Tobacco consumption and pancreatic cancer mortality: what can we conclude from historical data in Australia?

BACKGROUND: Tobacco consumption is an established risk factor for pancreatic cancer yet studies of long-term mortality trends have not statistically analysed this relationship. We sought evidence for this relationship based on an analysis of long-term population-level data in Australia.
METHODS: Pancreatic cancer mortality data from 1931, tobacco consumption data and fruit and vegetable consumption data for Australia were utilized. Log-linear Poisson regression models were used to analyse pancreatic cancer mortality from 1931 with cumulative cohort and lagged time-specific tobacco consumption data and fruit and vegetable consumption data.
RESULTS: Pancreatic cancer mortality rose steadily for males until it began falling from the 1970s, and continued rising for females until 2006. These trends correspond with a long-term rise in male tobacco consumption until the 1960s and a later peak for females. Our models show that cumulative tobacco consumption predicts pancreatic cancer mortality for both sexes but with time lags only being significant for males. Fruit and vegetable consumption provides a protective effect against mortality in some of the models.
CONCLUSION: The success of smoking reduction programmes in Australia has contributed to the decline in pancreatic cancer mortality for males, providing important evidence about the need for tobacco control measures in populations where it is still increasing. Continued declines in female tobacco consumption should lead to a reversal of the long-term rise in female pancreatic cancer mortality.