Leffondré et al did not consider a third possible approach to modeling smoking history: the use of one comprehensive smoking variable which allows for intensity, duration, and recency of smoking, including their interactions

2003

Leffondré et al did not consider a third possible approach to modeling smoking history: the use of one comprehensive smoking variable which allows for intensity, duration, and recency of smoking, including their interactions

2003

- Leffondré et al did not consider a third possible approach to modeling smoking history: the use of one comprehensive smoking variable which allows for intensity, duration, and recency of smoking, including their interactions
- Such an approach would avoid the problem of multicollinearity and is promising for obtaining a good model fit
- Assuming that lung cancer can be caused by a carcinogenic substance contained in cigarettes and cigarette smoke, the level of the carcinogen in the human body can partly be attributed to smoking
- Am J Epidemiol 2003;158:392–395 figure 2 of Leffondré et al.’s paper, the term should be logarithmically transformed, so a comprehensive smoking variable is given by X = ln
- It would be interesting to compare a lung cancer model using the proposed comprehensive smoking variable with models explored by Leffondré et al