A case-control study of the protective effect of alcohol, coffee, and cigarette consumption on Parkinson disease risk: time-since-cessation modifies the effect of tobacco smoking

Abstract

The aim of this study was to investigate the possible reduced risk of Parkinson Disease (PD) due to coffee, alcohol, and/or cigarette consumption. In addition, we explored the potential effect modification by intensity, duration and time-since-cessation of smoking on the association between cumulative pack-years of cigarette smoking (total smoking) and PD risk. Data of a hospital based case-control study was used including 444 PD patients, diagnosed between 2006 and 2011, and 876 matched controls from 5 hospitals in the Netherlands. A novel modeling method was applied to derive unbiased estimates of the potential modifying effects of smoking intensity, duration, and time-since-cessation by conditioning on total exposure. We observed no reduced risk of PD by alcohol consumption and only a weak inverse association between coffee consumption and PD risk. However, a strong inverse association of total smoking with PD risk was observed (OR=0.27 (95%CI: 0.18-0.42) for never smokers versus highest quartile of tobacco use). The observed protective effect of total smoking was significantly modified by time-since-cessation with a diminishing protective effect after cessation of smoking. No effect modification by intensity or duration of smoking was observed indicating that both intensity and duration have an equal contribution to the reduced PD risk. Understanding the dynamics of the protective effect of smoking on PD risk aids in understanding PD etiology and may contribute to strategies for prevention and treatment.

Figure 1. The modifying effects of time-since-cessation on the relation between smoking and PD.

The fitted excess OR model with a spline function for total smoking and effect modification by time-since-cessation. 95% confidence intervals were estimated via 100 bootstrap replications. A: The OR for different levels of total pack-years, plotted for 21 years-since-cessation. B: The OR for different levels of time-since-cessation, plotted for 15 pack-years of total smoking.

Figure 2. Odds ratios and 95% confidence intervals for time-since-cessation within categories of total pack-years of smoking.

A–E: Ever smokers were divided according to quintiles of total-smoking and the effect of time-since-cessation within those groups was plotted. ORs were located at the quintile-specific average time-since-cessation and based on unconditional logistic regression relative to never-smokers adjusted for age, sex and center. The lines represent the predicted ORs and 95% confidence intervals for different levels of time-since-cessation based on the excess OR model plotted for the quintile-specific average pack-years of total smoking.