Fate and exposure modeling in regulatory chemical evaluation: new directions from retrospection

Mark A. Bonnell *a, Angelika Zidek b, Adam Griffiths b and Don Gutzman a
aEnvironment and Climate Change Canada, 351 St. Joseph Blvd., Gatineau, Québec K1A 0H3, Canada. E-mail: mark.bonnell@canada.ca; Fax: +1-819-938-5140; Tel: +1-819-938-5085
bHealth Canada, Ottawa, Ontario, Canada

Received 26th October 2017 , Accepted 14th December 2017

First published on 14th December 2017

The development and application of fate and exposure modeling has undergone fundamental changes over the last 20 years. This has, in part, been driven by different needs within the regulatory community to address chemicals of concern using different approaches. Here we present a retrospective look at fate and exposure model application over the last two decades keeping an international regulatory perspective and using the Government of Canada's Chemicals Management Plan to illustrate concepts. We discuss the important role fate and exposure modeling has played to help address key data gaps when evaluating the risk of chemicals for both human health and ecological reasons. Yet limitations for more widespread model application within a regulatory context remain. Consequently, we identify specific data gaps and regulatory needs with an eye towards new directions for 21st century chemical evaluation. We suggest that one factor limiting greater model application is the need for increased awareness and agreement of what chemical exposure assessment encompasses within the risk assessment paradigm. This is of particular importance today because of the increased availability of computational and high-throughput data and methods for chemical assessment allowing evaluators to potentially examine exposure from site of release to site of toxic action, thus linking exposure with toxicology. We further suggest there is a need for discussion at a global level to promote the awareness of new tools and approaches available for fate and exposure modeling and suggest that this could be organized using the aggregate exposure pathways concept.

Environmental significance

The evaluation of chemical impacts is often challenged by the lack of empirically measured multimedia processes that contribute to exposures in human and non-human receptors. Fate and exposure models play a key role in helping decision-makers fill data gaps to better characterize these exposures and thus determine the risk potential of chemicals in the environment. This article takes a retrospective look at the regulatory application of environmental models that have or can be used to assist the characterization of exposures in human and ecological receptors while providing some insights on new directions for modeling within 21st century risk assessment.

1 Introduction

A quarter of a century ago much of the regulatory risk assessment world was concerned with the evaluation of what are now known as “legacy” chemicals. These chemicals include halogenated organics (e.g., polychlorinated biphenyls, hexachlorobenzene, dioxins), heavy metals (e.g., mercury, cadmium), and pesticides (e.g., dieldrin, heptachlor) now quite familiar to most environmental scientists. A global requirement to evaluate the environmental harm resulting from these chemicals, particularly from long-range transport in air, resulted in global and regional conventions such as the United Nations Environment Program (UNEP) Stockholm Convention1 and the United Nations Economic Commission for Europe (UNECE) Convention on Long-Range Transboundary Air Pollutants (LRTAP).2 Domestic approaches for evaluating existing substances were oriented around the assessment of data rich well studied chemicals. Timeframes for completion of an assessment were in the order of years. New substance programs, addressing substances entering into commerce, were largely in their infancy and were grappling with how to rapidly evaluate a variety of environmental chemistries, many of which were and are still, not well understood. New substances programs were designed to meet the need for rapid evaluation where data requirements for assessment are specified in new substance regulations and fulfilled according to imported or manufactured domestic tonnages notified by the chemical industry.3 As a consequence of both the wealth of data for existing substances and the rapid time frame allotted for new substance assessment based on prescribed data requirements, fate and exposure modeling was not often undertaken. This was due in part because there was not a perceived need for modeling (measured data from various media, human or other biota were often available) or time constraints in the assessment process did not easily allow for modeling. Fate and exposure models during this time period, such as the Equilibrium Criterion (EQC), SimpleBox, or ConsExpo models were also just coming into general regulatory awareness.4–6

2 Paradigm shift

At the end of the 1990s and early into the millennium, a public push for more rapid assessment of a broader range of existing substances started to become a widespread phenomenon. This was due to a greater public awareness of environmental and human exposure to industrial and non-industrial chemicals. This awareness in turn led to public concern that current regulatory practices for existing industrial substances, requiring multi-year evaluations, were not adequately addressing the expected level of exposure to the myriad of substances in commerce and thus it was perceived that public and environmental safety was potentially jeopardized. The result was that in the late 1990s the Canadian Government began to rethink the existing substance paradigm to better address this exposure concern. There was at this time a more vocal animal welfare movement, particularly in Europe, that advocated the banning of animal testing in favour of alternative non-animal methods, such as computational approaches. Consequently, governments began to examine entire inventories of industrial substances, both organic and inorganic. In Canada, for example, Environment and Climate Change Canada (ECCC) (then called Environment Canada) and Health Canada (HC) shifted away from evaluations of individual substances included on the Priority Substance List (PSL),7 a list of 69 substances largely derived using a nomination process to the prioritization for assessment of the ∼23[thin space (1/6-em)]000 substances on the Domestic Substances List (DSL); a process called categorization in Canada.8 The ecological and human health categorization of the DSL (1999–2006) eventually led to ∼4300 prioritized substances for which the Government of Canada formally committed under the Strategic Approach to International Chemicals Management (SAICM)9 to address under the Chemicals Management Plan (CMP) by the year 2020.10 Most (>90%) of the 4300 substances do not have a complete empirical data set for any type for chemical evaluation, thus there was and remains a heavy reliance on computational approaches to complete regulatory commitments. It was also during this time period that the Stockholm Convention was formally adopted by UNEP (2001) and eventually came into force in 2004 to deal with transboundary persistent organic pollutants of global concern.11 The assessment of pesticides and biocides was still conducted according to prescribed regulations by various governments and globally under conventions such as the Stockholm Convention. These regulations have not fundamentally changed to this day.

3 Impact of the shift on modeling approaches

3.1 Ecological modeling

The regulatory shift away from the lengthy evaluation of data rich legacy substances to a more rapid evaluation of relatively data poor existing substances and new substances had a fundamental impact on the approaches used to analyze and generate data for prioritization as well as for risk assessment. Prioritization of substances on the DSL (categorization) for ecological receptors was conducted using selected chemical properties in a “hazard-based” approach, that is, according to specific threshold criteria for persistence (P), bioaccumulation (B) and inherent toxicity (T) while prioritization for human health was conducted according to greatest potential for human exposure (GPE). Evidence of toxicity (e.g., carcinogenicity, mutagenicity, reproductive or developmental toxicity) was additionally incorporated. The persistence and bioaccumulation criteria used by ECCC for ecological prioritization of the DSL were directly adopted from the 1995 Government of Canada's Toxic Substances Management Policy12,13 (TSMP) and were later formalized in the Government of Canada's Persistence and Bioaccumulation Regulations.14 Importantly, on page eight of the 1995 TSMP policy document it states that “…exposure is an important element in evaluating environmental risk under the policy. Persistence and bioaccumulation can be used as qualitative surrogates for long-term exposure of environmental biota”. This is perhaps one of the first public policy statements where criteria for persistence and bioaccumulation are related to a spatial and temporal scale for exposure and can be used in place of, or alongside, monitoring data. The impact of domestic policy and subsequent use of criteria such as these, as well as global conventions such as the Annex D Criteria of the Stockholm Convention, was that modeling of fate and exposure for ecological receptors was now focused on specific intrinsic chemical properties and long-range transport in air rather than the more integrated approach conducted for the legacy substances or for pesticides. In fact, in 2001 global efforts began to emerge to address these intrinsic chemical properties in support of regulatory activities. An example of one such effort was the Organization for Economic Cooperation and Development's (OECD) Multimedia Modeling Expert Group which developed guidance for evaluating environmental persistence and long-range transport potential based on expert workshops sponsored by the OECD and UNEP in Canada and Europe15,16 and ultimately a consensus multimedia model for these properties.17,18 The OECD and UNEP initiatives also stimulated the development of approaches for using these properties for screening inventories of industrial chemicals.19–23 Similarly, for bioaccumulation, modeling efforts were and are still stimulated by the need to determine specific factors to compare to global and domestic criteria for aquatic bioaccumulation potential.24–29 It was soon realized that bioaccumulation model output is very sensitive to the metabolism rate constant input into the model and thus a focus on this parameter gained the attention of researchers worldwide.30–33 Given the fate data paucity for most industrial chemicals, bioaccumulation models became an essential tool for filling data gaps when Canada undertook the categorization of the DSL for persistence, bioaccumulation and inherent toxicity (PBiT). Almost all, approximately 97%, of the bioaccumulation results for categorization were model based. Model output used for the categorization of bioaccumulation was generated assuming no metabolism of the organic compound as a precautionary measure. Indeed, bioaccumulation models at this time were designed to provide ‘maximum bioaccumulation factors’ by default and few models existed that could accommodate corrections for known or estimated metabolism rate.24–27,29 Most bioaccumulation models were either structure–activity-based29 or simple linear regressions of observed bioaccumulation factors with the logarithm of the octanol–water partition coefficient.34 As a result of the no metabolism policy decision, Canadian categorization conclusions for bioaccumulation potential became biased towards substances with high values of the octanol–water partition coefficient. When the PBiT compounds from categorization were assessed for ecological risk during the first phase of the CMP (2006–2012), it became evident that bioaccumulation potential was almost always overestimated and required correction for metabolism rate. The categorization bioaccumulation outcome was one driver for increased domestic and international efforts to improve the estimation of fish metabolism rate29–33 as it became evident that model correction for metabolism has a dramatic effect on the bioaccumulation factor35,36 used for prioritization of chemicals or estimation of tissue residues.

Bioaccumulation models have also been developed for terrestrial environments as summarized in the findings of an international workshop on terrestrial bioaccumulation.37 Terrestrial models lag behind aquatic models in two important areas – definition of a default terrestrial food web and development of metabolism rate models for terrestrial receptors. Nonetheless, terrestrial bioaccumulation modeling can be an essential component of risk assessment when there is a need to account for the exposure from persistent and bioaccumulative compounds sorbed to biosolids applied to agricultural lands. Efforts to improve the in silico, in vitro and in vivo estimation of metabolism rate and dietary assimilation efficiency in ecological receptors continues to be a primary focus in bioaccumulation research38,39 and is regarded as high impact work by regulatory risk assessors. Finally, bioaccumulation models were also examined to determine if they could inform bioaccumulation potential in humans.40

3.2 Human modeling

In Canada, human health risk assessment of existing substances has focused on general population exposures via environmental media, food and consumer products, without consideration of occupational exposures. To date, occupational exposures in Canada have been considered to be covered under separate federal, provincial and territorial occupational health and safety legislation. Consumer exposure estimation to consumer products has tended to focus on product users. However, there is increasing awareness and need to consider emerging trends in consumer exposures (e.g., e-cigarettes and e-liquid, 3-D printers) as well as further consideration of bystander and post-application exposures. This is particularly the case for younger age groups, due to their differences in receptor characteristics (e.g., body weight, inhalation rate, body surface area, etc.) and behavioural patterns (e.g., mouthing, crawling, etc.) in comparison to adults. Other factors considered for exposure modelling include differences in product use by sex, route and duration of exposure, susceptible subpopulations, the critical hazard endpoint, and selection of sentinel scenarios. The OECD Task Force on Exposure Assessment (now known as the OECD Working Party on Exposure Assessment), was established in 1995 and had historically focused on environmental exposures. Only in the last few years (since 2008) has the scope of the group and its activities included a human health focus, including a better understanding of approaches and methods for estimating consumer product exposures.
3.2.1 Environmental models used for human exposure. In Canada, when lacking measured environmental data appropriate for a particular risk assessment, environmental exposures to humans from existing substances have tended to be estimated using mass-balance and aquatic down-the-drain or industrial emission rate modeling. Generic, conservative scenarios are often employed as a first tier in risk assessment of existing substances, with consideration of more site-specific parameters when refinement is required. Most non-scalable mass-balance models provide regional scale estimates of exposure, thus, as with ecological assessments, human health risk assessments have been complemented by local scale tools where general population exposures in the vicinity of point sources are foreseeable. Such local scale tools often require little to no physicochemical property data for use. For example, generic release scenarios for river dilution models often require only basic inputs such as emission rate, volumetric flow rates of effluent and river, and number of release days per year, though some inputs may be informed by the substance's properties. One challenge that human health regulators face when using regional-level mass-balance models is extrapolating to near-field estimates which more closely represent general population exposures (i.e. at the tap for drinking water downstream of a specific effluent discharge point, homes in the vicinity of industrial sites for ambient air). Validation of these models can also present challenges as modeled estimates based on level III fugacity simulations are steady-state regional estimates that do not reflect temporal or spatial variations. This can complicate comparisons to monitoring data, which may be targeted to specific point sources or are based only on limited sampling durations and times throughout the year.
3.2.2 Consumer exposure models. When exposures to consumer products were estimated for assessment of existing substances in Canada, 90% of exposures estimates were modeled given a lack of exposure data for these types of exposure scenarios relevant to the general population. Approximately 10–20% of existing substances in Canada with consumer product exposures, for which assessments were conducted over the last ten years, had exposures that were estimated using human biomonitoring data, often without supplemental exposure estimates being generated via modeling. In Canada, consumer exposure estimation for existing substances has most frequently been conducted using the Consumer exposure model (ConsExpo v4.1 or Web)41,42 or via manual calculation using algorithms contained therein. While the scenario defaults embedded in ConsExpo have generally been used to date for assessment of existing substances in Canada, Health Canada has expanded these defaults to include other survey data from North America. Such defaults were selected based on a review of data available in the literature for key defaults, which includes application of certain selection criteria (e.g., geographic location, year of survey(s), study size, etc.). Based on an analysis of consumer exposures assessed under the CMP over the last decade or so, direct exposures via products, specifically ‘leave-on’ products, were often found to be one of the higher sources of exposure compared to environmental media as well as other products. Parameters for which it was often challenging to identify data included the substance concentration in a given product as well as measured retention factors for rinse or wipe-off cosmetics and personal care products.

To address the lack of a batch-mode feature in the ConsExpo model as well as to build in novel cosmetic and personal care product defaults selected by Health Canada for ease of use, a new consumer exposure tool was developed. This tool, known as the Cosmetic Ingredients Database for Exposure Estimation (CIn-E2), contains exposure algorithms from ConsExpo and allows for batch-mode processing of multiple exposure scenarios and chemicals. Other tools that have been used sporadically in the past for human health risk assessment of existing substances in Canada include the American Industrial Hygiene Association's Industrial Hygiene Model (IH MOD),43 the US EPA's Wall Paint Exposure Assessment Model (WPEM),44 and the National Research Council Canada's Indoor Air Quality Emission Simulation Tool (IA-Quest).45 Work remains on determining how these tools and the numerous other consumer exposure tools available or currently in development, including the US EPA's Consumer Exposure Model (CEM v2.0)46 and the US EPA's Exposure Factors Interactive Resource for Scenarios Tool (ExpoFIRST v2.0),47 compare and potentially complement one other for use in risk assessment. There is a strong need for models that can offer tier 2 inhalation exposure estimates from niche product uses (e.g. do-it-yourself products, cleaning products), including a need to better define the domain of applicability and conservatism of various algorithms that feed into estimation of evaporation rates (e.g. Langmuir, Thibodeaux, etc.) as well as to incorporate post-application or bystander (non-user) exposures scenarios, particularly for children.

For human exposures to solid articles, such as building materials, textiles, foam objects, plastic toys and food packaging, manual calculation using algorithms identified from the literature is typically conducted to generate exposure estimates. Such algorithms may reflect dermal contact with the article, oral exposures via hand-to-mouth contact, product ingestion, direct object mouthing behaviours and migration to food items, and inhalation exposures resulting from off-gassing. When available, chemical-specific or analogue empirical emission data to air and migration data to various simulants (e.g., food, sweat, saliva), has been used. Recent model development in the area of estimating indoor dust concentrations include RIVM's DustEx tool48 and the US EPA's Indoor Environmental Concentrations in Buildings with Conditioned and Unconditioned Zones tool (IECCU v1.0).49 To date, tools for estimating emissions and migration from solid articles or for estimating indoor dust concentrations have seen little use in existing substances assessment in Canada. Under the CMP, validated models that provide refined exposure estimates for indoor environments, with a focus on indoor air, are highly sought after, including those that incorporate multiple sources (e.g. building materials) for a range of chemicals (e.g. SVOCs, VOCs). The development of generic scenarios to parameterize such models, such as default total release areas and masses for solid articles in a room (e.g., electrical and electronic equipment, building materials, etc.), are also needed for a screening context.

4 Data gaps and modeling needs for chemical evaluation

4.1 Scalable models

Although there has been a general uncoupling of fate and exposure from toxicology when using a hazard criteria approach to chemical evaluation, this has not generally been the case with substance-by-substance risk assessment, although hazard characterization is often better resolved than exposure characterization. The modeling of fate and exposure is often included in risk assessment where the determination of a measured or predicted environmental concentration (MEC or PEC) is required. Derivation of a PEC requires consideration of rates of emission, fate and distribution of the compound to determine the degree of organism contact in media in which the substance resides or is expected to reside. What is not consistent among regulatory and non-regulatory ecological risk assessment approaches, however, is the degree to which fate and exposure are integrated to derive a PEC. There are many cases where the environmental fate of a compound is determined, but no further linkages to exposure are made resulting in a mismatch between the PEC derived for the risk quotient analysis and the fate of the substance in the environment (e.g., for super hydrophobic chemicals). Integrated fate and exposure models such as the European Union System for the Evaluation of Substances (EUSES)50 have been used by member countries in Europe to evaluate substances long before REACH was implemented and can be used to derive PECs in various media. In North America, EQC, CHEMCAN51 and the Risk Assessment IDentification And Ranking (RAIDAR) model52–55 are examples of integrated fate and exposure tools used extensively by HC and ECCC for determining environmental media concentrations for ecological receptors and the general human population for both prioritization and risk assessment of new and existing substances at a regional scale. The RAIDAR model has also been applied to prioritize and rank personal care products for environmental risk assessment in Asian markets.56 However, ecological risk assessment will always involve exposure and fate assessment in the near field (i.e., near the point of discharge to the environment) and may only include far field considerations for those substances that are known or estimated to undergo long-range transport or which are widely dispersed to the environment from multiple points sources.

Other than EUSES, which is parametrized for Europe only, there are relatively few, if any, coupled fate and exposure models that can scale from local to regional to far field environments depending on chemical fate and distribution. USETox, a “nested” life cycle assessment model for human and ecological exposure is perhaps the exception.57 This is a major limitation to ecological assessment since in the absence of such tools, river dilution models for aquatic releases are used. The advantage of the dilution approach is that it is simple, transparent, suitable for the majority of industrial chemical releases and is designed for a local release whereas many multimedia models are designed at a regional scales (∼100[thin space (1/6-em)]000 km2). The disadvantage is that the dilution approach may not actually be protective in some cases (scenarios assuming immediate dilution by large receiving waters), nor does the approach easily accommodate the partitioning, persistence or the distribution of a compound over time. Therefore, the spatial and temporal scale of exposure is not fully accounted for in the assessment. This issue in chemical evaluation has been noted by regulators and researchers.58 One solution could involve the adaptation of multimedia models such as the Quantitative Water, Air, Sediment Interaction (QWASI) model, developed for fate and behaviour of substances emitted to the Great Lakes59 and then adapted for rivers.60 The QWASI or similar model could be further adapted to include a connected series of river reaches to address local aquatic emission scenarios to rivers and drinking water sources parameterized to a default environment but with the option for specific environments. However, the above example may not address near field dynamic (non-steady-state) situations. Consequently, a need yet remains for fate and exposure models to better accommodate near field dynamic situations (e.g., pulse type emissions, mixing zones) when it cannot be assumed that chemical kinetics are at steady state or equilibrium conditions assumed by many regional scale multimedia models.

4.2 Linking exposure with toxicology

Likely the majority of risk assessors would define exposure assessment to be the quantification of environmental media concentrations or dosages from exposure to consumer products, drinking water and foods. However, unlike human health assessment, it is less likely that many regulatory ecological evaluators would define exposure assessment to include estimation of dose or tissue residue concentrations on a whole body basis or at the site of toxic action of a selected organism. They would also not tend to include aspects of organism bioavailability. In other words, ecological exposure assessment stops at the interface with an organism, where the domain of toxicology begins. Consequently, many regulatory agencies purposely delineate exposure science from toxicology because they are characterized separately in risk assessment. This viewpoint is not entirely correct nor is it entirely incorrect because there is an overlap between exposure and toxicology at the interface of the organism, that is, when considering toxicokinetics. Toxicokinetics governs the concentration at the site of toxic action and is essential for more accurately estimating the internal exposure concentration resulting in adverse effects. Thus, toxicokinetics is essentially a fate assessment in an organism. Regardless of regulatory risk assessment structure, a central question arises for exposure science as to what the domain of exposure assessment should then include? Recently the concept of aggregate exposure pathways (AEP)61 has been promulgated in the scientific literature as a parallel and complementary approach to the adverse outcome pathway (AOP).62 An AOP can be used to organize toxicological information from molecular initiating events to adverse outcomes in individuals and whole populations. Like AOPs, AEPs organize exposure information along a pathway from site of release (source) to site of exposure within organisms (site of toxic action). The benefit of the AEP is that it can organize exposure processes and pathways data from site of environmental release to the site of toxic action (i.e., source to dose), including toxicokinetic data. The AEP provides a continuum into the AOP thus providing a needed bridge between exposure and toxicology (i.e., from release source to molecular initiating event). This has obvious implications for fate and exposure modeling because now key events along the AEP can be targeted by regulators and researchers and information generated to fill data gaps to provide coherent fate and exposure evidence for risk assessment. In particular, exposure and fate modeling can help determine the “state” of a chemical or other stressor from site of release to target site of exposure. Here state of a chemical stressor can imply multimedia transport or transformation processes along the AEP.61 The AEP concept may also provide a means to better accommodate aggregate and cumulative exposures in risk assessment. That is, because the AEP is inherently an aggregate of exposures, it can be applied to address the cumulative hazard from multiple chemicals from multiple exposure pathways if the toxicodynamics (at the site of toxic action) are known or can be confidently predicted. This is beneficial to risk assessment because cumulative approaches to both hazard and exposure assessment can be better aligned and better target key exposure sources which can be missed using a single chemical approach resulting in an under prediction of risk.

While physiologically based toxicokinetic modeling (PBPK) has been relatively well developed within the human health sciences, there has been comparatively little work done for non-human receptors. Models do exist, for example, for taxa such as fish.63 This modeling gap has obvious implications for further developing the AEP concept discussed above, but also for developing baseline exposures (doses) in key non-human receptors for regulatory use. Baseline exposure values can be compared to toxicological benchmarks (e.g., critical body burdens) or measures of activity from high through-put screening (HTS) in vitro assays from the TOXCAST/TOX21 database using the activity to exposure ratio concept (AER).64 The AER concept can help regulators establish a dose-based exposure baseline that is applicable, for example, for deprioritizing chemicals for further evaluation based on a selected margin of exposure set to an acceptable de minimus risk. Regulatory application, however, requires an adequate explanation and confidence with dose “scale-up” from in vitro systems to target tissues or whole organisms. Further ecological modeling is thus needed to support in vitro to in vivo extrapolation (IVIVE) for non-human receptors.

4.3 Linking bioavailability to route of exposure

Bioavailability has been an important consideration in human health risk assessment, particularly for dermal exposures when characterizing risk for a systemic endpoint using an oral or inhalation point of departure (POD) or toxicological reference value (TRV). Bioavailability reflects the uptake of the bioaccessible fraction of the substance which comes in contact with the lung, skin or gastrointestinal tract. Absolute bioavailability tends to be less via the dermal route than via the oral route,65 and is thus an important area for refinement when potential risk issues are noted in a risk assessment using a screening assumption of complete absorption. However, when bioavailability is applied as a refinement for dermal exposure scenarios in risk assessment, the relative bioavailability of the dermal route compared to the oral route when using an oral POD or TRV may be used,66 particularly for substances that are poorly absorbed orally, so as not to underestimate risk. Bioavailability may be expressed as an uptake fraction, an amount absorbed or a flux, with special consideration required for whether the dose conditions of the exposure scenario are best characterized as finite or infinite.67 While chemical-specific empirical bioavailability data are generally preferred in risk assessment, such data are often lacking, and when available their use may be complicated by differences between the exposure conditions used in the empirical study and those in the exposure scenarios of the risk assessment (e.g., differences in dermal loading, exposure duration, vehicle, etc.).68 Approaches for estimating dermal bioavailability in the absence of chemical-specific empirical absorption data include in silico tools, such as the Finite Dose Skin Permeation Calculator69 and IH SkinPerm,70 use of default uptake fractions from the literature or as established by regulatory bodies, such as the 10% default based on physicochemical properties set by the European Commission,71 and use of dermal maximum flux, such as has been used for cosmetics, fragrances and fragrance-like ingredients in the scientific literature.72–76

4.4 Exposure amortization

Exposure amortization or averaging is another important consideration in human health risk assessment of existing substances and is dependent on the close interplay between exposure, toxicokinetics and hazard profile of any given substance. Quantitative cancer risk assessment for (non-threshold) carcinogens has typically been conducted by calculating lifetime average daily doses (LADDs),77 which may be adjusted using age-dependent adjustment factors for carcinogens acting via a mutagenic mode of action.78 However, for risk characterization against longer-term non-cancer endpoints, or for carcinogens acting via a non-threshold mode of action, whether and how to incorporate exposure frequency for exposure scenarios involving short-term, intermittent (i.e., less than daily) or non-continuous exposures is an area of uncertainty. As a conservative starting approach, comparison of the event exposure to long-term hazard PODs or TRVs can been employed.79 If risk issues were identified from use of this starting approach, limited exposure averaging can be carried out in certain cases. Factors considered to gauge the appropriateness of exposure averaging include whether the critical hazard effect is local or systemic, whether the effect is highly dependent on the peak concentration, and windows of susceptibility.80 Exposure averaging may also be appropriate for risk characterization against acute or short-term hazard PODs or TRVs under certain conditions.80 For inhalation exposures, selection of an appropriate averaging time has also been an area of uncertainty, such as a peak, mean event, or 24 hour mean concentration. In some cases, the selection of averaging time may mimic the exposure regime; as an example, use of a 6 hour averaging time to mimic 6 hours of inhalation exposure administered in a hazard study per day. For risk characterization against acute PODs or TRVs, ECHA has proposed a convention of a 15 minute averaging time.79

4.5 Linking read-across to exposure

Increasingly, Quantitative Structure Activity Relationships (QSAR) and read-across have been used in to help inform the hazard characterization for a given substance or group of substances. This approach has been used in Canada for a number of substances (e.g. aromatic azo and benzidine-base substances, phthalate substance grouping) as part of the CMP as well as in Europe for REACH purposes81 and includes guidance for conducting read-across under the OECD.82 However, one could argue that the same approaches could be used for exposure. The idea of using of surrogate exposure data has been used for decades within the realm of pesticide registration. For example, the U.S. EPA and Health Canada use the Pesticide Handler Exposure Database (PHED) to estimate exposures for a range of exposure scenarios based on exposures measured for a subset of surrogate pesticides, however structures of each pesticide are not considered. In the absence of information on the use (including concentration) of a given substance in a product, searches using similar tools and approaches for the identification of analogues and ‘read-across’ to other use profiles and, to some extent, exposures are being investigated. Quantitative structure-use relationship (QSUR) models to identify potential exposure sources for candidate structures have been developed, albeit to a limited extent.83 These QSURs can predict the likelihood of a substance having a functional use commonly associated with certain consumer products. Further efforts to link functional use(s) of a substance and chemical structures or classes in a product to generic concentration ranges would advance priority setting activities and, in some cases, provide a more complete exposure picture to support regulatory decision making.

4.6 Incorporating exposure into chemical prioritization

A regulatory focus on hazard criteria to identify substances of concern from large chemical inventories or on a chemical-by-chemical basis is still important today for many regulatory agencies. The approach is precautionary and is ultimately intended to identify chemicals of highest regulatory concern. The question becomes whether this approach does in fact identify chemicals of highest concern if exposure is not directly integrated into the prioritization approach. Efforts can be spent regulating substances of high hazard concern, but which have very little exposure in the environment or to humans. Integrated fate and exposure models for ecological receptors and humans can be a useful means to integrate exposure descriptors into prioritization approaches or for simply supporting the de-prioritization of low tonnage substances.84 The impact of doing so can be dramatic. In the 2006 categorization of the DSL, criteria prescribed under CEPA included greatest potential for human exposure but only indirectly included exposure, as a function of PBT criteria, for ecological prioritization. This resulted in considerable disparity between substances deemed of human health priority versus those of ecological priority. In fact, only 12 of 194 compounds (∼6%) in the Government of Canada's challenge (the phase of the CMP that addressed substances of highest expected ecological and/or human health concern) were of both ecological and human priority resulting in an increase in total number of risk assessments to be completed by each department. In 2016, ECCC completed a further prioritization of 640 organic substances for the third phase of the CMP using the Ecological Risk Classification of organic substances (ERC) approach.85 The ERC incorporates multiple exposure and hazard descriptors largely derived from fate, exposure and ecotoxicological models. The 640 substances consisted of (P or B) and T substances as well as human health priorities from the original 2006 categorization of the DSL. The additional ecological prioritization resulted in slightly more than 80% of the 640 organic chemicals identified as being of low priority for further ecological assessment largely as a result of high margins of exposure. Importantly, greater parity with human health priorities, set in 2006, was gained with the ERC which produced 53% agreement between human and ecological priorities for the third phase of the CMP.86 This result was driven by greater parity with potential for exposure, but also closer agreement on toxicology. Prioritization approaches can therefore greatly benefit from inclusion of exposure descriptors and, by extension, exposure modeling to improve targeting of chemicals of highest concern.

4.7 Generating multimedia concentrations

One of the largest data gaps in ecological and human health risk assessment is measured concentrations in environmental media or in the general population.87,88 Most environmental media concentrations for risk assessment must be estimated as previously discussed, but these estimates may not be consistent with the multimedia mass-balance of the chemical in the environment. For example, assuming a steady-state, estimated water concentrations used for an aquatic PEC at a local or regional scale may not agree with measured air concentrations taken a local or regional scale. Consequently, multimedia concentrations are not in agreement with the steady-state mass balance for a chemical. While not without some uncertainty, mass-balance exposure models can help fill data gaps and provide the boundary conditions for environmental concentrations using an “inverse modeling” approach. This is a method for adjusting unmeasured causal parameters so that model output agrees with an observed value. An inverse modeling approach was recently applied to 30 CMP priority organic flame retardants (OFR) using the RAIDAR model.89 Emission rate to the environment could be adjusted in the model such that estimates of air concentrations closely agree with air monitoring data. Multimedia exposure concentrations at a regional scale, based on the model mass-balance, can thus be deduced. The results from this study helped to fill media concentration data gaps for CMP risk assessment of organic flame retardants, but also helped to target key media requiring future monitoring and surveillance. The AEP concept may also be of benefit here because it can provide the state of the stressor in all known media types and thereby help to determine pathways and media of exposure that contribute to the greatest hazard in an organism.

For human health exposures, estimated or measured concentrations in environmental media for a given substance can be substituted (and supplemented, as appropriate) with human biomonitoring data (HBM). HBM data are reflective of the absorbed dose into the human body and can provide a measure of integrated exposure from different exposure sources. Of the 2700 substances assessed during the challenge and second phase of CMP, approximately 230 had HBM data available in various biological matrices (or around 10%) with over half (62%) being organic. For the remaining 1550 substances in the third phase of CMP, it is estimated that 15–20% will have HBM data – the majority of which will be inorganic. A number of substances have been recently assessed using biomonitoring data, including assessments as part of Canada's CMP.90–93 An approach document outlining how biomonitoring data can be used for regulatory decision making has also been published using both qualitative and quantitative approaches.90,91 The increasing trend in the availability of HBM data has provided opportunities for applying this type of data in human health risk assessments of existing substances in a variety ways, including obviating the need to estimate exposures from specific sources for some assessments (e.g. consumer products, air, water, food, soil, etc.), although characterization of exposures by source may still be required for risk management purposes when relevant. However, even when HBM data are available, assumptions need to be made to estimate levels of exposures in the general population including adjusting for individual variability and kinetics (e.g., absorption, distribution, metabolism and excretion). Similar to environmental monitoring data, certain criteria need to be employed to interpret and evaluate whether the data are of sufficient quality to be used for estimating human health exposures. These criteria can include: geographic location; timing and handling of sample collection, the limit of detection or quantification; sub-population examined and sample size; age of study; and matrix.94 Chemical specific criteria are also critical and can include a thorough understanding of pharmacokinetics and metabolic pathways, combined with knowledge of the spatial and temporal conditions of exposures for a given substance is critical in determining if the biomarker of exposure is adequate for use in risk assessment.

4.8 Modeling exposures for human health – role of chemical commercial tonnages

Substance tonnage bands, such as those used for REACH registration or for new substance notifications, are a simple indirect means of considering an important aspect controlling exposure. Tonnages can be used to help generate environmental emission data for input into multimedia fate and exposure modeling as a relative means of comparing environmental media exposures between large numbers of chemicals. For example chemical tonnage was a key model input parameter used in multimedia fate and exposure models by ECCC for the ecological risk classification of organics discussed above. There are model limitations, however, when using chemical tonnage data to extrapolate to or prioritize substances when examining human exposures from consumer products. Based on a retrospective analysis of assessments completed during the Government of Canada's challenge initiative and the second phase of CMP, chemical tonnages reported to be manufactured or imported played a limited role when determining the potential for overall general human population exposure. More often than not, exposures estimated from consumer products were higher than those from environmental media, ranging from 0.04 to 4 mg per kg bw per day and from 0.0003 to 0.2 mg per kg bw per day, respectively. For consumer product exposures, more than half of the known higher probability exposures fell in the reported range of less than 100[thin space (1/6-em)]000 kg. In fact, for all substances with consumer exposures evaluated by Health Canada during this time, over 40% had reported commercial tonnages less than 10[thin space (1/6-em)]000 kg per year, with the majority reported as less than ∼1000 kg per year. However, a better correlation was found for tonnage bands with estimated human exposure from environmental media. Substances with known higher probability exposures from environmental media had reported tonnages that ranged from 0 to 157[thin space (1/6-em)]000[thin space (1/6-em)]000 kg per year, with half of the substances falling within a range of 0 to 1[thin space (1/6-em)]000[thin space (1/6-em)]000 kg per year. Despite this, modeled environmental exposures from food, ambient air, drinking water, indoor air or soil/dust were rarely a driver (or major contributor) of overall exposure to the general population (<10% of the time). In instances where a particular environmental medium was a driver for exposure, environmental monitoring data, either Canadian or foreign, were available and had been used to estimate exposures. However, it should be recognized that there may be different degrees of conservatism and uncertainty to various exposure estimates both empirical and modeled that can confound, at least in part, the identification of drivers.

4.9 Dealing with ionogenic chemicals & chemicals of very low volatility & water solubility

Most fate and exposure models are parameterized to deal with neutral organic substances. A review of the industrial chemical inventory in Europe however, has shown that approximately 50% of domestic inventories for industrial chemicals can contain substances that are ionogenic (ionizable) at relevant environmental pH.95 This finding agrees with the approximately 50% of ionogenic chemical categories considered to be of high ecological priority for the third phase of the CMP using the ERC approach.96 Many pharmaceuticals and most pesticides are also purposely ionogenic in order to provide an intended functionality. Given the paucity of data for industrial chemicals and many pharmaceuticals, there is an obvious need to better address the behavior of ionogenic compounds using fate and exposure models for both human and non-human receptors. This need has not gone unnoticed and research efforts have commenced recently to fill this data gap for some aspects of chemical evaluation such as bioaccumulation, sewage treatment plant and general multimedia fate and food web modeling.97–99 Health Canada recently developed a Threshold of Toxicological Concern (TTC) approach for substances remaining in the third phase of the CMP.41 Mass-balance modelling was conducted for approximately 85% of substances or representative structures in the TTC approach that were Mackay type 1 (i.e., vapour pressure >10−7 Pa and water solubility >10−6 g m−3), whereas an “environmental delivery” ratio to human receptors was used for the other approximately 15% of substances or representative structures that were not Mackay type 1.

Modifications made to mass-balance models to accommodate ionogenic substances performed for ecological evaluation of numerous industrial and pharmaceutical substances in Canada54,97–100 may pave the way for potential use of fugacity modelling for ionogenic substances in human health risk assessment of existing substances. However, uncertainty remains in handling of substances with multiple ionization centers and significant data gaps remain for some of the key parameters governing their environmental fate and distribution. Despite recent modeling efforts, there still remains significant data gaps and modeling needs for ionogenic compounds, particularly cationic substances for which partitioning behavior in the environment and biota may be well outside the domain of weak anionic and neutral chemical behavior.101,102 Many cationic substances also have “excess toxicity” not explainable by polar narcosis and it is suspected that interaction with biological surfaces may be the primary mechanism of toxicity (e.g., gill clogging, adsorption to algal cells, skin sensitization) as well as a very high affinity for the phospholipid bi-layer of cell membranes.103 For human health, ionization may also impact the dermal bioavailability of the substance, with ionized species generally exhibiting smaller permeability coefficients through the hydrophobic stratum corneum, though this difference is less marked when comparing maximum flux estimates.67 As maximum fluxes are proportional to solubility in the vehicle, one explanation could be that ionization promotes greater solubility of substances in polar vehicles than their neutral counterparts.

5 Towards the future

The data gaps and regulatory needs outlined in this manuscript are not intended to be exhaustive. Rather the needs identified here focus on factors that have high sensitivity in chemical evaluation based on regulatory introspection and retrospection over the last 15–20 years. While this discourse has focused on modeling efforts, it implies that model inputs, such as physicochemical properties data and half-lives and other parameters, also require effort to improve their measurement and prediction, particularly for challenging chemistries outside the domain of neutral organics. It is unknown, for example, if organometallics can be treated as discrete organic chemicals in the models or if and how model predictions can be extrapolated to substances of Unknown or Variable Composition, Complex Reaction Products and Biological Materials (UVCBs). Canada is already preparing for post 2020 chemicals management, while in the United States, reforms to the Toxic Substances Control Act (TSCA) have opened the door for consideration of alternative methods to prioritize substances for risk assessment. In Europe, the low tonnage registration (1–100 tonnes) will end on May 31, 2018 and a new round of evaluation and prioritization activities will commence by member countries and ECHA.

In all of these regulatory situations, it is certain that the majority of substances under consideration will have significant data gaps for regulatory evaluation, particularly as it relates to exposure. However, most jurisdictions will continue to struggle to identify when a given substance is used in a product, particularly imported products. This can include any consumer product or finished article that has no requirement to label or identify the chemical substance it may contain. Monitoring programs and published research to date have provided measured occurrence and concentration data on hundreds of substances in various media and matrices; however, in reality, the chemical space continues to expand into the tens and even hundreds of thousands of substances. Available test standards and methods for targeted analysis of samples are limited to a small fraction of the chemical universe. New methods, including suspect screening analysis and non-target analysis, currently under intensive research by some regulatory bodies, need to be further investigated to advance efforts in better understanding how to more accurately and efficiently prioritize substances for research and assessment, whether for substances that are commercially produced or the myriad of other substances that exist in consumer products, articles and in the environment (e.g., metabolites, degradation products, transformation products and naturally occurring substances). It seems apparent then, that there will be an increased global need for fate and exposure modeling to fill data gaps for these compounds both for regulatory and non-regulatory chemical evaluation. Ideally, as a first step, discussion is needed among regulators, researchers, industry and other interested parties regarding the definition of exposure for 21st century risk assessment. It would also be ideal to coordinate this effort internationally much like was done and accomplished from 2001–2005 under the OECD.


The views and opinions expressed in this manuscript are solely those of the authors and do not necessarily constitute the policies of Environment and Climate Change Canada or Health Canada.

Conflicts of interest

There are no conflicts to declare.


  1. UNEP, Final Act of the Conference of Plenipotentiaries on the Stockholm Convention on Persistent Organic Pollutants, United Nations Environment Program, Geneva, Switzerland, 2001, http://www.pops.int/, accessed September 2017 Search PubMed.
  2. UNECE, Convention on Long Range Transboundary Air Pollutants, Geneva, Switzerland, 1979, http://www.unece.org/env/lrtap/welcome.html, accessed September 2017 Search PubMed.
  3. Government of Canada, New Substances Notification Regulations (Chemicals and Polymers) Canada Gazette Part II, Ottawa, 2005, vol. 139, No. 19 Search PubMed.
  4. D. Mackay, S. Paterson, A. Di Guardo and C. E. Cowan, Environ. Toxicol. Chem., 1996, 15(9), 1627–1637 CrossRef CAS.
  5. D. Van de Meent, SimpleBOX: A Generic Multimedia Fate Evaluation Model, Rep. report number 672720 001, National Institute of Public Health and Environmental Protection (RIVM), Bilthoven, The Netherlands, 1992 Search PubMed.
  6. [ConsExpo Web] Consumer Exposure Web Model, Rijksinstituut voor Volksgezondheid en Milieu [National Institute for Public Health and the Environment], Bilthoven (NL), 2016, http://www.rivm.nl/en/Topics/C/ConsExpo Search PubMed.
  7. Government of Canada, Priority Substances List, https://www.ec.gc.ca/lcpe-cepa/default.asp?lang=En%26n=C6C230D5-1, accessed September 2017 Search PubMed.
  8. Government of Canada, Categorization of Chemical Substances, https://www.canada.ca/en/health-canada/services/chemical-substances/canada-approach-chemicals/categorization-chemical-substances.html, accessed September 2017 Search PubMed.
  9. SAICM (Strategic Approach to International Chemicals Management), http://www.saicm.org/, accessed September 2017.
  10. Government of Canada, International chemicals management: strategic approach, 2017, https://www.ec.gc.ca/international/default.asp?lang=En%26n=F0505BC6-1, accessed September 2017 Search PubMed.
  11. UNEP (United Nations Environment Program), Final Act of the Conference of Plenipotentiaries on the Stockholm Convention on Persistent Organic Pollutants, United Nations Environment Program, Geneva, Switzerland, 2001, http://chm.pops.int/Portals/0/Repository/COP4/UNEP-POPS-COP.4-28.English.PDF, accessed December 2017 Search PubMed.
  12. Government of Canada, Toxic Substances Management Policy, Cat. No. En40–499/1-1995, 1995, https://www.canada.ca/en/environment-climate-change/services/management-toxic-substances/policy.html, accessed December 2017 Search PubMed.
  13. Government of Canada, Toxic Substance Management Policy: Persistence and Bioaccumulation Criteria, Cat No. En40–499/2-1995E, 1995, https://www.canada.ca/en/environment-climate-change/services/management-toxic-substances/policy.html, accessed December 2017 Search PubMed.
  14. Government of Canada, Persistence and Bioaccumulation Regulations. Canada Gazette: Part II, 2000, vol. 134, No. 7, pp. 606–612, https://www.ec.gc.ca/lcpe-cepa/eng/regulations/detailReg.cfm?intReg=35, accessed December 2017 Search PubMed.
  15. OECD (Organization for Economic Cooperation and Development), Report of the OECD/UNEP Workshop on the Use of Multimedia Models for Estimating Overall Environmental Persistence and Long-Range Transport in the Context of PBTs/POPs Assessment, OECD Series on Testing and Assessment No. 36, OECD, Paris, 2002, http://www.oecd-ilibrary.org/environment/report-of-the-oecd-unep-workshop-on-the-use-of-multimedia-models-for-estimating-overall-environmental-persistence-and-long-range-transport-in-the-context-of-pbts-pops-assessment_9789264078505-en, accessed December 2017 Search PubMed.
  16. OECD (Organization for Economic Cooperation and Development), Guidance document on the use of multimedia models for estimating overall environmental persistence and long-range transport. OECD Series on Testing and Assessment. No. 45. ENV/JM/MONO(2004)5, 2004, http://www.oecd-ilibrary.org/environment/guidance-document-on-the-use-of-multimedia-models-for-estimating-overall-environmental-persistance-and-long-range-transport_9789264079137-en, accessed December 2017 Search PubMed.
  17. OECD (Organization for Economic Cooperation and Development), OECD Pov and LRTP Screening Tool, http://www.oecd.org/chemicalsafety/risk-assessment/oecdpovandlrtpscreeningtool.htm, accessed September 2017 Search PubMed.
  18. F. Wegmann, L. Cavin, M. MacLeod, M. Scheringer and K. Hungerbühler, Environ. Modell. Softw., 2009, 24, 228–237 CrossRef.
  19. A. Beyer, D. Mackay, M. Matthies, F. Wania and E. Webster, Environ. Sci. Technol., 2000, 34, 699–703 CrossRef CAS.
  20. F. Wania, Environ. Sci. Technol., 2003, 37, 1344–1351 CrossRef CAS.
  21. M. Stroebe, M. Scheringer, H. Held and K. Hungerbühler, Sci. Total Environ., 2004, 321, 1–20 CrossRef CAS PubMed.
  22. M. Stroebe, M. Scheringer and K. Hungerbühler, Sci. Total Environ., 2004, 38, 5665–5673 CrossRef CAS.
  23. K. Fenner, M. Scheringer, M. MacLeod, M. Matthies, T. E. McKone, M. Stroebe, A. Beyer, M. Bonnell, A. C. Le Gall, J. Klasmeier, D. Mackay, D. W. Pennington, B. Scharenberg and F. Wania, Environ. Sci. Technol., 2005, 39, 1932–1942 CrossRef CAS PubMed.
  24. F. A. P. C. Gobas, Ecol. Modell., 1993, 69, 1–17 CrossRef CAS.
  25. F. A. P. C. Gobas, B. C. Kelly and J. A. Arnot, QSAR Comb. Sci., 2003, 22(3), 329–336 CAS.
  26. J. A. Arnot and F. A. P. C. Gobas, QSAR Comb. Sci., 2003, 22(3), 337–345 CAS.
  27. J. A. Arnot and F. A. P. C. Gobas, Environ. Toxicol. Chem., 2004, 23(10), 2343–2355 CrossRef CAS PubMed.
  28. J. A. Arnot, W. Meylan, J. Tunkel, P. H. Howard, D. Mackay, M. Bonnell and R. S. Boethling, Environ. Toxicol. Chem., 2009, 28(6), 1168–1177 CAS.
  29. S. Dimitrov., N. Dimitrova, T. E. Parkerton, M. Comber, M. Bonnell and O. Mekenyan, SAR QSAR Environ. Res., 2005, 16, 531–554 CrossRef CAS PubMed.
  30. J. A. Arnot, D. Mackay and M. Bonnell, Environ. Toxicol. Chem., 2008, 27(2), 341–351 CAS.
  31. J. A. Arnot, D. Mackay, T. F. Parkerton and M. Bonnell, Environ. Toxicol. Chem., 2008, 27(11), 2263–2270 CAS.
  32. T. N. Brown, J. A. Arnot and F. Wania, Environ. Sci. Technol., 2012, 46(15), 8253–8260 CrossRef CAS PubMed.
  33. E. Papa, L. van der Wal, J. A. Arnot and P. Gramatica, Sci. Total Environ., 2014, 470–471 Search PubMed.
  34. W. M. Meylan, P. H. Howard, R. S. Boethling, D. Aronson, H. Printup and S. Gouchie, Environ. Toxicol. Chem., 1999, 18(4), 664–672 CrossRef CAS.
  35. J. A. Arnot and F. A. P. C. Gobas, Environ. Rev., 2006, 14, 257–297 CrossRef CAS.
  36. J. W. Nichols, M. Bonnell, S. Dimitrov, B. Escher, X. Han and N. Kramer, Integr. Environ. Assess. Manage., 2009, 5(4), 577–597 CrossRef CAS PubMed.
  37. F. A. P. C. Gobas, L. Burkhard, M. Bonnell, B. Doucette, K. Sappington, E. Verbruggen, B. Hope, J. Arnot and J. Tarazona, Integr. Environ. Assess. Manage., 2016, 12(1), 123–134 CrossRef CAS PubMed.
  38. J. C. Lo, D. J. Letinski, T. F. Parkerton, D. A. Campbell and F. A. P. C. Gobas, Environ. Sci. Technol., 2016, 50, 13299–13308 CrossRef CAS PubMed.
  39. J. A. Arnot and C. L. Quinn, Environ. Sci. Technol., 2015, 49, 4783–4796 CrossRef CAS PubMed.
  40. G. Czub and M. S. McLachlan, Environ. Sci. Technol., 2004, 38, 2406–2412 CrossRef CAS PubMed.
  41. Health Canada, Science Approach Document: Threshold of Toxicological Concern (TTC)-based Approach for Certain Substances, 2016, p. 54, http://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=326E3E17-1 Search PubMed.
  42. [ConsExpo] Consumer Exposure Model. ver. 4.1, Rijksinstituut voor Volksgezondheid en Milieu [National Institute for Public Health and the Environment], Bilthoven (NL), 2006, http://www.rivm.nl/en/Topics/C/ConsExpo, accessed October 2017 Search PubMed.
  43. [IH MOD] Industrial Hygiene Model. ver. 0.161, American Industrial Hygiene Association, Exposure Assessment Strategies Committee, Falls Church (VA), 2009, https://www.aiha.org/get-involved/VolunteerGroups/Pages/Exposure-Assessment-Strategies-Committee.aspx, accessed October 2017 Search PubMed.
  44. [WPEM] Wall Paint Exposure Assessment Model. ver. 3.2, US Environmental Protection Agency, Office of Pollution Prevention and Toxics, National Paint and Coatings Association, 2001, Washington DC, http://www.epa.gov/tsca-screening-tools/wall-paint-exposure-assessment-model-wpem, accessed October 2017 Search PubMed.
  45. [IA-QUEST] Indoor Air Quality Emission Simulation Tool. ver. 1.1, National Research Council Canada, Institute for Research in Construction, 2008, Ottawa (ON), https://www.nrc-cnrc.gc.ca/eng/solutions/advisory/ia_quest.html, accessed October 2017 Search PubMed.
  46. [CEM] Consumer Exposure Model. ver. 2.0, US Environmental Protection Agency, Office of Pollution Prevention and Toxics, 2017, Washington DC, https://www.epa.gov/tsca-screening-tools/cem-consumer-exposure-model-download-and-install-instructions, accessed October 2017 Search PubMed.
  47. [ExpoFIRST] Exposure Factors Interactive Resource for Scenarios Tool. ver. 2.0, US Environmental Protection Agency, Office of Research and Development, 2016, Washington (DC), https://cfpub.epa.gov/ncea/risk/recordisplay.cfm?deid=322489, accessed October 2017 Search PubMed.
  48. [DustEx] DustEx tool, Rijksinstituut voor Volksgezondheid en Milieu [National Institute for Public Health and the Environment], Bilthoven (NL), 2017, http://www.rivm.nl/en/Topics/C/ConsExpo/Related_tools/DustEx_tool/ Search PubMed.
  49. [IECCU] Indoor Environmental Concentrations in Buildings with Conditioned and Unconditioned Zones. ver. 1.0, US Environmental Protection Agency, Office of Pollution Prevention and Toxics, Washington DC, 2017, https://www.epa.gov/tsca-screening-tools/users-guide-and-download-ieccu-indoor-environmental-concentrations-buildings, accessed October 2017 Search PubMed.
  50. EUSES, European Union System for the Evaluation of Substances, https://ec.europa.eu/jrc/en/scientific-tool/european-union-system-evaluation-substances, accessed September 2017 Search PubMed.
  51. E. Webster, D. Mackay, A. Di Guardo, D. Kane and D. Woodfine, Chemosphere, 2004, 55, 1361–1376 CrossRef CAS PubMed.
  52. J. A. Arnot and D. Mackay, Environ. Sci. Technol., 2008, 42, 4648–4654 CrossRef CAS PubMed.
  53. J. A. Arnot, D. Mackay, E. Webster and J. Southwood, Environ. Sci. Technol., 2006, 40, 2316–2323 CrossRef CAS PubMed.
  54. J. A. Arnot, Updating the RAIDAR and FHX models to aid in the prioritization and assessments of chemicals including ionisable substances. Technical report prepared for Health Canada, Health Canada, Ottawa, 2011 Search PubMed.
  55. J. A. Arnot, Parameterize and apply the RISK Assessment Identification And Ranking (RAIDAR) model to screen approximately 1370 substances. Technical Report prepared for Environment and Climate Change Canada, ECCC, Gatineau, Quebec, 2014 Search PubMed.
  56. T. Gouin, R. van Egmond, O. R. Price and J. E. N. Hodges, Environ. Pollut., 2014, 165, 208–214 CrossRef PubMed.
  57. USETox, UNEP/SETAC scientific consensus model for characterizing human toxicological and ecotoxicological impacts of chemical emissions in life cycle assessment, http://www.usetox.org, accessed September 2017 Search PubMed.
  58. D. Mackay, D. Hughes, D. M. Romano and M. Bonnell, Integr. Environ. Assess. Manage., 2014, 10(4), 588–594 CrossRef PubMed.
  59. D. Mackay, M. Joy and S. Paterson, Chemosphere, 1983, 12, 981–997 CrossRef CAS.
  60. D. Mackay, M. Joy and S. Paterson, Chemosphere, 1983, 12, 1193–1208 CrossRef CAS.
  61. J. G. Teeguarden, Y.-M. Tan, S. W. Edwards, J. A. Leonard, K. A. Anderson, R. A. Corley, M. L. Kile, S. M. Simonich, D. Stone, R. L. Tanguay, K. M. Waters, S. L. Harper and D. E. Williams, Environ. Sci. Technol., 2016, 50(9), 4579–4586 CrossRef CAS PubMed.
  62. G. T. Ankley, R. S. Bennett, R. J. Erickson, D. J. Hoff, M. W. Hornung, R. D. Johnson, D. R. Mount, J. W. Nichols, C. L. Russom, P. K. Schmieder, J. A. Serrrano, J. E. Tietge and D. L. Villeneuve, Environ. Toxicol. Chem., 2010, 29(3), 730–741 CrossRef CAS PubMed.
  63. K. Krishnan and T. Peyret, in Ecotoxicology Modeling. Emerging Topics in Ecotoxicology (Principles, Approaches and Perspectives), ed. J. Devillers, Springer, Boston, MA, 2009, vol. 2, pp. 145–175 Search PubMed.
  64. B. A. Wetmore, J. F. Wambaugh, B. Allen, S. S. Ferguson, M. A. Sochaski, R. W. Setzer, K. A. Houck, C. L. Strope, K. Cantwell, R. S. Judson, E. LeCluyse, H. J. Clewell, R. S. Thomas and M. E. Andersen, Toxicol. Sci., 2015, 148(1), 121–136 CrossRef CAS PubMed.
  65. L. Geraets, J. G. M. Bessems, M. J. Zeilmaker and P. M. J. Bos, Regul. Toxicol. Pharmacol., 2014, 70, 54–64 CrossRef CAS PubMed.
  66. Health Canada, Federal Contaminated Site Risk Assessment in Canada. Part II: Health Canada Toxicological Reference Values (TRVs) and Chemical-Specific Factors, version 2.0, 2010, http://publications.gc.ca/collections/collection_2012/sc-hc/H128-1-11-638-eng.pdf Search PubMed.
  67. World Health Organization, Environmental Health Criterion 235: Dermal Absorption, 2006, http://www.inchem.org/documents/ehc/ehc/ehc235.pdf Search PubMed.
  68. US Environmental Protection Agency, Dermal Exposure Assessment: Principles and Applications. Interim Report. Exposure Assessment Group, EPA/600/8–91/011B, Office of Health and Environmental Assessment, U.S. Environmental Protection Agency, Washington, DC 20460, 1992 Search PubMed.
  69. National Institute for Occupational Safety and Health, Skin Exposures & Effects. Finite Dose Skin Permeation Calculator, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, 2013, http://www.cdc.gov/niosh/topics/skin/finiteskinpermcalc.html Search PubMed.
  70. American Industrial Hygiene Association, IH SkinPerm version 2.0, 2017, https://www.aiha.org/get-involved/VolunteerGroups/Pages/Exposure-Assessment-Strategies-Committee.aspx Search PubMed.
  71. European Commission, Guidanced Document on Dermal Absorption. Sanco/222/2000 rev. 7, Health & Consumer Protection Diretorate-General, European Commission, 2004, https://ec.europa.eu/food/sites/food/files/plant/docs/pesticides_ppp_app-proc_guide_tox_dermal-absorp-2004.pdf Search PubMed.
  72. F. M. Williams, H. Rothe, G. Barrett, A. Chiodini, J. Whyte, M. T. D. Cronin, N. A. Monteiro-Riviere, J. Plautz, C. Roper, J. Westerhout, C. Yang and R. H. Guy, Regul. Toxicol. Pharmacol., 2016, 76, 174–186 CrossRef CAS PubMed.
  73. J. Shen, L. Kromidas, T. Schultz and S. Bhatia, Food Chem. Toxicol., 2014, 74, 164–176 CrossRef CAS PubMed.
  74. R. Kroes, A. G. Renwick, V. Feron, C. L. Galli, M. Gibney, H. Greim, R. H. Guy, J. C. Lhuguenot and J. J. M. van de Sandt, Food Chem. Toxicol., 2007, 45, 2533–2562 CrossRef CAS PubMed.
  75. R. H. Guy, Chem. Res. Toxicol., 2010, 23, 864–870 CrossRef CAS PubMed.
  76. A. Garrigues, S. Grégoire, D. Duche and J. R. Meunier, Alternative approach to maximum flux for TTC applied to safety evaluation of cosmetic ingredients, ADMET – Cutaneous Absorption, L’ORÉAL Research and Innovation, Aulnay-sous-Bois (FR), 2011 Search PubMed.
  77. Health Canada, Federal Contaminated Site Risk Assessment in Canada. Part I: Guidance on Human Health Preliminary Quantitative Risk Assessment (PQRA), version 2.0, 2012, http://publications.gc.ca/collections/collection_2012/sc-hc/H128-1-11-632-eng.pdf Search PubMed.
  78. Health Canada, Federal Contaminated Sites Risk Assessment in Canada. Interim Guidance on Human Health Risk Assessment for Short-Term Exposure to Carcinogens at Contaminated Sites, 2013, http://publications.gc.ca/collections/collection_2013/sc-hc/H144-11-2013-eng.pdf Search PubMed.
  79. European Chemicals Agency, Guidance on Information Requirements and Chemical Safety Assessment. Chapter R.15: Consumer exposure assessment. version 3.0, 2016, https://echa.europa.eu/documents/10162/13632/information_requirements_r15_en.pdf Search PubMed.
  80. Health Canada, Federal Contaminated Site Risk Assessment in Canada. Part V: Guidance on Human Health Detailed Quantitative Risk Assessment for Chemicals (DQRAChem), 2010, http://publications.gc.ca/collections/collection_2011/sc-hc/H128-1-11-639-eng.pdf Search PubMed.
  81. ECHA (European Chemicals Agency), Grouping of Substances and Read-Across, https://echa.europa.eu/support/registration/how-to-avoid-unnecessary-testing-on-animals/grouping-of-substances-and-read-across, accessed October 2017 Search PubMed.
  82. OECD (Organization for Economic Cooperation and Development), Guidance for the Grouping of Chemicals Second Edition. OECD Series on Testing and Assessment. No. 194, ENV/JM/MONO(2014)4, 2014 Search PubMed.
  83. K. Phillips, A. McEachran, J. Sobus and K. Isaacs, Spring Meeting, American Chemical Society, San Francisco, CA, 2017 Search PubMed.
  84. Government of Canada, The Rapid Screening Approach, https://www.canada.ca/en/health-canada/services/chemical-substances/chemicals-management-plan/initiatives/rapid-screening-approach-chemicals-management-plan.html, accessed September 2017 Search PubMed.
  85. ECCC (Environment and Climate Change Canada), Science Approach Document: Ecological Risk Classification of Organic Substances, p. 74, https://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=A96E2E98-1, accessed September 2017 Search PubMed.
  86. M. Bonnell, Prioritizing and Assessing Organic Chemicals: Are PBT Criteria Still Useful Today?, Society of Environmental Toxicology and Chemistry, Pensacola, Florida, 2017 Search PubMed.
  87. D. C. G. Muir and P. Howard, Environ. Sci. Technol., 2006, 40, 7157–7166 CrossRef CAS PubMed.
  88. C. E. Cowan-Ellsberry, M. S. McLachlan, J. A. Arnot, M. MacLeod, T. E. McKone and F. Wania, Integr. Environ. Assess. Manage., 2009, 5, 662–679 CrossRef CAS PubMed.
  89. A. Falls and J. A. Arnot, Applying RAIDAR to Address Data Gaps for Assessing the Exposure and Potential Risk of Select Organic Flame Retardants to the Environment. Technical report prepared for Environment and Climate Change Canada, ECCC, Gatineau, Quebec, 2015 Search PubMed.
  90. Government of Canada, Science Approach Document – Biomonitoring-based Approach 1 for Beryllium, Vanadium, trichlorooxo Vanadium oxide, 2016, http://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=B0FA951B-1, accessed November 2017 Search PubMed.
  91. Government of Canada, Science Approach Document – Biomonitoring-based Approach 2 for Barium-containing Substances, Molybdenum-containing Substances, Silver-containing substances, Thallium-containing Substances, Inorganic Tin-containing Substances, 2016, http://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=D335D89F-1, accessed November 2017 Search PubMed.
  92. Government of Canada, Assessment Report - Triclosan, 2016, http://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=65584A12-1, accessed November 2017 Search PubMed.
  93. Government of Canada, Draft Screening Assessment – Phthalate Substance Grouping, 2017, http://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=516A504A-1, accessed November 2017 Search PubMed.
  94. A. Zidek, K. Macey, L. MacKinnon, M. Patel, D. Poddalgoda and Y. Zhang, Int. J. Hyg. Environ. Health, 2016, 220, 167–178 CrossRef PubMed.
  95. A. Franco, A. Ferranti, C. Davidsen and S. Trapp, Int. J. Life Cycle Assess., 2010, 15, 321–325 CrossRef CAS.
  96. Government of Canada, List of substances in the third phase of CMP (2016–2020), 2017, update. http://www.ec.gc.ca/ese-ees/default.asp?lang=En%26n=2A33EEC9-1, accessed September 2017 Search PubMed.
  97. A. Franco and S. Trapp, Environ. Toxicol. Chem., 2010, 29, 789–799 CrossRef CAS PubMed.
  98. J. M. Armitage, J. A. Arnot, F. Wania and D. Mackay, Environ. Toxicol. Chem., 2013, 32, 115–128 CrossRef CAS PubMed.
  99. A. Franco, J. Struijs, T. Gouin and O. R. Price, Integr. Environ. Assess. Manage., 2013, 9(4), 569–579 CrossRef CAS PubMed.
  100. J. A. Arnot and J. Armitage, Parameterization and application of the raidar model to aid in the prioritization and assessment of chemical substances. Technical Report prepared for Health Canada, Health Canada, Ottawa, 2013 Search PubMed.
  101. S. T. J. Droge and K. Goss, Environ. Sci. Technol., 2013, 47, 798–806 CrossRef CAS PubMed.
  102. S. T. J. Droge and K. Goss, Environ. Sci. Technol., 2013, 47(24), 14233–14241 CrossRef CAS PubMed.
  103. S. T. J. Droge, J. L. Hermens, J. Rabone, S. Gutsell and G. Hodges, Environ. Sci.: Processes Impacts, 2016, 18, 1011–1023 CAS.

This journal is © The Royal Society of Chemistry 2018