Use of a multistage model to predict time trends in smoking induced lung cancer

Abstract

Study objective - The aims were to use a mathematical model to predict the time course of smoking induced lung cancer, and to investigate to what extent the most recent increases in lung cancer mortality are due to cigarette smoking. Design - A mathematical model was developed and solved by simulation to construct detailed smoking histories of the US white male population given available prevalence data by age and cohort. A multistage carcinogenesis model was used to predict the time course of smoking induced lung cancer given the detailed smoking histories. Source of data and model parameters - The smoking prevalence figures were taken from work by Harris who calculated them using data collected in the Health Interview Survey. The parameters of the multistage model were taken from Whittemore who fitted the model to several sets of smoking and lung cancer data. Main results - The smoking model was used to construct detailed smoking histories of the US white male population from 1900 to 1985. In turn the multistage model was used to predict age and cohort specific smoking induced lung cancer mortality rates over this period. These results were compiled to predict the overall age adjusted trend in smoking induced lung cancer from 1970 to 1985. The model predicts a 12% decline in smoking induced lung cancer for this group over the 15 year period. Conclusions - The model calculations predict a 12% decline in smoking induced lung cancer for this group, during a period when the actual total rate of lung cancer increased by 26%. Taken together with the decline in average tar content in cigarettes over this period, and the relatively constant dose rate among smokers, these results strongly suggest that the recent increase in lung cancer among white males in the USA is due entirely or in large part to factors other than cigarette smoking.