Ae. Kunst et al., DETERMINANTS OF REGIONAL DIFFERENCES IN LUNG-CANCER MORTALITY IN THE NETHERLANDS, Social science & medicine, 37(5), 1993, pp. 623-631
Although regional differences in lung cancer mortality are likely to b
e attributable to regional differences in tobacco smoking, studies in
various countries found only weak relationships. This paper aimed at e
xplaining regional differences in lung cancer mortality in the Netherl
ands. In a first step, clues for the role of smoking were obtained fro
m a detailed description of regional mortality differences. These diff
erences were found to be strongly determined by cohort effects: they v
ary between birth cohorts, and have been stable for over 30 years. Reg
ional mortality differences reflect a diffusion of the lung cancer epi
demic from high-income regions to low-income regions. These findings a
re suggestive of a relationship with regional differences in trends in
cigarette smoking. In a second step, by means of multiple regression
analysis, mortality differences in 1980-84 were related to available d
ata on cigarette smoking and two other possible risk factors: work in
transport and manufacturing industry, and air contamination. The indep
endent variables referred to the 1970s. Positive associations with var
ious smoking measures were found for women, but for men the associatio
ns were weak or non-existent. Mortality differences among men 45-64 ye
ars were associated with work in transport and manufacturing industry.
Strong associations with air contamination were found for men and wom
en 65 years and older. Additional analysis showed that regional differ
ences in lung cancer among old men were strongly associated with smoki
ng in 1930, i.e. half a century before. Changes in the regional patter
n of tobacco consumption between 1930 and 1970 explain why smoking in
1970 is not associated with mortality differences in the 1980s. This s
hows that the explanation of regional lung cancer death rates sometime
s has to go far back in time. The strong association with air contamin
ation does not constitute firm evidence of an effect, but should be co
nsidered as a stimulus to carry out further analysis on the excess lun
g cancer mortality in polluted regions.