Background
It is well known smoking cessation reduces the risk of all-cause mortality; however, much of the existing evidence derives from conventional observational studies that are vulnerable to confounding, selection bias, and immortal time bias. Individuals who quit smoking often differ systematically from those who continue in terms of health status, socioeconomic conditions, and health-seeking behaviors, making causal interpretation challenging. Randomized controlled trials of smoking cessation with long-term follow-up on mortality are scarce and often infeasible for ethical and practical reasons.
 
Objective
We conducted a target trial emulation within the European Prospective Investigation into Cancer and Nutrition (EPIC) to estimate the causal effect of smoking cessation on all-cause mortality.
 
Methods
We specified a hypothetical target trial comparing smoking cessation versus continued smoking among current smokers at baseline. Eligibility criteria included people aged 35–60 years, no prior cardiovascular disease or treated diabetes, non-heavy alcohol consumption, and at least one year of potential follow-up. Time zero was anchored at the follow-up questionnaire, when smoking status was reassessed. Participants who were current smokers at baseline were classified as quitters or continuing smokers based on follow-up smoking status. Individuals who died before the follow-up questionnaire were excluded to avoid immortal time bias.
Follow-up accrued from time zero until death, administrative censoring, or a fixed horizon of 10 years. We constructed person–period data with one-year intervals and estimated discrete-time hazards using pooled logistic regression. Confounding was addressed using stabilized inverse probability of treatment weights (IPTW) derived from a propensity score model including age, sex, education level, alcohol intake, physical activity index, waist-to-hip ratio, smoking history, and study center. Centers without treatment variation were excluded to satisfy positivity. Robust standard errors were clustered at the individual level.
Risk differences (RDs) and risk ratios (RRs) were derived at prespecified time horizons. Nonparametric bootstrap resampling at the individual level was used to obtain confidence intervals for risk contrasts.
 
Results
Preliminary data included 23367 participants, of which 17791 current smoker and 5576 quitters,  contributing 202575person-years of follow-up. Compared with continuing smokers, smoking cessation was associated with a lower hazard of all-cause mortality during follow-up (interval odds ratio [OR]: 0.97, 95% CI: [0.82–1.09]. Standardized cumulative mortality risk at 10 years was lower among quitters than among continuing smokers, corresponding to an absolute risk difference of −0.0004 and a relative risk of 0.99.
 
Conclusions
In this target trial emulation within EPIC, smoking cessation was associated with lower all-cause mortality risk compared with those who continued smoking. Beyond its methodological contribution, these findings have important public health implications. Even modest reductions in mortality risk associated with smoking cessation, when applied at the population level, can translate into substantial absolute gains in survival. This underscores the continued need for comprehensive tobacco control strategies that integrate individual-level cessation support with population-based health promotion. From a public health perspective, strengthening and sustaining interventions that promote smoking cessation is therefore essential not only for clinical risk reduction but also for cancer prevention and the reduction of avoidable mortality at the population level.