|
|
|
|
|
|
|
|
| NIOSH Home > Safety and Health Topics >Skin Exposures and Effects >Occupational & Environmental Exposures of Skin to Chemicals- 2005> Abstracts |
|
|
 |
||||||||||||||||||||||||
Bayesian and Probabilistic Exposure ModellingN. Warren, Health and Safety Laboratory, Buxton, UK To date, most regulatory risk assessments for occupational exposure of skin to chemicals have followed a deterministic approach. Deterministic exposure assessments take single, usually conservative, values for each parameter in an exposure algorithm and provide a single estimate of systemic exposure. Probabilistic exposure assessments can be viewed as a further tier of refinement and provide a more detailed characterisation of systemic exposure. By replacing point estimates with distributions representing variability in work patterns, dermal exposure, personal protective equipment, dermal absorption and other physiological parameters, a probabilistic exposure assessment attempts to characterise the whole distribution of systemic exposure across the work population. More complex algorithms employing 2-dimensional Monte Carlo simulation can simultaneously evaluate both variability and uncertainty, and thereby, give risk assessors a more scientifically rigorous basis for their decision-making. Modelled uncertainties in systemic exposure can be very large reflecting the cumulative uncertainties in external dermal exposure, mitigation due to clothing or PPE and dermal absorption. In these situations Bayesian techniques that allow the synthesis of dermal exposure measurements with expert judgement and biological monitoring data may provide risk assessors with additional reassurance that margins of safety are being met. With a probabilistic exposure assessment typically sharing the same data and same underlying exposure algorithm as the deterministic approach deficiencies in the basic methodology can transfer across to the probabilistic assessment. Three key aspects of the methodology that will be discussed are: the biological relevance of modelled exposures, longitudinal exposure modelling and the selection of appropriate dermal absorption algorithms. Traditionally, occupational exposure assessments have tended to focus on determining systemic exposures resulting from a single exposure scenario, work-shift or day. Longitudinal modelling considers the profile of systemic exposure over a longer period – perhaps weeks, months or even years. Such an approach has several advantages. For chronic health-effects cumulative exposure (or equivalently average exposure over the relevant time period) provides a more appropriate exposure metric than a short-term daily dose and allows risk assessments to be based upon the probability of long-term over-exposure. Additionally, uptake of a chemical following dermal exposure can continue over a number of days so that systemic exposure in a 24-hour period is a composite function of the previous days exposures. These 'residual' contributions to systemic exposure are not captured by single-day assessments. The selection of dermal absorption algorithms is constrained partly by the availability of data with an appropriate dermal exposure metric. With dermal exposure measurements (and by implication predictive dermal exposure models) most frequently assessing total dermal mass of a chemical, many assessments adopt an approach based upon the percentage of the dermal dose absorbed. However, as this percentage is dependent upon both the amount applied per unit area and the duration of exposure, a probabilistic exposure assessment should ideally incorporate these inter-dependencies. More fundamentally, the total mass of chemical may not be the most appropriate exposure metric for determining systemic uptake, either because not all the mass is available for uptake or because it is distributed very heterogeneously. A more useful exposure metric might be the contaminated skin area or better still the area/concentration distribution. Whilst sampling techniques exist for both these exposure metrics (e.g. fluorescent tracers) most new exposure data are still expressed in terms of mass. A series of case studies will be presented to illustrate the use of probabilistic techniques in quantitative chemical risk assessment.
Content last modified: 10 April 2005
|
