ISEE/ISEA 2006 Conference Abstracts Supplement: Session Abstracts: Abstracts
*Department of Geography and †Department of Epidemiology & Biostatistics, Texas A&M University, College Station, TX
The overall goal of this epidemiologic case-control study is to estimate risk of developing specific childhood cancers associated with prenatal exposure to pesticides used in agricultural settings in Texas. During the study period, all but one of the 5 top crops by acreage in Texas were treated with chemicals that show evidence of possible human carcinogenicity, and many of these crops are sprayed from crop dusting airplanes, increasing potential for off-target drift. Cotton was chosen for evaluation of potential pesticide drift exposure because of the high-volume, repeated aerial spraying of this crop.
The study area comprised 12 Texas counties with intense cotton farming. To identify specific cotton fields during the study period, Landsat 5 TM data were multitemporally merged and then classified through supervised classification. Training data for the classification were obtained through land cover identification and location with GPS of more than 200 cotton fields. Once training and image classification was completed, an accuracy assessment was conducted to determine the effectiveness of the classification.
Meteorologic data were obtained for morning hours (likely spray time) for locations and time windows (months) of interest. Windows corresponded to times when potential in utero exposure could have occurred. Meteorologic data and other parameters were used as inputs for drift modeling (AgDrift v.2.03). Wind direction-specific model output was then integrated in a GIS for each cotton field and each time window, and potential exposures were assigned to individuals based on model-predicted pesticide deposition and air concentration at geo-coded birth residence.
Landsat 5 TM data effectively covered the study area and time period and frequent overflights allowed for the inclusion of multiple dates per year. Mutiltemporal data merging combined images obtained on different dates and proved effective for agricultural crop identification. Estimated exposure was primarily driven by wind velocity and direction. For some individuals, multiple fields contributed to exposure.
Discussion and Conclusions:
These GIS-based methods allow for the assessment of potential historical exposure to cotton pesticides of children living in farming communities and will allow for expansion of the study area. Information resulting from these exposure assessment tools is being used for epidemiologic analysis of risk of specific childhood cancers. Because causes of cancer in children continue to largely elude identification, leaving few avenues for developing prevention strategies, the possible role of agricultural pesticide exposure in the development of childhood cancers offers a tantalizing potential for preventing at least some of these cases.
© 2006 Lippincott Williams & Wilkins, Inc.