Modeling smoking history: a comparison of different approaches in the MARS study on age-related maculopathy

Smoking is an established risk factor for the development of age-related maculopathy (ARM), and its end stage, age-related macular degeneration (AMD). We evaluated the benefit of various smoking-related variables in modeling the association of smoking with ARM and AMD in a mixed sample of current, former, and never smokers.

This was a cross-sectional study in participants of the Muensteraner Altern- und Retina-Studie (MARS). Participants were classified according to the Rotterdam classification system as healthy, or having ARM or AMD. Using multinomial logistic regression techniques, the association with number of cigarettes, years of smoking, pack-years and time since cessation in former smokers were evaluated.

Mean age of the 982 participants (58.6% females) was 70.9 +/- 5.5 years. ARM was present in 483 (49.2%) and AMD in 285 (29.0%) individuals. The adjusted prevalence odds ratio (OR) in current smokers versus never smokers was 2.61 (95% confidence interval [CI] 1.34-5.09) for ARM and 3.94 (95% CI 1.91-8.14) for AMD. This effect decreased in former smokers with an OR = 0.55 (95% CI 0.33-0.99) per log-transformed time since smoking cessation for ARM and an OR = 0.52 (95% CI 0.30-0.90) for AMD.

By including a variable for time since smoking cessation, we were able to handle current, former, and never smokers in one model that estimates the association of smoking with ARM or AMD. Logarithmical transformation of the time since smoking cessation seemed to increase the model fit and to reflect a non-linear protective effect of smoking cessation on the onset of ARM and AMD in former smokers.