Background:
Smoking prevalence in Mexico is 20%; however, occasional smoking is common and in women there is an upward trend in tobacco use. Countries aiming to achieve the tobacco endgame (i.e., smoking prevalence <5%), need to generate the local evidence to drive smoking cessation policy. Given the limited evidence from randomized trials on smoking cessation and mortality and cancer incidence and the absence of estimates specific for Mexico, a country with a low-intensity and non-daily smoking pattern, observational data can be used to generate policy-relevant estimates.
Objective:
To estimate the effect of smoking cessation on all-cause and lung cancer-specific mortality among adult women.
Methods:
We used the Lung Health Study (LHS), a randomized study on smoking cessation with results on 15-year mortality, to specify a target trial emulation using observational data from the Mexican Teachers’ Cohort. This prospective cohort of 115,275 women was initiated in 2006-08 in Mexico. To identify incident smoking quitters we defined baseline as the first follow-up questionnaire in 2011. Inclusion and exclusion criteria were aligned to those used in the LHS. Deaths and causes of death were identified through a probabilistic linkage algorithm with national death registries specifically designed for Mexico. We used inverse probability weights and pooled logistic regression to estimate 12-year cumulative risks, risk ratios (RR), and risk differences (RD) of all-cause deaths and lung cancer deaths. We obtained 95% confidence intervals (95% CI) from nonparametric bootstraps with 500 replications. We explored the impact of smoking cessation in subgroups according to age and smoking intensity.
Results:
We identified 5,080 women (non-quitters 3,223; 892 quitters) who were smokers at baseline and had been followed for 12 years. We identified 170 deaths and 7 lung cancer deaths over the 12-year period. Among non-quitters, the 12-year cumulative risks of all-cause mortality was 3.29% (95% CI: 2.72–3.92). For quitters, the corresponding estimate was 2.93% (95% CI: 1.89–4.22). Compared with women non-quitters, quitting resulted in an 11% lower mortality risk (RR=0.89; 95% CI: 0.56–1.33). For every 1,000 women smoking cessation can prevent 4 deaths (RD=-3.7 per 1,000; 95% CI: -15.5, 10.2). For women <45 years of age the corresponding estimates were RR=0.59 (95% CI: 0.31–1.11) and RD=-8.7 per 1,000; 95% CI: -21.2–31.4). Results were null for women ≥45 years (RR=1.03; 95% CI: 0.59–1.75; RD=1.2 per 1,000; 95% CI: -15.5–10.2). Results did not vary according to smoking intensity (≤2 vs. >2 cigarettes per day). For lung cancer deaths we observed a RR of 0.75 (95% CI: 0.40–3.04) and a RD of -3.2 (95% CI: -14.5–1.77). Our estimates include wide confidence intervals; precision will be increased by the time results are presented in the meeting with increased mortality follow-up.
Conclusions/Implications:
We approximately emulated the LHS, an existing randomized clinical trial, using observational data. Our results were comparable to what was observed in the LHS, a trial that included high-intensity smokers. Smoking cessation in countries where low smoking intensity is common could prevent deaths particularly in young adults. Our estimates can directly inform policy-making.