Smoking and mortality in England and Wales, 1950 to 1976

Abstract

Errors in the certification of the cause of death complicate many studies of the presumed lethal action of cigarette smoking. Such errors can have a serious impact on the analysis of secular trends for specific diseases. In principle, secular trends have the advantage of being able to distinguish between causal, ‘non-causal’ (for example, constitutional) and mixed interpretations of mortality ratios. By studying secular trends in overall mortality, in relation to those of cigarette consumption, problems connected with the incorrect certification of the cause of death are avoided. Percentage changes in sex- and age-specific death rates for all causes of death in England and Wales are calculated, by 3-yearly periods, for the era 1950–1976. These changes in death rates are compared with corresponding trends in sex- and age-specific ‘constant tar’ and ‘current’ cigarette consumption in the United Kingdom. No obvious cause-and-effect relation can be discerned. The main problem is to explain the fairly consistent decrease in death rates in both sexes and all age groups during periods when cigarette consumption was either rising or falling. By considering the age-patterns of the difference in death rates rates between early and late periods the main cause of the fall in mortality in the lower age groups, from 20–24 yr up to 40–44 yr, appears to be the near-eradication of tuberculosis. In the higher age groups, the absence of appreciable structure in the age-pattern of the fall in mortality makes it difficult to identify the causes although up to 70–74 yr in males (but not females) a late mortality form of tuberculosis made a substantial contribution. The trends fail to support the hypothesis that smoking influenced mortality. Rose and Hamilton's randomized controlled intervention trial of the effect of intensive advice to stop smoking has also given no support to the idea that smoking is lethal. The available evidence for mortality in Swedish smoking-discordant pairs of twins (MZ and DZ) lacks statistical power but is consistent with the view that mortality ratios have a constitutional, rather than a (smoking) causal, basis. We should, perhaps, preserve an open mind about the supposed lethal action of smoking.