International Journal of Hygiene and Environmental Health
Development of a GIS-based indicator for environmental pesticide exposure and its application to a Belgian case–control study on bladder cancer
Introduction
Quantitative assessment of environmental exposure to contaminants in epidemiological studies is often hard to achieve, as this requires extensive environmental measurements combined with the assessment of individual behavioral and demographic factors. Reconstructing past exposure, needed to address adequately the latency period of some health effects, is even more challenging. Indicators of exposure are therefore often used as a proxy of exposure, requiring less information and at least enabling the relative assessment of exposures between individuals or the classification of individuals into groups (Nieuwenhuijsen, 2006). The use of geographic information systems (GIS) has become a useful instrument in linking and analyzing spatially resolved data. However, the accuracy of the basic data and the uncertainty and possible errors in the data processing need to be assessed to the extent possible (Nuckols et al., 2004).
The objective of the study presented in this article was the development of GIS-based indicators for environmental pesticide exposure. The indicators were used in a case–control study that investigated possible causes of bladder cancer in the province of Limburg, Belgium (Kellen et al., 2006). A literature search did not reveal published studies on non-occupational exposure to pesticides and the occurrence of bladder cancer. Studies on farmer's pesticide exposure are inconclusive with regard to bladder cancer risk. Some case–control studies found a positive association (Fincham et al., 1992; Forastiere et al., 1993; Lavecchia et al., 1990; Miller et al., 1978), while others found no evidence for a higher risk of bladder cancer in farmers (Claude et al., 1988; Kogevinas et al., 2003; Silvermann et al., 1983; Vineis and Magnani, 1985). However, as agriculture and fruit production is an important economic activity in the region from which cases and controls were recruited, the assessment of environmental pesticide exposure is justified. The dominant process of off-site pesticide migration is the so-called drift process during application, merely impacting surface waters within some hundred meters distance (FOCUS, 2001). Post-application losses to air are transported further, by some kilometers (EPPO, 2003; Siebers et al., 2003). Losses to air will have been higher in the past due to less-developed application technology. Leaching to ground water and presence in surface water cannot be excluded.
Earlier studies that report the use of GIS for estimating exposure to environmental pesticide use are available. Ward et al. (2000) defined 500 m buffers around the subject's residence at the time of interview in a feasibility study in Nebraska. The exposure measure combined area of the crops in the buffer, average distance of the subject's residence at the time of interview to the centroids of the crop fields and probability of pesticide use. Crop maps were reconstructed from 1984 satellite imaging and historical records. Xiang et al. (2000) used similar methods to generate 300 and 500 m buffers and to calculate area of crops within these buffers in a study in Colorado. Brody et al. (2002) extended these methods by incorporating historical information on pesticide use, meteorological information and pesticide application type to estimate distance and direction of drift processes, and the persistent nature of the pesticide, allowing to reconstruct exposure history between 1948 and 1995 for Cape Cod, MA.
Application of available methodologies in new studies is often not straightforward, as the available information often imposes limitations, requiring inventive use of previous developments. The measures of exposure developed within this study therefore build further on the approaches presented above, but we accounted for the information available. We used information at three scale levels: (a) information at the individual's level, such as distance to crop fields; (b) information at the level of the municipality, such as time-series of crop area; and (c) regional information, such as pesticide use. In addition to using historical records on land use and pesticide application, we also took into account address history of the individuals in the study. Two indicators were developed. The first indicator, called PESTcrop, represents the proximity to fruit and vegetable fields in the surroundings of residences. The second indicator, called PESTpest, represents potential pesticide exposure by pesticide class caused by the presence of agricultural fields in the surroundings of residences. The indicators were constructed at the individual's level, for a period of 20 years (1984–2004), thus accounting for changes over time of the subject's residence, crop area and pesticide use. The construction of the indicators is explained and the results of the application of the indicators within the case–control study are shown.
Section snippets
Study population
The cases and controls included in the study were limited to actual inhabitants of the province of Limburg, Belgium (Fig. 1). Cases were recruited by urologists and were defined as a patient with histologically confirmed transitional cell carcinoma of the bladder, diagnosed in 1999 or later. Controls were selected from the registry of inhabitants of the province of Limburg by simple random sampling among citizens above 50 years of age. Because of the strict privacy law in Belgium, the
Results
Basic values for the calculation of the crop indicator and the pesticide indicator are not independent from each other. A correlation can be seen between normalized crop area (km2/km2) and pesticide pressure (kg active substance/km2), as is shown in Table 2. The results show a very strong correlation between fungicides and the group other pesticides on the one hand and orchards on the other hand, indicating that fungicides and the group other pesticides are predominantly used in orchards.
Discussion
The objectives of the study were the development of a GIS-based indicator for environmental pesticide exposure based on available data and its application in a case–control study on bladder cancer. Two indicator types were developed. The first indicator (PESTcrop) used information on crop area, whereas the second indicator (PESTpest) combined information on crop area, pesticide dosage and relative risk. Both indicators are not independent from each other as certain groups of pesticides are
Conclusions
Two indicators to assess exposure to pesticides were developed within a GIS environment. The first indicator assessed crop pressure, whereas the second indicator assessed pesticide pressure. Crop area, degree of rural living, pesticide dosage and relative risk were quantified over time and combined into a yearly measure of exposure per subject's residence. Uncertainty and possible errors of the indicators and their limitations were evaluated. Although the indicators addressed elements
Acknowledgments
This research is financially supported by the Flemish Government, the Province of Limburg and the Limburg Cancer Foundation.
References (31)
- et al.
Pesticide exposure of children in an agricultural community: evidence of household proximity to farmland and take home exposure pathways
Environ. Res.
(2000) - et al.
Relative importance of risk-factors in bladder carcinogenesis
J. Chronic Dis.
(1978) - et al.
Investigation on downwind short-range transport of pesticides after application in agricultural crops
Chemosphere
(2003) - et al.
A geographic information assessment of birth weight and crop production patterns around mother's residence
Environ. Res.
(2000) - et al.
Using GIS and historical records to reconstruct residential exposure to large-scale pesticide application
J. Expo. Anal. Environ. Epidemiol.
(2002) - et al.
Breast cancer risk and historical exposure to pesticides from wide-area applications assessed with GIS
Environ. Health. Perspect.
(2004) - et al.
Occupation and risk of cancer of the lower urinary tract among men – a case–control study
Int. J. Cancer
(1988) - EPPO, 2003. Environmental risk assessment scheme for plant protection products – Chapter 3: Air. EPPO Bulletin, vol....
- et al.
Children's exposure to chlorpyrifos and parathion in an agricultural community in Central Washington State
Environ. Health Perspect.
(2002) - et al.
Patterns and risks of cancer in farmers in Alberta
Cancer
(1992)
Cancer among farmers in Central Italy
Scand. J. Work Environ. Health
Residence location as a measure of environmental exposure: a review of air pollution epidemiology studies
J. Expo. Anal. Environ. Epidemiol.
Dietary habit profile in European communities with different risk of myocardial infarction: the impact of migration as a model of gene–environment interaction. The IMMIDIET study
Nutr. Metab. Cardiovasc. Dis.
Cited by (21)
OBOMod - Integrated modelling framework for residents' exposure to pesticides
2022, Science of the Total EnvironmentPesticide exposures for residents living close to agricultural lands: A review
2020, Environment InternationalCitation Excerpt :Some epidemiological studies suggest an association between proximity to agricultural lands and a wide range of associated adverse health outcomes including birth-related outcomes (e.g., pre-term birth, fetal growth restriction, neural tube defects, hypospadias, gastroschisis and anotia) (Carmichael et al., 2016; Gemmill et al., 2013; Larsen et al., 2017; Meyer et al., 2006; Rappazzo et al., 2016; Rull et al., 2006b), childhood cancers (e.g., leukemia and lymphomas) (Carozza et al., 2009; Gómez-Barroso et al., 2016; Jones et al., 2014; Malagoli et al., 2016; Reynolds et al., 2005b; Rull et al., 2009), cognitive impairments (e.g., autism spectrum disorders, diminished intelligence quotient (IQ), verbal comprehension and attention, as well as hyperactivity and cognitive decline) (Coker et al., 2017; Corral et al., 2017; Gunier et al., 2017a; Gunier et al., 2017b; Paul et al., 2018; Roberts et al., 2007; Rowe et al., 2016; Shelton et al., 2014), respiratory outcomes (e.g., asthma) (Raanan et al., 2017), adult cancer (e.g., breast cancer and brain tumors) (Carles et al., 2017; El-Zaemey et al., 2013), Parkinson’s disease (Brouwer et al., 2017; Costello et al., 2009; Manthripragada et al., 2010; Wang et al., 2011; Wang et al., 2014) and amyotrophic lateral sclerosis (Vinceti et al., 2017). However these associations are generally weak or inconclusive and contradictory results have been observed in other epidemiological studies (Brody et al., 2004; Bukalasa et al., 2018; Carmichael et al., 2013; Carmichael et al., 2014; Clementi et al., 2007; Cornelis et al., 2009; Reynolds et al., 2004; Reynolds et al., 2005a; Shaw et al., 2014; Shaw et al., 2018). All the above-cited studies have limitations and weaknesses which could explain these inconsistent findings.
Passive exposure to agricultural pesticides and risk of childhood leukemia in an Italian community
2016, International Journal of Hygiene and Environmental HealthCitation Excerpt :Our a priori hypothesis was that for children living adjacent to cultivated fields, passive exposure to drift of the pesticides used for the specific crop types was not negligible (Lu et al., 2000; Rubino et al., 2012). We selected a 100-m buffer around homes since it represents the distance where substantial pesticide drift occurs according to some studies (Wittich and Siebers, 2002; Siebers et al., 2003; Martin et al., 2008; Wolters et al., 2008; Cornelis et al., 2009; Garron et al., 2009), taking into account that in Italy agricultural fields are sprayed only by ground-based devices and therefore larger drifts are difficult to assume (Tsakirakis et al., 2014). In particular, for arable crops, sprayers’ devices are kept in a horizontal position, very close to the plants, so the dispersion of pesticides is very limited.
Pre- and postnatal exposures to pesticides and neurodevelopmental effects in children living in agricultural communities from South-Eastern Spain
2015, Environment InternationalCitation Excerpt :Creatinine concentration in children's urine was determined using Jaffe's Method in a Hitachi 917 automatic chemistry analyzer. Exposure indices were developed to assess pre- and postnatal exposures to pesticides as described by Cornelis et al. (2009). Postnatal exposure spanned from birth to the time of neuropsychological assessment.
Pesticides exposure modeling based on GIS and remote sensing land use data
2015, Applied GeographyCitation Excerpt :When integrated with pesticide usage data within a geographic information system (GIS), such information could be further processed to model the exposure at a satisfactory level of accuracy (Nuckols, Ward, & Jarup, 2004). In addition, historical remote sensing imageries provide the possibility of recovering pesticide exposure information decades ago, which is especially useful for assessing exposure duration, a factor that influences chronic diseases such as Parkinson's Disease (Brody et al., 2001; Cornelis, Schoeters, Kellen, Buntinx, & Zeegers, 2009; Liou et al., 1997). Progresses have been made towards this goal in recent decades.
Mapping water and health: Current applications and future developments
2011, Current Opinion in Environmental SustainabilityCitation Excerpt :These include natural water bodies, such as surface water [8,10–17] and groundwater [6,18–22], drinking water [15,23–27], recreational water [13], waste water [28] and water for irrigation and aquaculture. Chemical [16,18,22,23,27,29–35], physical [26,36] parasitological [7•,8••] and microbiological contaminants [37,38] are a matter of concern and different water-related health outcomes. For instance, water bodies which provide breeding environments for disease vectors [39,40] and acute [13,37] and chronic [17,18,28,30,31,37–39] illness are investigated by GIS-based mapping and geostatistical analysis.