Smoking and death: the past 40 years and the next 40BMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6959.937 (Published 08 October 1994) Cite this as: BMJ 1994;309:937
- R Peto
- Imperial Cancer Research Fund Cancer Studies Unit, Nuffield Department of Clinical Medicine, Radcliffe Infirmary, Oxford OX2 6HE.
Smoking already kills about two million people a year in developed countries, half in middle age (35-69) and half in old age.1, 2 This number is still increasing as the death rate among women increases and populations grow larger and older (fig 1). Already smoking accounts for one sixth of the 11 million adult deaths each year in these populations. There are 1.2 billion people living in developed countries. If one sixth of their deaths continue to be caused by tobacco about 200 million of the adults and children now living in developed countries will eventually be killed by tobacco, and about 100 million of these will die while still in middle age.2
In developing countries there has recently been a large increase in the number of young men smoking. People in China, for example, now smoke about 30% of the world's cigarettes. This will have catastrophic effects next century as most other causes of death are likely to continue to decrease and the effects of tobacco to increase. If current smoking patterns persist - that is, if the smoking uptake rate among young adults continues to be substantial and the rate of stopping smoking at older ages continues to be low - by the time the children of today reach middle age smoking will be one of the largest causes of premature death in the world.
Over the next 40 years the annual number of deaths from tobacco will increase from about three million to more than 10 million (table I),2 yet 40 years ago the hazards were only just beginning to be recognised. The United Kingdom's Medical Research Council (MRC) supported much of the early research, and in 1957 it was the first national institution in the world to accept formally the evidence that tobacco is a major cause of death.
Evidence against smoking
In 1947 the MRC had called a conference to discuss the rapid increase in deaths in the United Kingdom attributed to lung cancer. Part of the increase was known to be an artefact of the improvements in the accuracy of death certificates that had resulted from better diagnostic methods. But since the increase was much steeper in men than in women changes in the real disease rates must also have occurred. Austin Bradford Hill had recently taken over as director of the MRC's Statistical Research Unit at the London School of Hygiene and Tropical Medicine, and one of his new recruits was Richard Doll. They began a large “case-control” study in 1948 in which the life histories of several hundred patients with lung cancer were compared with those of several hundred people without the disease.
Cigarette smoking was only one of several possible causes being investigated. (Doll himself originally thought the increase in cars and the tarring of the roads were more likely to be to blame.) But the results of this first study proved otherwise. The only big difference between those who had lung cancer and those who did not was that almost all those with lung cancer had been smokers. Doll and Bradford Hill published their results in the BMJ in 1950, and in the same year a parallel study by Wynder and Graham in the United States had independently reached similar conclusions.3, 4 This was the first clear evidence that smoking is a major cause of death.
Once it was shown that smoking was a cause of most deaths from lung cancer, the next step was to establish prospective studies in which people were first asked what they smoked and then followed for several years to monitor deaths not only from lung cancer but also from other causes. Parallel studies were again established independently in Britain and the United States.5, 6 The British study evolved into the first major prospective study of smoking and death in the world. It began in 1951, when Doll and Bradford Hill used BMA records to ask all the doctors in Britain what they smoked, and it still continues today, more than 40 years later. The American studies have tended to be shorter but to have a larger sample size. Indeed the American Cancer Society studies in the 1960s and 1980s both included over a million Americans.
By the mid-1950s, these prospective studies had shown strong associations between smoking and death not only from lung cancer (and later from other cancers) but also from respiratory disease, particularly chronic bronchitis, and vascular disease, particularly heart attacks.7, 8 Tobacco accounted for most of all the deaths from lung cancer, and in addition it caused an even large number of deaths from other diseases.
Although these early reports attracted much attention, in retrospect they greatly underestimated the hazards of prolonged smoking. When the 20 year follow up of the British doctors study was reported in 1976 the death rate in middle age from all causes was already twice as great in smokers as in non-smokers,9 but worse was to come. During the second half of the 40 year follow up (1971-91),10, 11 the death rate from all causes in middle aged smokers was three times that of non-smokers (fig 2). At last the true hazards of really prolonged smoking had been assessed reliably.
Similar findings have recently been reported from America. Two enormous prospective studies have monitored deaths in the early 1960s and deaths in the mid-1980s.1, 12 The lung cancer death rate in non- smokers stayed roughly the same between the 1960s and the 1980s, but the rate among male smokers doubled. The rate of lung cancer was 10 times greater than in non-smokers in the 1960s and 20 times greater in the 1980s.2 Presumably this is because men smoking cigarettes in the United States today have been doing so regularly throughout their adult life, whereas this was often not the case in the 1960s. For overall mortality in middle age, the ratio of the all cause rate among smokers to that among non-smokers has likewise risen from 2:1 in the 1960s to 3:1 in the 1980s,12 and most of this excess involves diseases that are likely to be caused by smoking (table II). This means that more than half of all the deaths of smokers in middle age (plus rather less than half of those in old age) are caused by tobacco. The old statement, based partly on the first 20 years of the study of British doctors, that “at least a quarter” of all regular cigarette smokers would be killed by the habit now needs revision. In fact, the proportion is about one half (box).2 By coincidence, the proportion of smokers who agree with the statement that “Smoking can't be all that dangerous, or the government would ban advertising” is also about one half.14
Hazards for individual cigarette user: 1990s British and American evidence2
The risk is great, especially among those who start smoking cigarettes regularly in their teenage years: about half of teenagers who keep smoking steadily will eventually be killed by tobacco (about a quarter in old age plus a quarter in middle age)
Those killed by tobacco in middle age (35-69) lose an average of 20-25 years of non-smoker life expectancy
Nationwide, tobacco is much the greatest cause of death. In non-smokers, cancer mortality is decreasing slowly and total mortality is decreasing rapidly
Most of those killed by tobacco were not particularly heavy smokers, but most did start in their teenage years
Stopping smoking works. Even in middle age, stopping smoking before having cancer or some other serious disease avoids most of the later excess risk of death from tobacco, and the benefits of stopping at earlier ages are even greater
Once the hazards had been shown the next step was public information. In 1962 the Royal College of Physicians published the first official report specifying the dangers of smoking,13 and this was followed two years later, with even greater impact, by the United States Surgeon General's report on tobacco.15 The Royal College of Physicians' report was written by Charles Fletcher, who had also instigated the 1959 MRC bronchitis survey,16 which provided the first detailed evidence of the respiratory benefits of stopping smoking.
Tobacco illustrates the importance of an approach to public health that is quantitative and highlights the value of large epidemiological studies. In one sense it represents a substantial success, since a moderate reduction in a big cause of death can save many lives. Moreover, although we have seen only a moderate reduction in use of tobacco in Britain and the United States over the past few decades, this could easily have been a moderate increase given the power and subtlety of the tobacco marketing methods now confronting children and teenagers.
In another sense, however, smoking represents a great failure in public health: more than 40 years after the hazards were first established, cigarettes are still responsible for 30% of deaths in middle age in Britain and the United States,2 and worldwide sales are increasing. More than $5bn a year is spent on global advertising and promotion (which, according to the tobacco manufacturers, has no effect on the proportion of children who choose to smoke).