Cigarette smoking and life expectancy.

Statistical relationships between smoking and death-rates may be expressed in a number of different ways, the two most common being either in terms of mortality ratios, i.e. the ratio of the death rate from any particular cause in smokers to the death rate from the same cause in non-smokers, or in terms of the differences between the numbers of deaths from various causes in smokers and non-smokers. The relative merits of these two approaches have been the subject of some controversy (Cornfield, Haenzel, Hammond, Lilienfeld, Shimkin, and Wynder, 1959; Berkson and Elveback, 1960), but it is probably true that the appropriateness of the method depends on the circumstances and purposes of the analysis. One aspect of this question of presentation is the kind of index that is the most meaningful to the layman. If we consider that smoking has an adverse effect upon health and wish to discourage the general public from smoking, then the evidence must be presented to them in such a way that they can readily balance the possible disadvantages of smoking against the advantages which the confirmed smoker feels sure that he enjoys. The Report of the Royal College of Physicians (1962) set down the risks in a variety of ways, namely the fractional risks of dying from all causes of death and from lung cancer in a decade, and the percentage of men aged 35 who may expect to die before the age of 65, but these fractions and percentages are not immediately translatable into concrete loss for an individual. Another possible method of presentation of the data is in terms of 'expectation of life', obtained from life tables of smokers and non-smokers. Such a method has not been widely used, although to the layman loss ofyears of life may seem a more concrete fact to face than a change in his mortality risk. It therefore seemed useful to calculate the life-expectancy of persons in different smoking categories using the age-specific mortality rates obtained in the various prospective surveys that have been made. The results of these calculations are presented in this paper.