Regional lung cancer death rates unrelated to smoking? The case of The Netherlands

It has been observed in various countries that regional variation in lung cancer mortality can hardly be explained by differences in current smoking. This paper addresses the question of whether mortality variation within. The Netherlands in 1980-1984 is due to differences in past instead of current smoking. A first indication of the role of past smoking is that, within male birth cohorts, regional mortality patterns have been very stable for over 30 years. Reliable data on smoking in 1972 explain 40% of the mortality variation among women, but only 2% of that among men. A crude indicator on smoking in 1930 explained 43% of the mortality variation among men aged 75+ years (correlation = 0.66). The lack of a relationship with smoking in 1972 appears to be due to a radical change in regional smoking differences, which caused smoking in 1972 to be unrelated to smoking in 1930; long time lags, so that these changes were not yet followed by changes in regional mortality differences. It is concluded that the explanation of regional lung cancer death rates sometimes has to go far back in time. Studying determinants of lung cancer by means of regional analyses requires a more detailed control for smoking history than has been usual