| It is hypothesized that the damage to health from smoking tobacco is principally the effect of residual damage caused by mutagenic substances, documented to be present in tobacco smoke, and that the gross effects of genetic damage are correctly modeled by the Cumulative Damage (CD) model, a compound Poisson model. Using data from the CPS-I study of the American Cancer Society, analyses are undertaken to demonstrate the fit of the CD model to mortality data for lung cancer, coronary heart disease (CHD) and cerebrovascular disease (CVD) in never smokers and current smokers, with separate analyses for men and women. The background rate of cumulative damage is fit to never smoker data. For smokers, an additional cumulative damaging process is fit for the period of smoking, proportional to cigarettes per day (CPD). Further, it is demonstrated that the fit of the CD models can be approximated by estimating models fit to Poisson assumptions, with variables of CPD, duration of smoking, and age, models more easily fit and evaluated than CD models. Finally, the models developed for current smokers are used to predict rates for former smokers, dependent on the fixed duration and CPD level of former smoking. These predictions are tested against the observed data for former smokers, demonstrating a generally excellent fit of the cumulative model to observed rates. Tests of alternate models show no significant pattern of decline of rates from the cumulative estimates with increasing years of cessation for any of the diseases tested. Tests of mean differences between predictions and observed rates generally showed no significant differences, except in the case of CHD in males, where the observed rates showed a constant negative difference of 30 deaths (-48, -12) per 100,000 person-years less than predicted rates, which can be interpreted as attributable to transient non-cumulative risk factors associated with smoking, the dominating portion of CHD risk fitting the cumulative pattern. Overall, these results provide strong support for the cumulative model of damage from tobacco smoke, which is taken as evidence of the general applicability of the cumulative model in understanding damage from mutagenic factors. |
