English Abstract In the Netherlands information about air pollution,
industrial activity, cigarette smoking and lung cancer mortality was
available for all 40 areas (COROP-regions) over the period 1969-1978.
Differences in lung cancer mortality between the areas could be due to
differences in smoking patterns, industrial exposure or air pollution. This
was investigated using weighted linear regression models. Gender specific
analyses were carried out, using the data on the 45-74 year old only.
Within these ages the directly standardized lung cancer mortality rates (per
100,000) ranged from 151.7 to 301.7 (men) and 6.1 to 41.1 (women)
between the regions. The regression model for lung cancer mortality and
smoking explained little of the variance in the data. The regression
coefficients from this model were used to "adjust" the mortality rates for
smoking, and the corrected data were regressed on industrial activity and
SO2, an indicator of air pollution. The regression for industrial activity
resulted in a better explanation of the observed variance. The branches of
industry with the highest regression coefficients were the petrochemical,
chemical and construction industry and relevant for men only. The
regression for SO2 resulted in a marginally better explanation of the
observed variance. There was a clear cross correlation between
air-pollution and industrial activity which obscures the results. The
correlations found in this investigation are not necessarily causal but are
consistent with other investigations. It is concluded that the value of
ecologic analysis is limited not in the least due to the poor quality and
availability of relevant data.