Duration and intensity when estimating smoking-attributable mortality: Development of a new method applied to the case of lung cancer in France

The standard method for estimating smoking-attributable mortality (SAM) does not account for smoking duration and time since cessation, which prevents analysts from running projections or what-if scenarios on changes in initiation and cessation behaviours. We propose a new method that combines the duration distributions observed in the population with empirical values for the duration effects on mortality from smoking and cessation, which are reported in the epidemiological literature. This new method is more data-demanding than the standard method, and we demonstrate its implementation on French lung cancer cases by creating pseudo-cohorts using pooled cross-sectional surveys (Baromètre Santé, INPES) from 1975 to 2010. We find that lung cancer mortality will increase by 50% until 2035, after which it will plateau. Our simulations show that halving initiation among teenagers would save 20,500 lives over the 50-year period between 2010 and 2060, while doubling the quit rate among adults would save 53,000 lives over the same period. This study statistically substantiates the intuitive belief that interventions and policies that increase cessation save more lives in the medium term than those aimed at decreasing initiation.