One does not need quantitative data to be convinced that tobacco smoking is harmful. In 1604, not long after tobacco smoking was introduced into Europe, King James I of England wrote A Counterblaste to Tobacco, in which he opined, “Smoking is a custom loathsome to the eye, hateful to the nose, harmful to the brain, dangerous to the lungs, and in the black, stinking fume thereof nearest resembling the horrible Stygian smoke of the pit that is bottomless.” However, this and other similar writings were largely ignored. Over the centuries, tobacco smoking became more and more popular. In Britain, cigarette smoking was made fashionable by officers returning from the Crimean War in the mid nineteenth century, and by the end of the Second World War, in Britain about 80% of men were regular smokers.

During the first half of the twentieth century, lung cancer had risen in prominence from being a relatively rare disease to being a major cause of death. It had not escaped the notice of medical experts that this rise had occurred in parallel with the rise of cigarette smoking, and some experts had linked the two rises, suggesting that cigarette smoke was a cause of lung cancer. This view was most prominent, perhaps, in Nazi Germany, largely because of Hitler’s personal view that tobacco (and alcohol) consumption were harmful. However, there were several other current theories on the cause of the rise of lung cancer, including the effects of motorised road vehicles on air pollution. A few scientific studies, looking for a relationship between smoking and lung cancer, had been carried out before the Second World War, and indeed had found some evidence of such a relationship, but generally their results were not particularly persuasive and had very little impact on the medical establishment or on public opinion.

The beginning of a major change in attitude can be dated back to 1950, when five studies were published giving clearer evidence of a link between smoking and lung cancer. Perhaps the most influential of these was carried out by the medical statisticians and epidemiologists Richard Doll and Austin Bradford Hill (both of whom were later knighted). This study took a sample of individuals with lung cancer from four hospitals in the UK, and another sample of individuals who were patients admitted to the hospitals for different reasons. It enquired about the smoking habits of the individuals. Hardly surprisingly, given how common smoking was at the time, a large majority of the individuals in both samples were smokers; but this proportion was considerably larger in the sample who had lung cancer. Doll and Hill were inclined to the view that the reason for this finding was that smoking caused lung cancer. However, other scientists pointed out that the findings were consistent with other explanations, for example that some other factor caused people to smoke, and, independently, also caused them to have lung cancer. A study of this general type is called a case-control study (because the people with the disease of interest are ‘cases’ and the group without this disease are the ‘controls’), and in such studies it can be difficult to rule out the possibility that some other factor is causing both the disease and the potential cause that is being investigated. The other four 1950 studies, all carried out in the USA, also had a case-control design.

Although these studies did not in themselves provide totally convincing evidence that smoking caused lung cancer, they did provide a spur to further, more persuasive, studies. One such study was again carried out by Doll and Hill. This study did not take people who already had lung cancer and look back into their past. Instead it was prospective; it took a group of people (a so-called cohort), and followed them up over time, recording their smoking habits and their health. In Doll and Hill’s study, the group of people followed up were British doctors, over 40,000 of them. Such a study takes time to conduct, because the effects of smoking on lung cancer take time, and it needs to study large numbers of individuals, because one needs a sufficiently large number of cases of lung cancer to arise to make the statistical conclusions valid. But even after two and a half years Doll and Hill’s prospective study had demonstrated large differences in lung cancer death rates between smokers and non-smokers. Doll and Hill found, further, that smokers were more likely than non-smokers to die of coronary heart disease. An even larger study, begun in 1952 in the USA by Hammond and Horn, followed up 190,000 men and provided similar findings.

Again the results of these studies were criticised, notably by the prominent statisticians Joseph Berkson and Sir Ronald Fisher. Fisher and Berkson did not dispute that the prospective studies had shown an association between smoking and lung cancer, but tried to point out in different ways that this association might not be causal. In other words, they tried to come up with other plausible explanations of the association, in which the smoking did not cause the lung cancer. However, further research — further case-control and cohort studies in many countries across the world, and other work on how tobacco smoke actually affects human or animal cells, eventually dismissed the objections of Fisher, Berkson and others.

During the late 1950s and early 1960s, the hypothesis that cigarette smoking causes lung cancer became more and more widely accepted. In 1962, the Royal College of Physicians of London published a report entitled Smoking and Health, which strongly took the line that smoking causes lung cancer, and in 1964, an advisory committee to the US Surgeon General published a report taking a similar line. Arguably it was these reports in particular that swung medical, and eventually public, opinion behind the view that smoking is seriously harmful to health.

Statistical and biomedical research on smoking and health has, of course, continued since the 1960s. Evidence for the link between smoking and coronary heart disease is now very strong, though coronary heart disease has a very complicated pattern of causation and is common even among non-smokers. It is generally accepted nowadays that smoking increases one’s chances of contracting something like 50 different diseases. Although many of the deaths attributable to smoking are due to lung cancer, many more of them are not. (Diseases of the heart and blood vessels, and chronic bronchitis are also high on the list.) The effect of life-long smoking on health is now known to be considerably more severe than Doll and Hill’s early studies indicated. However, it is worth bearing in mind that, even though the link between smoking and lung cancer is relatively clear-cut, it still took well over a decade after the publication of the first major studies before it was widely accepted that smoking causes lung cancer.

Further reading
A comprehensive, though in places rather technical, paper on the history or research into the effects of smoking, by Sir Richard Doll, one of the pioneers, is called ‘Uncovering the effects of smoking: historical perspective’, and was published in1998 in the Journal Statistical Methods in Medical Research, volume 7, pages 87–117.