Background: Cigarette smoking is implicated in a large number of diseases and other adverse health conditions. Among the dimensions of smoking are number of cigarettes smoked per day, duration of smoking, passive smoking, smoking of filter cigarettes, age at start, and duration elapsed since quitting by ex-smokers. The practice so far is to study most of these separately. We develop a simple index that integrates these dimensions of smoking into a single metric, and suggest that this index be developed further. Method: The index is developed under a series of natural assumptions. Broadly, these are (i) the burden of smoking monotonically increases with the cigarette-years but it is more severe in the beginning, (ii) start of smoking early in life is more burdensome than a late start, and (iii) the burden gradually reverses as the duration elapsed since cessation by ex-smokers increases. Result: The index so arrived is: S = (3 – a/15)*1/2*sqrt[sumof(pi*ni*xi) – 0.5] - y for S greater than equal to 0, and sumof(pi*ni*xi) greater than equal to 0.5; otherwise zero (use a =30 for a>30); where i = 1, 2, …, I, and I is the number of segments in life with different smoking pattern and a is the age at start of smoking, pi is the proportion of smoke inhaled in case of passive smoking (or adjustment for filter cigarettes or for other forms of smoking), xi is the number of cigarettes smoked for ni years, and y is the number of years elapsed since cessation by ex-smokers. Negative values of S are to be considered equal to zero. Examples are given that demonstrate the use of this index. Conclusion: Just as almost any other composite index, our index too could be good as a comprehensive measure of burden of smoking but not to study its individual dimensions. This measures the present burden in absolute sense and not the risk of smoking-related diseases. Like body-mass index, the smoking index may have good correlation with the risk of some diseases and poor for many others, depending upon the extent to which the risk of disease agrees to our postulations.


Statistical Models