Tiered approaches to exposure estimation: a basis for risk assessment

Public Health Significance of Urban Pests 511 These data often confirm previous findings and are used to put a more realistic perspective on exposure patterns. Fig. 14.5–14.7 give tiered approach sche- mes on the three different routes of exposure: inhala- tion exposure, dermal exposure and oral expo- sure.

14.7.2. Modelling residential exposures

In the absence of measured data on exposure or repre- sentative data on analo- gous substances, exposure must be estimated using recommended modelling approaches. To ensure that the predictions are realistic, all rele- vant exposure-related information on the substance should be used iteratively. General predictive models are available for generic substances and for specific scenarios. These models take into consideration the physical properties of active substances, such as the particle size of aerosols and volatility of liquids. Mathematical and empirical data- base models exist for a number of scenarios and tasks. Mathematical models for physical evapo- ration processes are nor- mally for a specific sub- stance and require data on such physical properties as saturated vapour pressure. Models for dispersive pro- cesses, such as spraying, typically apply to the pro- duct in use, emerging from the spray nozzle. Database models may be highly specific for exam- ple, an active substance discharged from a hand- held aerosol can or gene- ric for example, a product Pesticides: risks and hazards 510 mated from underlying physical processes, the physicochemical properties of the pro- duct, characteristics of the formulation and an understanding of the nature of contact with the chemical.

14.7.1.2. Tiered approaches to exposure estimation: a basis for risk assessment

The tiered approach is a logical stepwise approach to risk assessment that uses the avai- lable information to the optimum extent, while reducing unnecessary requirements for human exposure surveys or studies. Alternatively, the need for an exposure study can be justified through elimination of all other possibilities. Tiered approaches use increasin- gly sophisticated analyses, exposure controls and parameter sets. Initial tiers should pro- vide conservative assessments of exposure that are refined in subsequent tiers. The tiered process explained below is a standardized approach to the evaluation of risk. In residential settings, most risk assessments can be characterized as Tier 1 assessments. If a Tier 1 assessment does not adequately address the margin of safety for a pesticide use pattern, a Tier 3 level approach may be applied to refine the assessment, to eliminate uncertainty. Tier 2 assessments apply to professional applicators and occupational expo- sures only. In Tier 1, the assessor selects an indicative exposure value from an empirical database or mathematical model, or a reasoned worst case, or by selecting validated data from tasks likely to produce similar exposure distributions. For example, Tier 1 estimates must not take PPE into account. When the result of a Tier 1 exposure assessment produces an unacceptable outcome in risk assessment, a Tier 2 estimate is required. In Tier 2, the exposure estimate needs to state the default values – these are the assump- tions used in the absence of scientific data, and they are set on the basis of scientific infor- mation and for conservative purposes – selected and also all assumptions; the assessments may combine some chemical-specific data with standard default values or generic data. Tier 2 estimates are appropriate for a detailed exposure assessment of specialized pro- fessional users – for example, protective measures are supposed to be carefully observed. If the resulting exposure estimate produces an unacceptable outcome in the risk assess- ment, the exposure abatement measures may be successively refined and the exposure estimate revised, until the options for exposure reduction are exhausted. If after this remodelling the predicted exposure is still unacceptable, then a third iteration of the expo- sure assessment will be required. In Tier 3, the final tier of the assessment, valid estimates of human exposure are produ- ced through surveys or studies with the actual product or with a surrogate. Studies may need to cover an entire scenario and may include biomonitoring to show systemic uptake. The information is particularly useful in the case of a workforce that has been studied over a period of time and at a known fairly continuous level of exposure. For residen- tial settings, these types of studies are the exception and usually involve environmental monitoring for pesticide residue or air concentration levels instead of biomonitoring. Fig. 14.5. Tiered approach to estimation of inhalation exposure Source: Modified from EC 2002. Fig. 14.6. Tiered approach to estimation of dermal exposure Source: Modified from EC 2002. Public Health Significance of Urban Pests 513 rical models, they are likely to account for the many variables that influence exposure. Currently, no empirical models exist for predicting consumer exposures, since the avai- lable databases on exposure measurements are not sufficiently large. Pesticides: risks and hazards 512 in use, including propel- lant, held within the can. T he use of exposure models requires the selec- tion of various input para- meters. Insufficiently detailed information on exposure scenarios or lack of sufficient data may require the use of default values. Input data or default values used for the calculations must be clearly documented. Computer programs have been developed to imple- ment mathematical pre- dictive models and empiri- cal models. Statistical models have been develo- ped using available data and appropriate statistical methods. Model choice should be jus- tified by showing that the model uses the appropriate exposure scenario – for example, as judged from the underlying assumptions of the model. Expert judgement may be required to check the realism of the exposure value derived from a model, particularly if default or so-called reasonable worst-case values have been used. Modelling exposure can be performed either by taking discrete values point estimates or distributions for the model variables probabilistic modelling. Generally, exposure models fall into one of three types: mathematical mechanistic models, empirical or knowledge-based models, and statistical mathematical models. T hese models predict exposure levels from a mechanistic description of a process, an empirical database or statistical relationships.

14.7.2.1. Mathematical mechanistic models