Abstract
There has been growing concern about the health effects of agricultural pollutants among the rural populations, who by virtue of living near the sources of these pollutants are among the most exposed. The aim of this thesis was to make use of large national administrative databases to identify possible links
... read more
between living near sources of agricultural pollutants, namely pesticides used in crops and livestock farming emissions, and health of the Dutch rural population that could point to more specific (groups of) exposures for further confirmatory research. From the Dutch Environmental Longitudinal Study (DUELS), a national census-based cohort, we selected about 4 million subjects living in rural areas for whom we computed the average number of cattle, pigs, chicken, and mink present in 500m, 1000m, 1500m and 2000m around their residences. Using Cox proportional hazards regression and adjusting for potential confounders at individual and neighborhood level, we found evidence that living up to 2000m of pig farms was associated with increased respiratory mortality, namely from chronic lower respiratory diseases. Using the same cohort but restricted to ~3.1 million subjects, we studied the potential effects of living near pesticide-treated crops and cause-specific mortality. We estimated the area of specific crop groups cultivated within buffers of 50m, 100m and 250m around subjects’ residences and the amount of fungicides, herbicides and insecticides used within those buffers. Using the previous statistical approach, we observed overall decreased mortality risks but noted an increased risk for chronic lower respiratory diseases mortality and proximity to maize cultivation. We found no evidence of an association between the amount of pesticides used and cause-specific mortality. Using the Public Health Monitor national survey from 2012, we selected ~217.000 participants who lived in rural areas. As previously, we estimated area of specific crop groups cultivated near participants’ residences and investigated its association with mental and perceived health in a cross-sectional study. Area and individual-level potential confounders (including lifestyle factors, such as alcohol and drug use, smoking and physical activity) were included in logistic regression analyses. These revealed no evidence that residential proximity to pesticide treated-crops was associated with psychological distress or poorer perceived health. Finally, we explored associations between residential exposure to specific pesticides during pregnancy and birth outcomes using the Dutch birth registry. We estimated the amount (kg) of 139 active ingredients (AIs) used during pregnancy within buffers of 50m, 100m, 250m and 500m around residences of ~340.000 mother-infant pairs. Combining generalized linear models and a variable selection method, we observed that maternal residential exposure to linuron, glufosinate-ammonium, fluroxiypyr-meptyl, vinclozolin and picoxystrobin were associated with potentially adverse birth outcomes. Despite limitations regarding availability, level of detail and potential for misclassification of exposures, outcomes and confounders, administrative data are useful in exploratory epidemiological studies by providing large study populations in a time and cost efficient set-up. By granting statistical power to detect subtle effects and investigate rare exposures and outcomes, they helped identify relevant associations and narrow the scope of agricultural pollutants candidates for more targeted studies.
show less
