Early efforts to use mathematical models to predict dermal exposures to chemical substances can be traced to persons evaluating occupational exposures in agriculture. Relatively low vapor pressure and moderate to high lipid solubility are useful characteristics for pesticides. They are also characteristics that are likely to elevate the importance of dermal exposures relative to inhalation exposures. Hence mathematical prediction of dermal exposure was attempted earlier in agriculture than in many other industries. Initial models were relatively simple and heavily dependent upon empirical measurements. Multiple factors, including increased awareness of the transport of synthetic chemicals in the environment, changing definitions of acceptable risk, and mitigation of respiratory exposures in the occupational sector, have lead to consideration of a broader range of possible dermal exposure scenarios. In the United States, the Food Quality Protection Act (FQPA), which requires assessment of aggregate pesticide exposure, is an important driver of dermal exposure research. FQPA inspired residential exposure models feature relatively elaborate descriptions of human behavior linked to fairly simple models of surface-to-skin transfer and subsequent absorption. Not surprisingly, the latter components of these models draw upon prior experience in occupational agricultural exposure assessment. Also in the U.S. the necessity of setting soil cleanup standards at hazardous waste sites has fostered the modeling of dermal exposure to soils and concern over exposure to by-products of chlorine disinfection in drinking water and to groundwater contaminants has fostered modeling of dermal absorption of water-borne chemicals. Similar pressures are at work in Europe and reflected in significant initiatives and a rapidly expanding dermal exposure literature.

Dermal chemical exposures may lead to direct dermatological effects or to systemic absorption of those chemicals. Most dermal exposure modeling to date has focused on the latter question. In that context, its purpose is to facilitate the further step of predicting absorbed dose. To be useful, dermal exposure models should at least address the following three basic questions. 1) How much of the medium containing the chemical of interest or of the neat compound is transferred to the skin and how is it distributed? Characterization of distribution must include areal extent and should include evaluation of completeness of coverage and potential layering effects. Immersion in an effectively infinite source is an exception. In that case, exposure modeling is trivial and the problem is reduced to absorption modeling. 2) What is the thermodynamic activity of the compound on the skin surface? If the target compound is not present in neat (or unbound) form, the matrix in which it is found on the skin and the affinity of that matrix for the compound are critical determinants of absorption behavior. 3) How long does the compound of interest reside on the skin? Dermal absorption is time-dependent. Loss mechanisms that compete with absorption such as volatilization or removal by washing should be considered.

Mathematical treatment of each of these questions can range from relatively simple to very complex. Regulatory deadlines and data gaps impose constraints on modelers. An increase in modeling sophistication that is currently within reach involves treatment of inputs as stochastic variables. Historically deterministic models have been used in regulation. Disputes over the extent to which model predictions are or are not conservative have fostered increasing use of stochastic models. An additional question can be posed. 4) How temporally and demographically variable are the exposures and how uncertain are the measurements and sub-models used to characterize them? Characterization of the uncertainty of predicted exposures is necessary if models are to be tested against observations (which is essential) and if regulatory decisions based on those predictions are to be well founded.