Cigarette smoking continues to exert a major impact on morbidity and mortality. We aim to provide robust estimates of smoking dynamics and their effects on mortality in Tuscany, Italy, from 1993 to 2019, along with forecasts under alternative tobacco control policies (TCPs).
Smoking dynamics are modelled using a compartmental model combined with Monte Carlo (MC) simulations to propagate uncertainties. A variance-based global sensitivity analysis (GSA) quantifies the contribution of each input to output variance. TCPs are ranked according to the Surface Under the Cumulative Ranking Curve (SUCRA).
The MC approach produces results consistent with those obtained when only sampling variability is considered, although with wider uncertainty intervals (UIs). We estimate that in 2023 smoking caused 3,348 deaths (90% UI 2,761 to 3,942) among men and 1749 deaths (90% UI 1,397 to 2,888) among women aged over 65. By 2063, these numbers are projected to decline to 1477 (90% UI 1,081 to 2,144) and 1,436 (90% UI 742 to 2,138), respectively. According to the GSA, the time period used for calibrating the model is the main source of uncertainty in the model outputs, suggesting that the phenomenon has changed over time. The tobacco-free generation policy (TCP1) shows the highest SUCRA for smoking prevalence, while cessation treatments rank highest for mortality impact.
The approach shows promise for identifying key sources of uncertainty and guiding public health strategies. Although uncertainty limits the precise quantification of future dynamics and impacts, the TCP1 is likely to exert the strongest long-term effect on reducing smoking prevalence, while cessation interventions remain essential for achieving short-term reductions in smoking-attributable mortality.