Background: Smoking history is often di- or trichotomized into for example "never, ever or current smoking". However, smoking must be treated as a time-dependent covariate when lifetime data is available. In particular, individuals do not smoke at birth, there is usually a wide variation with respect to smoking history, and smoking cessation must also be considered.
Methods: Therefore we analyzed smoking as a time-dependent risk factor for cardiovascular atherosclerotic events in a cohort of 2400 individuals with familial hypercholesterolemia who were followed from birth until 2004. Excess risk after smoking-cessation was modelled in a Cox regression model with linear and exponential decaying trends. The model with the highest likelihood value was used to estimate the decay of the excess risk of smoking.
Results: Atherosclerotic events were observed in 779 patients with familial hypercholesterolemia and 1569 individuals had a smoking history. In the model with the highest likelihood value the risk reduction of smoking after cessation follows a linear pattern with time and it appears to take 6 to 9 years before the excess risk is reduced to zero. The risk of atherosclerotic events due to smoking was estimated as 2.1 (95% confidence interval 1.5; 2.9).
Conclusion: It was concluded that excess risk due to smoking declined linearly after cessation in at least six to nine years.