Smoking Cessation and Type 2 Diabetes: Application of Marginal Structural Model and G-Formula Method

Abstract

Smoking cessation has been proven to be associated with a reduced risk of chronic diseases, such as cardiovascular disease, chronic obstructive pulmonary disease, and cancer. However, its relationship with type 2 diabetes mellitus (T2DM) remains complex. Weight gain following smoking cessation has been identified as a key mediator that may offset some of the metabolic benefits of quitting. Previous studies have reported inconsistent findings, largely due to inadequate handling of time-varying exposures, confounders, mediators, and potential selection bias from informative censoring. This dissertation applies marginal structural models (MSMs) and the parametric mediational g-formula to address these methodological limitations and investigate the temporal relationships between smoking cessation, weight change, and T2DM risk. The dissertation is composed of three papers.Paper 1 presents a systematic review and meta-analysis of eleven cohort studies that examined the association between smoking cessation and T2DM risk, stratified by post-cessation weight change. Using random-effects models, we found that quitters who gained weight had a 71% higher T2DM risk compared to continuous smokers (HR = 1.71; 95% CI: 1.12–2.62), particularly among recent quitters (HR = 2.20; 95% CI: 1.27–3.82). In contrast, long-term quitters had a reduced risk (HR = 0.91; 95% CI: 0.87–0.95). Quitters without weight gain showed no increased risk if recently quit (HR = 0.99; 95% CI: 0.81–1.02), and a lower risk if long-term (HR = 0.84; 95% CI: 0.81–0.87). Compared to never smokers, recent quitters exhibited elevated T2DM risk regardless of weight change. These findings underscore the importance of both cessation duration and weight management in mitigating diabetes risk. Paper 2 utilizes MSMs with inverse probability weighting to estimate the causal effect of smoking cessation on T2DM risk using data from the Framingham Offspring Study (1983–2014). Among 1,880 ever-smokers, those who quit smoking had a 17% lower risk of T2DM compared to those who never quit (HR = 0.83; 95% CI: 0.61–1.12). Notably, individuals who maintained cessation for more than 10 years experienced a 31% reduction in risk (HR = 0.69; 95% CI: 0.49–0.96). Additionally, never smokers had a 26% lower risk compared to current smokers (HR = 0.74; 95% CI: 0.55–0.99). Long-term quitters had risk levels comparable to never smokers, reinforcing the long-term benefits of sustained cessation. Paper 3 applies the parametric mediational g-formula to estimate the interventional direct and indirect effects of smoking cessation on T2DM risk, with weight change as the mediator. The analysis included 416 current smokers at baseline (Exam 3) in the Framingham Offspring Cohort. Long-term cessation (20 years) was associated with a 6.77% absolute reduction in 30-year T2DM risk compared to continued smoking. Approximately 32% of the total effect at 1-year post-cessation was mediated through weight change, but this indirect effect diminished with longer cessation durations, approaching zero after 10 years. These findings suggest that the long-term metabolic benefits of quitting outweigh the short-term adverse effects of weight gain and support efforts to promote smoking cessation along with early weight management strategies. This dissertation demonstrates that while post-cessation weight gain may temporarily elevate the risk ofT2DM risk, long-term smoking cessation ultimately reduces that risk. The protective effect becomes more pronounced with sustained cessation, and the influence of weight gain as a mediator diminishes over time. These findings support the promotion of smoking cessation alongside early weight management strategies.