Dynamic Smoking Patterns and Risk of Parkinson Disease and All-Cause Mortality: A Competing Risk Analysis Approach.

Abstract

Smoking has been reported to be inversely associated with Parkinson disease (PD). However, the higher premature mortality among smokers may act as a competing risk, potentially confounding the inverse association. Because smoking behavior is dynamic, the long-term impact of changes among current smokers remains unclear. We investigated the association between longitudinal changes in smoking status and the risks of PD and all-cause mortality using a competing risk framework and an age-based time scale with left truncation. This large-scale retrospective cohort study included current smokers aged 40 years or older who participated in all 3 examination periods of the Korean National Health Screening. Based on longitudinal changes from the initial smoking status to 2 subsequent time points, participants were categorized into 4 groups: persistent smokers, recent quitters, sustained quitters, and relapsed smokers. Cumulative incidence functions for PD were estimated, with all-cause mortality as a competing event, and subdistribution hazard ratios (sHRs) with 95% CIs were obtained using Fine-Gray models. Data were obtained from 410,489 eligible participants (mean age 51.7 ± 9.0 years; 93.5% male). During a median 9.1-year follow-up, persistent smokers exhibited the lowest risk of PD. Both recent quitters and sustained quitters had higher PD risk than persistent smokers (sHR 1.60 [1.41-1.82] and 1.61 [1.42-1.81]), whereas relapsed smokers did not differ from persistent smokers (sHR 1.05 [0.87-1.28]). For all-cause mortality, recent and sustained quitters had 3% and 17% lower risks, respectively, compared with persistent smokers, whereas relapsed smokers showed no significant difference. The observed pattern of PD risk was suggested to be primarily associated with current smoking status rather than cumulative smoking exposure, as relapsed smokers and recent quitters, who had the same number of smoking time points, showed distinctly different risks. Furthermore, 1 time point (∼2 years) of short-term abstinence did not attenuate the protective association. Mortality was lowest in sustained quitters while recent quitters showed a marginal trend toward lower risk, supporting the benefit of early cessation. Interpretation should be cautious because smoking status was assessed at 3 time points, subsequent changes were unknown, and most participants were male.