Tobacco smoking and the British doctors' cohort

Abstract

In this issue of the BJC, Sir Richard Doll and colleagues from Oxford present findings from the 50 years of follow-up of British doctors in relation to cancer risk (Doll et al, 2005). There are many important aspects surrounding this article, some of which deserve wider and deep reflection. This has been a study that was completely innovative and ingenious in its construction and remarkable in the perseverance of its follow-up. When Richard Doll and Austin Bradford-Hill undertook this cohort study, they probably did not realise that they were setting a new paradigm for modern epidemiology, and choosing to do such a study among doctors was quite ingenious. Which group would be able to be followed to death by a variety of sources including via the Medical Register? The initial results (Doll and Hill, 1954; Doll and Peto, 1976) were highly significant and of great value in identifying a new and significant cancer risk, but the true worth of this study increased as follow-up increased and the flow of new information emerged. During the course of the follow-up, and in particular in the reports after 40 years follow-up (Doll et al, 1994) and 50 years of follow-up (Doll et al, 2004), the real impact of tobacco smoking on a wide variety of diseases and life expectancy itself was fully revealed. Half of the smokers die from a tobacco-related disease and half of these deaths occur in middle age. The impact of these deaths on the loss of nonsmokers life expectancy is enormous. Stopping smoking at any age is effective in reducing the loss of nonsmokers life expectancy, although this lessens off as age at quitting increases. The comparison presented here (Doll et al, 2005) is in many respects unique. Here, we have the opportunity to observe what really happens over a long period among those who are exposed to a carcinogenic risk and to compare it with a descriptive analysis of all the available published, epidemiological and mechanistic evidence. The International Agency for Research on Cancer (IARC) prepared a Monograph on Tobacco Smoking initially in 1986 (International Agency for Research on Cancer, 1986). When this was recently revised (International Agency for Research on Cancer, 2004a) in 2004, there was new information available to increase the numbers of cancer types deemed to be causally related to tobacco smoking (Figure 1). In 11 of 13 cancer types considered by IARC to be causally related to tobacco smoking, and which could be identified on death certificates, Doll and colleagues found them to be significantly related to smoking (Doll et al, 2005). For the two remaining sites (nasopharynx, nose and nasal cavity), deaths in the doctors’ cohort were sparse, although there was a suggestion that there could well be an association (Doll et al, 2005).