Direct Estimation of Death Attributable to Smoking in Switzerland Based on Record Linkage of Routine and Observational Data

Abstract

In Switzerland, estimations of smoking-attributable deaths were based on age- and sex-adjusted hazard ratios (HRs) from foreign cohorts, precluding consideration of country-specific properties and adjustment for confounding. In order to overcome this, we analyzed recently available individual data from Switzerland. We included 17,861 individuals aged ≥16 years who participated between 1977-1993 in health studies and were anonymously linked with the Swiss National Cohort. Adjusted Cox regression was used to calculate mortality HRs. Smoking status at baseline was categorized into never-smokers, former smokers, and current light or heavy smokers (<20 or ≥20 cigarettes/day). As covariates, we selected education, marital status, lifestyle, alcohol consumption, and body mass index. We differentiated between cardiovascular disease (CVD), cancer, and noncancer-non-CVD deaths. Smoking-attributable deaths were estimated with a HR-based approach and with age-specific prevalence rates and mortality estimates from 2007. Smoking men and women not only had an increased risk for all-cause (HR and 95% confidence interval vs. never-smokers: 1.71 [1.53-1.90]; 1.54 [1.36-1.75]), CVD (1.72 [1.43-2.06]; 1.50 [1.19-1.90]) and cancer (1.87 [1.56-2.25]; 1.58 [1.30-1.93]), but also for noncancer-non-CVD death (1.57 [1.29-1.89]; 1.58 [1.30-1.93]). Former smoking men had an increased risk for all-cause (1.16 [1.03-1.31]) and cancer death (1.35 [1.10-1.65]). Multivariate adjustment only slightly modified the association between smoking and mortality. Overall, 7,153 deaths per year could be attributed to smoking. Smoking is an important avoidable health burden in Switzerland, and its consequences may persist for decades after quitting. This stresses the need for putting more efforts in strategies aimed at preventing the onset of smoking.