Agricultural Chemical Exposures and Childhood Cancer

J.R. Nuckols, principal investigator

Ground Spray Equipment The State of California in the USA maintains a Pesticide Use Reporting Database (CPUR) that is one of the most comprehensive in the world. Data from CPUR is being used with increased frequency in epidemiological studies to evaluate the potential health effects of agricultural pesticides. We evaluated (1) the spatial reporting accuracy of the CPUR for 10 crops with the highest pesticide applications, (2) the utility of historical crop maps to refine the location of agricultural pesticide applications within the reporting unit for CPUR (an approximately one square mile section), and (3) the of effect of such a refinement on exposure classification for an epidemiological study. Our study was conducted in 6 counties in California’s Central Valley, one of the major agricultural and pesticide use regions in the USA. We used a geographic information system (GIS) to compare the location of pesticide applications by crop as reported in CPUR, with the location of the same crop in a California Department of Water Resources (CDWR) crop map database for each section in two counties (N = 3,856). We found good correspondence between crops reported in CPUR and the CDWR crop maps. Agreement between the two datasets for broad crop categories (i.e. row/field, orchard/vineyard) was >80% in 1988, and >87% in 1991, indicating a likely improvement in reporting accuracy over time. Agreement for individual crops ranged from 73% to 95%.

We also used a GIS to assign CPUR pesticide use to specific crop fields in the CDWR database and created a crop map metric that estimated pesticide use at a higher spatial resolution than CPUR. We then compared categorical (exposed/unexposed) and quantitative (pounds/ mi2) pesticide use predicted by the two metrics within a 500 meter radius (0.3 mi2 ) buffer around 577 residences from a childhood leukemia study conducted by the California Department for Health Services (CDHS). 500 meters is a distance used in previous studies of pesticide exposure due to primary drift from ground spraying operations. We evaluated the metrics for 6 pesticides with high use in the study area: simazine and trifluralin (herbicides), dicofol and propargite (insecticides), and methyl bromide and metam sodium (fumigants). We found high level of agreement and a statistically significant association between the two metrics in their categorical assignment of exposure for all of the 6 pesticides, except metam sodium for which the data was too skewed to make these determinations. However, we also found significant differences between the two metrics in their estimates of application amounts in the residential buffers (pounds/mi2) for all 6. When we restricted the analysis to residences classified as exposed, we observed very large differences in predicted pesticide use for all 6 pesticides. In all cases, the amounts were much lower for the CDWR-based metric. For example, median 25th and 75th percentile values were orders of magnitude lower for simazine and propargite, and greater than 5-fold lower for methyl bromide. Our results suggest that using CPUR data alone to quantify potential agricultural pesticide exposure may substantially overestimate the magnitude of pesticide applications within 500 meters of residences. To verify the optimal exposure metric for exposure assessment, the metrics should be validated, by comparing them with environmental and biological measurements of pesticides.

View the Project Report in PDF Format ( 1.4 MB download )