Search

Search Results

• 1. [Article] Leukemia incidence and benzene air pollution in Portland Oregon

  Benzene is a widespread urban pollutant and monitoring results show concentrations are particularly high in Portland, Oregon. Nationally, benzene air pollution is the largest contributor to total cancer ...

  Citation × Citation Citation Abstract Copy Title: Leukemia incidence and benzene air pollution in Portland Oregon Author: Voss, Robert W. Benzene is a widespread urban pollutant and monitoring results show concentrations are particularly high in Portland, Oregon. Nationally, benzene air pollution is the largest contributor to total cancer risk from air pollutants. We used data from several data sets including: modeled benzene distribution at the census block group level from the Oregon Department of Environmental Quality, ten-year leukemia incidence from the Oregon State Cancer Registry, and demographic data from the U.S. Census to examine the correlation of benzene with acute myeloid leukemia (AML), total myeloid leukemia (TML) and total leukemia subgroups. Modeled benzene values were averaged to census tracts using GIS methods. We performed Poisson regression on data for all census tracts and also aggregated data to quartiles of benzene concentration and modeled exposure to determine incidence rate ratios (IRR). We mapped the distribution of benzene, and agestandardized leukemia rates to examine spatial patterns which may not be evident in statistical analysis. Our regression results found no significant association between benzene and leukemia at the census tract level. AML IRR was slightly elevated for females at the second, but not the third and fourth quartiles of benzene modeled exposure. We found non-significant elevations in AML IRR for those less than 19 years of age at the second and fourth, but not the third quartiles of benzene modeled exposure. We did not observe a monotonically increasing trend toward higher incidence of leukemia with higher modeled benzene exposure. Nor did we see a tendency toward a stronger association in the myeloid types of leukemia. Similar modeling results were produced for perchloroethylene, a cooccurring air toxin that is not considered a leukemiagen. Although the general spatial pattern of leukemia incidence does not appear visually similar to the benzene distribution, some census tracts near point source emissions of benzene had AML rates in the highest bracket. There were some census tracts with elevated TML in the under 19 year old age group near major highways. Our overall result shows no association between benzene and leukemia at the census tract level. Census tract level ambient air pollution data such as ours is homogenous within census tract measurement units. Most census tracts in our study area are between one and seven miles across so variability of ambient exposure within a few hundred yards of major roads was not captured in our air pollution data. This misclassification of exposure, along with misclassification of case residence in time and place limits the utility of an ecologic design such as ours to detect small changes in disease incidence. The magnitude of cancer risk from ambient benzene is very small, so any misclassifications of exposure or case residence may render even large datasets incapable of detecting elevated incidence rates. Future efforts to investigate health effects of ambient benzene or other mobile source pollutants could incorporate road and residence location as well as modeled air pollution data at the census tract level to achieve more definitive results. Title: Leukemia incidence and benzene air pollution in Portland Oregon Author: Voss, Robert W. Benzene is a widespread urban pollutant and monitoring results show concentrations are particularly high in Portland, Oregon. Nationally, benzene air pollution is the largest contributor to total cancer risk from air pollutants. We used data from several data sets including: modeled benzene distribution at the census block group level from the Oregon Department of Environmental Quality, ten-year leukemia incidence from the Oregon State Cancer Registry, and demographic data from the U.S. Census to examine the correlation of benzene with acute myeloid leukemia (AML), total myeloid leukemia (TML) and total leukemia subgroups. Modeled benzene values were averaged to census tracts using GIS methods. We performed Poisson regression on data for all census tracts and also aggregated data to quartiles of benzene concentration and modeled exposure to determine incidence rate ratios (IRR). We mapped the distribution of benzene, and agestandardized leukemia rates to examine spatial patterns which may not be evident in statistical analysis. Our regression results found no significant association between benzene and leukemia at the census tract level. AML IRR was slightly elevated for females at the second, but not the third and fourth quartiles of benzene modeled exposure. We found non-significant elevations in AML IRR for those less than 19 years of age at the second and fourth, but not the third quartiles of benzene modeled exposure. We did not observe a monotonically increasing trend toward higher incidence of leukemia with higher modeled benzene exposure. Nor did we see a tendency toward a stronger association in the myeloid types of leukemia. Similar modeling results were produced for perchloroethylene, a cooccurring air toxin that is not considered a leukemiagen. Although the general spatial pattern of leukemia incidence does not appear visually similar to the benzene distribution, some census tracts near point source emissions of benzene had AML rates in the highest bracket. There were some census tracts with elevated TML in the under 19 year old age group near major highways. Our overall result shows no association between benzene and leukemia at the census tract level. Census tract level ambient air pollution data such as ours is homogenous within census tract measurement units. Most census tracts in our study area are between one and seven miles across so variability of ambient exposure within a few hundred yards of major roads was not captured in our air pollution data. This misclassification of exposure, along with misclassification of case residence in time and place limits the utility of an ecologic design such as ours to detect small changes in disease incidence. The magnitude of cancer risk from ambient benzene is very small, so any misclassifications of exposure or case residence may render even large datasets incapable of detecting elevated incidence rates. Future efforts to investigate health effects of ambient benzene or other mobile source pollutants could incorporate road and residence location as well as modeled air pollution data at the census tract level to achieve more definitive results. Close