Residential proximity to agriculture and risk of childhood leukemia and central nervous system tumors in the Danish national birth cohort.

BACKGROUND: Living in an agricultural area or on farms has been associated with increased risk of childhood cancer but few studies have evaluated specific agricultural exposures. We prospectively examined residential proximity to crops and animals during pregnancy and risk of childhood leukemia and central nervous system (CNS) tumors in Denmark. METHODS: The Danish National Birth Cohort (DNBC) consists of 91,769 pregnant women (96,841 live-born children) enrolled in 1996-2003. For 61 childhood leukemias and 59 CNS tumors <15 years of age that were diagnosed through 2014 and a ~10% random sample of the live births (N = 9394) with geocoded addresses, we linked pregnancy addresses to crop fields and animal farm locations and estimated the crop area (hectares [ha]) and number of animals (standardized by their nitrogen emissions) by type within 250 meters (m), 500 m, 1000 m, and 2000 m of the home. We also estimated pesticide applications (grams, active ingredient) based on annual sales data for nine herbicides and one fungicide that were estimated to have been applied to >30% of the area of one or more crop. We used Cox proportional hazard models (weighted to the full cohort) to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the association of childhood leukemia and CNS tumors with crop area, animals, and pesticide applications adjusted for gender and maternal age. RESULTS: Sixty-three percent of mothers had crops within 500 m of their homes during pregnancy; winter and spring cereals were the major crop types. Compared to mothers with no crops <500 m, we found increasing risk of childhood leukemia among offspring of mothers with increasing crop area near their home (highest tertile >24 ha HR: 2.0, CI:1.02-3.8), which was stronger after adjustment for animals (within 1000 m) (HR: 2.6, CI:1.02-6.8). We also observed increased risk for grass/clover (highest tertile >1.1 ha HR: 3.1, CI:1.2-7.7), peas (>0 HR: 2.4, CI: 1.02-5.4), and maize (>0 HR: 2.8, CI: 1.1-6.9) in animal-adjusted models. We found no association between number of animals near homes and leukemia risk. Crops, total number of animals, and hogs within 500 m of the home were not associated with CNS tumors but we observed an increased risk with >median cattle compared with no animals in crop-adjusted models (HR = 2.2, CI: 1.02-4.9). In models adjusted for total animals, the highest tertiles of use of three herbicides and one fungicide were associated with elevated risk of leukemia but no associations were statistically significant; there were no associations with CNS tumors. CONCLUSIONS: Risk of childhood leukemia was associated with higher crop area near mothers' homes during pregnancy; CNS tumors were associated with higher cattle density. Quantitative estimates of crop pesticides and other agricultural exposures are needed to clarify possible reasons for these increased risks.