Statistical modelling of lung cancer and laryngeal cancer incidence in Scotland, 1960-1979.

The ratio of cancer occurrence among smokers compared to nonsmokers has been generally reported to be greatest for carcinomas arising in the lung and larynx. This report presents the results of an investigation into mechanisms underlying the temporal trends in the incidence of these two forms of cancer in Scotland over the 20-year period, 1960-1979. The purpose of the study was to establish the independent effects of age, calendar time, and birth cohort on the observed pattern of incidence by adopting an alternative approach to this classical problem which can be employed when data are available as individual records. The most economic model found to explain the data involved the independent effects of site, sex, and age and different cohort effects in each sex. The cohort effect was more pronounced in men than women in older cohorts but, conversely, stronger in women than in men in younger cohorts, i.e., among those born most recently. This observation accords with the observed increase in the use of cigarettes in young women relative to young men. There appeared to be a small effect of calendar time which was the same in both sexes and both sites, although this main effect, it must be emphasized, could be eliminated from the model on statistical grounds. The effect can be interpreted to indicate the increase in the coverage of cancer registration in Scotland which has taken place since 1960. These results indicate that lung cancer risk in Scottish males will continue to fall, with the risk in women lagging 20 years behind, although the levels in women will probably not reach those attained by men.