Estimations of smoking-attributable mortality in Spain at a regional level: comparison of two methods

To estimate and discuss smoking-attributable mortality (SAM) for the 17 regions in Spain among the population aged ≥35 years in 2017, using two methods. A descriptive analysis of SAM was conducted using two methods, the prevalence-independent method (PIM) and the prevalence-dependent method (PDM). Observed mortality was obtained from the National Institute of Statistics; smoking prevalence from three National Health Surveys; lung cancer mortality rates from the Cancer Prevention Study-II; and relative risks from five US cohorts. SAM and percentages of change were estimated for each region overall, by sex, age and cause of death. In 2017, tobacco caused 56,203 deaths in Spain applying the PIM. Using the PDM the number of deaths was 4.4% (95% CI: 3.4–5.5) lower (53,825 deaths). Except in four regions, the PIM estimated a higher overall SAM and the maximum percentage of change was 18.6%. Overall percentages of change were higher for women (15.7% 95% CI: 12.6–19.0) and for cardiovascular diseases–diabetes mellitus (13.8%; 95% CI: 11.5–16.2). At the national level, both methods estimate similar figures for SAM. However, the difference in estimates appears at the subnational level. Differences were higher in subgroups with lower smoking prevalence and for causes of death with periods of induction shorter than those for lung cancer. •Few studies have compared smoking-attributed mortality with a dependent or independent method of prevalence in regions.•Dependent or independet method of prevalence can be used to estimate smoking-attributable mortality at a regional level.•In subgroups with low smoking prevalence (i.e women and 75+ age group), the differences between both methods are greater.•For causes with shorter induction period than lung cancer (i.e cardiometabolic diseases), differences in estimates are higher.•The independent method is an alternative when prevalences are not available or the source is unreliable.