, for its findings had been carefully confirmed. Doll later emphasised scrupulous confirmation of research was a crucial professional principle. Himsworth, he said, was right to ensure there were no special London factors in the study.
Doll's later meticulous approach to the structure of studies, to the mathematics they use and to the collection, analysis and quality of information needed to render them valid, sprang from this and earlier work. Through his integrity and lifelong insistence on the highest standards, Doll won the respect of colleagues and scientists throughout the world.
A former smoker, Doll gave up when he saw his research results. He also showed that smoking caused premature death from cardiovascular disease, and that a bland diet was of no benefit to people suffering from peptic ulcers, thus releasing them from a tedious, dull and pointless regimen. He did major work on the safety of the contraceptive pill, and showed that low-level radiation, as exemplified by the useless radiotherapy given to men with ankylosing spondylitis, caused a dose-related leukaemia.
As he aged, his autocratic perfectionism led to conflicts with many scientists and medics who wanted quicker answers than Doll could allow on available information - such as his uncertainty until 1993 about increased child thyroid cancer after Chernobyl. However, what he questioned, quite properly, was the quality of the underlying information. This conservative caution did not in fact reflect a political conservatism, for Doll and his wife were both quite leftwing early on, and associated with various progressive causes. Environmental pressure groups thought him so dismissive of early evidence or of a small number of cases as to be defending the interests of government and industry. Yet, as Doll pointed out, the mathematics of statistical studies is uncertain.
One of Doll's earliest papers, published in 1936, showed that many medical decisions at the time were based on studies of groups of patients so small as to be statistically meaningless.
In the 1950s Doll and Bradford Hill launched a long-term prospective study of the effects of smoking on British doctors (a group of about 20,000), which confirmed the connection with lung cancer. It also showed how risk related directly to the extent of smoking and how chronic bronchitis and coronary disease were also, according to this and other studies, caused by smoking - as was quickly accepted in the US.
Doll was more aware than many of his critics that an association, however powerful, does not necessarily imply causation. He never claimed a causal relationship between smoking and lung cancer on the basis of either the first case-control studies or the prospective study. Only when all other available information was brought into the picture did he feel that a true causal relationship had been shown.
His assessment included the historical sex-ratio difference of smoking and of later disease, the absence of any increase in lung cancer deaths in countries where smoking was rare or had started late (like Iceland and Norway), and the rapid increase of lung cancer incidence in countries where smoking had started early.
However, even when it was shown that all populations with a long smoking history suffered a large increase in lung cancer, yet nowhere was there a high incidence of lung cancer among non-smokers, the causal relationship continued to be challenged. This was a great disappointment to Doll. Until the media were convinced, he excused the failure of the public to respond, but he could not excuse the inactivity of governments.
In Britain it took a quarter of a century before tobacco was taxed on health grounds. Doll's letter of congratulation to Denis Healey, then Chancellor of the Exchequer, earned a personal reply as increases in taxation so rarely brought praise to No 11. But Doll was disappointed by the failure of successive governments to follow up Healey's initiative and, in later years, by a worldwide failure to ban tobacco advertising.
From the early 1950s Doll argued that the need was not to convert addicts, but to prevent children starting to smoke. Unless the health message was clearly spelled out in government action and endorsed in the example of teachers and parents, it was unreasonable, he said, to be disappointed that children continued to smoke.
The tobacco problem remained Doll's main concern until the end of his life, but he also studied industrial causes of cancer, domestic dangers such as radon in houses, and the problems of low-level radiation exposure for the general population. In 1993, presenting the 40-year follow-up findings of the study of smoking and health among British doctors - incontrovertible evidence - he spoke of a new study predicting the future effects of smoking in mainland China. Having been approached by the Chinese government he felt that, perhaps at last, the message was getting through.
He was interested in other problems - at the MRC, on the scientific committee of the International Agency for Cancer Research, and at the ICRF Cancer Research Unit, Oxford. Long after his formal retirement, working jointly with Professor Richard Peto (his colleague over many years and, in a sense, his heir), he set up a major study to disentangle conflicting evidence about cholesterol levels and heart disease.
Tall, slim, patrician in bearing and often sharply critical, Doll could be formidable. But he relaxed easily and had a great sense of humour. He enjoyed explaining how, had it not been for too much Trinity College ale on the night before the last paper in the open scholarship, he might have been a mathematician. But he failed the scholarship and was offered an exhibition. "The beer I had that night was the best drink I ever had," he quipped in later life. "I decided to go into medicine and I have enjoyed every minute since."
He had always admired his father's work and dedication as a GP in London and, in the event, went straight from Westminster School to St Thomas's, graduating in 1937. While still a student he wrote of the need for proper mathematical tests, such as the Chi Squared test, in statistical studies of disease.
After a year as casualty officer and house physician at St Thomas's, a voluntary research post came up under the professor of medicine at the Royal Postgraduate Medical School, Hammersmith. Doll was asked to develop a method for measuring the concentration of vitamin B1 in urine, part of an investigation into a possible connection between B1 deficiency and heart failure.
This project, which proved to be too difficult, nevertheless convinced Doll that research was truly for him. But before could take up the job as house physician then offered to him, he was called up as a member of the supplementary reserve, which he had joined after the Munich crisis. He served with the Royal Medical Corps in France and the Middle East throughout the second world war. His diary as a battalion medical officer of the retreat to Dunkirk, terse, observant and shrewdly aware of tragi-comedy, was published in the British Medical Journal in 1990.
He described life on a hospital ship in the Mediterranean during the north African and Italian campaigns as "the happiest years of the war", largely because Rommel and German U-boat commanders paid full respect to the Red Cross on the high seas, provided the hospital ships sailed under full lights.
In 1944 he came home as an invalid with a tubercular kidney. "Fortunately, I only lost one," was his later comment, adding that he had continued working as a physician during his convalescence. After discharge and some disappointment, he approached the MRC, and through Dr Joan Faulkner (whom he married shortly afterwards; they had a son and a daughter) found a research post at the Middlesex Hospital, studying occupational factors in the development of gastric and duodenal ulcers.
This launched both his glittering career and a lifelong powerful marriage. Within a year he joined Bradford Hill in developing ways of improving epidemiological techniques. Thirty years later, honoured throughout the world, he was still modestly seeking better ways of unravelling the connections between lifestyle, occupational exposures and disease.
He remained mentally alert, academically productive, and in good health, apart from being a little frail, until his death. His wife died in 2001. In 2002 he gave evidence on behalf of Mrs Margaret McTear, who sued tobacco manufacturers after her husband died, aged 48, of lung cancer after smoking 60 cigarettes a day since he was nine. Doll testified that the industry had been aware of the risks since around 1950, when he published his results and was immediately visited by two men from Imperial Tobacco.
Among many prizes, he received the UN Award for Cancer Research in 1962, the BMA Gold Medal in 1983, the Royal Society Royal Medal in 1986 and the Helmut Horten Foundation Award in 1991. Doll was made regius professor of medicine at Oxford University in 1969, retiring in 1979. It was mainly through his efforts that Green College, Oxford, was founded; it has a largely medical bias. He was knighted in 1971 and became a Companion of Honour in 1996.
Though he would have been reluctant to acknowledge it, Doll was both a great doctor and the greatest epidemiologist of our time.
· William Richard Shaboe Doll, epidemiologist, born October 28 1912; died July 24 2005
This obituary has been revised and updated by Caroline Richmond since the author's death in 1998
