LAUSR

Tobacco smoking is a major leading preventable cause of mortality. The Peto-Lopez method and its variants are commonly used for estimating smoking-attributable mortality in population in high-income countries. The method assumes the average cumulative exposure of tobacco smoking in the population can be approximated using lung cancer death rates. However, providing estimates at an infra-national level is often much more challenging because the observed number of deaths for lung cancer are generally too low to use directly the method. The aim of this study was to estimate the number of deaths attributable to smoking at the regional level in France in 2015. Population attributable fractions were computed using the Peto-Lopez method modified by Parkin which combines mortality statistics, lung cancer death rates and adjusted relative risks of deaths associated with smoking. To estimate the proxy for smoking prevalence in regional level by sex and age group, we pooled lung cancer mortality on several years according to the size of the regional populations and we smoothed the non-linear relationship of age with the estimates of the proxy. The attributable fractions were estimated for the 13 metropolitan regions and 4 French overseas regions. In whole France, we estimated that 19% of deaths among men and 7% among women were attributable to smoking in 2015. Deaths attributable to smoking ranged across metropolitan regions from 17% to 23% among men and from 5.5% to 8.7% among women. In overseas territories, estimates presented much more variability with a range from 6% to 18% for men and from 1.2% to 4.3% for women. We highlighted disparities in regional mortality attributable to smoking. Producing estimates of the burden of smoking at an infra-national level is a tool offering many perspectives for better analyzing territorial inequalities and ultimately to further guide more effective localized interventions to reduce tobacco consumptions. We provide a straightforward solution to estimate the infra-national burden attributable to tobacco from a usual method. Our results highlight the disparities in mortality attributable to smoking across regions in France in 2015.