There is a growing demand for guarantees of the safety of the food supply for the purposes of protecting consumer health and facilitating international trade. It is therefore extremely important that risk assessors and regulatory authorities have data and tools that allow insight into all aspects related to the safety of the food supply, including consumer exposure to food chemicals such as food additives, pesticide residues, micronutrients, mycotoxins and so forth.
Many of the current methods for estimating exposure to food chemicals are limited by the fact that they do not take account of variability and uncertainty in food chemical occurrence and concentrations.
Therefore, the objectives of this project are:
- To develop a comprehensive set of mathematical algorithms, purpose built to take account of all the necessary components for stochastic modelling of a variety of food chemicals and to develop appropriate computer software.
- To conduct a multi-centre study, using existing national data, to explore the influence of input distributions on model output for the key components of a stochastic model of food chemical intake (i.e. food intake, chemical occurrence, chemical concentration, market share, brand loyalty, correlated foods).
- To generate databases of true intakes of (i) food additives, based on brand level food consumption and ingredient composition, (ii) pesticide residues, based on duplicate diets and (iii) nutrients, based on biomarker studies.
- To assess the validity of the developed stochastic modelling software against true intakes, to conduct a comprehensive sensitivity analysis of validated models, and to compare these intakes against those derived using current approaches to exposure assessment.
- To provide a comprehensive set of practical guidelines for the appropriate use of stochastic modelling of food chemical intake and to provide guidelines on the correct interpretation of the output of stochastic modelling.
- To actively communicate all research findings to national authorities and scientific bodies as well as standardisation bodies involved in food chemical exposure assessment at regular stages through the project and, furthermore, to incorporate their feedback in the development of the software and guidelines.
Following acquisition of staff, the protocols for the various workpackages were finalised at the first plenary meeting. At this first meeting there was also a training workshop on the use of @ RISK and BestFit (two commercially available software packages used for fitting distributions and modelling of data). By month 3 of the project, a website providing information on the objectives and participants involved in the project, had been established. Workpackage 2, comprising a series of numerical experiments to explore how mode of inputting data can influence the output of stochastic models was still in progress by the end of the first reporting period. While 6 out of the 7 partners were engaged in this task involving the influence of input components on model output, they were also involved in workpackage 3. This workpackage entails the collection of primary data (Partners 5, 6 & 7) and the collection of ancillary data (Partners 1, 3 & 4) necessary to furnish existing food intake databases in order that each partner can engage in the validation studies. The validation of stochastic modelling of exposure to food additives, pesticides or micronutrients against true intakes based on brand level databases (Partners 1, 4 & 6), duplicate diets (Partners 5 & 7) or biomarkers (Partner 3) will comprise the work of workpackage 4 commencing at month 18. A primary set of algorithms for modelling food chemical intakes has been developed by Partner 2. These algorithms were then incorporated into a software programme which through the course of the first reporting period has been developed and advanced based on feedback from the partners and their results arising from the work undertaken in workpackage 2. An extensive list of relevant end-users have been identified and contacted with respect to the Monte Carlo project.
During the second reporting period for this project, the following outcomes have been achieved:
- Based on feedback from the partners, the software has been advanced to (I) facilitate upload of all databases, of varying structures, identified as being necessary for modelling and (ii) incorporate the algorithms necessary for running the models.
- Research into the selection of input data and distributions and identification of correlations and dependencies for probabilistic modelling of food chemical exposure was completed. The results of this research were compiled as a report and posted on the project website (www.iefs.org/montecarlo). Interested parties and potential end-users were contacted by e-mail to make them aware of the report.
- Partners completed the work of generating detailed databases of exposure (P1 and P4 fully recoded existing national food consumption databases to brand level and assessed the presence and concentration of selected additives in these brands. P3 compiled information about the major food sources of selected nutrients, variability in content and factors influencing bioavailability for an existing survey of 444 Dutch adults; P5 and P7 completed duplicate diet studies of approximately 250 Dutch and Spanish infants respectively that were then analysed for selected pesticide residues; P6 completed a survey of sweetener intakes in 337 Italian teenagers over 3 periods of 4 consecutive days of recoding)
- Partners designed validation protocols for WP4 and commenced work on the validation studies.