Performance of different exposure assessment approaches in a study of bitumen fume exposure and lung cancer mortality

Background We compared performance of different exposure assessment approaches in a cohort study of cancer risk among European asphalt workers. Methods Three bitumen fume exposure indices (duration of exposure (years), average exposure (mg/m(3)) and cumulative exposure (mg/m(3) *years)) and two latency models (with and without a 15 year lag) were considered for an association-between lung cancer mortality and bitumen fume. Results There was no association between lung cancer risk and either duration or cumulative exposure. However, there was the suggestion of an increase in lung cancer risk accompanying rise in average exposure. Only models with average bitumen fume exposure (with or without lag) improved model fit, albeit to the same extent. Conclusions Constructing quantitative indices of exposure intensity was justified because they produced the greatest improvement in fit of models that explored possible relationship between bitumen fume exposure and lung cancer risk. The identified associations require further investigation. (C) 2003 Wiley-Liss, Inc.