Estimation of human intake of pesticides from all potential pathways

The uses of pesticides and other similar products are carefully regulated in Great Britain. Before a product can be used it must undergo a review of its toxicity and potential human exposure to ensure that the risks are acceptable. Exposure during application and in other occupational situations is assessed along with people living near to fields or who just happen to be in the vicinity of fields being treated, i.e. bystanders. Consideration is also given to the possible risks to consumers from eating food containing small amounts of pesticide residues. However, the regulatory processes generally only consider individual products and do not take into account exposure to other pesticide active ingredients with similar toxicity, to other products at different times or to the consumption of many different foods that may each contain pesticide residues. This project has aimed to make an assessment of the exposure to pesticide mixtures from all potential pathways by creating a mathematical model of pesticide exposures from multiple active ingredients in food, or where exposure comes from occupational or bystander scenarios. We have devised a suitable theoretical framework for the exposure modelling, building upon existing models and data. We have collected together data on pesticide residues in food, food consumption, recipes for processed foods, patterns of exposure in agriculture, estimates of the numbers of people employed in agricultural uses of pesticides, numbers of people who may be bystanders when spraying of pesticides takes place in agriculture, and estimates of exposure to agricultural pesticides – both from applicators and bystanders. A total of 21 pesticides were initially selected for the project, although we were unable to identify food residue data for three of these and so they were dropped from the list for study. The compounds were selected to be representative of substances that have shown anti-cholinesterase activity or which were oestrogen agonists. It was not intended that we select every single compound with these toxicological effects but rather that we selected a representative selection of such materials. The selection was based on the usage of the compounds in the UK, and on their occurrence as residues in food. Compounds of particular concern such as beta oestradiol were also included in the list, which was agreed with FSA at the beginning of the project. The selected pesticides were used in agriculture, as biocides and in a small number of instances as veterinary medicines. There was a great deal of information available, but equally there were a number of areas where there were little or no data to assist in developing the model. For example, there were data on the levels of exposure likely to occur in the use of biocides containing the same active ingredients, i.e. non-agricultural pesticides, but there was no information about the usage of these materials or the numbers of people exposed. There were no data to develop the model for relevant veterinary medicines, although only one of the selected pesticides was used in these products. There is also no reliable information about the effect of processing on residues on food. In the model we have chosen to assume that processing may have a variable impact on residues with between no reduction and complete removal. There were also difficulties with the interpretation of the very low levels of contamination that are present in foods. In the majority of cases there was no recorded residue, but for some of these situations there could have been contamination detected but it was below the Reporting Limit or there may have been contamination present below the analytical detection limit. In both of these cases there would have been very low levels of pesticides present rather than none. We devised a simple algorithm to estimate the residues in these cases. The model for food consumption uses summarised data on pesticide residues in combination with data from the National Diet and Nutrition Survey along with data on the basic food components in common processed foods, i.e. recipe data. Data on occupational exposure were obtained from the EUROPOEM model and from periodic surveys of pesticide usage carried out for PSD. Data on the numbers of potential bystanders were estimated separately. The model produced simulations of internal dose using a simple single compartment pharmacokinetic model. It simulates the pathways from ingestion (food consumption) or skin exposure (in occupational settings) to the internal dose for each chosen compound separately. The compound mixture internal dose is then estimated as the sum total of the individual compound dose estimates. In this way it was possible to combine data from different exposure pathways in a way that took account of the likely residence of the compound in the body. To assess the effect of the mixture of pesticides we have also simulated the body mass of the “people” and used the Acceptable Daily Intake (ADI) to normalise the internal dose. We have selected to use ADI rather than the Acceptable Operator Exposure Level (AOEL) or the Acute Reference Dose (ARfD). In practice the AOEL and ARfD are generally quite similar and the exact choice of value would have limited impact on the results of the study. For acute exposures the ARfD would probably be more appropriate, although for simplicity we have used the ADI throughout the report. Exposure was seen to occur irregularly throughout the year, regardless of whether the source was from food or from use of pesticides in agriculture, although in the latter case there were some seasonal effects. Internal dose was highest for application of pesticides in agriculture (farmers and contractors), next highest for bystanders and lowest for consumers. Internal dose for child consumers was less than for adults. The maximum dose estimates for individual pesticide compounds from food consumption were all much less than the corresponding ADI dose and the aggregate exposure normalised to the ADI dose was also much less than unity, i.e. below an “aggregate” ADI dose. Exposure estimates associated with occupational exposures provided aggregate dose estimates that were in many cases higher than the aggregate dose equivalent to the ADI, particularly for farmers and contractors. Our simulations suggest that there may be some people in the population living near to spraying activities who are bystanders and those who are occupationally exposed who may have unacceptably high exposures, but this conclusion is dependent on the accuracy of the EUROPOEM model and it may be that as a regulatory model it overestimates the true exposure received by individuals.The impacts upon the regulatory systems in place in the UK were assessed, based on the recommendations of the COT report and the availability of data and information used for the modelling. To respond to the issues of possible human health effects of exposure to a mixture of pesticides the regulatory framework would need a more co-ordinated approach to prioritise the need for risk assessments to be carried out for exposure to more than one pesticide. However such as assessment would need to consider all sources and routes of exposure and could only practically be undertaken as a periodic review of groups of compounds of concern. In cases where exposure to a mixture of compounds was considered to result in a harmful dose there would be a need to regulate for this. In some cases this may require regulatory authorities to consider the removal or restriction of uses of certain compounds that are used in pesticides currently authorised or approved for use. In such cases there could be a conflict of interest, particularly when compounds with very different uses were being evaluated, such as the case where a compound is used for veterinary medicine purposes and also as an insecticide to prevent vector borne crop viruses. Regulatory decisions would need to be taken as to whether it is feasible to restrict the use of one or both products, the implications for restricting use, and the availability of alternative control methods.The information currently available to regulators is not adequate to allow the risk assessment to be carried out without some degree of associated uncertainty. A limited sensitivity analysis suggested that the conclusions from our simulations are not unduly sensitive to these uncertainties, particularly in the case of pesticide residues in food. To assess the risk of exposure to mixtures, more detailed information on the sources and routes of exposure, particularly for dietary exposure, would be helpful. This has consequences for the way residue surveillance programmes are organised together with a reduction in the reporting limits for residues in food products. Data would be needed on the patterns of use for biocides and veterinary medicines comparable to the patterns that exists for agricultural pesticides in the UK.