Objective: Some studies have suggested that residential proximity to high traffic areas is associated with increased risk of childhood cancer, although the epidemiologic evidence to date has been mixed. This study takes advantage of available information on population-based cancer reporting and various spatially assigned indices of traffic in a sufficiently large and heterogeneous area to obtain reasonably stable estimates of risk associations.
Methods: The time period 1988-1994 included a total of 7143 newly diagnosed cases of childhood cancer and 46 million child-years of observation in California. Rate ratios, estimated via Poisson regression (with adjustment for age, sex, and race/ethnicity), were computed for estimated traffic level as measured by spatial information on neighborhood vehicle density, road density, and traffic density.
Results: Compared to area air monitoring data, traffic density estimates were the most strongly correlated with measures of benzene and 1,3-butadiene. Rate ratios at the 90th percentile of traffic density (neighborhoods with over 320,700 vehicle miles traveled per day per square mile) were 1.08 (95% Cl 0.98-1.20) for all cancers in children, 1.15 (95% CI 0.97-1.37) for the leukemias, and 1.14 (95% CI 0.90-1.45) for the gliomas. There was also little or no evidence for rate differences in areas characterized by high vehicle or road density.
Conclusion: These data suggest that childhood cancer rates are not higher in high traffic neighborhoods, but future studies which can better refine timing and measures of exposure are needed to more directly address the question of etiologic risks.