Contents
Figures...... vi Tables...... vii Objectives...... viii Audience...... viii Acknowledgments...... ix
PART 1: BASIC CONCEPTS IN ENVIRONMENTAL HEALTH RISK ASSESSMENT...... 1
Chapter 1: Introduction to environmental health risk assessment...... 3 1.1 What is risk assessment?...... 3 1.2 When to undertake risk assessment...... 4 1.3 Types of risk assessment...... 5 1.4 The distinction between risk assessment and risk management...... 6 1.5 Evaluating risk assessment methods...... 6 1.6 Risk assessment models and formats...... 7 1.7 The five stages of environmental health risk assessment (EHRA)...... 7 1.8 Risk assessment frameworks...... 9 1.9 Tiered approaches to EHRA...... 12 1.10 Deterministic versus probabilistic estimates in EHRA...... 13 1.11 Environmental health risk assessment and health impact assessment...... 14 1.12 Environmental health risk assessment and the ‘precautionary principle’...... 14 Chapter 2: Problem formulation and scope...... 16 2.1 Identifying and describing issues within existing environmental conditions...... 16 2.2 Identifying and describing susceptible populations...... 17 2.3 Identification of potential exposure pathways...... 22 2.4 Potential interactions between agents...... 22 2.5 Identifying potential management options that may mitigate exposure...... 23 2.6 What risk and other technical assessments are necessary to evaluate risk and discriminate between potential risk management options?...... 23 Chapter 3: Hazard identification and dose–response assessment...... 24 3.1 Introduction...... 24 3.2 Hazard identification...... 24 3.3 Dose–response assessment...... 26 3.4 The prime role of animal studies in risk assessment...... 26 3.5 Dose scaling...... 26 3.6 Dose–response curve shape, construct and analysis...... 28 3.7 Hormesis...... 32 3.8 Dose–response modelling...... 32 3.9 Benchmark dose approach...... 33 3.10 Threshold versus non-threshold responses in risk assessment...... 36 3.11 Threshold versus non-threshold approaches – implications for risk assessment...... 39 3.12 Science policy and the selection of threshold and non-threshold models...... 40
ENVIRONMENTAL HEALTH RISK ASSESSMENT i Chapter 4: Exposure assessment...... 42 Chapter 7: Reviewing and appraising an environmental health risk assessment report...... 95 4.1 Introduction...... 42 7.1 Problem formulation and scope...... 95 4.2 Terminology used in exposure assessment...... 42 7.2 Hazard identification...... 95 4.3 Planning an exposure assessment...... 45 7.3 Dose–response checklist...... 96 4.4 Development of a conceptual site model...... 47 7.4 Selection of guideline values...... 97 4.5 Approaches to quantifying exposures...... 50 7.5 Data presentation...... 97 4.6 Exposure assessment calculations...... 52 7.6 Exposure assessment...... 98 4.7 Data analysis and evaluation...... 53 7.7 Risk characterisation...... 100 4.8 Individual exposure, dose and risk estimates...... 54 7.8 Confirmation of utility...... 100 4.9 Issues in exposure assessments...... 55 7.9 General assessment and report presentation...... 100 4.10 Exposure assessment of volatile agents...... 56 4.11 Exposure duration...... 56 PART 2: ADDITIONAL GUIDANCE ON SELECTED ISSUES...... 103 4.12 Adjustments for sensitive sub-populations...... 58 Chapter 8: Data collection and analysis...... 105 4.13 Default values for exposure assessments...... 58 8.1 Environmental distribution...... 105 4.14 Sources of exposure assessment data...... 64 8.2 Environmental sampling and analysis...... 105 Chapter 5: Risk characterisation...... 66 8.3 Analytical methodologies...... 107 5.1 Introduction...... 66 8.4 Quality assurance of data used in site-specific health risk assessment...... 108 5.2 Key principles in environmental health risk characterisation...... 66 8.5 Data analysis and presentation...... 110 5.3 Quantitative and qualitative risk characterisation...... 67 8.6 Some principles of graphical representation...... 112 5.4 Risk estimation...... 68 8.7 Censored data...... 113 5.5 Toxicological reference values (TRVs) in common use...... 69 Chapter 9: Evaluation of toxicity data...... 114 5.6 Determination of NO(A)ELs, ADIs (RfD) and TDIs for humans...... 70 9.1 Toxicity testing – major in vivo study types...... 114 5.7 Toxicological reference values derived using a non-threshold approach...... 72 9.2 Guidance on evaluating and interpreting toxicity tests...... 115 5.8 Age-specific adjustment factors...... 73 9.3 Evaluating the weight of evidence and considering the toxicology database in toto...... 115 5.9 Combining risk estimates...... 73 9.4 Developments in toxicity testing...... 116 5.10 Target risk levels...... 74 Chapter 10: Use of epidemiological data...... 118 5.11 Principles for setting risk-based environmental health guideline values...... 75 10.1 Introduction...... 118 5.12 Guidance on selecting sources of toxicological data and environmental health criteria...... 79 10.2 Types of epidemiological study...... 119 5.13 Chemicals with no available toxicity data...... 81 10.3 Investigation of apparent clusters...... 121 5.14 QSAR and read across techniques...... 82 10.4 Bias and confounding: key concepts in environmental epidemiology...... 122 5.15 Uncertainty and sensitivity analysis...... 83 10.5 Assessing the relationship between a possible cause and an outcome...... 122 5.16 Interplay of scientific judgement and science policy...... 85 10.6 The strengths and limitations of observational epidemiology versus experimental toxicology...... 124 Chapter 6: Community engagement...... 88 10.7 Critical evaluation of published research...... 127 6.1 Introduction...... 88 10.8 Undertaking health studies...... 131 6.2 Risk perception and risk communication in the context of community engagement...... 88 Chapter 11: Assessment of carcinogens...... 136 6.3 Risk communication – things to know and things to avoid...... 91 11.1 Methods for the hazard identification of carcinogens...... 137 6.4 Risk communication – understanding conflicts...... 92 11.2 Guidance on assessing carcinogens...... 137 6.5 Planning risk communication...... 92 11.3 Classification of carcinogens...... 138 11.4 When is a carcinogenic response relevant to humans?...... 141 11.5 IPCS program on chemical carcinogenesis...... 142 11.6 Quantitative risk assessment of carcinogens...... 144 11.7 An alternative approach to quantitative carcinogen risk assessment?...... 145 ii ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT iii Chapter 12: Assessing multiple routes and sources of exposure...... 146 17.6 Standards Australia model of risk management...... 186 12.1 Risk assessment of chemical mixtures...... 146 17.7 Risk management associated with genetically modified organisms...... 187 12.2 Assessment of representative mixtures...... 147 17.8 Supporting professional organisations...... 187 12.3 The toxicity equivalence factor approach...... 147 Chapter 18: International programs and guidance on EHRA...... 188 12.4 Summation of risk estimates...... 150 18.1 United States programs...... 188 12.5 Component elimination or simplification...... 151 18.2 United Kingdom programs...... 190 12.6 Biomarker approach...... 151 18.3 International programs dealing with chemical safety...... 191 12.7 International programs and reviews...... 151 18.4 Programs for developing alternatives to animal testing...... 193 12.8 Other reviews of chemical mixtures HRA...... 152 18.5 Integration of human health and ecotoxicology...... 194 Chapter 13: Exposure modelling...... 153 Appendix 1: Guidance on the evaluation of toxicity studies...... 195 13.1 Using point estimates and probability distributions...... 154 Appendix 2: Extracts from a 2010 NHMRC discussion paper on health-based targets for microbial safety of 13.2 Monte Carlo analysis...... 154 drinking water supplies...... 204 13.3 Integration of exposure with EHRA outcomes...... 160 Abbreviations...... 208 Chapter 14: Biomonitoring...... 161 Glossary...... 214 14.1 Biomarkers...... 161 References...... 225 14.2 Choice of tissue...... 163 14.3 Choice of a test...... 164 14.4 Influences on biological monitoring results...... 166 14.5 Exposure and biological monitoring results...... 166 14.6 Abnormal results...... 167 14.7 Health monitoring...... 167 14.8 Biomonitoring and blood lead...... 167 Chapter 15: Microbiological risk assessment...... 169 15.1 Introduction...... 169 15.2 Definitions...... 169 15.3 General principles...... 170 15.4 Microbiological risk assessment – paradigms and frameworks...... 170 15.5 Expression of microbiological risk...... 170 Chapter 16: Guidance on route-specific EHRA...... 172 16.1 Assessment of site contamination National Environmental Protection Measure...... 172 16.2 Air pollutants...... 172 16.3 Food contaminants...... 174 16.4 Water contamination...... 175
PART 3: REGULATORY CONTEXT...... 177
Chapter 17: Australian regulation of chemical hazards...... 179 17.1 Overview of the chemical risk assessment agencies...... 179 17.2 National Industrial Chemicals Notification and Assessment Scheme...... 180 17.3 Australian Pesticides and Veterinary Medicines Authority (APVMA)...... 182 17.4 Role of the Office of Chemical Safety (OCS) in public health risk assessment...... 183 17.5 Role of the Department of Sustainability, Environment, Water, Population and Communities (DSEWPC) in environmental risk assessment...... 186 iv ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT v FIGURES TABLES Figure 1: Environmental health risk assessment model...... 9 Table 1: Factors influencing the absorption of chemicals...... 20 Figure 2: A revised outline of the interlinked processes of EHRA...... 10 Table 2: Factors influencing the distribution of chemicals in the body...... 20 Figure 3: Expanded illustration of the major exposure pathways, potentially exposed groups leading to potential health outcomes...... 11 Table 3: Factors influencing the rate of metabolism of chemicals...... 20 Figure 4: Elements of a tiered approach to EHRA...... 12 Table 4: Factors influencing the elimination of chemicals from the body...... 20 Figure 5: Conceptual representation of the proposed IPCS risk assessment framework...... 13 Table 5: Dose conversion from mg/kg (ppm) in diets to mg/kg/day in animal toxicity studies...... 27 Figure 6: Potential sources of information used to identify and characterise environmental hazards, leading to characterisation Table 6: Explanations of terms used in assessing dose and exposure...... 43 of mode of action (MoA), dose–response models and susceptible populations...... 25 Table 7: Summary of suggested exposure factors for adults...... 59 Figure 7: Hypothetical curve for an animal carcinogenicity study...... 29 Table 8a: Summary of suggested exposure factors for 2–3-year-old child (male and female combined)...... 61 Figure 8: Different dose–response curves for different effects from a hypothetical substance...... 29 Table 8b: Summary of suggested exposure factors for 1–2-year-old child (male and female combined)...... 62 Figure 9: A conceptual model that integrates individual risk dose–response, background disease incidence and variation in Table 9: Summary of non-age-dependent exposure factors...... 63 susceptibility into population dose–response...... 30 Table 10: Environmental health criteria derived from a threshold approach...... 70 Figure 10: Extrapolation from the individual risk to population risk incorporating uncertainty estimates...... 31 Table 11: Current uses of the TTC approach in chemicals regulation...... 81 Figure 11: Conceptual models describing low-dose linearity or non-linearity of individual and population dose–response curves...... 31 Table 12: Useful versus not useful graphics...... 112 Figure 12: Examples of the potential variability when different models are used to extrapolate risk...... 33 Table 13: Study designs in environmental epidemiology that use the individual as the unit of analysis...... 120 Figure 13: Graphical illustrations of the benchmark dose...... 34 Table 14: Applications of different observational study designs...... 121 Figure 14: Analysis of data on formaldehyde toxicity using different mathematical curve-fitting models...... 35 Table 15: Advantages and disadvantages of different observational study designs...... 121 Figure 15: Decision-making process for choosing dose–response data in risk assessment of carcinogenic substances...... 39 Table 16: Guidelines for assessing causation...... 123 Figure 16: Decision tree for choosing a threshold or non-threshold model for risk assessment...... 41 Table 17: Relative ability of different types of study to ‘prove’ causation...... 124 Figure 17: Representation of dose and exposure...... 44 Table 18: Checklist for evaluating published research...... 131 Figure 18: Exposure pathways...... 46 Table 19: Toxicity equivalence factors (TEFs) for dioxin congeners...... 148 Figure 19: Representation of a CSM flow chart for potential airborne exposures from an industrial site...... 48 Table 20: TEFs values for PAHs...... 149 Figure 20: CSM site diagrams representing potential exposure pathways from a contaminated site...... 49 Table 21: Some key variables for which probability distributions might be needed...... 158 Figure 21: Components of exposure assessment...... 50 Table 22: Substances likely to be suitable for biological monitoring...... 165 Figure 22: Illustration of potential variability in exposure patterns...... 57 Table 23: Chemicals regulation in Australia...... 180 Figure 23: Proposed subdivision of default uncertainty factors to be used in risk assessment...... 72 Table A1: Calculation of pathogen reductions for a source water containing 1 organism per litre of Cryptosporidium, Rotaviruses Figure 24: Decision tree for the development of risk-based environmental health GVs...... 76 and Campylobacter...... 206 Figure 25: Graphical depiction of the range of variability in toxicological endpoints that may be used in an EHRA...... 98 Figure 26: Graphical depiction of the range of variability in odds ratios describing the effects of water disinfection by-products on birth defects...... 99 Figure 27: Example of a structured EHRA report for air emissions from an industrial facility...... 102 Figure 28: Contour modelling of ground-level concentration (GLC) data...... 111 Figure 29: Assessing the relationship between a possible cause and an outcome when an association is observed...... 123 Figure 30: Principles of the Monte Carlo method...... 155 Figure 31: Proposals for integrating exposure with outcomes of EHRA...... 160 Figure 32: Flow charts for deriving BEs for chemicals pharmacokinetic data are available for both animals and humans along with toxicity data from both species (a) or only from animals (b)...... 162 Figure 33: Choice of biological tissue or fluid for a hypothetical substance...... 163
vi ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT vii Due to the complexity and scale of the The assistance of other contributors to OBJECTIVES environmental health risk assessment AUDIENCE ACKNOWLEDGMENTS the 2002 edition was acknowledged in process a concise ‘cookbook’ is not that edition. Since some of the text was This enHealth document provides a This enHealth document is primarily This enHealth document is an update practicable, although this document carried through into the 2012 revision, national approach to environmental health intended to be used by: environmental of the original enHealth guidance attempts to provide pragmatic and user- in a revised format, their contributions are risk assessment. health agencies reviewing risk document Environmental health risk friendly advice. Similarly, the situation- once again acknowledged with thanks. assessments; people preparing risk assessment: Guidelines for assessing specific issues are often sufficiently Risk assessments are being assessments for environmental health human health risks from environmental complex and situation-specific that a The steering committee that oversaw undertaken for a wide variety of agencies; and those regulatory agencies hazards, first published in 2002. This manageable and complete algorithm the development of the 2012 update projects by governments and industry. reviewing risk assessments. It is also 2012 update has been funded by the for decision making cannot be drafted; comprised: Environmental health agencies need to be intended to be of assistance to a enHealth committee through a contract however, the document provides a series able to assess their content and approach broader audience seeking information with the Australian Centre for Human Ms Jan Bowman (chair), Dr Kevin of guidelines and checklists to assist against a benchmark. The document about processes of environmental risk Health Risk Assessment (ACHHRA) at Buckett, Dr Peter Di Marco, Dr Liz the decision-making process. Where presents a general environmental assessment in Australia. Monash University. The principal author Hanna, Dr Andrew Langley, Mr John possible, the document is prescriptive health risk assessment methodology of the update is Professor Brian Priestly, Dempsey and Dr Roscoe Taylor, with about certain aspects of risk assessment. applicable to a range of environmental Risk assessors should have a basic director of ACHHRA and president of secretariat assistance provided by Having specific requirements for the health hazards. The focus is on chemical grounding in epidemiology, toxicology the Australasian College of Toxicology the Environmental Health Unit of the content of investigations and having hazards in the first instance, but the and chemistry. and Risk Assessment (ACTRA), with Department of Health, Victoria. them presented in uniform, coherent core methodology can also be applied assistance in collecting and reviewing and logically developed reports will to physical (e.g. radiation, noise) and relevant literature by Dr Jennifer Ong, an All these people provided editorial enable more efficient, accurate, timely microbiological hazards. The core ACHHRA research associate. comment and assistance through the and transparent decision making and methodology is intended to be able to preparation of the draft revisions, with a greater consistency of environmental accommodate specialised ‘modules’ that Toxikos Pty Ltd, under the direction of further comments from staff of SA Health health decision making across Australia. will deal with issues such as physical and its principal toxicologist Dr Roger Drew, and Victorian Department of Health However, contemporary paradigms microbiological hazards and mixtures provided significant input into sections (Dr David Cunliffe, Dr Kateryna Babina, of risk assessment acknowledge that as they become available. The links dealing with exposure assessment, Dr Richard Evans and Mr Robert stakeholders may sometimes impose to risk management and community including a substantial re-drafting of Mckenzie). unrealistic demands on the available consultation/risk communication will be the Australian exposure factor guidance science of risk assessment and that identified. The document emphasises document initially drafted by Dr Andrew data gaps and uncertainties may limit International peer review on the draft the importance of prior planning and Langley. Dr Roger Drew also provided the options for establishing all the document was provided by Professor appropriate scoping in the design valuable assistance with planning the available risk management options. Steve Hrudey, University of Alberta, and phase of a risk assessment. It further review of the document, including Such knowledge gaps should not deter ChemRisk LLC, San Francisco (Ellen notes that appropriate consultation advice to ensure the document adopts decision makers from considering the Donovan and Hugh Scobie). Box 1 in with all stakeholders, but particularly a user-friendly approach to health risk range of options within an appropriate Section 5.16 describing the regulatory with decision makers, is essential to assessment and addresses the pragmatic science-based framework, but this experience with chloroform was drafted ensure the conceptual models and needs of risk assessment professionals. should always be done with a full and by Professor Hrudey. methodologies used are adequate to Mr John Frangos and Ms Tarah Hagen frank acknowledgement of the inherent address the desired outcomes. provided significant input into the revised uncertainties. Australian exposure factor guidance document.
The original 2002 version of the enHealth guidance was prepared by Dr Andrew Langley with assistance from various colleagues, including Mr Jack Dempsey, Dr Les Davies and Dr Jim Fitzgerald, who prepared the hazard identification section. Professor Steve Hrudey, Dr Andrew Langley, Dr Hugh Davis and Dr Les Davies provided extensive written editorial advice augmenting the comments made by people at the two workshops that reviewed the original document.
viii ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT ix Part 1: Basic concepts in environmental health risk assessment Chapter 1: Introduction to environmental health risk assessment
Virtually all aspects of life involve There are uncertainties related to which are often not explicitly laid out in exposure to risks (National Research risk assessment and it is important to legislation or regulations. The objective Council – NRC 2008). Understanding make the best possible use of available of this enHealth document is not to the nature of risk, including the way information. It is equally, if not more enunciate specific science policy relating people perceive threats to their health important, to be able to explain to to EHRA but to provide information to and the rational and emotive factors stakeholders in the EHRA processes how risk assessors on different approaches that govern that perception, is vital these uncertainties have been identified to EHRA methodology, and to provide to developing appropriate ways to and managed. guidance on how to use default values at manage environmental health risks. various stages of an EHRA. The difficulties Risk assessment can be a useful tool in Risk assessment gathers and organises in establishing such defaults within a managing environmental health risks. information and enables: science policy context are discussed in some detail in Section 5.16, where there •• risks at a point in time (including is a discussion on the selection of ‘target baseline risks) and changes in risk 1.1 risk’ in the EHRA of carcinogens. over time to be estimated and to WHAT IS RISK establish whether action is necessary Risk assessment may be done as ASSESSMENT? •• assessments of new and different a relatively rapid ‘desktop’ study or types of risk ‘screening’ study for simple issues, or may Risk assessment is the process of •• the identification and comparison of be a large and complex process where estimating the potential impact of a different factors that affect the nature there are significant health concerns. chemical, physical, microbiological or and magnitude of the risk These processes may be designated as psychosocial hazard on a specified •• issues to be prioritised according to Tier 1, 2 or 3 processes (see Section human population or ecological system their levels of risk 1.9). There are numerous models of risk under a specific set of conditions and for assessment to suit the many contexts in •• health guidance values (GVs) to be a certain time frame. which risk assessments are undertaken. estimated for environmental hazards Even limited measures of the level of risk that can be used and will adequately The scope of environmental health risk can be valuable for identifying complex protect public health, as a preface assessment (EHRA) can cover health cause-and-effect processes and the most to setting risk-based standards for impacts of: efficient means of addressing the risks. regulatory exposure limits as well as •• chemical pollutants and contaminants clean-up standards In this context, the methods used in in air, water, soil and food •• a comparison of the potential health EHRA are inherently conservative1 and •• pathogenic microbiological impacts of various environmental highly protective of public health. This contaminants in food and water health interventions (thus enabling is especially true of ‘screening’ type risk •• radiation sources cost-effectiveness estimates) assessments, which tend to use the •• electromagnetic fields (EMFs) •• risk-based policy making and most conservative assumptions about consistent, transparent appraisal and exposure and risk. These are generally •• climate and climate change recording of public health risks termed Tier 1 risk assessments. A conservative approach is also taken In all cases of the above impacts, priority •• questionable theories, methods and when the EHRA is used as a basis for is attached to evaluating the potential data to be challenged and addressed establishing environmental guidelines human health impacts. This update of by providing a clearly documented or standards. Conservatism is often enHealth guidance on EHRA focuses and open process (Covello & built into an EHRA by using exposure primarily on hazardous chemicals (and to Merkhofer 1993). estimates that represent ‘worst case’ or at a lesser extent, microbiological hazards). least the upper percentiles of parameter Risk assessment relating to radiation Risk assessment is significantly distributions, rather than mean, average hazards, EMFs and climate change are influenced by science policy or typical values. Furthermore, exposure covered elsewhere. considerations (see NRC 2008 for an outline of American EHRA policies). is usually considered to be constant over a substantial period of time (sometimes Risk assessment is intended ‘to provide Science policy on EHRA in Australia complete information to risk managers, is somewhat fragmented, with various specifically policymakers and regulators, Commonwealth and state or territory 1 In this context, ‘conservative’ is intended to imply a authorities applying risk assessment cautious approach to evaluating and managing the so that the best possible decisions uncertainties inherent in a risk assessment, which are made’ (Paustenbach 1989 p. 28). policies and default approaches, reduces the probability of harm occurring.
ENVIRONMENTAL HEALTH RISK ASSESSMENT 3 an entire lifetime), whereas many •• There is an emphasis on cancer risk to risk assessments will be needed for •• situations where vulnerable 1.3 1.3.2 Qualitative and quantitative environmental exposures are episodic, the possible neglect of other adverse products, processes, situations and populations may be affected by risk assessments and may decline over time due to loss or effects, for example, reproductive and activities where there is a plausible environmental health issues such as TYPES OF RISK degradation of the contaminant. developmental outcomes. case that there could be an increased the location of schools The level of risk can be described either qualitatively (i.e. by putting risks into •• In some situations, there may be risk of significant health consequences •• legislative or policy changes ASSESSMENT The conservatism in EHRA can for the human population from the categories such as ‘high’, ‘medium’ or insufficient scientific knowledge •• designating housing setbacks from sometimes lead to the development of product, process, situation or activity. 1.3.1 ‘low’) or quantitatively (with a numerical to be able to perform credible risk industry and transport corridors estimate). Practical guidance on how risk-based GVs that are so far below assessments. A risk assessment can also be used to Individual and population risk the capacity of contemporary analytical inform the selection of the safest option •• where health impact assessments are assessments to manage risks is the approach taken •• Risk assessment can be perceived techniques that compliance monitoring when making decisions about how to undertaken. in AS/NZS ISO 31000:2009 (Standards to be tailored to provide a desired or Risk assessments generally make becomes impossible or impractical. In achieve a particular aim. A screening Australia, 2009) and in the Risk analysis predetermined outcome (NRC 1994). risk estimates for defined groups or some cases, conservative risk-based GVs level comparative risk assessment could Risk assessment is inappropriate when framework used by the Office of the populations. The term ‘receptors’ is often may be driven to levels below background •• Excessive emphasis is given to the be used to compare the risks associated it is a ritual rather than a meaningful Gene Technology Regulator to manage used to designate people who may be concentrations, casting doubt on the process of risk assessment rather than with various options when, for example, process and should not be undertaken risks associated with genetically modified exposed to an environmental hazard, and credibility of the process. its content. formulating a particular product or when: organisms (GMOs) (OGTR 2009). (See to whom the EHRA would be directed. •• The risk assessment process can controlling pests. Sections 5.3, 17.6 and 17.7.) •• there is no data or an insufficient Identification of ‘receptor’ locations and It is important that assessors, users, become so ‘bogged down’ (NRC 2008) amount of data pathways by which they might be exposed regulators and members of the public that it takes far too long to achieve Examples are: Current risk assessment methods do not •• it is clear that the proposal, situation is an integral part of any EHRA. recognise risk assessment may not useful or timely outcomes. enable accurate quantitative estimates •• new additives to food or potable or or activity is seen by health and other always provide a compelling or definitive of risk for low levels of exposure to •• The risk assessment process is used recreational waters experts as having few potential risks to Individual risks are usually estimated environmental hazards. Numerical outcome. Some of the criticisms of risk as a ‘whitewash’ or used to justify the •• introduction of a new chemical health for a hypothetical person with assumed estimates of risk can be presented, but assessment are as follows: continuation or increase of polluting under the NICNAS (National •• risks may be likely, but the evidence is characteristics for various durations of caution must be exercised in assigning activities. •• Default values and assumptions Industrial Chemicals Notification and already well documented and it may exposure (e.g. per year or per lifetime) or strict meaning to the numbers: are not realistic – a series of such •• The efforts in risk assessment may be Assessment Scheme) program (see be possible to develop evidence-based for different locations. The hypothetical unrealistic values or assumptions inappropriately distributed in cases Section 17.2) recommendations without the need for individual is designed to represent the ... a number is a number is a number compounds the inaccuracy so that where enormous effort is spent on average person in the situation or the •• assessment of a contaminated site a comprehensive assessment ... and yet exactitude should not be risks may be seriously overstated complex modelling in cases where maximally exposed person. However, •• there is an inability to take action or it confused with accuracy. or understated if the default values some targeted data collection could •• assessment of a major planning such risk estimates cannot be targeted to is too late to take action (Langley 2003 p. 166) are too conservative or insufficiently provide much more relevant and development, especially where hazards a specific person. The distinction between conservative, respectively. credible evidence. are anticipated •• there are insufficient resources ‘there is a risk’ and ‘I am at risk’ is often Complexity of the exposure conditions, •• assessment of pollution impacts at difficult to explain to both the public and •• Interactions between agents (i.e. •• the proposal is clearly politically or variability in the environmental agents and Tal (1997) indicates that environmental existing facilities by regulators, especially when discussing mixtures of agents) and the variability socially unacceptable. exposed populations, and any inherent groups identify a number of problems very small probability estimates and this of response between individuals are •• changes to climate, landform, limitations in toxicological data may limit with the way risk assessments have been can lead to serious misunderstanding commonly unknown and may be geography or demography that Of relevance to risk assessment is the accuracy of numerical risk estimates. practised, including disempowerment among stakeholders about the meaning insufficiently taken into account. may impact on disease vectors and Bardwell’s reference (cited in Thornton While a degree of quantification may be and potential regulatory delays. Risk & Paulsen 1998 p. 799) to a study that of quantitative risk estimates (McAuley •• The use of default values and parasites possible for some components, such as assessments should be designed and indicates that ‘about 90 per cent of real & Hrudey 2006. In the case of a lottery, assumptions may become too rigid •• situations where environmental data collection and exposure assessment, undertaken in ways that minimise world problem solving is spent: a winner may be found, despite the so that situation-specific data is not standards or guidelines are unavailable it is important that all uncertainties are these pitfalls. small odds of winning, whereas in applied. •• solving the wrong problem; reflected in the EHRA outcomes. Further •• environmental changes that will most quantitative risk assessments the discussion of qualitative and quantitative •• The nature of the population to whom increase traffic flow and may increase •• stating the question so that it cannot probability of anyone being at risk is small risk assessment appears in Chapter 5. the risk assessment is to be applied 1.2 the risk of injury or air pollution, such be answered; and the probability of a specific individual regarding its exposure characterisation as new traffic corridors •• solving a solution; being at risk is very much smaller. or susceptibility is often poorly defined. WHEN TO UNDERTAKE •• changes where impacts on •• stating questions too generically; or •• The uncertainties of risk assessment RISK ASSESSMENT environmental health factors may be Population risk may relate to the number •• trying to get agreement on the answer are often inadequately described, for permanent and irreversible of adverse health effects (e.g. fatalities, before there is agreement on the example, specific point estimates are The issues identification phase (see cancers or illnesses) in a population over •• changes that may impact on the question’. given that do not recognise uncertainty, Chapter 2) will determine when to microbiological or chemical safety of a specified period of time or the rate of or simplistic upper-bound estimates of undertake a risk assessment. The food chains and food supplies adverse effects for a given location or sub- uncertainty are used. need to undertake a risk assessment population (Covello & Merkhofer 1993). •• situations where there is a high level will be influenced by situation-specific of public interest in or concern about factors. As such, the following list is environmental health issues indicative and not exhaustive. In general,
4 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 5 Risk assessors should generally strive to: Consultation with the community to •• effectiveness (e.g. usefulness of Further details of the US approach to 3. Dose–response assessment 1.4 identify their concerns is clearly an results; range of applicability across EHRA are outlined in Chapter 18. •• generate a credible, objective, realistic 4. Exposure assessment for the relevant important component of both risk different risks and problem areas; the THE DISTINCTION and scientifically balanced analysis population assessment and risk management. generalisability of the conclusion to Recent summaries of approaches to BETWEEN RISK •• present information on the separate See Chapter 6. other problem areas; and effectiveness EHRA in the Australian context are: 5. Risk characterisation. components of the risk assessment and efficiency of linkage with other ASSESSMENT AND RISK •• guidance on health risk assessment of •• explain the confidence in each Scientific judgements and policies must types of methods). Some of the key factors and questions contaminated sites in schedule B(4) of assessment by clearly delineating be clearly identified. Inevitable gaps that must be taken into consideration at MANAGEMENT the National Environmental Protection strengths, uncertainties and in knowledge will be filled by scientific each of these stages include the following: Measure (NEPM) – currently under Risk assessment is a process that informs assumptions, along with the impacts judgements and policies. These must 1.6 review and being updated (National the risk management process. Risk of these factors (e.g. confidence limits, be clearly identified so that others may 1. Issue identification RISK ASSESSMENT Environment Protection Council – assessors and risk managers should use of conservative/non-conservative understand the role of judgement in NEPC 2010) •• What are the true drivers for the issue be sensitive to the distinctions between assumptions) on the overall interpreting the evidence. MODELS AND FORMATS being assessed? (e.g. there is no risk assessment and risk management. assessment. •• guidance on ambient air quality point in doing a quantitative cancer The enHealth framework for EHRA A variety of models are used for risk standards-setting (National Health and risk assessment if the real concern is (see Figure 1) clearly differentiates The risk assessors should do this 1.5 assessment in Australia by government Medical Research Council – NHMRC cognitive impairment of children, and if risk assessment and management as without considering issues such as cost agencies and consultants. Many of 2006) – which includes a review of EVALUATING RISK the latter cannot be addressed by risk separate but interlinked processes, of remediation, feasibility or how the these models are based on paradigms health-based approaches to hazard assessment, then another approach with risk management following the scientific analysis might influence the ASSESSMENT METHODS for risk assessment first outlined by assessment of air pollutants. This led may be necessary). risk characterisation phase of a formal regulatory or site-specific decision (United the US National Academy of Sciences to a comprehensive Methodology for risk assessment. States Environment Protection Agency Criteria for evaluating risk assessment in 1983 in the seminal work Risk setting air quality standards in Australia •• Are intervention strategies available to – US EPA 1995a). However, it is likely methods (Covello & Merkhofer 1993) assessment in the federal government: (NEPC 2011), which includes detailed manage the outcomes of the EHRA The development of risk management that a more thorough EHRA process (e.g. include: managing the process (NRC 1983). information regarding risk assessment (e.g. containment of contaminated soil, plans is outlined in detail in AS/NZS ISO moving to a Tier 2 or Tier 3 analysis) may This document laid the foundation for in the specific context of air pollutants. chlorination of water, pasteurisation of •• the logical soundness of the method 31000:2009. The important elements provide the risk manager with a suite of contemporary risk assessment processes food)? (e.g. its justification based on of a risk management framework are options for managing the identified risks. (including quantitative methodologies) An extensive discussion of the different theoretical arguments or scientific •• Have transport mechanisms whether it: This should assist in determining the most and established different approaches to framework models for human health knowledge, and the validity of been adequately considered (e.g. cost-effective set of actions. assessing carcinogenic risks versus non- risk assessment in use in Canada and meteorological factors affecting air •• evaluates the external and internal the underlying methodological carcinogenic health effects. the US can be found in Jardine et pollution, vectors for communicable contexts of the organisation tasked assumptions) Risk assessment processes should be al. (2005). This review compares the diseases)? with implementing new or existing risk •• completeness (e.g. whether it can models, emphasising the ways in which coherent and transparent. It is important Through the 1980s, 1990s and 2000s •• Are there factors that could affect management plans or policies address all aspects of the problem the basic framework can be changed that the basis of the decision making these risk assessment paradigms were persistence (e.g. photolysis and •• provides for accountability and and the degree to which it excludes to accommodate different regulatory is clearly documented. This formal formalised (and continue to be updated) volatilisation of chemicals, desiccation transparency in the decision-making issues because they are hard to settings, and the extent to which different record should be clear, comprehensive in a number of US guidance documents, of micro-organisms)? process and concise, and include a summary accommodate) which include: models emphasise the importance of •• Has the risk assessment been initiated •• ensures that resources are made of the key data that has influenced •• precision and accuracy (e.g. reflected consultation with stakeholders and •• Risk assessment and management: as the result of a breakdown of available to measure and report on risk the risk assessment and an appraisal in the confidence level associated with socioeconomic analyses. framework for decision making (US public health measures (e.g. flooding and risk mitigation procedures of its quality (Advisory Committee on the results or the biases resulting from EPA 1984) affecting waste control and potable Dangerous Pathogens – ACDP 1996). undue weight being given to specific •• establishes internal and external 1.7 water treatment)? Further guidance on compiling an EHRA interests or considerations and the •• Science and decisions: advancing risk communication and reporting assessment (NRC 2008) mechanisms report is in Chapter 7. sensitivity of results to untested or THE FIVE STAGES OF untestable assumptions) •• Risk assessment guidance for 2. Hazard assessment •• ensures that there are audit processes Risk assessment information is only one superfund (‘RAGS’ documents), ENVIRONMENTAL HEALTH •• Have the severity and reversibility of appropriate to the evaluation of the risk •• acceptability (e.g. compatibility of several kinds of information used for with existing processes; whether it published successively from 1989, and RISK ASSESSMENT health effects been considered? management strategies which continue to be updated today decision making. The risk management is viewed as rational and fair; the •• Is there any interaction between the •• provides effective processes for decision will be determined not only by level of understanding for all parties •• Guidelines for carcinogen risk (EHRA) identified hazards and other agents in collecting feedback and information for the risk assessment but a range of other affected by it; and the confidence and assessment (US EPA 2005a) the environment? continuous improvement factors, including ‘technical feasibility familiarity of those who will use it) The historical development of formalised •• Exposure factors handbook (US EPA •• Is the onset of health effects immediate •• develops monitoring and review (e.g. treatability, detection limits), EHRA has resulted in the process being •• practicality (e.g. the level of expertise, 1997a), with an updated version or delayed? While health-based processes at the implementation stage economic, social, political,’ and legislation categorised into five distinct stages: time and input data required) released for public comment in guidelines generally assume long-term of all risk management plans and when determining whether to regulate October 2009. 1. Issue identification continuous exposure, and are usually strategies. and, if so, to what extent (US EPA based on chronic (preferably lifetime) 1995a p. 2). 2. Hazard identification
6 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 7 dosing in animal studies, there may 5. Risk characterisation this myth in the risk communication Figure 1: Environmental health risk assessment model be circumstances where data from an process by explaining its inaccuracy 1.8 •• Has genetic variability in the acute or short-term toxicity test may because of variability and uncertainty, one exposed population (or in the source RISK ASSESSMENT be more appropriate to use in the should not lose sight of the fact that an toxicological data) been adequately risk assessment (e.g. adverse effects exceedence of a standard or guideline or FRAMEWORKS accounted for? associated with irritancy). other indicator of ‘safety’ by a derived risk •• Are there individual host The framework model that •• Is there is a critical window of assessment number should always trigger characteristics (e.g. age, gender, encompasses the five stages of EHRA exposure? This is often associated further consideration of the situation body weight, pre-existing poor health, and their interlinkage with stages of with chemicals that modify foetal being assessed. Such consideration could immune status, nutritional status, risk management and stakeholder development, either during gestation include refinement of the assumptions, previous exposures or reproductive consultation was first proposed for or in the early postnatal period modelling or input values, and the status) that need to be considered? use in Australia in the 2002 enHealth when critical neural or organ magnitude of safety factors. document, which is revised in this system developmental processes •• Are there population characteristics document. Conceptually, the five stages are occurring. It is also likely that (e.g. herd immunity and social Part of the reason behind this false are closely linked and dependent on the epigenetic and hormonal disturbance behaviours for communicable perception is a lack of understanding preceding stages. mechanisms act mainly during critical diseases, social mobility for exposure of what the numbers generated in an exposure windows. to air and soil contaminants, EHRA really mean. This has possibly recreational patterns for exposure to been compounded by the dichotomy that The original model is illustrated in •• Has the carcinogenic and/or contaminated recreational waters) that has developed in the approach to cancer Figure 1. The terminology is similar genotoxic potential of the identified need to be considered? and non-cancer endpoints. The recent to terminologies used by other major hazards been addressed? models. •• Has the risk estimate been expressed review of EHRA methodology by the US quantitatively or qualitatively and, National Research Council (NRC 2008) 3. Dose–response Note that stakeholder consultation is if quantitative, is it a finite risk proposes harmonising approaches to considered essential at all stages of •• Is appropriate dose–response data estimate based on extrapolation of these two types of endpoints, including the EHRA process, and that a review/ available, and has the data been the dose–response relationship, or the use of BMD methodology to derive reality check should be built into the appropriately scaled in translation from is it an acceptable daily intake (ADI) a point of departure (POD) for risk critical stages of hazard and exposure animal to human? or tolerable daily intake (TDI), based estimation, and assigning a finite risk assessment, and risk characterisation, on application of safety/uncertainty/ estimate to both a cancer risk estimate •• Has the potency of the agent been to ensure the outcomes have not been modifying factors to a no observed and the calculation of a derived reference determined for both acute and chronic distorted by inappropriate choice of adverse effect level (NOAEL), lowest dose (RfD) (see Section 3.9). The danger dosing? data inputs. observed adverse effect level (LOAEL) in this approach is that it introduces •• Does a threshold or non-threshold or benchmark dose (BMD)? additional challenges in communicating model best describe the data? the meaning of small probability risks, Various revisions to this basic model of EHRA are set out in the NRC (2008) All of these issues will be addressed in where previously, assessments based 4. Exposure update of approaches to risk assessment. more depth in Chapters 2 to 16. The on a threshold approach were explained This NRC document outlines a more •• What is the duration, timing, frequency terminology ADI, TDI, NOAEL, LOAEL without recourse to citing finite risk holistic framework (see Figure 2), which and consistency of exposure? and BMD is explained in more depth numbers. The NRC recommendations emphasises the importance of problem in the Glossary and their derivation in have also been criticised (Goldstein 2010) •• Are exposures continuous, intermittent formulation and planning as precursors Section 5.6. on the basis that EHRA methodology has or episodic, or do they show clear worked well and is not in need of any to the formal steps of quantitative risk patterns? assessment. It further reinforces the view One of the concerns about some such ‘improvements’. •• Are there are relevant past, current or that the outcomes of the EHRA process stakeholder perceptions of current future exposure patterns to consider? are critical in better informing the risk EHRA methodologies is that an At the present time, where EHRA is management stages, and providing for •• Have all exposure routes (ingestion, impression may be given that the derived practised in Australia, assigning an ADI stakeholder consultation and review to inhalation, dermal) have been risk assessment number, whether or TDI carries no such implication that it occur at all stages. considered? based on extrapolation or an ADI/TDI is associated with any finite level of risk. •• Are exposures intergenerational approach, can be taken as a ‘bright The operating definition is that an ADI or While the model described in Figure 1 or cumulative, or should they be line between possible harm and safety’ TDI represents ‘an estimate of the intake has served Australia well over the past aggregated? (NRC 2008 p. 8) or, in other words, the of a chemical which, during a lifetime decade, it is recommended that the more separation between safe and unsafe of exposure, appears to be without holistic model (Figure 2) provides a more exposures. While it is important to dispel appreciable risk, on the basis of all facts known at the time’. structured and informative framework for EHRA in Australia from now on.
8 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 9 Figure 2: A revised outline of the interlinked processes of EHRA The relationship between the expanded are an expanded version of the fifth stage Various elements of the framework can framework and the framework first of the 2002 framework. All phases of the be expanded to illustrate the critical described in 2002 is that Phase I and new model have more description of the individual components. the planning stage of Phase II are aligned key issues that need to be addressed at with the former ‘issue identification’ each stage, although it is likely that in For example, Figure 3 illustrates the stage, the Phase II ‘Planning and conduct some risk assessments, answers to some multiplicity of exposure sources, exposure of risk assessment’ elements (hazard questions will be obvious, while others pathways, receptors, endpoints and identification, dose–response assessment may need a full and detailed approach. measurements (metrics), which could be and risk characterisation) are aligned included in conducting an EHRA. with comparably named elements of As in the former framework, stakeholder the former framework, while the engagement at all phases is emphasised Phase III ‘Risk management’ elements as a critical element.
Figure 3: Expanded illustration of the major exposure pathways, potentially exposed groups leading to potential health outcomes
Adapted from: NRC 2008.
Solid lines indicate pathways usually considered. Other pathways shown may not be considered in conventional EHRAs. Adapted from: NRC 2008.
10 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 11 which must be substituted for knowledge In the screening step, usually the There is a danger that the tiered Figure 5: Conceptual representation of the proposed IPCS risk assessment framework 1.9 in the risk assessment process may be maximum concentration of each structure common to some regulatory TIERED APPROACHES reduced (NEPC 2010). chemical in the full list of chemicals risk assessments can lead to a negative or other agents that might pose a risk impression that the only possible TO EHRA A risk assessment progresses from at the site are compared to relevant outcome contemplated by going to a Tier 1 to Tier 2 when the less-refined risk national or international guidelines. higher tier is a relaxation of remediation Because of the cost and complexity estimates at Tier 1 may be unacceptable, This is conservative as the maximum requirements rather than an unbiased of contemporary formal EHRAs, and further assessment is needed. concentration is presumed to be present approach allowing for more evidence circumstances may suggest a tiered Progression from Tier 2 to Tier 3 is driven at all times in all situations for this step. leading to more onerous remediation approach to formulating a site- or issue- by potentially unacceptable risks at Tier 2. If a chemical or other agent is found to requirements. This perception needs to specific EHRAs. The simplest approach Tier 3 provides more detailed and specific exceed the guideline value then that be offset by clearly explaining that the (Tier 1) would be an initial screening-type focus on risk-driving factors. chemical is classified as a chemical only purpose of going to a higher tier evaluation of risks using conservative of potential concern and a detailed is to reduce uncertainty by including default exposure parameter estimates and In Australia, there is often no clear assessment should be triggered. more realistic estimates of exposure. comparison with published health-based break between the different tiers. The Going to a higher tier may also aid risk guidelines. Tier 2 and Tier 3 processes investigations and risk assessment In the detailed assessment step, the management by better targeting the risk would involve collecting additional proceed until the level of information chemicals are assessed more fully, and management options. exposure data and a more detailed is appropriate for the decision making this may include exposure scenario analysis of dose–response data, possibly required. It is common for most risk modelling, fate and transport modelling or Conservative exposure settings and including calculation of target tissue assessments, regardless of which tier, further investigations to better understand assumptions (as in Tier 1 assessments) doses or translating animal doses into to have a screening step and a detailed the situation under investigation and to may not be realistic for the site under human-equivalent dose estimates. assessment step. refine the assumptions to make them consideration as they are based on more realistic. Such assessments usually generic assumptions and parameters The tiered approach in risk assessment is include consideration of the maximum that are not likely to be realistic. A common in many jurisdictions, although case and an average case. Tier 2 assessment may be used to the number of tiers and their precise produce more appropriate values by usage may differ. For example, Health amending the assumptions to reflect Canada uses the term Preliminary actual site conditions. Where available, Figure 4: Elements of a tiered approach to EHRA Quantitative Risk Assessment (PQRA) to data on biodegradation of contaminants refer to what would otherwise be called and bioavailability of chemicals should be Reproduced from IPCS 2009b with permission from WHO. Tier 1, while the terms Tier 2 and 3 are considered (see Section 4.2.1). Exposure allocated to site-specific risk assessment factors (and assumptions) should reflect Note that the IPCS framework includes is that it is simple, easily understood, and (SSRA). the scenarios under consideration. an additional tier (Tier 0) that is not therefore widely applied in EHRA practice. included in the Australian guidance. In However, an inadvertent consequence Figure 4 is a schematic depiction of some The tiered approach is expanded and the IPCS context, Tier 0 would encompass is that many of the point estimates used of the elements that might comprise Tier 1 outlined in more detail in the International initial crude exposure estimates that are are chosen at the upper end of their likely to Tier 3 EHRAs. Programme on Chemical Safety less well defined than those used in an ranges. This can lead to compounding (IPCS) framework for risk assessment Australian Tier 1 assessment. of the conservatism in a model, and Tiered risk assessment is particularly of combined exposures to multiple consequently in the EHRA outcome. relevant to the EHRA of contaminated chemicals (IPCS 2009b). A flow chart Sensitivity and uncertainty analysis are sites, and is discussed in more depth describing the IPCS framework 1.10 used to overcome this disadvantage, and in schedules B(4) and B(7) of the (Figure 5) emphasises integration of DETERMINISTIC provide increased understanding and contaminated sites NEPM (NEPC 2010). information on mode of action (MoA) clarity on which values are risk-driving; The tiered approach allows the problem and gradual refinement of exposure VERSUS PROBABILISTIC this is itself a useful part of the risk under consideration to be assessed at assessment throughout the tiers. It assessment (NEPC 2010). an appropriate level of complexity. The applies a separate tiered approach to ESTIMATES IN EHRA degree of health protection achieved both hazard and exposure assessment. Probabilistic techniques can overcome the A deterministic approach means that is equal at each tier. As the amount of It also notes that the hazard assessment potential for compounding conservatism, input values in an exposure model are data and assessment detail increases component could be based on individual and may provide for a better descriptor expressed as single values or point and the conceptual understanding of components or incorporate dose-additive for the uncertainty associated with the estimates. These are intended to be site conditions (i.e. the conceptual site approaches for mixtures (see Chapter 12 various input parameters, and also ‘best estimates’ of the value of the input model) is refined, the level of uncertainty for discussion of toxicological assessment provide estimates of the statistical limits Adapted from: NEPC 2010. variables. The advantage of this approach decreases. In turn, the amount of caution of chemical mixtures). of underlying parameter distributions.
12 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 13 This can be useful for the risk manager ... a combination of procedures or •• Applying the precautionary principle •• It provides for taking appropriate the fact that Australian agencies use a to decide on the extent to which ‘outliers’ methods by which a policy, program 1.12 is essentially a matter of making actions in the face of uncertainty or precautionary approach to managing may influence the EHRA process, and or project may be judged as to the ENVIRONMENTAL HEALTH assumptions about consequences and lack of complete knowledge of the uncertainty in the risk assessment and provide a basis for deciding the limits effects it may have on the health of likelihoods, and then using standard potential risks. management of hazardous chemicals. applicable to protecting the extremes of a population. RISK ASSESSMENT AND procedures of risk assessment and •• There needs to be consideration of the Advocacy to adopt a precautionary the population distributions to which the THE ‘PRECAUTIONARY management to inform decisions on economic and practical feasibility of approach to EHRA is reflected throughout EHRA applies. In other words, HIA is a systematic how to address the hazard or threat. implementing the risk management this enHealth document. process to assess the actual or potential, PRINCIPLE’ •• Invoking the precautionary principle measures that may be suggested by Probabilistic risk assessment and direct or indirect, effects on shifts the burden of proof in invoking the precautionary principle. methodologies have been reviewed by the health of individuals, groups or The precautionary principle (PP) demonstrating presence of risk or •• Application of the precautionary Bogen et al. (2009). These methods communities arising from environmental was first formalised in 1984 at the degree of safety towards the hazard principle is closely intertwined with can be used to assess and manage conditions or hazards arising from First International Conference on the creator. The presumption is that the social equity issues. Protection of uncertainty, inter-individual heterogeneity policies, objectives, programs, plans or Protection of the North Sea and has since hazard creator should provide, as a more vulnerable, socioeconomically and other sources of variability. activities. It looks at both potential health been integrated into several international minimum, the information needed for disadvantaged or ethnic segments benefits and health impacts from an conventions and agreements. The decision making. of the community who may be at Monte Carlo analysis is one probabilistic activity or situation. It is usually a process UN Conference on Environment and •• Decisions reached by invoking and greater risk from environmental tool that has been promoted for use in undertaken as part of an environmental Development (UNCED – 1992 Rio applying the precautionary principle hazards is an important element of EHRA because it replaces deterministic impact assessment for a significant declaration from Agenda 21) interpreted should be actively reviewed, and risk management. estimates of individual parameter inputs project and looks at both positive and a precautionary approach to chemicals revisited when further information with probability distribution functions negative impacts on health. HIA is management in the following statement: that reduces uncertainty becomes In EHRA, consideration of the describing the variability of those input generally undertaken in the early stages available.’ precautionary principle is particularly parameters. The probabilistic exposure of project planning in order to predict ‘In order to protect the environment, relevant during the risk management model can be run through thousands of and facilitate avoidance of potentially the precautionary approach shall be In Australia, the 1992 Inter Governmental stage. Risk assessment provides a iterations, with values for each parameter negative health impacts, to promote widely applied by the States according Agreement on the Environment (IGAE) process for applying the precautionary selected randomly on the basis of their more positive health impacts and to to their capabilities. Where there are
14 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 15 Chapter 2: Problem formulation and scope
The purpose of this stage of EHRA is to Examples of issues are: •• biological monitoring (e.g. of children’s •• the hazard may have single or multiple risk assessment. The methodologies for formulate the problems to be considered blood lead levels or Ross River virus sources (e.g. atrazine contamination ‘aggregate’ risk assessment are set out in 2.2 •• community concerns over emissions by the risk assessment and to clarify the antibody levels) of a drinking-water supply from a Chapter 4. from a smelter or other industrial IDENTIFYING AND proposed scope. It corresponds with chemical spill versus particulates in facility •• disease surveillance (e.g. of salmonella Phase I of the framework outlined in types for food poisoning, skin cancer ambient air arising from diesel engines, In the case of ‘cumulative’ exposure, the DESCRIBING SUSCEPTIBLE •• community outrage over the proposed Figure 2 and with ‘issue identification’, the rates, pregnancy outcomes) wood stoves and environmental methodologies are more complicated, POPULATIONS first stage of the original risk assessment development of a communications tobacco smoke) particularly since cumulative exposure •• health monitoring (e.g. of lung framework depicted in Figure 1. tower risk assessments are intended to address function testing to detect the onset of •• there are concerns about a range of Within the general population there •• how contaminated sites are managed the interactions of multiple agents or environmentally caused asthma) potential health effects from various may be sub-groups potentially Essentially, this means addressing the •• development of new standards for hazards stressors, not all of which are necessarily more susceptible to the effects of following points: •• epidemiological studies (e.g. of chemical agents. It could include water quality, including use of a new •• there is variable and often superficial environmental chemicals than others particular disease rates in certain biological or physical agents that could •• What is the concern? water treatment chemical or new uses information on exposure and the level in the population. Human variability populations such as workers) to modify the toxicity of the environmental for recycled water of health problem (also called intra-species, i.e. inter- •• Why is it a concern? identify previously unknown hazards chemicals under consideration. •• changes to a food standard that permit individual, variability) may arise through •• How urgent is the concern? •• information about analogous hazards. •• there is a context of public anxiety, higher levels of exposure or introduce toxicokinetic or toxicodynamic variability anger and impatience Further guidance on assessing multiple •• How do stakeholders perceive the new chemicals into the food chain. (Dybing & Soderlund 1999). Both concern? 2.1.1 •• different stakeholders may have chemical exposures (i.e. mixtures of these areas of variability may be due to different perceptions of the issues chemicals) is outlined in Chapter 12. ‘Hazards’ relate to the capacity of a Phases of issue identification acquired and/or inherent factors that for example, a stakeholder group This will include identifying and specific agent to produce a particular type may make a person more susceptible to Issue identification comprises several comprised of workers at a smelter who describing: of adverse health or environmental effect. 2.1.2 environmental pollution. phases: are also nearby residents may have The environmental agents of concern may Identifying chemicals of potential •• issues associated with existing complex perceptions 1. Identification of environmental health concern Sensitivity of individuals is also likely to environmental conditions include physical, chemical, biological or issues (or an individual issue) and •• the hazards may be compared with be affected by age, sex, nutritional and social factors. It is quite likely that the issue and/ •• susceptible and/or vulnerable determining whether there are hazards other environmental hazards affecting pregnancy status, and combinations or hazard identification stages will find populations likely to be exposed •• Physical factors include heat, cold, amenable to risk assessment – this the community; this component of these (IEH 1999c). It is therefore a large number of chemicals whose •• potential exposure pathways noise, mechanical hazards, solar will involve demarcating ‘hazards’ from of the appraisal will be affected by imperative that the issue identification presence in the environment and toxicity radiation, ionising radiation (e.g. ‘issues’ and may require environmental objective data (e.g. of different disease stage considers whether any of these •• potential management options that may give rise to adverse health outcomes. X-rays) and non-ionising radiation (e.g. sampling. rates) and subjective perceptions factors could influence the outcome of may mitigate exposure The issue then becomes which of them microwaves), noise and vibration. by the stakeholders (Presidential/ the EHRA. •• the risk and other technical 2. Putting the hazards into their must be designated as chemicals of •• Chemical factors include synthetic and Congressional Commission on Risk assessments necessary to evaluate risk environmental health context potential concern (COPC) (which must naturally occurring substances. Assessment and Risk Management – A distinction needs to be drawn between and discriminate between potential risk (clarification and prioritising of be addressed in the formal EHRA) or P/CCRARM 1997). the use of the terms ‘susceptibility’, management options. •• Biological factors include viruses, problems and hazards). whether any of them can be readily prions, bacteria, parasites and vermin. eliminated from further consideration. ‘vulnerability’ and ‘sensitive groups’ 3. Identifying all the chemicals of In relation to assessment of multiple •• Social factors include poverty, within a population of ‘receptors’ under potential concern (COPC) (i.e. exposure routes and sources, US 2.1 unemployment cultural values and consideration in an EHRA. prioritising those chemicals that need regulations define two types of exposure The tiered risk assessment screening effects on access or amenity. process (see Section 1.9) may enable IDENTIFYING AND to be fully considered in a quantitative that need to be considered in a risk Susceptibility: refers to intrinsic biological risk assessment) assessment: COPC to be discriminated from among DESCRIBING ISSUES Examples of hazards include the a much larger number of environmental factors that can increase the health risk capacity of: 4. identification of potential interactions •• aggregate exposure: the analysis of contaminants, and may point to the need of an individual at a given exposure level. WITHIN EXISTING between agents. exposure to a chemical by multiple for a more advanced (Tier 2 or 3) risk Examples of susceptibility factors include: •• benzene to cause leukaemia pathways and routes of exposure assessment. genetic factors, late age and early life, and ENVIRONMENTAL •• solar radiation to cause skin cancer prior or existing disease. At this stage it often becomes apparent •• cumulative exposure: the combined CONDITIONS •• salmonella to cause vomiting and that the setting for the risk assessment is risk estimate where exposure occurs It may also be possible to establish Vulnerability: refers to human populations diarrhoea. a situation where: simultaneously or consecutively to that chemical concentrations in the ‘Hazards’ need to be distinguished from environment are so low that exposures at higher risk due to environmental •• there are multiple, interacting hazards multiple chemicals that exert toxicity ‘issues’. The determination of the issues are unlikely to exceed a generic threshold factors. Examples of vulnerability factors Hazardous agents may be identified from rather than an isolated hazard – through a common mechanism. is necessary to establish a context for of toxicological concern (TTC) and can include poverty, malnutrition, poor the range of data sources, including: perhaps the contaminant affects the risk assessment, and assists the therefore be discounted. The application sanitation, climate change and stress •• environmental monitoring (e.g. of food, multiple environmental media (e.g. In the case of ‘aggregate’ exposure, process of risk management. Issues of a TTC approach to risk assessment is associated with mental health diseases. air, water and soil) lead smelter emissions contaminating the requirement is no more than would have dimensions related to perceptions, be normally done in a conventional discussed in Section 5.13. soil, air, water and food) Sensitive groups: refers to populations science, economics and social factors. •• emissions inventories (e.g. the National EHRA, where all potential exposure with both susceptibility and vulnerability Pollutant Inventory) pathways should be considered in the factors.
16 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 17 2.2.1 some cases this is proportional to the used. Particularly, each assessment •• disposition of the agent within the Special consideration has been advocated Cognitive impairment is common in Epidemiological principles degree of disease severity) and extent of should consider whether early childhood body, for example, transit time, pH and by the US EPA for assessing risks the very old or those with age-related exposure. In many situations, acquired exposure to carcinogens is relevant for the enzyme activity in the gut are different associated with early-life exposure to pathology (e.g. Alzheimer’s) and affects A sensitive sub-population is one where susceptibility (e.g. illness or old age) shifts site or activity being investigated and, if for children as are tissue–chemical mutagenic carcinogens (see Section their abilities to recognise, interpret and an adverse response to an environmental a person towards the ‘sensitive’ tail of the so, whether it is covered in the guideline bindings 5.8). US legislation mandates the react to acute and chronic environmental pollutant occurs at concentrations population dose–response curve for the proposed for use. application of an additional 10x safety/ hazards. They are higher consumers of substantially lower than that affecting the •• liver function related to detoxification pollutant. These individuals nonetheless matures after birth, as does the renal uncertainty factor in the derivation of an pharmaceuticals and there is a potential majority of the population. Another sub- remain part of the continuum of the 2.2.2 RfD for pesticides where studies indicate interaction with these pharmaceuticals population that may be considered to be excretion of foreign compounds overall population dose–response and Risk assessment and children developmental neurotoxicity or other toxic and other agents. ‘sensitive’ is one where the consequences experience effects similar to others but •• differences in gut microflora effects that could be associated with of exposure are more significant than Children may differ from adults in a at lower exposures to pollutants, or more •• the immaturity of children’s immune early-life susceptibility. 2.2.4 in the majority of the population. For intense effects at equivalent exposures range of behavioural and physiological systems example, children may be considered Risk assessment and gender (NHMRC 2006). parameters that may need to be taken •• differences in the clearance of While Australian environmental health a sensitive population because any into account in the risk characterisation chemicals – the higher clearance of authorities have not enunciated specific Gender differences may need to irreversible adverse effects may influence phase of risk assessments. Genetic variability can make an important certain chemicals from the body in policies relating to applying these US be taken into consideration when their health throughout their life. The contribution to human variability, such children compensates in part for the early-life risk assessment strategies, identifying potential exposure pathways elderly, especially those with specific as in the form of polymorphic genes The principal factors causing these greater sensitivity for their developing additional precaution tends to be in the exposure assessment phase and comorbid effects such as cardiac or for metabolism or tissue repair from potential differences are: organ systems (Renwick 1999) but applied on a case-by-case basis when characterising potential adverse health respiratory failure, may also constitute a toxic insult. Although it has long been for some other chemicals, clearance justified by relevant data. While the US effects in the risk characterisation phase sensitive group because the secondary •• growth, development and maturational recognised that genetic polymorphism rates may be lower early-life risk assessment policies are of the risk assessment process. consequences (e.g. pneumonia, plays an important role in driving the not automatically adopted in Australia, •• children’s greater potential future •• exposure factors – the surface area to worsening cardiac failure) may be more variability in xenobiotic metabolism, and they have been incorporated into the There are anthropometric (e.g. height, durations of life, which is relevant body mass ratio will change markedly serious than in the remainder of the genetic polymorphisms have been used development of health investigation levels weight, body surface area) and body to the potential for accumulation or with ageing. In the newborn, the ratio population (NHMRC 2006). as biomarkers of potential effect (Scherer (HILs) for benzo(a)pyrene (BaP) in the composition differences (e.g. fat exceeding latency periods is typically 0.067 (m2/kg) decreasing to 2005), this awareness has typically not revision of the contaminated sites NEPM content, muscle mass) between males 0.025 in an adult. While the respiratory The identification of sensitive sub- translated into quantitative use of the data •• dietary differences – children can eat (see Section 16.1) and they should be and females that may affect exposure volume remains fairly constant at populations may be guided by: in risk assessment or standard-setting much greater quantities of particular considered where there is good evidence concentrations of agents from different 10 ml/kg/breath, the surface area of (Haber et al. 2002; US EPA 2002a). foods (particularly dairy products, soft that such an approach is relevant. pathways. These differences may also •• clinical history (e.g. the presence of the alveoli increases from 3 m2 in an This is likely due to data gaps in our drinks and some fruit and vegetables) influence the absorption, distribution, diseases such as asthma, cardiac infant to approximately 75 m2 in an understanding such as: than adults on a body weight basis metabolism and elimination of xenobiotics failure, chronic bronchitis or cystic adult and the respiration rate drops 2.2.3 (Rees 1999) and have a significant influence on toxicity fibrosis, which may exacerbate •• prevalence of polymorphism from 40 breaths per minute to 15 Risk assessment and older persons •• placental transfer of contaminants and (Silvaggio & Mattison 1994). Some of the sensitivity to environmental pollutants) breaths per minute (Snodgrass 1992). •• lack of a defined link between genetic accumulation in breast milk can result For the ageing, there is a decrease of factors that influence these processes are •• clinical evidence of hyper- polymorphism and an adverse effect in exposures which are unique to the functional reserve in the physiological summarised in Tables 1–4. responsiveness (e.g. using A general discussion of the issues relating •• extent of induction/inhibition through prenatal and postnatal states and psychological systems. Distribution of methacholine or more specific to risk assessment for children may be co-exposure with other substances, chemical agents is affected by changes Men and women differ in many lifestyle challenge tests to assess susceptibility •• behavioural factors, for example, found in Calder (2008), Roberts (1992) lifestyle or diet in body composition with age: body fat and occupational exposure factors to irritant air pollutants) children’s play activities may put and US EPA (2005b; 2006a), with a more increases and body water decreases with •• relative contribution of multiple enzyme into more frequent contact with (e.g. alcohol drinking and cigarette •• demographic factors (e.g. the elderly extensive review of the physiological, age. The clearance of renally eliminated systems smoking) dietary patterns and how they soil contaminants and they are also pharmacokinetic, behavioural, genetic compounds is reduced because of or very young) spend their time. These factors may •• allelic frequencies for major ethnic more likely to indulge in soil eating and exposure factors that may alter the changes in renal function. Liver function •• genetic factors (e.g. cystic fibrosis). influence the exposure and effect of an groups behaviours (pica) sensitivity of children to environmental can be reduced in the elderly affecting agent on the individual. •• available parameters for toxicity From a physiological standpoint, •• large numbers of low-frequency alleles hazards (Hines et al. 2010). biotransformation of chemical agents. assessment, for example, techniques Increased sensitivity to the central any person who has decreased •• absence of chemical-specific For many chemical toxicants there are for assessing dizziness, intelligence The potential impact of these differences nervous system in the ageing population functional reserve in an organ system phenotype data. important differences between males and hearing impairment are different highlights the need for agent-by- from many drugs has been reported is theoretically less able to cope with and females in experimental studies. between children and adults agent appraisal. However, it should be (Crome 1999). Changes will occur to the additional environmental stressors, be Guideline values used in risk assessments Calabrese (1985) identified 200 toxicants recognised that the derivation of some immunological system often resulting in they ‘non-chemical’ or ‘chemical’ in should normally have been developed •• biochemical and physiological where toxicological analysis of animal types of toxicological reference doses reduced immunocompetence. nature. As with other areas of medicine in such a way that most sensitive sub- responses, for example, children studies suggest there are important (e.g. ADI or TDI) envisage whole-of-life and toxicology, whether a particular populations are protected. Some older have a higher metabolic rate, differences between males and females in more limited ability to control body exposure and may therefore be relevant Ageing populations are very heterogeneous individual will respond adversely to a guideline values do not specifically the expression of toxicity. certain environmental stressor depends address this issue. Risk assessors should temperature, more rapid growth rate for assessing risks associated with early- in terms of their general health. For those upon the relative balance between the check that sensitive sub-populations and a higher percentage of water in life exposure stages. with impaired health, there may be a variety extent of physiological compromise (in are covered in the guidelines being the lean body tissue of conditions present.
18 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 19 Table 1: Factors influencing the absorption of chemicals There have been reports of differences Access of an agent to the foetus is lifestyle factors should be clearly identified when comparing men and non-pregnant determined by its lipophilicity, molecular in risk assessments. Specific lifestyle Parameter Physiological difference Effects on toxicokinetics women in their response to toxic levels weight or ionic nature. Generally the factors that may have an effect on risk Gastric juice pH M < F < pregnant F Absorption of acids/bases modified by change in pH of lead, beryllium and benzene. Gender more lipophilic a chemical is the more assessment include: differences have also been reported to likely it is to cross the placental barrier. Gastric juice flow M > F > pregnant F Absorption modified by decreasing flow •• tobacco smoking occur from exposure to ionising radiation, For large molecules like polymers, size Intestinal motility M > F > pregnant F Absorption increases with decreasing motility noise and vibration, and extreme generally prevents their passage across •• diet Gastric emptying M > F > pregnant F Absorption and gastric metabolism increase with decreasing gastric emptying temperature changes (i.e. heat and cold the placental barrier. Most teratogenic •• hobbies effects are also dose-related; that is, stress) (Hunt 1982). •• recreational drug use Dermal hydration Pregnant F > M, F Altered absorption in pregnant F the larger the dose, the more likely and •• excessive use of prescribed Dermal thickness M > F Absorption decreases with increasing dermal thickness 2.2.5 severe the effect. High-dose exposures to polychlorinated biphenyls (PCBs) have medications. Body surface area M > pregnant F > F Absorption increases with increasing body surface area Risk assessment and reproductive been associated with foetal abnormality. Skin blood flow Pregnant F > M, F Absorption increases with increasing skin blood flow status Tobacco smoking will affect the exposure assessment component of the risk Pulmonary function M > pregnant F > F Pulmonary exposure increases with increasing minute volume The human reproductive system is Timing of exposure is particularly susceptible to environmental factors important. The critical period for assessment process because there will Cardiac output M > pregnant F > F Absorption increases with increasing cardiac output that can produce a variety of adverse organogenesis is in the first trimester be an increase in background exposure effects during the production of ova (between days 18 and 55 of gestation). to substances found in smoke, such Table 2: Factors influencing the distribution of chemicals in the body (oocytogenesis) and viable sperm This is the time of greatest cell as cadmium, cyanide and polycyclic (spermatogenesis) on fertilisation, on differentiation, and environmental aromatic hydrocarbons (PAHs). Parameter Physiological difference Effects on toxicokinetics implantation within the uterus, and agents may have a profound effect on Plasma volume Pregnant F > M > F Concentration increases with increasing volume growth and development of the embryo development at this stage. Tobacco smoking also affects the toxicity and foetus. assessment component. Maternal Total body water M > pregnant F > F Concentration decreases with increasing body water The extent of the toxicity effect will also cigarette smoking and passive smoking Plasma proteins M, F > pregnant F Concentration fluctuates with changes in plasma proteins and protein binding Reproductive status is also influenced by depend on the genetically determined have been associated with respiratory detoxification mechanisms (i.e. enzyme illness, acute toxicity and cardiotoxicity Body fat Pregnant F > F > M Body burden of lipid-soluble chemicals increases with increasing body fat the extent of exposure and adverse effects from occupational and environmental systems) of individuals. among newborns. Furthermore Cardiac output M > pregnant F > F Distribution rate increases with increasing cardiac output agents. Teratogenesis (abnormal epidemiological studies have shown development of the embryo and foetus) is Exposure of environmental or evidence of synergistic interaction Table 3: Factors influencing the rate of metabolism of chemicals a risk for the foetus that may be exposed occupational agents can also occur at between human carcinogens and long- to environmental agents. The principal the postnatal stage. The production of term cigarette smoking. The best studied milk during nursing and breastfeeding Parameter Physiological difference Effects on toxicokinetics factors that determine an agent’s risk interactions have included joint exposure is one pathway for the excretion of of teratogenicity and which need to be to tobacco and radon and tobacco and Hepatic metabolism Higher BMR in M, fluctuating hepatic Metabolism generally increases with BMR contaminants such as lead, mercury, metabolism in pregnant F considered in risk assessment include asbestos. Results from epidemiological PCBs and organochlorine pesticides (Goldfrank et al. 1990): studies of joint exposure to radon and Extra-hepatic metabolism Metabolism by foetus/placenta Metabolism fluctuates (e.g. DDT) stored in other body tissues. cigarette smoke have shown an additive Plasma proteins Decreased in pregnant F Elimination fluctuates with changes in plasma proteins and protein binding •• the nature of the agent Kinetic processes, such as absorption, or possibly a multiplicative increase in •• access of the agent to the foetus distribution and elimination, will influence the number of cancers induced and a BMR = Basal metabolic rate. the passage of agents into breast •• the onset and duration of exposure synergistic decrease in the latency period milk. Milk has a high fat and protein for tumour induction. Epidemiological •• the level and duration of dosage Table 4: Factors influencing the elimination of chemicals from the body concentration and lipid-soluble or protein- studies have shown that asbestos and •• the genetic constitution of the foetus bound contaminants pass readily to milk tobacco administered together can and are dissolved in or bound to the milk Parameter Physiological difference Effects on toxicokinetics •• the timing of the exposure in relation to produce an increased incidence in fat and protein (Hunt 1982). Renal blood flow, GFR Pregnant F > M > F Renal elimination increases with increasing GFR the stage of foetal development. lung cancer that is greater than from the administration of either agent alone Pulmonary function M > pregnant F > F Pulmonary elimination increases with increasing minute volume Substances that inhibit mitosis (e.g. 2.2.6 and the interaction is considered to Plasma proteins Decreased in pregnant F Elimination fluctuates with changes in plasma protein and protein binding antineoplastic agents such as vincristine Risk assessment and lifestyle factors be multiplicative by most investigators and vinblastine) are also of particular (NRC 1994). GFR = glomerular filtration rate Lifestyle factors may have an impact risk to pregnant women and exposure to (Tables 1–4 adapted from Government/Research Councils Initiative on Risk Assessment and Toxicology, 1999) on individual risk assessments and such agents may lead to teratogenicity population risk assessments if the activity Diet will also influence the stages of the and embryotoxicity. The female foetus is is widespread. For this reason, where risk assessment process, particularly sensitive to toxic chemicals or to other the influence of these factors can be the toxicity and exposure assessment agents affecting gametogenesis, which in distinguished, the potential influence of stages. Interactions between toxic humans finishes by the seventh month. metals and essential metals from the
20 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 21 diet have been known to affect the risk a reduced exposure to zinc. Individuals the risk assessment proceeds and new There are several potential types of risk assessment will be inappropriate of toxicity. Absorption of toxic metals who consume barbecued foods can be information and perspectives emerge. interaction between hazardous agents: 2.5 because the solutions to the problem will from the lung and gastrointestinal tract exposed to relatively large amounts of not be based on addressing risk but on •• additive, where the combined effect IDENTIFYING POTENTIAL may be influenced by the presence of PAHs from the pyrolysis of fats and other The issue identification processes addressing other factors such as social of two or more agents is equal to the an essential metal or trace element if the food components during the cooking relating to exposure may be aided by the MANAGEMENT OPTIONS and political concerns. sum of the individual effects toxic metal shares the same homeostatic process. Populations (e.g. general development of appropriate conceptual (e.g. 2 + 3 = 5) – an example THAT MAY MITIGATE mechanism. Examples are lead and population and fishermen) who consume site models (CSMs), which delineate Risk management also needs to is cholinesterase inhibition from calcium, and cadmium and iron. Other seafood may be exposed to heavy metals the exposure sources and potential be understood within the inevitable simultaneous exposure to two EXPOSURE dietary interactions include an inverse such as mercury in fish and zinc in pathways leading to human exposures. constraints that it will operate when risks organophosphorus insecticides relationship between protein content of shellfish (e.g. oysters). Identification of exposure pathways using While risk management should be are found to be small. Risk management the diet and cadmium and lead toxicity. CSMs and the modelling and quantitative •• synergistic, where the combined considered to be a process separate from inevitably involves trying to steer a Vitamin C in the diet also reduces lead The exposure to a hazard may also be description of them are discussed in more effect of two or more agents is much or dissociated from risk assessment, a sensible course between making a Type 1 and cadmium absorption. influenced by lifestyle and hobbies. detail in Chapters 4 and 13. greater than the sum of the individual formal EHRA is likely to identify those (false positive) or Type 2 (false negative) For example, the amount of time effects (e.g. 2 + 2 = 20) – examples hazards and exposure pathways that error (Hrudey & Leiss 2003). In other Different types of food will have different spent indoors (e.g. in the home, work Where multiple exposure pathways are risk of lung cancer from asbestos make the greatest contributions to the words, risk management for small and amounts of agents and hence cause environment/office, factory), outdoors may include background exposure not and smoking and the hepatotoxicity of overall risk. This may include assessment uncertain risks involves trying to decide a range of toxic effects depending on or travelling in a car, bus, aeroplane specifically associated with the source carbon tetrachloride and ethanol of possible future risks associated with between taking action when none is dietary habits. For example, the major or train will also influence the amount under consideration in the EHRA, •• potentiation, where one agent alone continuing or expansion of existing required or failing to take action when it pathway of exposure to many toxic metals of exposure of agents and the risk to consideration needs to be given to does not have a toxic effect but, when operations. Information derived from an is required. in children is food and children consume health (e.g. lead, benzene levels in the allocating a permissible component of the given with another agent, results in EHRA will be useful to risk managers in more joules per kilogram of body weight car, cosmic radiation in aeroplanes, exposure under consideration. Allocating a much greater toxic effect from the formulating and prioritising risk mitigation than adults do. Furthermore, children etc.). Hobbies such as pistol shooting in specific proportions of the ADI/TDI to other agent (e.g. 3 + 0 = 8) – an options. These may include: 2.6 have a higher gastrointestinal absorption indoor shooting ranges, antique furniture account for background or other sources example is risk of cancer from an •• closing down/ceasing use of WHAT RISK AND of metals, particularly lead. restoration, lead soldering, boat building of exposure is discussed in more detail in initiator and a promoter (tobacco the hazard source altogether or and lead lighting can also result in an Section 5.11.1. smoke contains both) OTHER TECHNICAL substituting with a less hazardous Alcohol ingestion may influence toxicity increased exposure to lead (Lead Safe •• antagonistic, where the combined material where minimisation of any indirectly by altering diet and reducing 1997). House renovating can result Other issues relating to identifying ASSESSMENTS ARE effect of two or more agents is less further environmental contamination is essential mineral intake. The ingestion of to an increase exposure to hazardous exposure pathways and quantitation of than the sum of the individual effects required NECESSARY TO EVALUATE alcoholic beverages (ethanol), fats, protein, agents such as lead and asbestos. Other pathway inputs are discussed further in – an example is risk of cyanide •• cleaning up a contaminated site calories and aflatoxins has been implicated hobbies involving paint stripping using Chapter 4. toxicity from cyanide after receiving RISK AND DISCRIMINATE using the EHRA outcomes and/or in carcinogenesis (Klaassen 1996). methylene chloride can cause exposure to an antidote such as Kelocyanor statutory instruments to guide the BETWEEN POTENTIAL its metabolic breakdown product (carbon (Klaassen 1996). Home-grown produce (e.g. vegetables) monoxide), and car maintenance can 2.4 level of clean-up required – this may RISK MANAGEMENT require a combination of in situ hazard has been associated with contamination also result in an increase in exposure to The potential hazards from interactions POTENTIAL INTERACTIONS treatment or containment, or removing of heavy metals such as lead, arsenic hydrocarbons and heavy metals. between chemicals are widely discussed OPTIONS? the hazardous material to another site and cadmium. This may be of particular BETWEEN AGENTS but there are no generally accepted A formal EHRA will generally provide concern when assessing contaminated methods for predictive appraisal of •• sealing off the contaminated 2.3 sufficient information to identify the sites. The NEPM processes used to There may be interactions between the interactions as part of the risk assessment environment to prevent further access major contributors to risk and enable the develop health investigation levels (HILs) physical, chemical, biological and social process. Some contemporary approaches by human receptors IDENTIFICATION OF risk manager to prioritise the needs for incorporate four different exposure hazards that need to be identified and to the health risk assessment (HRA) •• preventing ongoing release to the risk mitigation. Having then identified scenarios, one of which includes POTENTIAL EXPOSURE considered as part of the risk assessment. of chemical mixtures are discussed in environment or denying development technically feasible risk management potential for home-grown vegetables as For example, malnutrition may increase Chapter 12. plans that could increase such release. an exposure source. The potential for PATHWAYS the absorption of cadmium and hence options, the next phase would be to lipophilic chemicals to be stored and/or the risk of renal dysfunction. A high zinc undertake an economic analysis of When issues have been identified, a In some instances, the hazard and bioaccumulate in meat and poultry tissue intake may reduce the gastrointestinal the costs and benefits associated with preliminary qualitative risk assessment need for action will be so obvious to all (e.g. meat, fat, skin) and eggs (egg yolk) absorption of cadmium, reducing the risk each of these risk management options. may be carried out to prioritise issues stakeholders that risk assessment will be is an important consideration in regulating from high environmental levels. People The economic assessment processes for a more detailed study. This will undertaken only to determine the effect pesticide residues and environmental who carry the sickle cell anaemia gene are beyond the scope of this enHealth consider the likelihood of exposure and cost-effectiveness of the various contaminants in foods (see Chapter 17). have a reduced risk of malaria, while document, but some guidance may and the possible consequences, taking management options. In this situation, the people with the genetic condition of be found in enHealth monographs on into account things such as biological costs of undertaking a risk assessment The type of diet can also influence the Wilson’s disease have a greatly increased economic evaluation of environmental plausibility, evidence of exposure and to determine whether action is necessary risk of exposure to hazardous agents. risk from environmental copper. health issues (enHealth 2003). community concerns. There may be are considerable. In other instances, Individuals who are vegetarians will have multiple iterations of hazard appraisal as
22 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 23 Chapter 3: Hazard identification and dose–response assessment
•• Acute toxicity: Studies investigating the Key issues include: •• whether the mode of toxic action is Various sources of information are needed 3.1 3.2 effects of single doses of a substance. well understood – knowledge of the to identify and characterise environmental •• nature, reliability and consistency of The LD test, or medium lethal dose mode of action (MoA) is becoming hazards (Figure 6). Integrating INTRODUCTION HAZARD IDENTIFICATION 50 human and animal studies test are typical examples. The standard increasingly important in interpreting information on MoA, exposures (including The two elements of risk assessment Hazard identification examines the acute toxicity studies also include tests •• the availability of information about the carcinogenic responses (see Chapter background exposures) and identifying discussed in this chapter are: capacity of an agent to cause adverse for: acute oral, dermal and inhalational mechanistic basis for activity 11) and assessing the risk of chemical susceptible populations are all important health effects in humans and other toxicity; eye irritation; skin irritation; •• the relevance of the selected animal mixtures (see Chapter 12). factors in determining the correct use of •• hazard identification (using toxicity animals (US EPA 1995a). It is a qualitative and skin sensitisation. studies to humans dose–response information. test data) description based on the type and quality •• Sub-chronic toxicity: Short-term, •• dose–response assessment. of the data, complementary information repeat-dose studies, generally having Figure 6: Potential sources of information used to identify and characterise environmental hazards, leading to characterisation (e.g. structure–activity analysis, genetic an exposure duration up to 90 days in of mode of action (MoA), dose–response models and susceptible populations These elements are identified as part of toxicity, pharmacokinetic) and the weight rodents. The main purpose of sub- Phase II of the expanded framework for of evidence from these various sources. chronic testing is to identify any target EHRA outlined in Figure 2. organs and to establish dose levels for Hazard identification uses: chronic exposure studies. There are essentially two levels of hazard identification commonly undertaken in •• animal data – this is usually assessed •• Chronic toxicity: Studies lasting for the risk assessments in Australia. For many by toxicological methods greater part of the life span of the test animals, usually 18 months in mice risk assessments developed as part •• human data – this is usually assessed and 2 years in rats. Chronic studies of environmental protection licensing, by epidemiological methods when are particularly important for assessing planning processes or contaminated sites groups of people are involved, or by potential carcinogenicity. assessments, the hazard identification toxicological methods when using case component may simply identify the studies and acute chamber studies •• Reproductive toxicity: Studies designed relevant national or international guideline (both qualitative and quantitative to provide general information about values for each chemical that may be toxicity information is evaluated in the effects of a test substance on present. For risk assessments undertaken assessing the incidence of adverse reproductive performance in both male by national chemicals regulators or effects occurring in humans at and female animals. those setting national guidelines, the different exposure levels) •• Developmental toxicity: Studies in assessment will generally involve a full •• other data – this includes data such as pregnant animals that examine investigation of the international toxicity structure–activity data or in vitro data the spectrum of possible in utero literature relevant to the chemical, assessed by toxicologists. outcomes for the conceptus, including including an appraisal of the dose– death, malformations, functional response relationships that underpin any The data may come from a range of deficits and developmental delays. derived guideline values. This chapter sources such as ad hoc data, anecdotal More recent developments extend the focuses on an understanding of dose– data, case-report data and data collected period of dosing and/or observation response relationships that provide insight from epidemiological registries (including into the neonatal period, to assess into the development of health-based cancer or pregnancy outcome data). In potential neurobehavioural effects and guideline values, while Chapter 5 outlines other potential post-partum toxicity. the sources of guideline values and other each instance, the quality of the study information that can be used in the risk design and methodology, as well as the •• Genotoxicity: Studies designed to characterisation process. resulting data, will need to be rigorously determine whether test chemicals can assessed. perturb genetic material to cause gene Additional detail on the design and or chromosomal mutations. interpretation of animal-based toxicity Section 5.12 provides guidance on sources •• Other tests: Specific tests developed for tests is included in Chapter 9, along of toxicological information and where to endpoints such as neurobehavioural with a discussion of some of the newer find health-based guideline values. There toxicity, developmental neurotoxicity techniques for hazard assessment (in is also guidance on what to do when no and various in vitro tests (e.g. skin vitro and in silico techniques, genomics, suitable toxicological data appears to be absorption, irritancy potential and structure–activity analysis and ‘read- available (see Section 5.13). endocrine-related endpoints), which across’ from comparable substances) aim to reduce or eliminate the in vivo where the toxicity database for a In the case of data derived from use of animals, on the grounds of new industrial chemical may be less experimental studies in animals (see addressing animal welfare issues. comprehensive than other types of Chapter 9), a comprehensive data regulated chemicals. package will generally consist of: NRC 2008. Reprinted with permission from the National Academies Press, Copyright 2009, National Academy of Sciences
24 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 25 3.3 see Chapter 10 for more detail), animal chambers) may provide data whose 3.5.1 based on body weight to the ¾ power. Another conversion commonly needed studies have several advantages that relevance certainly exceeds that derived Inter-species scaling of doses The same factor is used for children in EHRA is changing the expression of DOSE–RESPONSE may be exploited in the risk assessment from animals. However, ethical constraints because it is slightly more protective concentration of substances in air from process (see Chapter 9): limit the levels of exposure that can Where animal studies provide the dose– than using children’s body weight. This ppm to mg/m3 (the preferred units). ASSESSMENT be used in such tests, and they would response data used in EHRA, scaling adjustment factor is used because The conversion equation is: •• Exposures can be controlled so that certainly be precluded where there is an the dose to provide an HED is a critical it represents scaling of metabolic groups are consistently exposed to Dose–response assessment examines expectation of an adverse outcome. step in the process. For many years, rate across animals of different size. 3 MW known amounts of chemicals. mg/m = ppm × the quantitative relationships between the simplest form of conversion or dose Because the factor adjusts for a V exposure and the effects of concern. •• The studies may be supplemented scaling has been expression of the parameter that can be improved on where V (volume of 1 g mole) = RT/P The determination of whether there is a by useful data on bioavailability, 3.5 dose on an equivalent mass basis – e.g. and brought into more sophisticated hazard is often dependent on whether a target tissue dose and possibly other as mg/kg body weight. This simplistic DOSE SCALING toxicokinetic modeling when such data R is the universal gas constant (so that dose–response relationship is present. toxicokinetic parameters. scaling approach is still very widely used become available, they are usually in EHRA, although other forms of dose V = 24.5 at 25°C and 760 mm Hg) •• The incidence of disease in groups The dose administered is a critical preferable to the default option. and MW is the molecular weight of Important issues include: treated with defined dose levels can conversion may provide better estimates component of the dose–response of the HED. the substance. •• the relationship between the be ascertained using a combination of relationship. To calculate a dose For inhalation exposure, a human extrapolation models selected and observational, clinical measurements relevant to EHRA, it may be necessary equivalent dose for adults is estimated 3.5.2 (blood and urine analysis growth) Scaling doses on the basis of body by default methodologies that provide available information on biological to undertake dose scaling to convert surface area (SA) has been used for Route-to-route scaling mechanisms and post-mortem tissue pathology the doses used in animals to a human- estimates of lung deposition and assessment. animal–human dose conversion by many internal dose. Often the toxicological data is not •• how appropriate datasets were equivalent dose (HED) or human- pharmaceutical regulatory agencies, available for the most appropriate route of selected from those that show the equivalent concentration (HEC). These advantages are, to some extent, although a direct conversion based Where oral doses are expressed in parts exposure for humans. For example, only range of possible potencies both in offset by a lack of knowledge of the on pharmacokinetic data is usually per million (ppm) in the diet or drinking oral carcinogenicity data may be available laboratory animals and humans In the animal studies that form available for toxicological assessment of following. the mainstay of much toxicological water, the dosage needs to be converted from which reference values have been •• the basis for selecting inter-species therapeutic agents. Where toxicokinetic to mg/kg body weight using appropriate calculated (e.g. cancer slope factor – •• The adverse effect observed may or information used in EHRA, the dose is scaling factors to account for scaling data is available for scaling conversion for estimates of food or water consumption CSF), whereas environmental exposure by may not be relevant to be extrapolated generally known and controlled. The doses from experimental animals to other types of chemicals, this is also the and body weights (see WHO 1987; oral, dermal and inhalational routes may to humans (i.e. is the effect species- route of administration in animal studies humans preferred approach, although such data Faustman & Omenn 1996). be important. Thus, extrapolation from specific because of some basic and the frequency of dosing may not •• relevance of the exposure routes may be less available to risk assessors one route of exposure to another may be difference in physiology or metabolic be directly relevant to the exposure used in the studies to a particular engaged in EHRA. Conversion factors commonly used for necessary; this needs to be assessed on function). conditions associated with environmental assessment and the interrelationships exposures. such conversions are cited in Table 5. a case-by-case basis depending on the of potential effects from different •• The dose–response relationship in The 1986 US EPA cancer guidelines available data. animals is not so relevant for low- (and many other US EPA guidance exposure routes For example, many animal studies Note that, while these conversion factors dose extrapolation (i.e. since such notes) recommended a SA-based scaling may provide rough estimates of dose, it One important consideration in route-to- •• the relevance to the assessment of use dietary admixture to provide a 2/3 studies use relatively high doses factor of bw for converting oral doses the expected duration of exposure continuous intake of test chemical and is always preferable to calculate actual route extrapolation is determining whether in order to be able to demonstrate in animal–human-equivalent doses. In and the exposure durations in the it is then necessary to measure food doses from feed intake and chemical the adverse health effects are localised to dose-related toxicity with an animal 2005, the US EPA guidance on scaling studies forming the basis of the intake (or use standardised food intake analytical data, where available. the exposure site or whether they are a sample size kept to practical limits, proposed altering the SA-based scaling consequence of systemic distribution. dose–response assessment conversion factors) to convert the dietary 3/4 the extrapolation to lower doses may factor to bw , although many of the dose- concentration to ‘dose’. Volatile chemicals •• the potential for differing susceptibilities be compromised if such responses scaling conversions found in the literature or dusts administered by the inhalational 2/3 in population sub-groups are only likely to be seen at very high will have used the earlier bw factor. The route (nose-only, head/nose-only or in an Table 5: Dose conversion from mg/kg (ppm) in diets to mg/kg/day in animal toxicity studies •• dose averaging/averaging exposure. doses). recent update of US guidance on dose– exposure chamber) may have intermittent response assessment in risk assessment Species Weight (kg) Food consumed 1 ppm in food 1 mg/kg/day •• The results have been obtained in exposure schedules (e.g. 6 hours/day healthy and genetically homogeneous (NRC 2008) also recommends a dose- per day (g) = mg/kg/day = ppm in food 3.4 5 days/week) that are different from those scaling factor of bw3/4. Further detailed strains of animals (usually rodents) associated with environmental exposures. Mouse 0.02 3 0.15 7 and are therefore less likely to be guidance on the use of body weight and THE PRIME ROLE OF Rat – young 0.1 10 0.1 10 representative of the human population SA conversion factors to produce HED ANIMAL STUDIES IN These factors can be taken into estimates is found in US guidance (US – adult 0.4 20 0.05 20 with its genetic variance and variable consideration in using the data to EPA 2005a p. A7): Guinea pig 0.75 30 0.04 25 RISK ASSESSMENT background health status. calculate a dose equivalent to human exposure (see Section 4.11 for further As a default for oral exposure, a Rabbit 2 60 0.03 33 Studies in which humans have been While a risk assessment may be able to discussion of time scaling). human equivalent dose for adults Dog – dry chow 10 250 0.025 40 exposed to chemicals, either in access data from controlled exposure is estimated from data on another – moist chow 750 0.075 13 studies using either animals or human occupational settings or in controlled species by an adjustment of animal Monkey 5 750 0.05 20 (including epidemiological studies; laboratory experiments (e.g. inhalation applied oral dose by a scaling factor
26 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 27 If the effects are localised at the exposure assumed that both the oral and dermal A procedure for deriving an HED for Mathematical models may be used to fit Figure 7: Hypothetical curve for an animal carcinogenicity study site and not a consequence of the routes will deliver an equivalent target inhaled particles and gases has been the experimental data to smooth curve, systemic distribution of the agent, then it tissue dose. The US EPA guidance described by Di Marco and Buckett (1996). but, as shown in Figure 7, the data points is not appropriate to extrapolate the dose recommends the adjustment when: from which the curve is derived may be to a different route of exposure. If the scattered around this relationship. •• the toxicity data from the critical effects are consequent to absorption and 3.6 oral dosing study is based on an systemic distribution of the agent, then The shape of the dose–response curve administered dose, and DOSE–RESPONSE CURVE dose scaling between routes of exposure below the experimental range can have needs to account for the bioavailability of •• there is a scientifically defensible SHAPE, CONSTRUCT multiple shapes depending on the model the agent by the different routes. database that shows the used. The choice of the model should, gastrointestinal absorption from a AND ANALYSIS where possible, be based on mechanistic Therefore, bioavailability is an important medium similar to the one used in the information. consideration when extrapolating the critical study (water, feed or gavage Elaboration of dose–response relationships is a fundamental element of contemporary applied dose to different routes of vehicle) is significantly less than It is obvious that different dose–response risk assessment. The availability of human exposure. However, additional factors 100 per cent (i.e. <50 per cent). relationships can be elaborated where data on dose–response is often quite may need to be considered, such as The 50 per cent cut-off is proposed the chemical causes more than one limited unless well-structured, ethically physiological differences between species since it is thought to reflect the intrinsic form of toxicity (e.g. growth retardation, justified and controlled experiments have when extrapolating, for example, from variability in data from absorption hepatorenal toxicity or neurotoxicity) been conducted, Epidemiological studies inhalational exposure in animals to oral studies. (see Figure 8). The dose–response may be useful for providing estimates exposure in humans or vice versa. The relationship furthest to the left (i.e. of whether health risk in a population assessor should include information about There are relatively few substances for occurring at the lowest dose range) exposed to an environmental chemical, or the bioavailability of the chemical agent which GAFs have been determined. will usually drive the risk assessment group of chemicals, has been increased in the experimental studies in the final US EPA guidance (US EPA 2004b) has process, unless it is considered that in comparison to an unexposed group or report. There is further discussion of the tabulated GAFs for a few selected metals the effect (e.g. mild liver enlargement; population. In some cases, the studied importance of assessing bioavailability in and metalloids, of which only five out of changed enzyme activity with no obvious cohorts may be categorised in relation Section 4.2.1. nine listed have GAFs below 50 per cent. toxic consequence) is not considered to to exposures to one or more chemicals be ‘adverse’. However, enzyme activity in the environment. However, such In cases where bioavailability data is not Route-to-route dose scaling is mainly changes, such as the release of lactic human data is rarely, if ever, as complete Adapted from: Levy & Wegman 1988. available, important clues may be gained done for adjusting doses used in animal dehydrogenase from tissues into blood, for the purposes of constructing dose– from the physical and chemical properties studies, but it is an equally important are commonly used as a surrogate for response relationships as that available Figure 8: Different dose–response curves for different effects from a hypothetical substance and physical state of the agent (e.g. consideration where human data is used. measuring organ toxicity, so would be from controlled toxicity studies using liquid, solid, gas). relevant to risk assessment. 3.5.3 animals because they are based on Unless specific dermal toxicity data is Other factors in dose scaling estimates of exposure rather than actual measured exposure. In large part, the 3.6.1 available, risk assessments involving the Individual versus population risk dermal route of exposure may necessitate For inhalational exposure, doses default database used in most health risk 3 the use of toxicity data derived from expressed as mg/m or ppm may need assessments relies on extrapolating human There is an extensive discussion of the another exposure route (usually oral). to be converted to mg/kg body weight in health risks based on dose–response conceptual models and implications The toxicity reference dose (e.g. TDI, RfD) the test species by calculations based on relationships from animal studies. for health risk assessment when can be adjusted as follows: the physical properties of the agent and extrapolating from dose–response minute volumes and respiration rates There are different ways of characterising curves that describe individual risk TDIdermal = TDIoral × GAF of the animal (Kennedy Jr & Valentine dose–response relationships. In risk to those that characterise risk in a 1994; US EPA; 1994). However, more assessment, dose–response relationships population (NRC 2008). Differences (GAF = gastrointestinal absorption factor) recent US EPA guidance (US EPA 2009a) based on experimental studies in animals in individual sensitivity can markedly proposes that such conversion is not are often characterised as shown in influence the shape of the population A similar adjustment approach for non- needed for airborne substances where Figure 7. The dose or exposure scale is dose–response curve. For example, threshold toxicity replaces TDI with CSF. the toxicological information suggests a generally logarithmic while the response differences in sensitivity in a vulnerable dose–response threshold (see Sections scale represents the proportion of the group or bimodal differences in The basis for application of a GAF to 4.6 and 4.12). tested animals that respond at the doses sensitivity can flatten out the lower part an oral dose is outlined in the US EPA used. This scale may also be converted of the dose–response curve (Figure 9). guidance on dermal risk assessment The process for converting doses derived to a probit scale based on the properties Additionally, extrapolation of population (US EPA 2004b). The adjustment is from animal studies to an HED or of a Gaussian distribution curve. Such risk can be influenced by variability in the not necessary where oral bioavailability concentration (HED/C) is detailed in US a transformation can linearise a large background exposure or spontaneous approaches 100 per cent because it is EPA guidance (US EPA 2005a). part of the dose–response curve.
28 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 29 incidence of the disease that is using upper-bound probability estimates curved at the bottom end. Where the Figure 10: Extrapolation from the individual risk to population risk incorporating uncertainty estimates characterised in the individual dose– of the population risk (see Figure 10). population curve intersects the axes will response curves or by the assumptions be determined by whether the relationship made about whether low-dose effects are Individuals will vary in the lowest dose is assumed to have a threshold above the best expressed by linear or non-linear that can initiate the response, so that background level of exposure (non-linear functions. dose–response curves relating to more population extrapolation in Figure 11, susceptible individuals will be furthest to Model 2) or a non-threshold relationship It is also possible to incorporate estimates the left, while more resistant individuals for both individuals and populations of uncertainty associated with varying will have dose–response curves further to (Figure 11, Model 3). background exposure and heterogeneity the right. This translates into a population of biological response and susceptibility, dose–response relationship that may be
Figure 9: A conceptual model that integrates individual risk dose–response, background disease incidence and variation in susceptibility into population dose–response
NRC 2008. Reprinted with permission from the National Academies Press, Copyright 2009, National Academy of Sciences.
Figure 11: Conceptual models describing low-dose linearity or non-linearity of individual and population dose–response curves
NRC 2008. Reprinted with permission from the National Academies Press, NRC 2008. Reprinted with permission from the National Academies Press, Copyright 2009, National Academy of Sciences. Copyright 2009, National Academy of Sciences.
30 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 31 3.7 some sites in rats were actually lower 3.8 3.8.1 Figure 12: Examples of the potential variability when different models are used to at low doses than in controls, implying Mathematical models used in risk extrapolate risk HORMESIS a protective effect. The study went on DOSE–RESPONSE extrapolation to note that cancer risk, as reflected The basic tenet of dose–response in case-control studies in the Finnish MODELLING The BMD approach requires that all the assessment in EHRA is that the population, appeared to decrease slightly experimental data is used to derive a relationship between dose and toxic in those groups showing higher levels There are various ways of managing ‘best fit’ in determining the shape and effect is monotonic (i.e. response of dioxin body burden, where dioxin dose–response data using mathematical position of the dose–response curve. A always decreases as the dose or exposure through food is the dominant equations to derive dose–response curves variety of mathematical models have been exposure is lowered). exposure matrix. that fit the experimental data. They developed to facilitate such curve fitting. assume that the toxic effect results from They may be subdivided into: the random occurrence of one or more The basic concept of hormesis is that While the concept of hormesis attracts •• statistical or distribution models vigorous debate in the scientific literature biological events. These are known as dose–response curves in toxicology –– log-probit are non-monotonic. That is, they (Calabrese 2005; 2008; Thayer et al. stochastic events (Klaassen 1996). may resemble a J-shape or possibly 2006; Mushak 2007) and has been –– logit Mechanistically derived models have been U-shape at low dose and that, as the addressed to some extent in the most –– Weibull dose or exposure gets lower, the risk recent US guidance on risk assessment particularly used for cancer modelling •• ‘mechanistic’ models of harm may actually increase rather (NRC 2008), there is no consensus that (especially based on radiation exposures). than decrease, at least over part of the basic concepts of risk assessment The simplest form is a ‘one-hit’ linear –– one-hit model in which only one ‘hit’ or critical the dose–response range. In fact, it (i.e. that it is possible to define low doses –– multi-hit has been argued that non-monotonic or exposures where risk is either negligible cellular interaction results in the alteration –– multi-stage dose–response relationships are quite or non-existent) should be modified to of a cell. This model would propose that a common in toxicology, and that there are accommodate the concept of hormesis. single molecule of a genotoxic carcinogen –– linearised multi-stage UK EA 2009. Contains Environment Agency information © Environment Agency and database right. would have a ‘minute but finite chance rational mechanistic and toxicokinetic –– stochastic two-stage model The extent to which the debate over of causing a mutational event’ (Klaassen explanations for the phenomenon (Moolgavkar–Venson–Knudson) (Connolly & Lutz 2004). hormesis has polarised the scientific 1996). From these models more (continuous or dichotomous) (Slob community is well set out in a number complex models based on multi-hits or •• model enhancement 3.9 2002; Sand et al. 2002; 2003; 2006). Hormesis theory presents important of commentaries (e.g. see Axelrod et multi-stage events have been derived. •• time–tumour response Imprecision in exposure estimates based Although conceptually based on biological BENCHMARK DOSE implications for risk assessment, since al. 2004; Calabrese 2004; Kaiser 2003; •• physiologically based toxicokinetic on epidemiological studies can lead to mechanisms, most of these models extrapolation of risk to very low levels of Renner 2004). models APPROACH a derived BMD estimate that is biased exposure may be confounded by such do not rely on independently validated towards a higher and less protective level •• biologically based dose–response a fundamental change in shape of the A separate but related consideration is parameters describing the mechanisms, The BMD approach has been used (Budtz-Jorgensen et al. 2004). models. dose–response curve. The issue becomes the concept of the acceptable ranges of but rely on fitting curves to empirically in dealing with both cancer and non- even more complex when it has been oral intakes for essential trace elements, observed data. cancer endpoints. It is described in The BMD corresponds to a A schematic diagram illustrating shown that, for the same chemicals, both where there is a need to ensure EHC170 and a modified version for use predetermined increase (between 1 and the critical effect on risk levels if an hormetic and non-hormetic responses tolerable intakes do not fall below the More recently these models have with carcinogenic soil contaminants was 10 per cent but commonly 5 per cent) inappropriate extrapolation model is may be observed in different tissues minimum requirements. WHO (1996) been adapted to take into account described in NHMRC (1999a). of a defined effect in a test population. chosen is shown in Figure 12. (Borak & Sirianni 2005). regards iron, zinc, copper, chromium, information based on knowledge of the Mathematically it is the statistical iodine, cobalt, manganese, molybdenum relevant physiology and toxicokinetics There has been more recent discussion lower confidence limit on the dose that For a more extensive coverage of the One of the more controversial aspects and selenium as unequivocally essential (physiologically based toxicokinetics/ of the utility of the BMD approach in the corresponds to that predetermined mathematical concepts and models of hormesis theory is that part of the low for human health. However, it is not pharmacokinetics modelling – PBPK). literature (for reviews, see Falk-Filipsson increase, derived by extrapolation used in quantitative risk assessment, dose–response relationship may describe clear that the toxic effects associated These models take into account the et al. 2007; Travis et al. 2005; Sand et from the upper confidence level of refer to monographs by David Vose a phenomenon where the risk of harm is with vitamin deficiency and overdose effective dose at the target organ. al. 2008) and various guidelines have the dose–response curve (Figure 13), (2008) or Dennis Paustenbach (2002), actually less than in unexposed (controls); is an appropriate example of hormesis. A further development has been to been issued by national bodies regarding although some agencies are using a best or more general monographs on the that is, low dose exposure may have a The adverse effects associated with make generalised mechanistic models it applications in risk assessment (e.g. estimate rather than a lower confidence principles of environmental health risk beneficial or protective effect. vitamin overdose are different from take into account specific biological see US EPA 1995b, 2000b; Appel et al. limit (IEH 1999b). assessment (Fjeld et al. 2007; Robson those associated with vitamin deficiency. processes. An example of these 2001; Health Council of the Netherlands biologically based dose–response models & Toscano 2007). The case for beneficial effects of toxic Therefore, it is inappropriate to depict 2003; IPCS 2009c). The influence of is the Moolgavkar–Venson–Knudson chemicals at low doses has been these different adverse effects as a model- and dose-level selection has model that uses a two-stage model for supported to some extent by analyses continuous dose–response curve or as been evaluated, as well as assumptions carcinogenesis (Klaassen 1996). of cancer rates for dioxin). Tuomisto et an example of hormesis. about the underlying nature of the data al. (2005) reported that cancer rates for
32 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 33 Figure 13: Graphical illustrations of the benchmark dose The stylised example (Figure 13a) Figure 14: Analysis of data on formaldehyde toxicity using different mathematical curve-fitting models
shows the derivation of the BMD05, the dose estimated to result in a 5 per cent increase in a defined effect. It also shows the derivation of the lower confidence limit (BMDL), which some agencies
prefer as the estimate of BMD05. The dose–response relationship (Figure
13b) shows the derivation of the BMD05
(9.3 µg/kg) and BMDL05 (8.3 µg/kg) for some real toxicity data – malformations in mice after maternal dosing with TCDD (Sand et al. 2008, using data from Birnbaum et al.1989). It should be noted that, in this case, the figures derived from the BMD approach are between the NOAEL (6 µg/kg) and LOAEL (12 µg/kg) that are predicated by the dose selection in the study.
Fully maintained software for using the BMD approach sanctioned by the US EPA is available from the US EPA National Center for Environmental Assessment (NCEA) website at
As part of the NHMRC process for evaluating the mean BMD methodology (see Section 11.7), the CSIRO developed software that could calculate (a) The fitted curve using the Weibull model and (b) a table showing the goodness of fit (GOF) and derived BMD estimates for a number of different a mean BMD from most of the above curve-fitting models. mathematical models, and derive goodness-of-fit parameters that could Particular advantages of the BMD •• is able to be rigorously described When the benchmark response is within aid selection of best-fitting curves. approach include that it: or near the experimental range of the An illustration of the application of the •• allows potency comparisons between data, the corresponding values of the CSIRO software modelling for a specific •• takes into account information from the endpoints and between studies. benchmark doses are not greatly sensitive chemical of potential concern (COPC) is entire dose–response curve rather than to the choice of the model used. The shown in Figure 14. focusing on a single test dose such as Its disadvantages are that it may not best scientific choice of a model would is done with the NOAEL approach (i.e. be possible to define the shape of the be a biologically based mechanistic For developmental toxicity the BMD uses all available relevant information) dose–response curve because of limited 05 model. Sand et al. (2002; 2003; 2006) values have been similar to statistically dose groups or the number of animals per •• uses responses within or near the confirmed that a BMD range within 5–10 derived NOAELs for a wide variety group. It also requires greater statistical experimental range rather than relying per cent provides an estimate that is of developmental toxicity endpoints expertise than the NOAEL-type approach on extrapolations to doses considerably less dependent on the dose–response (a) stylised to show the extrapolation of BMD05 and its lower confidence limit (based on upper (Klaassen 1996). BMD approaches are (IEH 1999b). below the experimental range model used and the variation in the data. confidence limit of incidence) also being developed and tested in regard •• uses a consistent benchmark-response Use of a BMD provides a more data- However, the method used to calculate (b) illustration of BMD calculation from real data on malformations induced by TCDD Sand et al. to acute inhalation toxicity (Fowles et al. 05 1999), to the relationship between the level that crosses a range of studies sensitive and less model sensitive confidence intervals can influence the (2008) using data from Birnbaum et al. (1989) and endpoints precision of the BMD estimate (Moerbeek BMD and the MTD (Gaylor & Gold 1998), endpoint than using BMD01 (Klaassen and to addressing statistical procedures •• is less influenced than NOAEL 1996; NHMRC 1999). et al. 2004). Also, study designs with available for calculating BMDs and approaches by the arbitrary selection more dose levels improve precision, even their confidence limits for non-cancer of doses (Crump 1984) when they result in fewer test animals endpoints (Gaylor et al. 1999). per dose group (Slob et al. 2005).
34 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 35 An important proviso is that where the 3.10.1 Numerical estimates of risk probabilities the chemical of concern has been shown and influenced by such things as the age 3.10.3 data fit is relatively poor (more than Threshold approaches are generated by fitting one or more to have genotoxic potential. For non- and breeding (strain) of the animals used, Outputs from a non-threshold 18 per cent coefficient of variation), mathematical models to the data in threshold dose–response relationships, the experimental conditions (e.g. caging risk assessment the ability to estimate either a BMD A threshold occurs when the dose or the experimental dose range and the excess incidence (i.e. incidence and feeding conditions and other aspects or a NOAEL is more compromised. exposure has not reached a critical extrapolating the upper 95 per cent corrected for background) of induced of animal husbandry), the rigour of the The outcomes of a non-threshold risk level sufficient to trigger a response. confidence limit of the curve fitting to the cancer is assumed to be zero only at pathological investigations, the magnitude assessment are either: The concept is well established in relation low environmental exposure doses. For zero exposure. of the doses used, especially when 3.10 to receptor-mediated mechanisms, when (i) the dose describing at which the example, low-dose extrapolation using escalated towards a dose level defined chemicals produce a predetermined a certain degree of receptor occupancy is a linear model is a default approach for Low-dose linearity assumes a positive as the maximum tolerated dose (MTD), THRESHOLD VERSUS required to trigger a response. A threshold risk level – note that this requires cancer risk assessment in the United slope of the dose–response curve and even the way the test chemical is some judgement on what constitutes NON-THRESHOLD may also appear to occur when biological States (US EPA1996a; 2005a) and is one upward from zero dose and implies that administered (gavage or via feed). mechanisms act to reverse the toxic an acceptable level of risk (what approach which has been used (perhaps a single, irreversible genetic event at the may constitute an ‘acceptable’ or RESPONSES IN RISK effects. These can include a saturable inconsistently) by WHO for genotoxic initiation stage of carcinogenesis leading It follows that the numerical data capacity to inactivate a toxic chemical by ‘target’ risk estimate is discussed ASSESSMENT carcinogens in deriving drinking-water to transformation of a cell is sufficient derived from any one study may not be in Chapter 5); or metabolism or excretion, or where tissue guidelines (WHO 1993a; 2006a). by itself to lead to the development of reproducible in another repeat study of damage can be repaired up to the point (ii) the estimated risk level at any A longstanding convention in risk cancer. The major difficulty in this debate the same design. It is also important to of irreversibility or below a critical dose or particular dose. assessment has been the different The revised US EPA guidelines for cancer is the impossibility of experimentally acknowledge this fact, since there is a level of damage. treatment of dose–response relationships risk assessment state that: testing the shape of the dose–response general belief that numbers generated The parameter from which (ii) is derived where either a threshold or a non- curve at extremely low doses (Purchase in a toxicity study are inviolable A threshold may also be apparent where is the cancer slope factor (CSF) or unit threshold relationship is assumed. ‘A linear extrapolation approach & Auton 1995). or sacrosanct as inputs into a risk the level of cell damage or cell death does is used when the mode of action assessment. risk factor (URF), which is the probability not reach a stage where tissue damage is (or risk) of the response (e.g. cancer) There may even be a dichotomy in information is supportive of linearity or A transformed cell that has acquired the apparent, or where the tissue function is per unit of intake (usually expressed in understanding of the meaning of a mode of action is not understood.’ potential to develop into a tumour will Furthermore, the extrapolation compromised. It may be possible to ‘kill’ mg/kg body weight per day) or exposure ‘threshold’. The toxicologist interprets probably realise that potential only rarely methodology used to estimate the disease individual cells without consequence for ‘When adequate data on mode of concentration, respectively, over a lifetime a ‘threshold’ as a dose or level of (US EPA 1996a), most likely because incidence (or risk) at dose levels well the organism as a whole. action provide sufficient evidence to of exposure. exposure where the toxicity response or of the natural large-scale repair of DNA below the high dose levels actually used support a nonlinear mode of action adverse effect measured is no greater damage and other defence mechanisms in the study is quite model-dependent. The approach with these models is to for the general population and/or It should be noted that the CSF is derived than the background. That is, there is of the body (DOH 1991; Abelson The requisite knowledge or understanding derive exposure limits such as an ADI, a any subpopulations of concern, a conservatively and is based on a linear no measurable incremental risk. On 1994). Furthermore, while it is generally of the toxicological mechanisms that provisional tolerable weekly intake (PTWI), different approach – a reference dose/ extrapolation to zero dose from the upper- the other hand, an epidemiologist may accepted that mutagens and mutations account for nuances in the shape of the tolerable daily intake (TDI) or RfD (see reference concentration that assumed bound estimate of a dose at which the consider a threshold to be a point where play a role in the development of cancer, dose–response relationship at low dose Section 5.6). This approach makes no nonlinearity – is used.’ incidence of cancer can be predicted the incidence of disease begins to carcinogenesis is more than mutagenesis, may be incomplete. attempt to calculate a level of risk at low with a number of non-mutagenic as (e.g. a BMD – see Section 3.9). The URF exceed background, especially when the ‘When the mode of action information exposures. Rather, it derives a dose that well as mutagenic events taking place The above uncertainties about the is similarly derived using concentration as background incidence is not zero. indicates that the dose–response is apparently without effect in a human during the process (see Section 11.4). quantitative data derived from an animal the exposure measure. The implication is population or suitable animal model, function may be adequately described that the relationship that best describes This concept of a threshold is not The shape of the dose–response curve testing program need to be borne in and then applies a factor to derive an by both linear and nonlinear approach, the low-dose behaviour of the dose– necessarily the same as a NOAEL. A at any one of these steps, not just the mind when extrapolating risk estimates exposure that has a high likelihood that then the results of both the linear and response curve is linear in this region, NOAEL is defined on the basis of it being mutagenic events, can influence the within the same experimental species, no effect will occur in the general human nonlinear analyses are presented.’ although it may become non-linear at the highest dose used in a toxicity test shape of the dose–response curve for let alone when attempting to extrapolate population. higher doses. where there is no appreciable increase ‘Absent data to the contrary, the the carcinogenic response. Factors, such from one species to another (e.g. to over controls in the incidence of the default assumption is that the as genetic make-up, lifestyle and other humans). The resultant numerical risk 3.10.2 Earlier versions of US EPA guidance adverse effect. The NOAEL is therefore cumulative dose received over a environmental factors may also have a estimates may need to be extensively Non-threshold approaches on non-threshold dose–response influenced by dose selection and other lifetime, expressed as a lifetime modifying influence on the processes of qualified by reasonably large statistical extrapolation referred to the output parameters of the experimental system Non-threshold models assume linearity average daily dose or lifetime average carcinogenesis. confidence limits. Risk assessment may parameter from the linearised multi-stage and a true threshold could be set at a between the lowest experimentally derived daily exposure, is an appropriate conventionally focus on the extreme ends models as q1*, defined as: different level if more appropriate doses dose and the zero dose (the origin). This measure of dose or exposure.’ This can introduce some complexity of these confidence limits in order to be had been selected in the defining study implies there is a calculable probability (US EPA 2005a pp. A8–A10) into the quantitative aspects of the risk protective of more vulnerable members An upper bound, approximating a (see Section 5.6). of an adverse effect (risk) no matter how assessment. The incidence of cancer in of the population. This can inject quite 95 per cent confidence limit, on the small the dose. This does not mean there Studies where the significant endpoint unexposed or control animals is rarely a degree of unfounded conservatism increased cancer risk for a lifetime is no dose that could be considered safe includes a neoplastic change zero, and finite values may even approach into a risk assessment, especially when exposure to an agent. This estimate, unless safety must be equated with zero (carcinogenesis) are usually assumed 100 per cent. More of a problem is that ‘worst-case’ estimates for several different usually expressed in units of proportion risk (Hrudey & Krewski 1995). to represent a non-threshold dose– the background or spontaneous incidence parameters are incorporated into the risk (of a population) affected per mg/kg/ response relationship, especially where can be quite variable from study to study, assessment methodology. day, is generally reserved for use in the
36 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 37 low-dose region of the dose–response daily dose so that they can survive for It is notable that the latest US EPA Figure 15: Decision-making process for choosing dose–response data in risk relationship, that is, for exposures the 2-year duration of the experiment to guidelines on carcinogenic risk assessment of carcinogenic substances 3.11 corresponding to risks less than 1 have an opportunity to contract cancer. assessment (US EPA 2005a) make no THRESHOLD VERSUS NON- in 100. For most carcinogens tested, the MTD is reference to either CSF or q1*. The a very high dose, which may not be far above definition is now simply linked to THRESHOLD APPROACHES Dose selection in non-threshold models removed from an acutely toxic dose. the term ‘slope factor’. – IMPLICATIONS FOR RISK has been discussed by Lovell and Thomas (1996), who suggest that the estimate Krewski et al. (1993) showed in an Bodies such as WHO, US EPA and ASSESSMENT of q1* is so dependent on the doses analysis of bioassay results for 191 California EPA Office of Environmental selected that it is almost independent individual carcinogens plotting the Health Hazard Assessment have all The important conceptual distinction of, or at least insensitive to, the actual upper-bound estimate for the q1* versus developed CSF and URF for carcinogens. between non-threshold methods and tumour incidences in the dose groups. the MTD that these values were highly However, the values can vary by an order those that derive an acceptable exposure Specifically, the highest dose in an animal negatively correlated (r = –0.941) for of magnitude depending on the studies from the NOAEL using a safety or bioassay has overwhelming influence values that spanned nine decades in MTD used and the model chosen to derive uncertainty factor is that this approach on the estimate of q1*, thus leading to and ten decades in CSF, a result that the factor. makes no attempt to determine a finite the overestimation of risk at very low could not conceivably be achieved from level of risk at low exposures, whereas doses, with the extent of overestimation 191 truly independent experiments. The step-wise process for deciding on non-threshold methods make an increasing as the environmental exposure the dose–response data to adopt for estimate of the risk at low exposures using becomes lower. Typically, the highest This finding shows that carcinogens with the EHRA of carcinogens (or potential (usually linear) extrapolation from a point dose in a carcinogenicity bioassay is the a very high MTD (low toxicity) had a very carcinogens) is set out in Figure 15. This higher in the dose–response relationship maximum tolerated dose (MTD), a dose shallow q1* and carcinogens with a very decision-making process recommends (see benchmark dose methodology in that causes no more than a 10 per cent low MTD (high toxicity) had a very steep use of a BMD approach to selecting Section 3.9). decrease in body weight and no other q1*. A primary determinant of the q1* a POD for risk assessment once a overt toxicity. The MTD is very much for any of these chemicals was its MTD, decision has been made on classification An NRC report (NRC 2008) greater than doses expected from non- given the procedure used for determining of the COPC, as a carcinogen and a acknowledges this important distinction, occupational environmental exposures. the CSF. Krewski et al. (1993) described carcinogenic risk assessment approach but also advances an argument that the Therefore, the dose that is the least how this outcome was created by is warranted. Where appropriate BMD two processes may be harmonised by relevant to environmental risk assessment the relatively small range of possible data is not available, alternative dose– redefining the RfD/RfC as risk-specific has the greatest influence on low-dose outcomes from the cancer bioassay once response data should be sourced, which dose estimates describing the proportion risk estimates. the dose, which according to the range of may include the use of CSF (for genotoxic of a population likely to be susceptible MTDs varied over a much larger range, carcinogens) and ADI/TDI (for non- below the adjusted NOAEL. An extremely insightful finding about was determined. genotoxic carcinogens). the meaning of the q1* estimated in The NOAEL is assumed to be the this manner was revealed by Krewski et Non-threshold models currently in use threshold dose for the effect. Both al. (1993). This analysis considered the are inflexible and generally do not take threshold and non-threshold approaches relationship between the q1* calculated account of the complexities of the events have advantages and disadvantages. using the linearised multi-stage model between exposure to an agent and the or equivalent no threshold model and induction of a neoplasm. Risks estimated The advantages of the threshold the MTD, the maximum tolerated dose. at doses below the range of experimental approach are that the NOAEL is relatively Because carcinogen bioassays are data can vary considerably depending on easy to determine, and the process is very expensive experiments (costing the model used, even though the various simple to use, easy to understand and millions of dollars if the full protocols mathematical models used generally allows the use of expert judgement. In the are followed), which are limited in the fit the experimental data equally well few cases where epidemiological data has number of animals and dose levels that (Crump 1985; Paustenbach 1995). The become available, the ADIs derived by can be tested, there is a practical need numerical expression of the estimated this method have been validated (Lu & to maximise the potential for detecting level of risk falsely gives the impression Sielken Jr 1991). a carcinogenic response in the typical that it represents an exact measure of 2-year duration of the experiment. This actual risk. This numerical expression Additionally, the approach has been need has been met by doing a range- provides little or no information on the applied seemingly in a consistent finding experiment to determine the uncertainties related to the estimated level fashion by WHO over three decades in maximum dose of carcinogen that the of risk, nor does it allow comparison with deriving ADIs for pesticides. The safety- experimental animals can tolerate as a values for non-cancer health effects. factor approach remained essentially unchanged until 1994 (WHO 1994a),
38 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 39 although a number of articles were The choice of the NOAEL is limited to one With advances in biological knowledge, theoretically much greater values than Figure 16: Decision tree for choosing a threshold or non-threshold model for published suggesting modifications of the doses included in the experimental the classification and assessment of that associated with most environmental risk assessment or improvements (e.g. Calabrese & design. The biological no-effect dose may carcinogenic risk is now being guided by chemicals. It is also apparent that Baldwin 1994; Calabrese & Gilbert 1993; occur at this dose or possibly at a lower mechanistic, pharmacokinetic and other some chemicals can influence the Crump 1984; Johannsen 1990; Lewis dose that is not included in the study. The relevant data. The US EPA undertook a quite complex multi-stage process of et al. 1990). closeness with which the selected NOAEL review of its science policy with regard carcinogenesis by modifying events truly reflects the actual no-effect dose to carcinogenic risk assessment (US downstream from the initiating genetic A potential disadvantage of relying on the has an obvious impact on the degree EPA 2005a) and incorporated evaluation defect (i.e. via epigenetic or promoter- NOAEL as the starting point for threshold- of protectiveness in the derived ADI, of MoA and its use in framing the type mechanisms), or even by simply type risk assessment is the precision with PTWI or RfD. Furthermore, the precision risk assessment approach, as well as amplifying the spontaneous development which the real NOAEL can be estimated with which the NOAEL can be assessed suggesting a more narrative approach to of cancers by increasing the rate of cell (see Section 5.6). Another possible is influenced by the biological effects classify carcinogenicity. The use of MoA turnover, sometimes following a significant limitation is the need to consider that a monitored, the number of animals in the information and its impact on determining cell-damaging event. finite background level of disease may test groups, the spontaneous incidence how to make low-dose extrapolation give an impression of non-linearity at of the adverse effect, and the criteria in EHRA was reviewed at a ‘state-of- The US National Research Council, low dose. used to determine when the incidence in the-science workshop on issues and working in conjunction with the US EPA, a test group exceeds that in the controls approaches in low-dose extrapolation’ in the National Institute of Environmental Additional limitations of the threshold (Renwick & Walker 1993). April 2007 (White et al. (2009). Health Sciences (NIEHS) and the approach include: the NOAEL is often National Academy of Sciences (NAS), perceived as a biological threshold, While the US EPA approach continues has carefully enunciated the science whereas it is a threshold limited by the 3.12 to rely almost exclusively on the non- policy decisions that underpin the use experimental protocol; risk is expressed SCIENCE POLICY AND threshold, low-dose extrapolation for of threshold and non-threshold risk as a fraction of the guidance dose (e.g. cancer risk assessment as in the past, assessment approaches in the United ADI); it makes limited use of the dose– THE SELECTION OF there is a growing acceptance that a States. The outcomes of this project response slope; the choice of safety threshold approach may be valid where were communicated through various factors has been arbitrary to some extent; THRESHOLD AND NON- the scientific data justifies an assumption consultations and reports, culminating in and the process does not generate a THRESHOLD MODELS of non-linearity at low dose (e.g. where the release of the seminal report Science NRC 2008. Reprinted with permission from the National Academies Press, Copyright 2009, range of estimates of risk, but rather a cytotoxicity is a necessary precursor to the and decisions: advancing risk assessment National Academy of Sciences single estimate of a dose below which no In deciding between a threshold or non- carcinogenic response). in December 2008 (NRC 2008). adverse effects are likely to be produced. threshold approach to risk assessment, When a non-threshold model is Actual risk might be as large as the it is important to recognise that one is It may be argued that the impetus for Figure 16 is a flow chart outlining the adopted the resulting prediction may upper bound, but could be lower Because it provides numerical estimates entering the realm of science policy. Such applying non-threshold methodology step-wise process whereby decisions can be useful for demonstrating a plausible and could even be zero. It is not of risk at all doses, the non-threshold policy is commonly applied to genotoxic to carcinogenic risk assessment was be taken about whether to use a threshold upper bound of cancer risk, but the case that risk assessors, at least approach, in principle, has the potential carcinogens by many regulatory agencies the initial premise that all carcinogens or non-threshold approach. estimates by these methods are not those who truly understood the advantages (if the estimates are correct) around the world, although it is being less are mutagens (Ames et al. 1973). One intended to be used as estimates of problem of low-dose extrapolation, of allowing computation of comparative rigorously applied in contemporary risk mutation or one DNA damage event No Australian environmental health “expected” cancer risk. In this regard, have ever claimed that risks risks in the sub-experimental range, assessments to carcinogens for which was considered sufficient to initiate the authorities have enunciated a specific a quote from an award speech by Joe predicted in this fashion are known which may be a useful tool in risk there is reasonable evidence that they process that leads to the development policy on when a threshold or non- Rodricks, one of the creators of the to be accurate, even ignoring the management and communication- act through non-genotoxic mechanism(s) of cancer. A more contemporary threshold approach should be used U.S. policy approach for carcinogens is uncertainties introduced by the potency comparisons between chemical (NRC 2008). hypothesis is that cancer formation in EHRA. However, it is common agents at a particular risk level and requires a series of mutagenic events, practice, accepted by most state instructive (Rodricks 2007): fact that most risk assessments are based on animal, not human data. estimates of the increased risks if a The fact that a distinction may be made perhaps in a defined sequence, not just and territory jurisdictions, that a non- The linearized multistage model particular dose is exceeded. It has between a genotoxic and a non-genotoxic a single event of DNA damage. threshold approach consistent with that was selected because it seemed This enHealth document supports the been argued (McMichael 1991) that mechanism for a carcinogenic response used by the US EPA should be used to have some basis in the leading application of sound scientific principles risk estimates using this approach will be based on the available evidence. This premise that carcinogenicity and when assessing carcinogenic risk for mechanistic hypotheses regarding to assessing carcinogenic risk. These approximate those seen in humans However, it does not mean that a non- genotoxicity are inextricably linked has genotoxic carcinogens. This implies that the carcinogenic process, and also principles are articulated further in in some cases and where there are genotoxic carcinogen does not affect been questioned, even by Ames himself the endpoint of the EHRA is an estimate because it seemed highly likely Chapter 5, and include consideration of disparities they are overestimates of the genetic material of the cell under (Ames 1987), who has repeatedly pointed of risk that needs to be compared that the model – because of its carcinogenic MoA in determining whether the risks. some circumstances or that a genotoxic out that many animal carcinogens are with an acceptable or ‘target’ risk level ‘linearization’ at low dose – would a threshold or non-threshold approach is effect is the only event required for the not genotoxic, and that many naturally (see Section 5.10). not underestimate low dose risk, more suitable for the EHRA. Both the threshold and non-threshold development of cancer by a genotoxic occurring processes (dietary elements that it would, in fact, place an methods are likely to be unduly carcinogen. and even oxygen itself) can produce upper bound on low-dose risk. influenced by the selection of doses. a mutagenic yield in cellular DNA at
40 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 41 Chapter 4: Exposure assessment
•• a range of exposure factor data The information in this chapter, and in Table 6: Explanations of terms used in assessing dose and exposure 4.1 relevant to Australia the complementary Australian exposure INTRODUCTION •• where appropriate, default point factor guidance document, includes Term Explanation estimates or, in some cases, probability material published in some key US and Bioavailability A generic term defined as the fraction of a contaminant that is absorbed into the body following dermal contact, Exposure assessment requires a distributions of exposure data for use IPCS guidance documents on exposure ingestion or inhalation. It is expressed as the ratio (or percentage) of the absorbed dose (systemic dose) to the determination of the magnitude, in exposure assessments. assessment, including: administered dose. frequency, extent, character and duration •• Public health assessment guidance Absolute bioavailability The mass of a contaminant that is absorbed and reaches systemic circulation following dermal contact, ingestion of exposures in the past, currently and in Basic elements to consider in planning an manual (1992), United States Agency or inhalation. the future. There is also the identification exposure assessment are: for Toxic Substances and Disease Relative bioavailability The comparative bioavailability of different forms of a chemical or for different exposure media containing the of exposed populations and potential Registry (ATSDR) exposure pathways. Environmental •• Purpose: the reason the study is being chemical expressed as a fractional relative absorption factor. In the context of environmental risk assessment, relative monitoring and predictive models can be undertaken and how the results will be •• Guidelines for exposure assessment bioavailability is the ratio of the absorbed fraction from the exposure medium in the risk assessment (e.g. soil) to the used to determine the levels of exposure used. (1992), US EPA absorbed fraction from the dosing medium used in the critical toxicity study. at particular points on the exposure •• Scope: exactly what are the study •• EHC210 Principles for the assessment Bioaccessibility Generically, it is the ability for a chemical to come into contact with the absorbing surfaces in an organism. It is pathways. The contaminant intakes from areas, the population to be assessed, of risks to human health from exposure related to solubility and dissolution, since absorption usually can only occur from a liquid or gaseous phase, and not the various pathways under a range of compounds and media to be to chemicals (WHO 1999b) from a solid phase. It is defined as the fraction of a contaminant in soil that is soluble in the relevant physiological scenarios, including worst-case situations, measured. •• EHC235 Dermal absorption (WHO milieu (usually the gastrointestinal tract) and available for absorption. This can be assessed by validated in vitro test systems. There are only a few such test systems and these have been found to be applicable to only a limited can then be estimated (US EPA 1989). •• Level of detail: what level of accuracy 2006c) number of contaminants. In conjunction with bioavailability, it can be a significant factor determining the amount of a is required in the estimate of the •• EHC237 Principles for evaluating substance that might be absorbed from soil at a contaminated site. Exposure assessment is one of the exposure for this to be meaningful, health risk in children associated with Exposure Concentration or amount of a particular chemical that reaches a target organism, or system or (sub)population in more critical and complex areas of risk given the level of knowledge available exposure to chemicals (WHO 2006d) assessment. Due to the complexity and about the toxicological links between a specific frequency for a defined duration. Exposure is usually quantified as the concentration of the agent in the •• Exposure factors handbooks (2009b), scale of the EHRA process, a concise exposure dose-effect and risk. What medium integrated over the time duration of contact. US EPA. ‘cookbook’ on exposure assessment is are the resource constraints and how Exposure concentration The exposure mass divided by the contact volume or the exposure mass divided by the mass of contact volume, not practicable. Similarly, the issues are can resources be most efficiently used? depending on the medium. often sufficiently complex and ‘situation- It also includes exposure factor •• Approach: what methods will be used Exposure duration The length of time over which continuous or intermittent contacts occur between a chemical and the exposed specific’ that a manageable and complete information presented in the proceedings to determine exposure and do these of the five National Workshops for population. algorithm for decision making cannot accurately represent pathways of the Health Risk Assessment and Exposure event The occurrence of continuous contact between chemical and exposed population. be drafted. This chapter attempts to exposure that will affect risk. What is summarise useful guidance on exposure Management of Contaminated Sites, the nature of sample collection? (e.g. data developed from research by the Exposure frequency The number of exposure events within an exposure duration. assessment to assist the decision- How many are needed? From where? making process. Exposure assessment South Australian Department of Human Exposure route or pathway The way a chemical enters an organism after contact (e.g. by ingestion, inhalation or dermal absorption). How frequently?) How will the data be Services and data sources from the The pathway usually describes the course a chemical or physical agent takes from a source to an exposed organism. is identified as part of Phase II of the handled, analysed and interpreted (US expanded framework for EHRA outlined international literature. An exposure pathway describes a unique mechanism by which an individual or population is exposed to chemicals EPA 1992). or physical agents at or originating from a site. Each exposure pathway includes a source or release from a source, an in Figure 2. exposure point, and an exposure route. If the exposure point differs from the source, a transport/exposure medium This enHealth document, along with 4.2 (e.g. air) or media (in cases of inter-media transfer) is also indicated. Where possible, the information is the companion Australian exposure prescriptive about certain aspects factor guidance document aims to TERMINOLOGY USED IN Exposed population The people who may be exposed to the contaminant. Synonymous with ‘receptor’. of exposure assessment. Having assist with this process by collating and Dose The amount of a substance available for interaction with metabolic processes or biologically significant receptors after specific requirements for the content tabulating data that could be useful to EXPOSURE ASSESSMENT crossing the outer boundary of an organism. of investigations and having them estimate exposures in an EHRA, and to Potential dose Amount of a chemical contained in material ingested, air breathed or bulk material applied to skin. presented in uniform, coherent and fill any knowledge gaps where default The terminology used to define exposure and the factors that can influence the logically developed reports will enable assumptions may need to be made. Using Applied dose Amount of chemical in contact with the primary absorption boundaries (e.g. skin, lungs, gastrointestinal tract) and more efficient, accurate, timely and data from the Australian exposure factor extent of exposure by various pathways available for absorption. guidance document is discussed further are explained in Table 6. transparent decision making and a Internal (absorbed) dose The amount of a chemical penetrating across an absorption barrier or exchange boundary via either physical or greater consistency of environmental in Sections 4.13 and 4.14. However, it biological processes. health decision making across Australia. is emphasised that inputting data based on valid measurements of parameters Target tissue (biologically Amount of chemical available for interaction with any particular organ or cell. The aim is to provide: describing the exposure scenarios under effective) dose consideration are always preferable to •• details on conducting appropriate using the default assumptions that are a The terminology relating to dose in Table 6 is represented graphically in Figure 17. exposure assessments common feature of Tier 1 assessments.
42 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 43 Figure 17: Representation of dose and exposure and different forms of a given metal can •• pollution impact assessment at existing be absorbed to a different extent. For 4.3 facilities example, a metal in contaminated soil PLANNING AN EXPOSURE •• changes to climate, landform, may be absorbed to a greater or lesser geography or demography that extent (but generally somewhat lesser) ASSESSMENT may impact on disease vectors and than when ingested in drinking water parasites (US EPA 2007a). 4.3.1 identifying release of a chemical or •• situations where environmental standards or guidelines are unavailable The overall dermal bioavailability from agent to the environment soil should be chosen on a chemical- •• environmental changes that will A chemical or other agent will be specific basis following a review of the increase traffic flow and may increase released to the environment from a scientific literature. The US EPA has the risk of injury or air pollution, such facility, situation or process in a variety published default dermal bioavailability as new traffic corridors of ways. The first step in planning an Adapted from: US EPA 1992. factors for 23 chemicals (US EPA 2007a). •• changes that may impact on the exposure assessment is to work out Additional dermal bioavailability and microbiological or chemical safety of how the chemical or agent gets into the bioaccessibility data is detailed in Section food chains and food supplies It is normal practice in Australia to from the media consists of two oral bioavailability of a chemical from environment. 3.4 of the Australian exposure factors •• situations where there is a high level estimate the exposure to a chemical major processes, bioavailability and environmental media will be at least the guidance document. Toxicity studies As discussed in Section 1.2 examples of of public interest in or concern about or other agent that people are likely to bioaccessibility. Together these terms same as the bioavailability of the chemical generally use applied dose rather than when EHRAs may be undertaken include: environmental health issues receive for a specific situation, process represent the amount of a substance in toxicity experiments underlying the actual measures of the absorbed dose •• situations where vulnerable or facility. This is then compared that may reach the systemic circulation derivation of the guideline value (GV) and a default bioavailability of 100 per •• new additives to food or potable or populations may be affected by with the dose likely to not cause an of a human ‘receptor’ following exposure (i.e. a relative bioavailability of 100 per cent was assumed for oral studies of recreational waters environmental health issues such as effect (estimated in the effect or to an environmental contaminant. cent) (UK EA 2009, US EPA 2007a). highly water-soluble substances. The toxicity assessment phase) in the risk Bioaccessibility refers to how much of •• new and existing chemicals the placement of schools NRC (2003) prefers to use the term assessment characterisation step to determine if the the chemical dissolves into bodily fluids, Oral bioavailability relates to the fraction of ‘bioavailability processes’ to encapsulate •• legislative or policy changes •• contaminated sites assessment situation or facility poses an unacceptable such as inside the stomach, enabling the an orally administered dose of chemical the mechanisms involved in the •• designating housing setbacks from risk to people. This is likely to be chemical to reach and cross biological that reaches the systemic circulation dissolution, transport and absorption of •• assessment of major planning industry and transport corridors a conservative way to address the membranes and enter the circulatory (RIVM 2009). environmental contaminants by a receptor developments assessment of risk given that it is likely system. Bioavailability then refers to how •• where health impact assessments are organism. •• assessment of hazardous to overestimate the target tissue dose much of the dissolved chemical can cross undertaken. The term ‘relative bioavailability’ developments for the actual organ or cell where effects the absorption barriers. refers to a comparison of absolute Some factors that influence bioavailability will occur. bioavailabilities. Therefore, it is the ratio are listed below. Essentially: bioavailability = of the bioavailability of a substance in one 4.2.1 bioaccessibility × absorption exposure context (i.e. physical chemical Significance of bioavailability and matrix or physical chemical form of the – mass distribution of soil on the skin surface bioaccessibility The significance of understanding substances) to that in another exposure bioavailability and bioaccessibility in context (commonly an administered dose – mass of chemical in the soil matrix Many substances are able to tightly bind assessing exposure is discussed in more in an experimental animal study). to environmental matrices such as soil detail in the Australian exposure factors – soil properties (e.g. particle size, moisture) Bioaccessibility or sediment. In most situations, there guidance handbook, with chemical- Estimates of bioaccessibility and overall Bioavailability will be little or no information about specific data summarised where it is bioavailability (i.e. bioaccessibility plus – properties of the soil-bound chemical agent the bioavailability of a contaminant in available. Much of the following text is absorption) can be determined from the situation under investigation. It is, extracted from Sections 3.4 and 4.1 experimental studies – in vitro systems – environmental conditions therefore, normal practice to assume of that companion document, and mimicking biological conditions for 100 per cent bioavailability. If reliable definitions of the various terms describing bioaccessibility estimates and in vivo – properties of the skin Absorption through skin information is available it can be used bioavailability and bioaccessibility are (whole animal) models for bioavailability. – the residence time of the soil on skin to justify the use of a value other than defined in Table 6. 100 per cent. The companion Australian The processes of bioaccessibility and exposure factors guidance document Oral exposure is commonly the main route absorption affect the bioavailability of all includes more extensive guidance on for entry of chemicals into the body. Given chemicals from environmental media, where specific bioavailability data may its importance, in the absence of specific but are of special importance for metals. be available. bioavailability data, it has been common This is because metals can exist in a practice to conservatively assume the variety of chemical and physical forms, The bioavailability of the substance
44 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 45 In all these situations, chemicals may Fate of a chemical or other agent A summary of exposure pathways or 4.3.4 CSMs are site- and scenario-specific •• characteristics of the exposed be released directly into food, air, water, describes the reactions it may undergo routes and their interrelationships is Identifying potentially exposed and describe the pathways by which populations – exposed populations soil or waste. They may also indirectly or once released. A chemical can react shown in Figure 18. Interventions to populations chemicals transfer from environmental may be humans residing or working accidentally be released to these media. with other chemicals like humic reduce exposure may be made at a sources (e.g. soil, groundwater, airborne at the site or adjacent areas, acids, be broken down by chemical single point or multiple points. To potentially pose a risk it is necessary emissions from an industrial facility). future occupiers of the site after In each of these situations, consideration processes like hydrolysis or photolysis, to have an exposed population for the A CSM may be based on diagrams or redevelopment, or environmental of how the chemical or other agent is be broken down by micro-organisms A fundamental concept of risk assessment agents of concern. Part of the information flow charts, but it must be supplemented populations such as ecosystems used and how it may be released to the in the environment, or persist for an is that there must be plausible evidence of to be collected in developing a conceptual with detailed information on COPC in receiving environments such as environment is required. How often, extended period. What actually occurs an exposure pathway linking the source of site model is whether people may live concentrations in various media, natural surface waters. how much, what happens in unusual in a specific situation will depend on the contamination and the exposed population. or interact with the agent of concern transfer characteristics, and receptor circumstances like plant malfunctions, characteristics of the environment into Where this linkage exists, an assessment or with media contaminated with the characterisation. CSMs are particularly important in EHRA and how variable the release rate may be, which it is being released. of the nature and significance of the agent of concern. Often this pathway is of contaminated sites but they are also all must be considered. exposure pathway is required to determine quite obvious but the possibility of less A CSM is generally a written description useful for elaborating exposure pathways 4.3.3 the level of risk and the extent to which the obvious exposure pathways should not of the site that is accompanied by a associated with airborne pollutants from 4.3.2 Identifying exposure pathways proposed pathway is ‘complete’. be overlooked. schematic, graphical interpretation industrial sites. Identifying fate and transport of a that depicts what is known or has been Exposure pathways are those processes A pathway may be considered ‘complete’ There are situations, however, where it inferred about the site. CSMs are an 4.4.2 chemical or agent that take a chemical or other agent from where there is documented evidence of can be quite difficult to establish whether important tool for visualising the pathways Examples of CSM diagrammatic Once a chemical or other agent reaches its point of release to the environment a source (i.e. presence of one or more people might be exposed. When the by which human exposure to chemicals representations the environment, its fate and transport through to a situation where a person can COPC at the site under consideration behaviour is rare or limited in extent in the from a variety of environmental sources needs to be considered. be exposed. So it may be direct exposure for the EHRA); there is evidence that general population establishing whether may occur. The CSM usually includes There are many ways of depicting a because a chemical may be added to food the COPCs are actually released from there are actually likely to be people who diagrammatic representations of elements CSM using diagrams and flow charts. Transport away from the release point which is then consumed. It may be indirect their sources; that there are transport meet the criteria can be difficult. For of the pathways between source and Two examples are shown in Figures 19 is fairly obvious. If released to air, then exposure where a chemical is released to pathways and mechanisms that could example, determining whether a particular receptors, but it must always include and 20. the air containing that chemical or other the environment and people are exposed convey the COPCs from source to the location is actually a spot where people more detailed textual descriptions of the agent may be blown away and diluted. at some temporal or geographic distance various sites where ‘receptors’ are like to go fishing and so could be exposed characteristics of each source, pathway If released to water in a river, then the from the initial release point via an located; and that the potential for via consumption of the fish they catch. and receptor. chemical may flow away and be diluted exposure pathway consisting of more than human contact has been established quite a distance while release to a lake or one step. Often the exposure pathways for contaminated environmental media 4.4 It is important to be clear whether an tiny creek may stay close to the release will be fairly obvious, but there may be a (air, water, soil food, surfaces) at each exposure pathway is ‘completed’ (i.e. point for an extended period. If released range of situations such as the movement point in the transport chain. DEVELOPMENT OF A there is reasonable evidence that there to groundwater then it may not flow far. of contaminated groundwater or volatile are ‘receptors’ who would actually be If released to soil or waste, then it may be chemicals from contaminated groundwater The development of a conceptual site CONCEPTUAL SITE MODEL exposed in the given scenario) or whether trucked away from the original source and that are less obvious. These less obvious model (CSM) may be quite useful the exposure pathway is ‘potential’ (i.e. it reach a whole new location. pathways also need to be evaluated. in identifying and quantifying the Once all the information discussed may need to be considered in a holistic exposure pathways. above is collected, it should be collated EHRA), but where the contribution to together to enable the development of a overall exposure may be so slight or conceptual site model. limited that it would have little impact on Figure 18: Exposure pathways the risk estimate. 4.4.1 Conceptual site models A detailed conceptual site model should Developing a conceptual site model include information on the following: (CSM) can materially assist the process •• the contaminants – concentration, of understanding how human ‘receptors’ distribution and media in which they may be exposed to chemicals from occur (soil, water, sediment or air) relevant environmental sources. The CSM •• physical characteristics of the describes the sources of contamination, environment for contaminated sites – the pathways by which contaminants including soil type, porosity, potential may migrate through the various preferential pathways, vadose zone environmental media and the populations thickness, groundwater gradient and (human or ecological) that may potentially Adapted from: McKone 1993. velocity, and hydraulic conductivity of be exposed (NEPC 2010). the saturated zone
46 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 47 Figure 19: Representation of a CSM flow chart for potential airborne exposures from an industrial site Figure 20: CSM site diagrams representing potential exposure pathways from a contaminated site
Reproduced with permission from the contaminated sites NEPM Schedule B(7) (NEPC 2010).
48 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 49 4.5 be made to accurately characterise the on the use of modelling in exposure 4.5.2 Each of these methods is a separate •• the time frame (e.g. retrospective, population or individuals for whom the assessment. Determinants of exposures entity using different information, and current or prospective). APPROACHES TO exposure assessment is relevant. so each one can be useful in verifying or 4.5.1 The principal determinants of the level of validating the results of other methods Modelling may be used to estimate exposure are: QUANTIFYING EXPOSURES Direct measurement of the exposures Measurement of exposure (ATSDR 1992). the concentration that people may be of the (potentially) affected population •• concentration of the agent in the exposed to when measurement is not Accurate and useful exposure assessment An initial requirement for exposure provides the best exposure data relevant medium Commonly, chemical levels will be practical or possible in the time frame assessment is an understanding of the but this is not always available or requires a detailed understanding both •• exposure duration measured at the point of release to the required. See Chapter 13 presence (or absence) of an agent and its practicable and default exposure factor of the strengths and weaknesses of the environment, as this is likely to be the concentrations and distribution, including data are often required. exposure assessment techniques, and •• exposure frequency point where concentrations are highest The initial release of a chemical may any fluctuations over time. Guidance on (Langley 1993a p. 90) the specific exposure factors used in the •• exposure fluctuations (continuous or and so it will be the easiest to measure. be modelled for facilities that are yet sampling and analysis of environmental assessment. intermittent) Such data may also be available from to be built. These will be based on the media is summarised in Chapter 8. Figure 21 outlines the integration of •• whether exposure pathways are monitoring required by regulation. More engineering of the facility and the way Direct measurement of the exposures direct and indirect measurements of completed or potential. information is provided in Chapter 8. chemicals are to be used. Accurate and useful exposure assessment exposure. Most EHRA processes rely on of the (potentially) affected population requires a detailed understanding of both provides the best exposure data, indirect estimation of exposure, using Monitoring environmental media can It is important to have a good Measured data for the release of a the strengths and weaknesses of the but this is not always available or environmental monitoring data and provide useful indications of exposure understanding of the strengths and chemical from a facility may be available exposure assessment techniques, and practicable (except perhaps at the Tier 3 models to quantify chemical transport levels, providing there is a strong weaknesses of the sampling design but models may commonly be used to the specific exposure factors used in the level) and default exposure factor data is through the identified exposure pathways. correlation between environmental media and the analytical methods used in any describe the transport of the chemical assessment. Considerable effort needs to often required. Chapter 13 outlines further information levels of exposure and personal doses. measurement. away from the point of release, such as air dispersion models, or to describe the An understanding of transport and fate fate of the chemical in the environment Figure 21: Components of exposure assessment A powerful method of direct exposure assessment is using biomarkers as part of models for the agent(s) in question is through consideration of half-lives, effect a biological monitoring process, but this also important. Transport and fate will be of organic carbon or other characteristics. approach has its limitations. The role of affected by (Fiksel & Scow 1983): In developing sampling plans for chemical biomonitoring is discussed in Chapter 14. •• environmental exposure medium (e.g. agents and assessing exposure, an air, surface water, soil, groundwater or understanding of the movement of 4.5.3 biota) Quantification of exposure chemical agents within and between •• geographic scale (e.g. global, national, environmental compartments and the The quantification of exposure can be regional or local) effects of environmental partitioning will done in three ways: •• pollutant source characteristics be necessary (see Section 4.9.1). •• The exposure can be measured (e.g. continuous, intermittent or at the point of contact (the outer instantaneous releases from industrial, boundary of the body) while it is taking residential and commercial point or place, measuring both exposure area sources) concentration and time of contact •• the nature of the risk agent (e.g. and integrating them (point of contact whether it is a specific agent or group measurement). of agents) •• The exposure can be estimated by •• the receptor population (e.g. humans, separately evaluating the exposure animals, plants, micro-organisms and concentration and the time of contact, habitats, as well as specific sub- then combining this information populations exposed to high levels (scenario evaluation). of the agent or who are particularly •• The exposure can be estimated sensitive to exposure) by dose, which in turn can be •• exposure routes (e.g. ingestion, dermal reconstructed from internal contact or inhalation) indicators (biomarkers, body burden, •• environmental conditions (e.g. excretion levels) after the exposure pH, presence of organic matter, has taken place. clay content, temperature and meteorological) Adapted from: National Academy of Sciences (NAS) 1991.
50 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 51 information. Examples are: empirically derived, the latter are usually time span of the life stage; for example, chemical characteristics of the substance 4.6 preferable providing the experimental for young children from birth to the it may be difficult or impossible to EXPOSURE ASSESSMENT For oral ingestion: conditions are appropriate. sixth birthday the AT is 6 years (US accurately assess past exposures (e.g. EPA 1989). to volatile hydrocarbons in the surface C (mg/kg) x × AoF x × IGR (mg/d) x × EF (d/yr) x × ED (yr) x × CF (10–6) CALCULATIONS I (mg/kg/d) = Volatilisation of organic compounds from stratum of soil). It may only be possible 365 (days in a year) × AT (yr) × BW (kg) water may need to be taken into account. 4.7 to make some crude classification of A generic formula for calculating chemical In this case, AoF is an oral absorption factor or bioavailability estimate (unitless) and Blando and Cohn (2004) describe an exposure as ‘high’, ‘medium’ or ‘low’. intake from various media has been ingestion rate (IGR) of the medium (e.g. food, soil, water) replaces CR. For water intake, approach using equations defined by the DATA ANALYSIS AND A ‘crude dichotomy’ of exposure may be formulated and described in US RAGS-A the units for C and IGR are expressed in mg/L and L/d respectively. US EPA (1994). EVALUATION all that is possible (e.g. ‘exposed’ versus guidance on risk assessment (US EPA ‘not exposed’). 1989). The generic formula is: For inhalation of volatiles: The averaging time (AT) will depend on 4.7.1 3 3 the nature of the adverse effect (e.g. Errors in estimating historical exposures C × CR × EFD × 1 × CF C (mg/m ) × IR (m /h) × LR × ET (h/d) × EF (d/yr) I = I = acute or chronic) being assessed and Nature of the exposure assessment may have significant consequences for BW AT 365 × AT (yr) × BW (kg) the exposure scenario (e.g. short periods Risk models for carcinogens, in particular the accuracy of the risk characterisations. I = intake of chemical (usually expressed IR = inhalation rate; LR is a lung retention factor (unitless); ET = exposure time or long term) predicted from the issues that of the US EPA, use lifetime time- If the exposure estimates are being used as mg/kg bw/day) identification phase of the risk assessment weighted average doses to determine in epidemiological studies it may mean Note that this equation is based on the US EPA RAGS-A approach (US EPA 1989). or the relevant conceptual site model. the dose–response relationships. This is that it is impossible to determine whether The most recent US EPA guidance (RAGS-F) proposes that there is no need to calculate based on using data from lifetime animal there is a dose–response relationship or C = average chemical concentration in an intake rate based on concentration, inhalation rate and lung retention. RAGS-F •• For non-threshold adverse effects studies to determine dose–responses and an accurate level of association between media over the exposure period (e.g. recommends calculating a modified exposure concentration (EC), which is then (e.g. genotoxic carcinogenesis) and mg/L, mg/kg or mg/m ) hence slope factors. Consideration must the level of exposure and a health effect 3 compared directly with the RfC derived from toxicological studies. threshold effects where exposure be given to the nature of the exposures (US EPA 1992). may be assumed to be over a lifetime CR = contact rate the amount of that occur, as in many instances Using the RAGS-F approach (which is the recommendation of this updated enHealth (e.g. via food or drinking water) the contaminated media contacted per unit exposures may be episodic or quite US EPA (1992) provides several document), the equation converts to: AT value most commonly used is 70 time or event (e.g. L/day) variable. In some instances, an exposure references detailing approaches for 3 years. It should be noted that where Exposure concentration (EC) = C (mg/m ) × ET (h/d) × EF (d/yr) at a critical period may be of more determining and estimating past EFD = exposure frequency and duration the exposure is less than lifetime, this AT (yr) concern than average exposure over a exposures. (how long and how often exposure occurs) calculation may underestimate risk 3 long period. EFD may be based on the product For inhalation of dusts I = C (mg/kg) × IR (m /h) × ET (hr/d) x × EF (d/yr) if the effect does not depend upon of two parameters 365 × AT (yr) × BW (kg) × PEF (m3/kg) lifetime exposure. 4.7.3 If exposure ends, this must be taken Dealing with data gaps EF exposure frequency (e.g. days/ PEF = particle emission factor •• For adverse effects considered to into account in the exposure assessment, year) and ED exposure duration exhibit a threshold, the AT depends as it may mean risk ceases (e.g. the Constraints in time resources and For dermal contact with soils (e.g. years) on the nature of the toxicity data and risk of trauma when hazardous machinery money always lead to data gaps. Ways the assumed period of exposure. of addressing these data gaps include BW = body weight, usually averaged over C (mg/kg) × AH (mg/cm2/d) × SA (cm2) × AF × EF (d/yr) × ED (yr) × CF (10–6) is fenced) or decreases (e.g. the I = Where a shorter period better reflects (US EPA 1992): the exposure period (e.g. kg) 365 × AT (yr) × BW (kg) decreasing risk of lung cancer related the expected exposure paradigm, to the period since smoking ceased). •• collecting new data – this is the AT = averaging time period over which the AH = soil adherence; SA = surface areas of skin exposed; AF = skin absorption factor a shorter averaging time may be In these situations, account must be preferred option, although it must exposure is averaged (e.g. hours, days, used. For example, where exposure taken of the influence of these episodic be recognised that the time taken months, years) For dermal contact with water at a domestic dwelling may be exposures rather than using a lifetime to collect new data may compromise 2 2 from contaminated soil, vapour CF = conversion factor, if units in above DAevent (mg/cm /event) × SA (cm ) × EV (events/d) EF (d/yr) × ED (yr) average exposure. getting a resolution of the problem I = from contaminated groundwater or parameters don’t match 365 × AT (yr) × BW (kg) if the need for a risk assessment emissions from a local industry, an DA = dose absorbed per event; EV = event frequency (events/d); EF = exposure 4.7.2 is urgent The first term in the above equation event AT value of 30 years (the average Assessing past exposures •• using models to estimate exposure calculates an intake based on frequency frequency (d/yr) period of residency) may be values and duration of exposure, adjusted for considered suitable (see Section 7.1 There are often considerable difficulties body weight, while the second term The above equation is the first of a series provided in RAGS-E guidance from the US of the Australian exposure factor in assessing historical exposures. It •• inserting conservative assumptions adjusts the intake on the basis of the EPA (US EPA 2004b; Eq 3.1) for chemical intake via water exposure. Equations are guidance handbook). Similarly, ATs may be possible if there have not been – however, the assessor should be selected averaging time, and incorporates also provided for DAevent, the choice of equation depends on the relationship of the for occupational exposures may be disturbances of the environmental media aware of the flow-on consequences an adjustment where units in the exposure time (ET, in the DAevent equations) with the time to reach steady-state water restricted to the anticipated time and and the substance is relatively inert (e.g. of utilising conservative assumptions, individual terms do not match. concentrations. In estimating intake of organic chemicals from contact with water, frequency of exposure, or assumed the amount of lead in soil is unlikely particularly a series of such the risk assessor may also wish to incorporate consideration of bioavailability (skin working lifetime in a particular to change). However, in a situation assumptions Where specific routes of exposure are absorption) (Eq 3.8, US EPA 2004b). Reference values for many of the equations industry. For risk assessments in where there has been disturbance •• using professional judgement – considered, this basic equation may be parameters are available for a large range of compounds in Appendix B, US EPA which exposure may be for a particular of the environmental media (e.g. soil although this should depend on modified to incorporate route-specific (2004b); it should be noted however that many are calculated rather than being life stage, the appropriate AT is the movements) or changes to physico- extensive experience rather than
52 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 53 anecdotal information and the assessor 3. Robust methods: The robust method •• Has the appropriate dose for use in the physiological or metabolic factors. It may Partitioning will reflect the fact that must account for such judgements in combines the observed data above dose–risk relationship been identified? not be possible to identify such people 4.9 substances tend to move to the the uncertainty analysis the detection limit with extrapolated It must be ascertained whether and for this reason the conservatism ISSUES IN EXPOSURE environmental compartment for which •• narrowing the scope of assessment below-limit values to compute summary reference data applies to absorbed/ built into EHRA should provide they have the most affinity (Calamari if the data gaps appear to be in one statistics. In contrast to the distribution internal/potential/applied/effective protection for more susceptible sub-groups. ASSESSMENTS 1993; 1999). Transformation may occur pathway or exposure route – this method, the actual data above the doses and the relevant data is used High-exposure people are usually defined in any environmental compartment. All exposure pathways must be is likely to be the least satisfactory reporting limit is used to fit a distribution appropriately in the model. This may as being ‘above the 90th percentile of the considered for health risk assessment, option if such a route or pathway is rather than assuming a distribution. require allowances for variations in population distribution, but not higher than Fugacity modelling (assessment of although one exposure pathway may important, or if there is insufficient bioavailability. Bioavailability data from the individual in the population who has the escaping tendency of a chemical the highest exposure’ (US EPA 1992 be dominant. As the total amount of a knowledge to determine the This method has the advantage that animal studies should not necessarily from one environmental phase to p. 22901). chemical absorbed by a person’s body significance of the pathway. estimates of extrapolated values can be be considered to be similar for another) enables an estimation of directly retransformed and summary humans. influences the risk to health, exposure which compartment will contain most statistics computed in the original units, There is a need for caution when applying assessment must take into account of the agent and where the highest 4.7.4 Given that these two factors have been thereby avoiding transformation bias. the concept of ‘worst case exposures’. all sources of exposure irrespective of concentrations in the ‘unit of world’ are Dealing with censored data adequately resolved, estimates of Also, this method is not as sensitive to These exposures are often based on the whether these are from food, water, the (Mackay 1991). Mackay’s ‘unit of world’ exposure, dose and risk can then be Heyworth (1991) provides a summary of the fit of the distribution for the largest accumulation of a range of unlikely but workplace, outdoor air, or a combination is a hypothetical box 1 km square and undertaken for individuals or narrow the three essential methods for dealing observations because actual observed individually plausible scenarios. Such of these and other sources (Langley 6 km deep that includes air, terrestrial sub-populations. Often the high-end risk with censored data: data is used to fit the distribution. worst exposure cases are often worse than 1991a; IEH 1999a). and aquatic biomass, soil, water and will need to be estimated because this any remotely plausible case because they sediment. Because environmental will often be a driving force for any risk 1. Simple substitution methods: Simple The probability plotting method can represent a ‘hypothetical individual In large-scale contamination (i.e. data is typically distributed on a log management measures. substitution methods refer to those used to fit the distribution in robust and an extreme set of conditions [that] regional), more exposure pathways will scale, the approximate estimates methods that substitute a single value, methods can be computed quite will usually not be observed in actual be involved than in small-scale (very provided by fugacity models are very The US EPA points out ‘the high end populations’ (US EPA 1992 p. 22901). such as one-half the detection limit readily by most commercially available localised) contamination. useful for distinguishing where a segments of the exposure, dose and The term ‘worst case exposure’ is to be for each censored value. While these statistical packages. The US EPA contaminant is likely to reside, even if the risk populations may represent different contrasted with the more practical term methods are commonly used they website includes ProUCL 4.0 software Children usually receive a higher absolute distribution estimates are not individuals’ (US EPA 1992 p. 22921). ‘maximum exposed individual’, which have no theoretical basis. The choice for managing data that contains exposure to environmental agents per completely accurate. Simpler fugacity This is due to variations in bioavailability describes ‘an individual that does, or is of the substitution value is essentially data requiring censoring along with unit body weight than adults because of models assume equilibrium, which in absorption, intake rates, susceptibility and thought to, exist in the population’. arbitrary and the estimates of summary technical support documents (Singh behavioural and physiological factors (e.g. many cases does not apply, but the other variables between the segments of statistics will be biased by these et al. 2007). See
54 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 55 •• dispersion models, which predict In relation to modelling issues of volatiles Currently, field monitoring data is the most Figure 22: Illustration of potential variability in exposure patterns reductions in concentrations from in groundwater, there are more extensive appropriate to use in assessing exposures point sources based on assumed discussions of such modelling in: to volatile substances. Environmental mathematical functions or dispersion fate and modelling characteristics properties of the chemical •• Review of schedule B(4) of the present problems for using short-term contaminated sites NEPM (NEPC •• transport models, which predict field monitoring data. This is particularly 2010) concentration changes over distance marked for the decay of exposures to and can represent dispersion, •• CRC-CARE technical reports on the finite sources of volatile substances. The biochemical degradation and development of health screening validity of assumptions of biodegradation absorption. levels for petroleum hydrocarbons in and finite source are difficult to assess contaminated soil and groundwater without an empirical database showing 4.9.2 (CRC-CARE 2009) paired measurements of indoor air versus sub-slab and soil concentrations that test Environmental persistence •• Evaluating vapor intrusion pathways at hazardous waste sites (ATDSR 2006) the applied assumptions and exposures The terrestrial, aquatic and atmospheric in building developments on potentially •• OSWER draft guidance for evaluating fate of agents needs to be considered contaminated land. the vapor intrusion to indoor air Figure 22 shows some of the potential variations due to short-term conditions or This equation is equivalent to the area as part of the exposure assessment. pathway from groundwater and soils exposure scenarios that may occur activities can be missed. under the exposure curve. The area The agents may be relatively inert (e.g. The failure rate of vapour intrusion models (subsurface vapor intrusion guidance) over a period of time. It demonstrates under the dose–response curve (AUC) asbestos) or subject to biodegradation (as routinely applied by consultants in (US EPA 2002b). This document is the difficulties in describing a variety of The issue of whether estimates of long- that measures the total delivered dose to and abiotic degradation. Persistent Australia) is currently unknown. To avoid expected to be updated by late 2012. exposure scenarios with a single number. term or short-term exposure should be a target tissue is an analogous concept. substances, or those with long half-lives potential public health and financial used in the risk assessment may also be However, not all substances follow this in the environment or biota, may provide impacts, vapour modelling should be For groundwater the commonly used If fluctuations in exposure are relatively informed by the nature of the predicted simple relationship, and a more general opportunities for exposure to increase appropriate and reasonably conservative. models (e.g. the Johnson and Ettinger large, the risk assessor may need to health effects. For example, short-term exponential relationship of: over time, such as if there are movements Demonstration of conservatism may be (1991) model) for assessing vapour consider whether averaging the exposure exposure estimates (particularly peak in the population, as can occur with achieved by analysis of model uncertainty intrusion into indoor spaces from soil over a relevant exposure period could exposures) may be more relevant where Cnt = K residential redevelopments of an old in which sensitive parameters are or groundwater combines convective underestimate risk, especially where an acute effect (e.g. sensory or mucosal industrial area. identified and the influences of changes and diffusive parameters into a one- acute effects associated with peaks in irritation) is the effect driving the risk in model input values are compared n is a chemical-specific parameter that dimensional model describing these exposure could result in more critical assessment. The source of the data and the relevant with the output from the assumptions is also specific for a specified health transfer functions. Some of the limitations effects on health. environmental comparisons with of an exposure scenario considered endpoint. of these models have been discussed Australian conditions should be taken into realistic. The risk manager may then 4.11.2 by Turczynowicz and Robinson (2007). account. For example, Australian soils choose a modelling scenario that provides 4.11.1 Adjusting exposure duration This is required to describe the effects They can overestimate or underestimate and climatic factors may result in different an appropriate level of precaution. Short-term versus long-term of concentration and exposure time for indoor air concentrations depending It is possible that toxicological data against environmental persistence for some Arguably, the ultimate test of modelling exposure data some toxicological endpoints (Ten Berge on whether attenuating factors such which inhalational exposures is to be pesticides in Australia compared with appropriateness is collection of indoor air et al. 1986). as biodegradation, or finite versus Short-term data applies to data benchmarked has been developed from North American or northern European and measurement of the contaminants of infinite sources, are taken into account. collected over minutes, hours, days studies where the duration of exposure conditions. concern. This of course requires attention Guidance on using such data for setting Under-estimation may occur when or months. Long-term data refers to is different to the exposures modelled or to sampling and statistical issues to air quality standards (NHMRC 2006) factors such as preferential pathways years or ‘lifetimes’. Long-term exposure expected in an environmental health risk ensure confidence in the resulting data. recommends that, as a general rule, 4.10 and non-diffusive vapour ingress (e.g. estimates based on a ‘snapshot’ of data assessment. The conventional approach averaging times should be consistent advection) are addressed. The Johnson at a particular point in time may not to adjusting the toxicological benchmarks with or, where practical, shorter than the EXPOSURE ASSESSMENT and Ettinger model has been used 4.11 accurately represent the fluctuations that is the application of Haber’s Law, or a elicitation time for the effect of concern. OF VOLATILE AGENTS extensively in Australia. For example, occur during exposures lasting years or variation of that approach. This assumes This is an appropriate and conservative with appropriate modification, it was EXPOSURE DURATION a lifetime. Populations are not randomly that concentration and time of exposure are approach for public health protection. used for the assessment of petroleum mobile or static, and using short-term equally important in producing the effect. Exposure assessment for volatile Therefore: hydrocarbon vapour intrusion in the CRC- Exposure is rarely constant over time. data to estimate long-term exposures agents may need to be undertaken CARE development of health screening It can vary substantially over periods tends ‘to underestimate the number of Haber’s Law states the product of •• for compounds whose effects require under a variety of circumstances and levels (HSLs) (CRC-CARE 2009) and in ranging from a few minutes, to hours, people exposed, but to overestimate the concentration (C) and the time of chronic exposure, the averaging time media (e.g. soil/landfill, water, spills, the revision of the HIL for total petroleum days or weeks. Variation is common the exposure levels to the upper end of exposure (t) is equal to a constant level should be a year consumer goods) and will often depend hydrocarbons (TPH) in the contaminated where the source emits the COPCs the distribution even though the mean or severity of response (K) for a specific •• for pollutants whose effects are rapid on modelling. It is beyond the scope sites NEPM (NEPC 2010). intermittently, e.g. from the stack of an will remain the same’ (US EPA 1992 toxicological effect. Therefore: in onset and/or readily reversed, the of this enHealth document to discuss industrial facility. p. 22917). Conversely, the fact that averaging time should be of the order modelling of volatile agents in detail. long-term data often tends to even out C × t = K of days, hours or less. exposures can mean that significant
56 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 57 Ideally, the averaging time will be exposure characteristics it is necessary other population exposures to risk Table 7: Summary of suggested exposure factors for adultsa determined from the experimental data to apply appropriate adjustment factors agents (Covello & Merkhofer 1993). used to identify and measure the key to the exposure assessment. Where Where specific data is unavailable (or Suggested value adverse effect. However, issues such the inhalational exposure route involves inadequate) to quantify exposures by all Parameter Average (95th percentile)a Units Comment as the practicality of measurement, the children, note the previous advice on the relevant pathways, the risk assessor Anatomical and physiological parameters (Sections 2.1.4 & 2.2.4)b desire for uniformity or the possibility of EHRA-proposed adjustments, taking usually relies on modelling or other means 78 (107) kg M & F combined* short-term spikes within the averaging into consideration the different breathing of calculating the exposure amounts period may influence the final regulatory rates and volumes applicable to children, and fluxes. Such exposure models and 85 (114) kg M Body weight time frame assigned to an air standard. as well as their activity patterns. Guidance calculations may, in turn, need to make 70 (100) kg F on the selection of suitable respiratory extensive use of default values to estimate For acute systemic adverse effects, rates, volumes and activity patterns for the various model inputs where direct 70 kg Lifetime average M & F combined toxicokinetic information on the active children is summarised in Sections 5 measurements are absent. 169 (181) M & F combined compound (either the parent molecule and 6 of the Australian exposure factor Body height 176 (188) cm M or metabolite) may inform selection guidance document. Tabulated data on human anatomical of an appropriate averaging time. For and physiological parameters and human 162 (174) F example, where an adverse effect that It should be noted that recent US activity patterns relevant to Australia has * Note these heights and weights may differ for uniform populations of a specific race (Sections 2.2.1 & 2.2.2)b is not dependent upon accumulation guidance on inhalational exposure for been compiled in the Australian exposure b of toxicity is identified in a clinical or children (US EPA 2009a) recommends factor guidance document (AEF). Dermal exposure parameters (Section 3.2.4) animal study, but is observed some a different approach that does not 20,000 (24,000) M & F combined. See Table 3.2.3b for surface area of specific body parts. time after steady-state blood or body- necessarily adjust exposures based on Historically, many of these types of Total skin surface area 21,000 (25,000) cm2 M burden concentrations are achieved, the different breathing and activity patterns. data have been sourced from various averaging time for a standard could be The basis for this recommendation is that international sources, although the 19,000 (23,000) F determined by adjusting the experimental US guidance on chemical-specific data amount of relevant Australian data 6,300 (7,900) M & F combined* observational period (often this is the for toxicity assessment recommends that, sources has been increasing. Australian Exposed skin surface 6,700 (8,100) cm2 M* same as the exposure period) according where appropriate (e.g. for mutagenic data that may provide a useful source of area to the half-life of the active molecule. carcinogens; see Section 5.8), early-life default estimates for air, water, soil and 5,900 (7,500) F* It takes approximately five half-lives to exposures be factored into the relevant food-based risk assessment have been * These defaults approximate the sum of forearms, hands, lower legs and feet. The actual exposed body parts should be used as in indicated by the achieve steady-state conditions for blood reference dose assessments. Where this juxtaposed with overseas data to allow exposure scenario. Section 3.2.4b or body burden concentrations of the is done, the US EPA considers that it is no an appreciation that not all overseas- pollutant. This means that an air standard longer necessary to adjust the exposure sourced data truly reflect the current Oral exposure parameters averaging time less than the equivalent estimates for age-related ventilation rates Australian population. In some cases, Lifetime tap water (i.e. community supply) intake. Includes water used in food of five half-lives will confer additional or body weights (see Section 4.6 for the it may be necessary to use defaults that preparation. Excludes commercially purchased bottled water and water intrinsic Drinking water intake conservatism if the numerical value RAGS-F exposure equations). are more population- and site-specific 2 to purchased food and beverages (i.e. milk). M & F combined of the standard has been established (e.g. air quality data specific to a region). L/d Less than lifetime tap water intakes. (gender-specific data on a NOAEL identified from the longer While it is recognised that some Stakeholder consultation may be useful (Water intake may be much larger with high activity ± tropical or arid areas). not available) 1.2 (2.8) experimental observational period. Australian air quality GVs-based inhalation in establishing such site-specific data. Can be used for pregnancy but 50% increase during lactation. risk assessments may have been carried (Section 4.1.3)b Further detail on the approach to adjust out using the previous approach, updated Another use of default parameter 1,400 M & F combined* inhalational exposure times, including enHealth guidance in this document estimates is in initial screening adjustment to the application of Haber’s recommends the RAGS-F approach be assessments or ‘back of the envelope’ Food intake 1,550 g/day M* Law, are detailed in NHMRC (2006). used from now on in Australia. appraisals to establish whether there is a 1,200 F* need to move to site-specific appraisals. * Average food intakes not including beverages (e.g. juices, tap water, coffee) but including milk for ≥19 yrs; upper intakes from recent Australian food surveys are not readily available. For intakes of individual food groups see Section 4.3.4b and Tables 4.3.1a,b,c.b 4.12 4.13 Tables 7 to 9 summarise the ADJUSTMENTS FOR DEFAULT VALUES recommended default parameters for Soil ingestion 50 (60) mg/day Section 4.5.3b adults, children and non-age-dependent Incidental water Average (upper estimate) SENSITIVE SUB- FOR EXPOSURE exposure factors from the Australian ingestion while 25 (125) mL/hr (Section 4.6.3)b exposure factor (AEF) document. Internal swimming POPULATIONS ASSESSMENTS references in these tables refer to relevant sections in the AEF. Identifying sensitive groups that may be As outlined earlier in this chapter, exposed to an agent of concern is often exposure assessment is the process of a critical component in an EHRA. Where measuring or estimating the intensity, such groups are likely to have different frequency and duration of human or
58 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 59 Suggested value Table 8a: Summary of suggested exposure factors for 2–3-year-old child (male and female combined) a Parameter Average (95th percentile) Units Comment Note: Values are average or 95th percentile (in parenthesis)a Inhalation exposure parameters Parameter Suggested default Units Comment and internal reference Inhalation rate M & F combined (Section 5.1.3)b. (long term-exposures) 15 (20) m3/day For specific inhalation rates by activity or for short-term exposures, see Section 5.1 Anatomical and physiological parameters (Table 5.1.2)b Body weight 15 (17) kg Section 2.2.4b
Activity patterns Body height 96 (106) cm Section 2.1.4b Average (upper estimate) Total time indoors 20 (24) hr/day Dermal exposure parameters (Section 6.2.3.2)b Total skin surface area 6,100 (7,000) cm2 Section 3.2.4. See Table 3.2.5 for specific body part datab. Time indoors Average (upper estimate) 2 b 20 (24) hr/day b Exposed skin surface area 2,300 (2,700) cm Section 3.2.3, 3.2.4 (at home) (Section 6.2.3.2) Approximate average value for Australian adults (Section 6.2.3.2)b. Oral exposure parameters Time spent outdoors 3 hr/day Upper estimate not available. Drinking-water intake 0.4 (1) L Tap water intake. Includes water used in food preparation.(Section 4.2.3)b. 0.5 For general population* Food intake (excludes Upper percentile not available. For intakes of individual food groups, see Section 1,100 g/day Time spent swimming hr/day beverages except for milk) 4.4.4 and Table 4.4.2b. 1.5 For people who swim regularly* 50, * Estimates for Australians assuming all outdoor sports activity is swimming (Section 6.2.4.3)b. Central tendency for outside soil. Soil ingestion (100), mg/day (Reasonable maximum, outside soil)
M = Male (adult) b [100] [Central tendency outside soil + indoor dust]. (Section 4.5.3) F = Female (adult) Incidental water ingestion 50 (~ average) a The summary tables provide suggestions for possible values for use in screening risk assessments. An average (i.e. central) and reasonable mL/hr Section 4.6.3b maximum value is provided. The latter is in parenthesis and when data permits is the 95th percentile, otherwise it will be an ‘upper’ estimate as while swimming 150 (~ upper estimate) indicated. In general, an ‘upper estimate’ is a reasonable maximum value. The specific sections in the AEH should be consulted for additional explanations. It is the ultimate decision of the risk assessor to choose the most appropriate value to use on a case-by-case basis. Wherever possible, Inhalation exposure parameters b data which is specific for the risk assessment scenario, chemicals and receptors of concern should be used ahead of the values in this table. 3 Section 5.1.3 . For specific inhalation rates by activity or for short term exposures, Inhalation rate 9.5, (15.9) m /day b When separate values for males or females are not provided in the summary tables the recent data used for generating the tables did not contain see Section 5.1; Table 5.1.2 this information. Older agency publications (e.g. from Australia, US EPA, Canada, the Netherlands) may have such data and the risk assessor Activity patterns should seek and justify the use of this information as needed. 13 (37) b The references to sections and tables refer to those in the Australian exposure factors guidance document. Frequency of hand to (indoors) contacts/ Section 6.1.1.3b mouth 5 (20) hr (outdoors) Mouthing duration Varies hrs/d Mean and maximum values differ by object mouthed (Section 6.1.1.3)b Upper estimate not available 21.9 (total) hrs/d (Section 6.1.2.3)b Time spent indoors 16 (21.6) hrs/d Section 6.1.2.3b (at home) Time spent outdoors 2 hrs/d Playing on sand/gravel 0.9 hrs/d Average. Upper percentiles not available (Section 6.1.2.3)b Playing on grass 1 hrs/d Playing on dirt 0.8 hrs/d Time spent swimming 23 hr/year Average value. Upper estimate not available (Section 6.2.4.3)b
60 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 61 Table 8b: Summary of suggested exposure factors for 1–2-year-old child (male and female combined) Table 9: Summary of non-age-dependent exposure factors
Note: Values are average or 95th percentile (in parenthesis)a Note: Values are average or 95th percentile (in parenthesis)a
Parameter Suggested default Units Comment and internal reference Parameter Suggested value Units Comment Anatomical and physiological parameters Anatomical and physiological parameters Body weight 11 (13) kg Section 2.2.4 82 [*] Male and female combined (Section 2.4.1)b Body height 81 (86) cm Section 2.1.4 Life expectancy 79 yrs Male (Section 2.4)b 84 Female (Section 2.4)b Dermal exposure parameters *Upper estimate not available. Many national and international agencies use 70 yrs as the assumed lifetime exposure to environmental agents. Total skin surface area 5,300 (6,100) cm2 Section 3.2.4. See Table 3.2.5 for specific body part data. Exposed skin surface area 1,600 (1,900) cm2 Section 3.2.3, 3.2.4 Dermal exposure parameters Applicable for outdoor and indoor residential child and adult exposures Oral exposure parameters mg soil/cm2 Soil adherence 0.5 (1.7) (Section 3.3.1)b. skin Drinking-water intake 0.3 (0.9) L Tap water intake. Includes water used in food preparation (Section 4.2.3). For specific activities and body parts see Tables 3.3.3, 3.3.4, and 3.3.5b. Food intake (excludes 720 (1,700) g/day For intakes of individual food groups, see Section 4.4.4. Organics: 1 Chemical-specific. beverages except for milk) Dermal bioavailability Inorganics: 0.0001 These table values are to be used only when other reasonable information is not 50, Central tendency for outside soil. (Reasonable maximum, outside soil) Unitless available. Soil ingestion (100), mg/day Organics: 1 [100] [Central tendency outside soil + indoor dust]. (Section 4.5.3) Dermal bioaccessibility Inorganics: Bioaccessibility of inorganics from soil or other media can be approximated with b No default experimental tests (Sections 3.4 and 4.0) . Incidental water ingestion 50 (~ average) mL/hr Section 4.6.3 while swimming 150 (~ upper estimate) Shower and bath Central estimate (upper estimate) for adults and children (Section 3.5.5)b 1 (2) #/day frequency Inhalation exposure parameters Shower duration 8 (16) mins Section 3.5.5b Section 5.1.3. For specific inhalation rates by activity or for short term exposures, Inhalation rate 8.0 (12.8) m3/day Shower volume 72 L b see Section 5.1;Table 5.1.2 Volume and flow rate of non-water saving shower (Section 3.5.5) . Shower flow rate 9 L/min Upper estimates not available. Activity patterns Bath duration for adults and children combined (Section 3.5.5)b. 20 (63) Bath duration 21 mins Upper percentile not available. Frequency of hand (indoors) contacts/ Section 6.1.1.3 Bath volume Insufficient data L Insufficient data (Section 3.5.4)b. to mouth 14 (42) hr (outdoors) Oral exposure parameters Mouthing duration Varies hr/d Mean and maximum values differ by object mouthed (Section 6.1.1.3) Organics: 1 Oral Chemical-specific. Upper estimate not available 22.6 (total) hr/d bioavailability Inorganics: (Section 6.1.2.3) No default These table values are to be used only when other reasonable information is Time spent indoors Unitless not available. 17.8 (24) Organics: 1 hr/d Section 6.1.2.3 Bioaccessibility of inorganics from soil or other media can be approximated with (at home) Oral bioaccessibility Inorganics: experimental tests (Sections 3.4 and 4.0)b. Time spent outdoors 1.4 hr/d No default Playing on sand/gravel 0.7 hr/d Inhalation exposure parameters Average. Upper estimates not available (Section 6.1.2.3) Playing on grass 1.1 hr/d The residential value is midpoint of range for ‘closed’ Australian dwellings. Air Residential: 0.6 changes will be higher with open doors/windows, ceiling fans and air conditioning. Playing on dirt 0.9 hr/d Building air exchange Commercial: no #/hr A single value is not suggested for commercial buildings. rate Time spent swimming 21 hr/year Average value. Upper estimate not available (Section 6.2.4.3) recommendation Ur estimates not available (Section 5.2.4)b. a See Footnote (a) to Table 7. For screening risk assessments and establishing guidelines the most sensitive receptor is assumed to be a 2–3-year-old Indoor particle (Table 8a) or a 1–2-year-old (Table 8b). No recommendation #/hr Markedly differs from house to house. No suggested value (Section 5.3.1)b. deposition rate b The references to sections and tables refer to those in the Australian exposure factors guidance document. Houses: 190 Floor area of residential Average values. There is a wide range of values for houses and other types of Other: 120 m2 dwelling dwellings (Section 5.4.1)b. Upper estimates not available. All: 180
62 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 63 Parameter Suggested value Units Comment It is important that the risk assessor The Australian Bureau of Statistics can consult the text and, if necessary, the provide a range of Australian data. Houses: 460 Average values. Assumed ceiling height is 2.4 m. There is a wide range of Air volume of residential primary information source, prior to Other: 280 m3 values for houses and other types of dwellings (Section 5.4.1)b. Upper estimates dwellings using any of the summary information The US EPA has developed an exposure All: 420 not available. contained in Tables E1 to E7. factors program to further advance the Uptake (product science of exposure assessment in risk of retention and Chemical-specific value should be used if available Australian exposure factor information assessment. The US EPA Exposure Default 100% unitless b absorption) of inhaled (Section 5.5) . has been sought and juxtaposed with factors handbook (US EPA 1997a) is contaminants overseas data to allow an appreciation a comprehensive collation of exposure 10 3 Background particulate PM ; 17 (39) μg/m National average of all urban monitoring sites – 50th percentile, (95th percentile). of the fact that not all overseas data information used in US risk assessment PM ; 7 (16) levels for urban 2.5 Proportion of PM that is PM . reflects sectors of the current Australian practices. This is continually updated, Ratio; [0.4] [unitless] 10 2.5 ambient air (Section 5.6.2)b. population. If Australian information is and the most recent update as released not available, overseas data may be used, for consultation in October 2009 and Fraction of indoor dust 50 (100) % Average (maximum) (Section 5.7)b but will require justification in the risk published early in 2012. from outside soil assessment as to why they are applicable Activity patterns in Australia. The American Industrial Health Council’s Exposure factors sourcebook (1994) Time spent in transit 1 hr/day Total time by all travel modes (Section 6.2.3.2)b. Upper estimate not available. It should also be appreciated that the provides examples of probability Approximate median information may not be current at the distributions for a range of exposure Frequency of swimming 52 (150) d/yr (upper estimate). time the risks assessor consults this factors. These largely relate to the US Actual frequency will depend on the Australian locality (Section 6.2.4.3)b. document. Some values may be more population. These, and similar US-based than a decade old and Australian data, should only be used if they can Residence and population mobility parameters demographics and behaviour may have be demonstrated to be relevant to the Duration of residence 10 (35) yr Section 7.1.3b. changed from the time the information Australian population. was first gathered. For example, there a See Footnote (a) to Table 7. For screening risk assessments and establishing guidelines the most sensitive receptor is assumed to be a 2–3-year-old. are currently many more people of Asian, b The references to sections and tables refer to those in the Australian exposure factors guidance document. Indian and African descent residing in Australia. People are more mobile and, due to water restrictions in most states It should be noted that body weight document is 78 kg, with 85 kg and 70 kg and territories, shower durations and is a parameter used to adjust dose recommended for only males or females 4.14 garden irrigation are different from and exposure in many aspects of risk respectively, and 70 kg recommended SOURCES OF EXPOSURE 10–20 years ago. These examples assessment. Because of demographic for whole-of-life body weight average highlight the necessity to make differences, various authorities around (see Table 7). ASSESSMENT DATA sure exposure parameter values are the world have used slightly different contemporary and ‘fit for purpose’. It default values for age-related body weight In Australia, it is generally assumed It is recommended that the Australian is the risk assessor’s responsibility to in their risk assessment equations. For that the most sensitive individual is the exposure factor guidance (AEF) ensure this is so. example, the default generic value used 2–3-year-old child (enHealth 2003, document be used as the primary by Health Canada (1994) is 70 kg, a 2004; NEPC 1999). Data is provided source of exposure data. While the Other useful sources of information value commonly found in many health throughout the Australian exposure factor information contained in this document and data (of a somewhat older vintage) risk assessment documents, including guidance document for a 2–3-year-old is not intended to be a comprehensive include: the Australian drinking water guidance child. However, in Table 8b, information compendium of exposure parameters, •• Exposure scenarios and exposure (NHMRC, NRMMC 2004). The WHO is also provided for a 1–2-year-old child. it has been compiled with a view settings (Taylor & Langley 1998) figure is 60 kg and the US EPA (1997a) The risk assessor should determine which to providing guidance to Australian Exposure factors handbook lists 71.8 kg age bracket most closely resembles risk assessors. While a number of •• Exposure factors in risk assessment (increased to 80 kg in the 2009 update). the most sensitive individual for their ‘recommendations’ have been made (Langley & Sabordo 1996) exposure scenario. regarding parameter values in the •• 1996 Australian exposure factors Both the Australian and US handbooks text (summarised in Tables E1 to E7), (Langley, Taylor & Dal Grande 1998). provide extensive tables of age-related It is important to recognise that many risk assessors and others using the body weights and growth curves. The toxicological reference values and information should check to ensure the default adult (combined sexes) body environmental guidance values will have suggestions presented are suitable for the weight value recommended in the been derived using exposure factors that scenarios they are evaluating. Australian Australian exposure factor guidance are now outdated. data should be used where it is available.
64 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 65 Chapter 5: Risk characterisation
reference value or a health-based 5. Risk characterisations (and risk Differentiation of the approaches used consider the probability of both types of 5.1 guideline value, usually one that has been 5.2 assessments) should be undertaken in qualitative and quantitative risk error. Other desirable features include INTRODUCTION set using the same principles of health KEY PRINCIPLES IN using methodologies outlined in this assessments are informed by definitions a lack of ambiguity, ease of calculation, risk assessment set out in these enHealth enHealth document, noting that that have been developed for each of clarity (especially for the presentation Risk characterisation is the final step in guidelines (see Section 5.5). ENVIRONMENTAL HEALTH methodologies may be revised as these two processes. of the results) and broad acceptability the risk assessment process that: RISK CHARACTERISATION needed to maintain consistency with of the reasoning for the determined The final risk characterisation is limited best scientific practice. Reports should Qualitative assessment: An inquiry classification. •• integrates the information from follow a consistent general format (see by the available data, and this should be There are a number of key principles for process that generates non-numerical hazard identification, dose–response Chapter 7), bearing in mind the need discussed in the uncertainty assessment. health risk characterisation: data, providing an ‘understanding of Numerical estimates of risk may be an assessment and exposure assessment The process requires considerable to recognise the unique characteristics a social or human problem, based on outcome of a quantitative assessment, •• discusses chemicals of potential expertise. If data is collected and 1. Protection of human health is the of each specific situation. building a complex and holistic picture but with the qualification that these are concern (COPC) and quantifies analysed according to the principles and primary objective. Human health risk formed with words, reporting detailed mathematical constructs incorporating 6. Risk assessors should review the risks associated with these specified guidelines in this enHealth document, the assessment is generally undertaken views of informants and conducted in various degrees of uncertainty. Numbers most up-to-date scientific literature chemicals process will become more transparent with an appreciation that the health a natural setting’ (Creswell 1994); or ‘a may give a misleading implication of relevant to the risk assessment under and consistent. Some parts of the risk risk assessment is part of a larger classification process, where objects or accuracy, especially when based on poor •• identifies the contributions to risk from consideration and to the toxicological assessment process such as ‘data assessment that encompasses materials are assigned to some class or uncertain information (Langley 2003). all the relevant exposure pathways, profile of the identified COPC. collection’ and ‘exposure assessment’ ecological risk assessment. on the basis of tests made against The numbers generated should never and aggregates these risk estimates Information in appropriately peer- will be, at least in part, quantitative However, actions based on the established or implied criteria’ (Ellison be portrayed as being highly precise or •• considers the possibility that multiple reviewed articles should be accorded and possibly based on modelling or risk characterisation taken should et al. 1998) accurate (IEH 1999b). The risk level COPCs may have cumulative effects, greater weight than information in extrapolations from measured data. always adequately protect public should never be expressed in a way that and considers options for best articles that are not peer-reviewed. health and the environment, putting integrating the effects of combined These guidelines are intended to assist Quantitative assessment: The application suggests a greater degree of precision than these responsibilities before all other exposures (see Chapter 12) the qualitative process of determining 7. Variations in risk assessments of a set of scientifically measurable, is warranted by the data, for example, a whether environmental health intervention considerations. as a result of particular statutory reproducible and mathematically sound risk level of 4.73 × 10–6 (i.e. using three •• describes the risks to individuals and is required or not required. requirements, resource limitations, significant figures rather than 5 × –610 ) is 2. Risk assessments should be data values to estimate value, probability populations in terms of nature, extent and other specific factors should probably meaningless in the context of an transparent (Schreider et al. 2010). and associated risk of loss. and severity of potential adverse be explained as part of the risk EHRA outcome (Langley 2003). Risk characterisation may involve The nature and use of default values health effects characterisation. For example, a comparing environmental data, exposure and methods, assumptions and policy In quantitative risk assessment, reporting •• provides an evaluation of the overall reason will be required to explain why data, intakes and biological monitoring judgements in the risk assessment of a measurement is an approximation Variability associated with the identified quality of the assessment and certain elements are incomplete. results with established criteria, including should be clearly identified and or an estimate of the value of the subject hazards, the nature of the exposed the degree of confidence the risk guideline values (GVs) established or documented. Conclusions drawn from being measured. Such a result should populations or groups of people assessors have in the estimates of risk published by authoritative sources. the evidence should be distinguished only be considered complete after it has (‘receptors’), and limitations in the and conclusions drawn; this should be 5.3 from policy judgements, and the been evaluated and the uncertainties in toxicological and exposure data will all based on appropriate uncertainty and Due to the complexities of the matter, influence of ‘scientific judgement’ QUANTITATIVE AND the measurement explained. There are contribute to these uncertainties. sensitivity analyses the risk characterisation process cannot made clear. different ways of measuring uncertainty. •• communicates results of the risk be reduced to a ‘cookbook’. In this QUALITATIVE RISK Statistical analysis allows for the The most conservative mathematical assessment to the risk manager context, the guidance in this document 3. Risk characterisations should include evaluation of random events and from models used in quantitative EHRA can a summary of the key issues and CHARACTERISATION •• provides key information for risk consistently recommends that the those arising from a systematic effect. be virtually insensitive to the actual choice of default parameters, GVs or conclusions of each of the other experimental data and should be viewed communication. The level of risk estimated in any risk risk assessment methodology must components of the risk assessment, Accounting for uncertainty in a qualitative only as a risk management solution, not a assessment can be described either include an assessment of their suitability as well as describing the nature and assessment is the acknowledgement risk assessment technique (IEH 1999b). Risk characterisation is identified as part qualitatively (i.e. by putting risks into for use in the EHRA at hand. In other likelihood of adverse health effects. that the original classification has been The extent to which manipulation of the of Phase II of the expanded framework for categories such as ‘high’, ‘medium’ or words, care must be taken to ensure that The summary should include a made on the basis of the available input data can influence the resultant EHRA outlined in Figure 2. ‘low’) or quantitatively (with a numerical published or derived health-based GVs description of the overall strengths evidence and that misidentifications risk estimates should be determined estimate). Current risk assessment are ‘fit for purpose’. and limitations of the assessment and may have occurred. There may be using ‘sensitivity analysis’ techniques The overall objective of the risk methods described in this enHealth conclusions. a lack of evidence in an observation (see Section 5.15). characterisation stage is to determine document provide quantitative estimates that has caused it to be placed in a that exposures to COPCs from the 4. To protect public health and the of risk but the precision of any such particular class, and this may result in Estimates do not have to depend environmental source under consideration environment an appropriate degree of estimate will be limited by the data either a ‘false positive’ or ‘false negative’ on the use of numbers to be useful; do not exceed a level considered to be conservatism must be adopted to guard available to use in the assessment. protective of human health. In practice, against uncertainties. There should classification. Methods used in describing ordinary language may be used to this means that the estimated total be a detailed description of the areas the identification of any classification indicate the level of risk. A finely exposure (including background where of uncertainty and an analysis of the should reflect the uncertainties associated divided ranking system can give a relevant) does not exceed a toxicological effects of these on any derived values. with the evidence, permit updating relatively accurate indication of quantity on the basis of further evidence, and without using numbers (ACDP 1996).
66 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 67 Clearly defined qualitative categories Qualitative risk characterisation may need term).2 The HQs for all exposure pathways •• HI/RIs should represent the exposure strictly additive, the total cancer risk Unless evidence for synergistic effects can enable reliable and effective risk to be put into context or compared with for each contaminant can be summed to pathways that have the potential to estimate might become artificially more is available, the potential for synergistic management decisions. other risks relevant to the community. produce a total hazard index (HI) or risk expose the same individual or sub- conservative as risks from a number of effects may be omitted from the However, if comparisons do not directly index (RI) (see Section 12.4). population, making sure to consider different carcinogens are involved. assessment. Additive effects are much Some attempts have been made to relate to alternative options, they may be areas of highest exposure for each •• It will often be the case that more common and are covered in risk integrate the concepts of qualitative and counterproductive and caution should be Hazard quotient (HQ) = pathway for both current and future substances with different weights of assessment by summing risks across qualitative risk estimates by assigning exercised in making such comparisons, Intake (mg/kg/day) land uses. All exposure pathways evidence for human carcinogenicity chemicals and pathways. explanatory statements to the various especially if like is not compared with like, Threshold TRV (mg/kg/day) should be summed unless information are included. The cancer risk qualitative categories of risk and, in or if comparisons are being used to imply is available that indicates the same equation for multiple substances 5.5 some instances, a numerical probability acceptability. Flippant comparisons are Hazard index (HI) = ∑ Hazard quotients individual or sub-population cannot be sums all carcinogens equally, giving range associated with these descriptors. certainly likely to be counterproductive exposed by a particular pathway. as much weight to Class 2 as to TOXICOLOGICAL Examples of this approach may be found (DOH 1998). Comparisons should be The HQ/RQs for all exposure pathways for Class 1 carcinogens. In addition, in an assessment by Biosecurity Australia used only where the evidentiary base all contaminants should be summed to 5.4.2 non-threshold TRVs derived from REFERENCE VALUES (2006) of risk estimates associated with and the method for risk estimation are produce a total HI/RI, unless evidence is Non-threshold risk estimation animal data will be given the same (TRVS) IN COMMON USE the importation of apples from New similar and where the uncertainties in available to show this is not appropriate. weight as non-threshold TRVs derived Zealand, and in WHO/FAO guidelines on all the comparative estimates are shown When summing these HQs, the following Where non-threshold TRVs are adopted from human data. As outlined in Chapter 3, both threshold risk characterisation of microbiological (Thomas & Hrudey 1997). should be taken into consideration: (that is, assuming a linear low-dose •• The action of two different carcinogens and non-threshold methodologies for hazards in foods (WHO 2009). relationship), risks are estimated as the •• HI/RIs should be calculated for each may, or may not, be independent. utilising dose–response data rely on It is important to consider contingent additional probability of an individual age group (category) separately. establishing a point of departure (POD) While the qualitative descriptors indicating risks. This requires not looking at risks developing cancer over a lifetime as a In practice, it will often be the case that for further analysis. For threshold the likelihood of the occurrence of event in isolation so that, for example, the •• HI/RIs should be calculated separately result of exposure to the carcinogen. there is insufficient information to make responses (all toxicological endpoints may amplify the understanding of simple risks of immunisation or chlorination are for chronic, sub-chronic and shorter- The estimated intake for each exposure a well-informed decision as to whether other than carcinogenesis), the POD is descriptors such as low, medium and considered in the context of the risks of duration exposures. pathway and non-threshold TRV are it is reasonable or not to sum ILCRs usually the NOAEL, LOAEL or a BMD. high risk, the linking of these descriptors not having immunisation or chlorination. •• Ideally, HI/RIs should be categorised multiplied to produce pathway-specific across either pathways or contaminants to quantitative probability estimates into groups of chemicals that estimates of increased lifetime cancer (see also Chapter 12) Where more This POD is then divided by a series of is a different matter. The quantitative induce the same type of effects or risks (ILCR). information is available, a decision to safety factors (SFs – sometimes termed probability estimates in the above 5.4 that act by the same mechanism assess contaminants or pathways as uncertainty or modifying factors) to examples were developed for quarantine of action. However, this process is However, for those carcinogens where RISK ESTIMATION independent and non-additive should be determine an acceptable or tolerable and food quality risk assessment not simple and requires a thorough appropriate benchmark dose data is supported with reference to the toxicology daily intake (ADI or TDI). The general scenarios. They are not necessarily understanding of the toxicology available and suitable for use, the risk Risk estimation combines the estimated of the contaminants concerned. This methodology for assessing the ADI or appropriate for linking qualitative of the chemicals concerned, and estimation method outlined above for intakes calculated in the exposure assessment should be undertaken by TDI is outlined in Section 5.5.1. descriptors with probability estimates must only be undertaken by an threshold compounds applies. assessment with the toxicological an appropriately qualified toxicologist. of human health risks associated with appropriately qualified toxicologist. reference values (TRVs) (threshold and It is recommended that the following Table 10 describes the toxicological environmental chemical exposures, If this segregation is not performed ILCR = intake (mg/kg/day) × non-threshold where relevant) from the –1 approach should be followed under most reference values possibly generated nor have such linkages been endorsed carefully, an overestimate (or, less TRV(mg/kg/day) toxicity assessment to produce numerical circumstances. by the threshold approach. by enHealth or any other public health commonly an underestimate) of indices of likely health effect. The risk 3 authority for this purpose. the true risk could result. When ILCR = exposure concentration (mg/m ) × •• ICLR estimates should normally be estimation methodology differs for 3 –1 The term ADI is generally used for toxicological information is lacking or TRV(mg/m ) summed across pathways unless threshold and non-threshold compounds pesticides that are deliberately used Qualitative risk conclusions can also be unclear, it can be assumed that the specific evidence is provided that the due to the different modes of chemical on food or crops and may appear as used to avoid the false sense that the chemicals act by the same mechanism ILCR estimates from all pathways and same person/group of people cannot effect. residues in the food chain. The term TDI extent of the risk is known precisely. of action with summation of the HQ/ all contaminants should be summed be exposed by the different pathways. to produce a total additional increased is generally used when a chemical is a Using terms such as ‘high’, ‘medium’ or RIs. It should be noted that this will •• ICLR estimates should normally be 5.4.1 lifetime cancer risk, unless evidence food or environmental contaminant. Both ‘low’ may have different interpretations result in an overestimate of the true summed across contaminants unless Threshold risk estimation is available that suggests that this is the ADI and TDI are conceptually similar for different groups so they should be risk. See Section 12.4 for a wider specific evidence is provided by a not appropriate. When combining to the RfD. The difference in terminology clearly defined. Such definitions, or risk For threshold compounds, the intake discussion of the use and limitations qualified toxicologist that this is not the ILCR estimates, the US EPA (1989) arises because the terms have been categorisation matrices, can be found in for each exposure pathway is divided of the HI/RI approach. appropriate approach. some Australian guidance documents by the appropriate threshold, TRV identifies several limitations that may be coined by different agencies. •• It is recognised that synergistic (that (e.g. AS/NZS ISO 31000:2009; OGTR (allowing for intakes from other sources considered. These include: is, more than additive) effects are risk assessment framework, see where relevant) to produce a simple 2 Note that the WHO preferred terms are ‘risk •• As each non-threshold TRV is an upper possible; however, the practical Section 1.3.2). ratio, termed a ‘hazard quotient’ (HQ) quotient’ and ‘risk index’. The terminology HQ/HI is 95th percentile estimate of potency, difficulties of quantifying the synergy (commonly used/historical term) or ‘risk retained in here because of its widespread popular and because upper 95th percentiles in a risk assessment are significant. quotient’ (RQ) (WHO recommended use in the USA and Australia, including in the revised NEPM on contaminated sites assessment. of probability distributions are not
68 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 69 Table 10: Environmental health criteria derived from a threshold approach It is important to note also that in public These exposure limits are derived by first From the data available on humans and or occupational health risk assessments, determining the NOAEL or, if the NOAEL experimental animals, it appears that inter- Toxicity reference value Units Description or definition establishing a NOAEL is likely to be cannot be determined, taking the LOAEL species and intra-species differences are, Acceptable daily intake (ADI) mg/kg/day The daily intake of a chemical that, during a lifetime, appears to be without appreciable risk, on influenced by a consideration of the and dividing the value by factors to in general, less than 10, hence the often- the basis of all the facts known at the time (WHO 1994a). The term ADI is generally used for relevant route(s) of exposure and account for: used overall safety factor of 100 for these chemicals such as pesticides, which may be present in foods within their maximum residue level experimental design. two factors is conservative and adequately •• inter-species differences (extrapolating (MRL) because of their permitted uses in agriculture. The ADI and RfD are conceptually the same; protective of public health (Johannsen from animals to humans) the terminology differs because of development by different authorities. The selection of the NOAEL can be 1990; Renwick & Walker 1993). •• intra-species differences (differing Tolerable daily intake (TDI) mg/kg/day An estimate of the intake of a substance that can occur over a lifetime without appreciable significantly influenced by: health risk (WHO 1994a). This is conceptually the same as the ADI and RfD but used when the sensitivities between individuals) One of the outcomes of an IPCS program •• the selection of doses used in the substance is an unintended contaminant in food or an environmental medium such as air, water •• the severity of the adverse effect (IPCS 2005) to develop chemical- study – the ‘real’ NOAEL is likely to or soil. This terminology avoids the implication that the contaminant is ‘accepted’. specific adjustment factors (CSAFs) was lie somewhere between the apparent •• the quantity and quality of the to further refine the breakdown of the Reference dose (RfD) mg/kg/day An estimate (with uncertainty spanning perhaps an order of magnitude) of a daily oral exposure to NOAEL and LOAEL doses (if there is a scientific data. conventionally used 100x safety factor (US terminology) the human population (including sensitive sub-groups) that is likely to be without an appreciable relatively wide margin between doses risk of deleterious effects during a lifetime. by incorporating figures based on inter- used in the study, a higher NOAEL The general approach to calculating an species and intra-species toxicokinetic Reference concentration (RfC) mg/m3 An estimate (with uncertainty spanning perhaps an order of magnitude) of a continuous inhalation might have been obtained if doses had ADI/TDI follows the principles initially and toxicodynamic variation. The IPCS (US terminology) exposure to the human population (including sensitive sub-groups) that is likely to be without an been more appropriately spaced) outlined in the IPCS Environmental health appreciable risk of deleterious effects during a lifetime. criteria monograph No. 104 (WHO 1990). proposal for the default CSAFs is set out •• the number of test subjects in the The uncertainty inherent in extrapolation in Figure 23. dose levels – a smaller number of between and within species has test animals per dose in a study generally been dealt with by using safety The IPCS program recommends using Note that the US EPA envisages that, provisional ADI or TDI may incorporate an compromises the statistical power of while the RfD or RfC are generally only additional SF in the calculation because 5.6 (uncertainty) factors. chemical-specific data to replace being able to discriminate between default values where adequate data is used for non-cancer endpoints, it may be of the inherent uncertainty. DETERMINATION OF dose levels that produce an ‘effect’, necessary to derive both an RfD (or RfC) Historically, the most common overall available, and outlines the nature of compared with those where the data on toxicokinetic and toxicodynamic and a non-threshold cancer risk slope There are other sets of health-based NO(A)ELs, ADIs (RFD) factor used by a number of regulatory incidence of disease or toxicity variation that could be used. The factor where both cancer and non-cancer guideline values derived from occupational bodies is 100, comprised of 10 to is comparable to the ‘controls’ or program recognises that the combined endpoints need to be considered in the health and safety (OHS). These include AND TDIs FOR HUMANS account for uncertainties in inter-species untreated animals uncertainty factor (CUF) based on risk assessment. The only exception is threshold limit values (TLV), short-term extrapolation, and 10 to account for The determination of an acceptable •• the extent to which disease or CSAFs could be less than, or more where a carcinogenic response drives the exposure limits (STEL) permissible intra-species variability. An additional or tolerable daily intake (ADI or TDI ) toxicity associated with administration than, the common default value of 100, risk assessment, but it is considered that exposure limits (PEL), to be used in an factor of 10 is sometimes used if the involves establishing an overall NOAEL of the test agent can be discriminated but notes that this should be made a non-threshold approach is warranted environmental health risk assessment. NOAEL was not established in the for a chemical that is generally the lowest from disease processes that occur transparent to the risk manager. It still because of the proposed mode of action. While they are often derived using study and the LOAEL used instead, if NOAEL in the most sensitive species. naturally during ageing. This is recognises the need to add additional comparable methodology, extrapolating the study used to determine the ADI/ particularly true of neoplastic uncertainty factors when the quality of the The time base may be altered from daily from animal toxicity studies, human TDI must be based on a relatively This approach of using the lowest NOAEL responses, where it may be difficult studies is deficient or where significant intake to weekly or monthly intakes where exposure studies and epidemiological short-term study (e.g. 28–90 days) or is justified unless there is evidence of one to ‘score’ the number of neoplasms at data gaps occur. the exposure pathways or toxicokinetic studies, they are based on the protection if a large toxicological database has not or more of the following: different stages of the life span of the behaviour of the chemical warrant a of workers (who are on average healthier been assessed. Application of additional test animals, where the progression longer period of averaging. For example, than the whole community) during the •• from pharmacokinetic/metabolic factors needs careful consideration for The decision on the magnitude of through a series of pathological the term ‘tolerable monthly intake’ (TMI) course of a normal working shift and studies that the most sensitive new industrial chemicals, where the factors to use is predominantly based changes is not well delineated. is applied to dioxins because of the very a normal working lifetime. They may species shows a different toxicokinetic available toxicological databases may on expert or informed judgement. While Where the progression of toxicity is long half-life for elimination and the use use different levels of protection and behaviour than humans and is be less comprehensive than those for this approach to selecting the number time related and possibly reversible of body-burden estimates in humans and safety factors than used for the general therefore less relevant as a predictor new agricultural chemicals, proposed and magnitude of the safety factors if exposure ceases, the duration of animals to adjust intakes (Joint FAO/WHO community, such as tolerating relatively of human toxicity than another toxicity food additives or medicines (human and can appear to be somewhat arbitrary, treatment becomes a more critical Expert Committee of Food Additives – minor adverse effects. It would be unusual test species animal). The overall factor can range improved knowledge of the biological factor in the experimental design. JECFA 2002; OCS 2004). for OHS-based guideline doses to be •• that the toxic effect that has the lowest from 10 to 10,000, depending on the processes that cause inter- and intra- used in an environmental health risk NOAEL is not relevant for humans, or source and quality of data, the biological species variation (e.g. metabolic and An ADI or TDI is derived from the NOAEL other pharmacokinetic rate differences) Sometimes the word ‘provisional’ is assessment, although the risk assessor •• that the lowest NOAEL is derived from relevance of the endpoint, and the hazard (or LOAEL) as follows: have generally supported the choice of attached to the term (e.g. the provisional may need to be aware of potentially an inadequate or invalid study. assessment (carried out on a case-by- case basis). In general terms only, a safety the default safety factors. tolerable weekly intake – PTWI). This conflicting situations where it may not ADI or TDI = NOAEL factor of 10 would apply when appropriate is generally done when further data is be clear whether the risk estimates of an Thus it is emphasised that the full SF required to establish an acceptable or EHRA should be applicable to both the database must be used and all relevant human data were available. tolerable intake but a temporary GV general community and/or to workers findings correlated when determining the is required by the risk managers. A within an exposure scenario. most appropriate health endpoint.
70 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 71 Figure 23: Proposed subdivision of default uncertainty factors to be used in Where different default values are The step-wise process for deciding on Carcinogens identified by the US EPA as risk assessment 5.7 used (e.g. those recommended in the the dose–response data to adopt for having a mutagenic mode of action (as of TOXICOLOGICAL Australian exposure factor guidance the EHRA of carcinogens (or potential 2005) are discussed in US EPA 2005a. document – see Chapter 4), it may be carcinogens) is set out in Figure 15 This list includes benzo(a)pyrene, and the REFERENCE VALUES necessary to adjust CSFs derived by the (Section 3.10.3). This decision-making additional safety factors recommended DERIVED USING A NON- US EPA or in Integrated Risk Information process recommends use of a BMD by the US EPA have accordingly been System (IRIS) databases. approach to selecting a POD for risk incorporated into the derivation of HILs THRESHOLD APPROACH assessment, once a decision has been for benzo(a)pyrene in the revision to the Recent US EPA (2009a) guidance made on classification of the COPC as contaminated sites NEPM (NEPC 2010). The two toxicological reference values that (RAGS-F) indicates that inhalation risks a carcinogen and a carcinogenic risk may be developed using a non-threshold should be assessed by calculating an assessment approach is warranted. approach are: exposure-adjusted air concentration, Where appropriate BMD data is not 5.9 available, alternative dose–response •• cancer slope factor (CSF): This is the which is then used in EHRA risk COMBINING RISK data should be sourced, which may plausible upper-bound estimate of the characterisation equations. This means probability of a carcinogenic response that exposure estimates are no longer include the use of CSF (for genotoxic ESTIMATES per unit of intake over a lifetime; it is adjusted based on inhalation rate or body carcinogens) and ADI/TDI (for non- expressed in units (mg/kg/d)–1 weight, and the only difference in risk genotoxic carcinogens). Where there are several exposure estimates between a child and an adult is pathways, the incremental lifetime cancer •• unit risk factor (URF): This is an the exposure time. risk (ILCR) estimate is simply summed expression of carcinogenic potency 5.8 for each of the relevant pathways to in concentration terms, expressed as The application of these risk factors is get a combined risk estimate (see the probability of cancer per unit of AGE-SPECIFIC to calculate the probability of a finite Section 5.4). However, some caution an exposure medium (e.g. per µg/L of increase in cancer risk over a lifetime, ADJUSTMENT FACTORS should be exercised in adopting this water, per µg/m3 of air or ppm). according to the equation: simple summation approach (US EPA The US EPA has directed particular 1989). Since the CSF is an upper 95th The CSF is used in EHRA to estimate the upper-bound probability that cancer will Increased lifetime cancer risk (ILCR) = attention to the possibility that early-life percentile estimate of cancer potency, develop over a lifetime of exposure to a chronic daily intake (mg/kg/d) × exposure to a carcinogen may exacerbate simple addition of 95th percentiles is –1 chemical at a specific level of intake. It CSF (mg/kg/d) risk to the extent that the default approach strictly not correct. Such an approach based on a whole-of-life CSF or URF may can add unnecessary conservatism is the slope of a linear extrapolation from or the upper-bound estimate of a POD dose not be sufficiently protective (US EPA to the aggregate risk estimate. The to zero. Increased lifetime cancer risk (ILCR) = 2005a). The guidance is consistent with CSF are not weighted according to the exposure concentration × URF reviews of animal carcinogenicity bioassays strength of evidence that underpins Reproduced from IPCS 2009b with permission from WHO. The URF can be derived directly from relevant to the assessment of early-life their categorisation. All classes of inhalation or drinking-water studies The outcomes of cancer risk estimates susceptibility to carcinogens (Hattis et carcinogenic categorisation are given It is generally accepted that if the However, Gaylor et al. (1999), when depending on which media is being based on CSF or URF calculations are the al. 2004; 2005). US EPA guidance on equal weight, including those where magnitude of the overall safety factor commenting on the possible use of a assessed. Where such data is not directly prediction of an increased lifetime risk of early-life exposure to carcinogens for which either human or animal data (or both) available, these unit risks can be derived developing cancer. The intake estimate a mutagenic mode of action (MoA) has drive the categorisation. approaches or exceeds 5,000, this is benchmark dose approach (using BMD10) effectively an admission that there is as a point of departure, recommended by converting an oral CSF with units (or exposure concentration) must be been reasonably established has been –1 insufficient knowledge of the environmental the use of a default uncertainty factor of of mg/kg/d to a concentration of the averaged over the lifetime of expected summarised in the 2005 supplemental There may be different CSF estimates hazard under consideration and that 10,000 for irreversible adverse effects (e.g. substance in air, water or other media. exposure (default 70 years). The ILCR guidance (US EPA 2005c) and for a single chemical where the cancer the underlying data may be unsuitable cancer) with a smaller default uncertainty These extrapolations often assume default must be clearly presented so that the incorporated into the most recent RAGS-F data relates to different tumour sites. to support a risk assessment. US EPA factor of 1,000 for reversible adverse intake rates for the specified media (e.g. cancer estimate over 70 years cannot be guidance (US EPA 2009a p. 23). The EHRA usually uses the CSF that 3 practice has been to not recommend effects. The argument was based on the an inhalation rate of 20m per day of air, misrepresented as an estimate of annual predicts the highest risk. If cancer an RfD/RfC if the combined uncertainty fact that BMD approximates the LOAEL or ingestion of 2 L/day of water and a cancer risk. The guidance indicates that an additional potency and/or the type of tumour 10 body weight of 70 kg). factor exceeds 3,000 (NRC 2008). Where dose in conventional threshold-type risk safety factor should to be applied to produced differs according to the route a precautionary approach requires the assessments. To convert a lifetime to an annual risk mutagenic carcinogens as follows: of exposure, the aggregate risk may need The conversion equation most commonly application of such large uncertainty estimate is approximated by a simple to reflect this difference. used is: •• tenfold adjustment for exposures factors in setting a health-based guideline This update of enHealth guidance on division by the standardised lifetime during the first 2 years of life Inhalation URF (µg/m3)–1 = duration (70 years in most jurisdictions). For example, in the case of benzo(a) value, it is inevitable that when better EHRA commends the IPCS approach •• threefold adjustment for exposures information becomes available, the to selecting and justifying CSAFs and In reality for cancer, incidence will be pyrene (BaP), the representative CSF (mg/kg/d)–1 × 20 m3/d from ages 2 to <16 years of age consequent change in the numerical value recommends that it be adopted in EHRA greater in the later part of the 70-year carcinogenic polycyclic aromatic 70 kg BW × 1000 µ/mg (often an increase) can reduce community practice in Australia when relevant data window. •• no adjustment for exposures after hydrocarbon (PAH), the CSF used for confidence in its health-protectiveness. is available. (BW = body weight) turning 16 years of age. carcinogenic risk assessment in the
72 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 73 US IRIS database is 7.3 mg/kg/d–1 exposures to a point where a specified extrapolation over several orders combined risk of 10–5 may be considered Where health-based GVs are available •• What is the critical health effect? What (range: 4.5–11.7 depending on the level of risk will not be exceeded. of magnitude. Estimated risks are acceptable. The revision of the from multiple sources, the assumptions, is its nature, severity and reversibility? database used). The CSF used for BaP believed to represent only the plausible contaminated site NEPM (NEPC 2010) defaults and science policy that underlie •• Are interactions with other agents carcinogenicity assessment of B(a)P in While acknowledging that setting a level upper bounds and vary depending proposes a carcinogenic risk ‘target’ of them should be taken into consideration relevant? the contaminated sites NEPM review of ‘acceptable risk’ is often necessary for upon the assumptions on which they 10–5, irrespective of whether a single or and should be compatible or generally •• Is there sufficient data to establish (NEPC 2010) is 0.5 mg/kg/d–1 for the oral decision-making purposes, setting the are based. multiple chemical exposures contribute similar to relevant Australian practice. a GV? route. An estimated CSF by the dermal numerical value for such a risk level is a to the combined risk. route (25 mg/kg/d–1) has been derived socio-political matter, requiring extensive What sort of target risk levels have In setting GVs, apportioning exposures •• What population(s) will be affected? in data reported by Knafla et al. (2006), consultation with stakeholders, including been used in various risk assessment between media should take account of •• Are there any sensitive or particularly although these estimates have yet to be the community likely to be affected by paradigms? For carcinogens, a target risk 5.11 published relevant data on background susceptible sub-populations who endorsed by any regulatory agency. the environmental hazard and those level of 1 × 10–6 is the one most commonly PRINCIPLES FOR exposures and consider policies and are exposed, and is the derived GV responsible for managing or ameliorating used. The origin of the 10–6 level has been practices detailed in Section 5.11.1. This intended to be protective of these the risks. A socioeconomic or cost–benefit attributed to US regulators designating SETTING RISK-BASED apportionment should be documented for groups, or merely protective of the 5.10 each substance for which a GV is derived. analysis of the risk management options this level as a negligible or essentially ENVIRONMENTAL HEALTH average person? should also be part of the process. non-existent risk, from the legal point of •• Over what period of time will the TARGET RISK LEVELS Generic GVs developed using this view that de minimus non curat lex (the GUIDELINE VALUES population be exposed to the agent –6 methodology require endorsement by When the risk assessment uses a It is therefore important that all parties law does not deal with trifles) and 10 for which the GV is being set? appropriate national health bodies. threshold approach (e.g. derivation of an appreciate the real meaning of a ‘target’ is a convenient quantitative expression Elements of the risk assessment Situation-specific environmental health •• Is the GV to be generic (applying to ADI or TDI based on application of safety risk level – often expressed as something of the de minimus concept (Langley methodology described in this enHealth –6 criteria developed using this methodology many situations) or situation-specific? factors to a NOAEL or LOAEL), there are like 1 × 10 (one in a million). It should 2003). The current review of schedule document provide a framework for setting require endorsement by the appropriate •• What patterns of exposure are likely no implicit target risk levels associated not be taken to imply certainty that one B(4) of the contaminated sites NEPM risk-based environmental health GVs or health agency before being applied to a to occur? Are there likely to be short- with any derived environmental standards person will get the disease if there are cites a commentary by Kelly (1991) choosing appropriate GVs from published –6 particular situation. The use of ‘sensitivity or long-term fluctuations? Has the or guidelines. The use of safety factors at least one million people exposed. It sourcing the origin to the arbitrary 10 national and international sources (see analysis’ approaches (see Section 5.15) proposed guideline value appropriately to further reduce an exposure that is is simply a way of expressing risk, as a risk level used by the US Food and Drug Sections 5.11, 5.12). will provide some insight on the validity matched the expected exposure assumed to be without appreciable risk numerical expression of the likelihood Administration (FDA) in the 1970s to of the data inputs and the level of patterns with the health-based data over the specified period of time (usually of an event occurring under the defined regulate carcinogenic residues in foods. Toxicity reference values are generally uncertainty in the derived criteria. used to derive the benchmark? 70 years, representing an entire life span) conditions of exposure, based on Irrespective of the source, the general based on the most sensitive toxicological extrapolation of dose–response data. The understanding is that a 10–6 risk level •• Are there difficulties in getting is presumed to deliver an acceptable level effects, when known. When selecting A wide range of issues are considered precision of the risk estimate is subject is either essentially zero, or it can be relevant and accurate background of ‘safety’. toxicological reference values for by those establishing risk-based to many assumptions about the validity considered negligible in a regulatory sense. exposure data? comparison with exposure estimates, or environmental health GVs. A good of the extrapolation methods used, the For an individual COPC, the aim of the when using established health-based understanding of how a particular set of •• What is the relationship between degree of conservatism built into the However, it must also be acknowledged risk assessment is to determine whether environmental GVs, it is preferable to GVs has been calculated and what policy the derived tolerable intake and modelling (for example, if upper-bound that the target risk level has been varied the exposure exceeds an appropriate risk- use values that have been endorsed drivers are included in the considerations any background exposure? If the limits are used rather than means upwards to between 10–5 and 10–3 in based guideline value. This is generally or formally approved by the relevant is needed by those using GVs to ensure tolerable intake estimate is actually or central estimates). It is therefore different types of risk management expressed as a hazard quotient (HQ), Commonwealth, state or territory that the values chosen for a particular less than background, does this have important that risk communication situations. The higher risk levels are defined as the ration of the exposure environmental health agencies. If there risk assessment are appropriate to the implications for health consequences strategies address the need to explain the more commonly found in occupational divided by the GV (see Section 5.4.1). are no approved and relevant toxicological situation being investigated. Some of in the population, or is it because of meaning and derivation of such numerical exposure settings, or associated with the reference values or health-based the key issues that may have been the conservative way in which the When the risk assessment involves estimates of risk, including its likely level evaluation of contaminated sites. For environmental guideline values, advice considered when establishing GVs are: tolerable intake has been derived? combining exposures to multiple agents of conservatism and the background to example, the Dutch ‘intervention levels’ should be sought from an administrative In such circumstances, should any •• Why is a GV being proposed and is it the outcome of the risk assessment may the development of the target risk levels. for management of contaminated soils are authority at the relevant Commonwealth, actual experience or health information –4 necessary? be a hazard index (HI). In such a process, based on a carcinogenic risk target of 10 state or territory level. on the exposed community take the ‘target HI’ is generally assumed to Reinforcing the view that setting an (de Bruijn et al. 2001). •• Are there alternative means of precedence over a conservatively be 1. Chapter 12 includes discussion of ‘acceptable’ risk target for carcinogens Health-based GVs should be derived achieving the desired outcome? derived tolerable intake? is a process based on policy, rather than The ‘target’ risk level to which some The large improvements achieved the possible implications when the HI is using toxicological data or exposure •• What should be done if the tolerable science, WHO (1994a) stated that: Australian environmental regulatory by the Clean Air Act (1956) in the above or below 1. criteria from agencies or organisations intake is so conservative that it authorities aim is 1 × 10–6, although UK occurred without any air quality ... crude expression of risk in terms relevant to the state or territory (e.g. derives a guideline value so low that this may depend on whether the risk standards. When the risk assessment uses a non- of excess incidence or numbers of local or Commonwealth health agencies it cannot be measured or achieved is associated contamination of air, such as NHMRC or enHealth, or to threshold approach (as is the case for cancer per unit of the population at •• How will the GV be used? Is it simply a by reasonable risk management water or food, or whether the exposure which Australia is party (e.g. WHO). most quantitative carcinogenic risk doses or concentrations much less guideline or should it be considered to practices? Should environmental is associated with a single carcinogen Section 5.12 of this enHealth document assessments) it is implicit that any derived than those on which the estimates are be a standard? Standards often have health criteria ever be set to levels or is the outcome of multiple chemical provides some additional guidance on a environmental standards will attempt to based may be inappropriate, owing greater legal or regulatory standing that cannot be measured by exposures. In the latter case, a hierarchy of preferred alternative sources. protect the community by minimising to the uncertainties of quantitative than guidelines. contemporary analytical techniques
74 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 75 and, if so, how can such criteria be Figure 24: Decision tree for the development of risk-based environmental health GVs A slightly different approach will be B(7) of the contaminated sites NEPM •• What is the nature of the background managed in a regulatory context? required where acute exposures (NEPC 2010). exposures? Are they fixed or changing •• How can tolerable intakes, which are may cause the hazard of concern over time? Are they able to be altered? usually based on ingestion, be applied to become manifest. Examples are: 5.11.1 Are they voluntary (e.g. smoking) or to other exposure routes? setting microbiological standards for Allocating background exposures involuntary (e.g. ambient air pollution)? water and food where acute exposures To what degree should voluntary •• Has the tolerable intake been set in setting environmental health precipitate disease the setting of criteria background exposures be taken into using gavage or bolus administrations guidance values for sulfur dioxide in air where acute account? rather than administration as part exposures may cause exacerbation Chemicals designated as of potential of the diet? Has the interpretation of •• What is the background level of of asthma; and setting soil values concern (COPC) in an EHRA may also such tests been compromised by the exposure to the agent from media for nickel or chromium (VI), both of occur naturally in the environment, effects of a vehicle (e.g. an oil carrier) other than the one for which GVs are which may cause allergic reactions including chemicals like benzene, or the dietary matrix, such that the being set? How are these background from acute exposures in sensitised chlorinated dioxins or petroleum bioavailability of the test substance has or extraneous exposures managed in individuals. It is particularly relevant hydrocarbons. The particular situation been significantly altered? the standard-setting process? where the susceptible sub-population being assessed for risk may be due to •• How do you apportion exposures •• How is exposure occurring in the comprises a significant proportion the activities of people, but it will mean associated with such background? environmental setting, and to what of the total population, for example, that any exposed population may well What percentage of the total tolerable extent do exposures vary over time approximately 20 per cent of Australian have exposure to a chemical outside that intake should be used for establishing and place? children have asthma and 10 per cent being considered in the risk scenario a set GV? Interagency cooperation •• How do you deal with multiple of women are allergic to nickel. In these under consideration. It is important to will be required to enable appropriate pathways of exposure such as the situations, GVs (which, by definition, consider these background exposures apportionment. relatively high exposures in tobacco are based around long-term exposures) in a risk assessment to ensure that the smoke compared with dietary sources? will be irrelevant and unsuitable for situation being assessed does not result The toxicity reference value (e.g. TDI) use. However, the remainder of the risk in exceedence of a TDI, for example. •• Is the bioavailability of the substance gives a measure of the total daily assessment process will still be relevant known? While the amount of validated exposure to a substance that can be for establishing criteria. Background exposure is also important data on bioavailability from oral and to consider for chemicals that may tolerated for a lifetime. In the derivation dermal routes is steadily increasing, not occur naturally but are now of health-based GVs, a proportion of the For most criteria there is a significant a conservative default assumption so ubiquitous that exposures from assumed total exposure to a substance margin of safety between the GV and of 100 per cent bioavailability is sources other than the scenario under is attributed to the exposure medium typical exposures. Safety is enhanced usually assigned where such data is consideration are inevitable. under consideration (e.g. soil, water, food, by a further margin of safety arising not available. air) and the remainder to ‘background’. from the process by which GVs are set. •• Is there a background level of exposure For many agents, there may be several Such background levels of exposure However, for some agents health effects to the agent from media other than the exposure pathways. Copper will be can be associated with the chemical may arise from background exposures one for which GVs are being set? How found in water and food (and, of lesser concentrations present in the environment such as exacerbation of asthma from do you apportion exposures associated importance, in consumer products) so as a result of everyday activities (e.g. urban ozone exposures or cardiovascular with such background? How are these that setting a GV for copper in another emissions from motor vehicles) or natural diseases associated with exposure to background or extraneous exposures medium (e.g. soil) or a particular foodstuff sources (e.g. dissolution of mineral ultrafine particulates (PM and PM ). managed in the standard-setting 10 2.5 (e.g. shellfish) will need to take into deposits). Chemicals may be present in In these instances, a risk management process? (See Section 5.11.1.) account the range of other potential food, air, water and consumer products, decision will need to be taken about the exposure pathways. The exposure route and these all contribute to the quantity •• Who will be involved in setting the GV? incidence of adverse health effects in for dioxins likely to dominate in a risk of the chemical that a person might be the community. Further discussion about assessment is the intake of fatty foods. A decision tree detailing the use of EHRA exposed to on a daily basis. managing these types of environmental While dioxins in soil, air and water may to develop risk-based environmental health problems with no apparent make only a small direct contribution health criteria is provided in Figure 24. Apportionment of the toxicity reference threshold is contained in the NHMRC to intake, these sources need to be This model uses a GV (e.g. ADI) that value to a specific environmental guidance on setting ambient air quality considered in an EHRA in order to is apportioned between background medium can range from 100 per cent standards (NHMRC 2006). account for possible transfers into the exposures and exposures relevant to a (representing negligible background food chain. exposure from any other source) to 20 per particular exposure pathway (e.g. food, A useful source of equations used cent (i.e. 80 per cent to background) water, air or soil). This approach is based generally in toxicology is in Deralenko and Intakes may need to be apportioned where there is a significant background on chronic or subacute exposures. It may Hollinger (1995), while equations used between the different exposure intake from sources other than not be applicable for acute exposures such in EHRA processes specific for assessing pathways. The apportioning of intakes contaminated soil. The fraction as when dealing with a respiratory irritant. contaminated sites is in Schedule raises other questions: attributable to the source medium
76 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 77 under consideration is then used to No specific attribution factor for airborne Since the setting of default values for 5.11.2 •• theoretical, where the term is simply It may be that some international data calculate the GV. exposure has been determined in background is a somewhat arbitrary Using weight of evidence in setting applied as a conceptual framework. sources may be preferable in the Level 1 Australian guidance on setting air process, no specific recommendations environmental health guidance values hierarchy on the basis that they address Various default values have been used quality standards (NEPC 2009), but for AFs are made in this update of However, Weed (2005) noted that, in a specific aspects of toxicity in a more in attributing proportions of the tolerable air quality environmental health criteria enHealth guidance on EHRA. A case- The NHMRC guidance document on review of 276 papers published between detailed way. For example, International intake to specific environmental media. are often set on epidemiological data by-case analysis of the available data setting ambient air quality standards 1994 and 2004 (and which purported to Agency for Research on Cancer (IARC) For example, in Canada, 20 per cent where background exposures would should be made to determine what (NHMRC 2006) also includes a useful apply a WoE analysis), approximately 50 monographs (when current) may present a of total exposure is allowed from each be accounted for in the measured proportion of the tolerable intake discussion of the application of weight- per cent of them could be classified as deeper and more structured appreciation of food, air, water, soil and consumer health effects data. In guidance under estimate could be allocated to the of-evidence (WoE) criteria in standard- simply applying a metaphorical meaning of the carcinogenic potential for a specific products in setting source-specific development for management of exposure routes under consideration. setting. It is equally applicable across to the WoE concept, with no elaboration of COPC than other documents. environmental criteria. US EPA guidance contaminated sites (NEPC 2010), health If there is insufficient data available to the board in setting other health-based how the concept was applied. This means on aggregate and cumulative risk investigation levels (HILs) for specific make such a judgement, default values environmental criteria. The guidance that any study that professes to include While this section primarily refers only assessment (US EPA 2003a) does not substances use soil attribution factors of 20 per cent for any specific pathway notes that, where a range of evidence is WoE to underscore its outcomes needs to sources of toxicological and TDI data, include specific recommendations for varying from zero (representing negligible might be reasonable (as done in Canada), available to support a standard-setting to be carefully examined for the real rather than derived GVs, there may be attributing pathway-specific exposures, background exposure form any other but it is again emphasised that reasoned process, assessing the weight to be meaning of the claim. a need to encompass such GVs where but it does suggest that where a pathway source) to 80 per cent, where there is analysis is preferable to using arbitrary applied to any piece of evidence is a they have been based on a transparent contributes less than 1–10 per cent of a significant background intake from default values. critical part of the analysis. In many risk assessment process. In line with this the population-adjusted dose (PAD), it sources other than contaminated soil. cases, multiple sources of possibly 5.12 philosophy, published Australian ADIs may be discounted in an aggregate risk This background attribution process conflicting data will be available, and GUIDANCE ON should be used in risk assessments when assessment. Other US guidance refers No specific recommendations for is applied only to substances that choosing the most appropriate source for available, but other data may be used to a relative source contribution (RSC) route-specific background exposure demonstrate threshold dose–response standard-setting becomes a crucial step. SELECTING SOURCES OF with appropriate justification. Different factor, which defines the relationship attribution factors (AF) were made in characteristics. For substances exhibiting If the conflicting data results in different agencies are likely to have used differing between source-specific and background the original 2002 enHealth document, non-threshold characteristics (mainly reference doses or concentrations derived TOXICOLOGICAL DATA AND risk assessment and standards-setting exposures. In the case of setting drinking- although such a factor was included in carcinogens) background exposures are by different regulatory authorities, it ENVIRONMENTAL HEALTH methodologies and these differences water guidance (US EPA 2000c), the RSC some of the equations recommended for not considered in the EHRA, since the may be difficult to resolve the apparent should be appraised by the risk assessor. factor specified to account for non-water calculating exposure such as the equation risk estimate output is extra risk (i.e. risk conflicts unless there is sufficient clarity CRITERIA All documents, particularly those in the sources (except for carcinogens) was recommended in 2002 for the derivation above background). about the sources of data used, and how second and third levels, require rigorous recommended to range from 20–80 per of drinking-water GVs. they were used to derive the values, and The original 2002 version of this appraisal for relevance, validity and cent, with 20 per cent recommended The adjustment in the calculation is: some judgement can be made about the enHealth document included a accuracy. An important consideration is the relative weights to be given to the different as the default value in the absence of GL = RL × BM × AF × ED TDI(adjusted) = (1 – background) × TDI section that categorised sources of frequency with which documents and GVs specific data. In some specific cases, IR × UF × AT data sources. toxicological information and established are updated via a periodic review process. where human exposure data has been Where data suggests that background environmental health criteria according available, the RSC has been set at GL = the guideline concentration of exposure is essentially negligible In reviewing the concepts that underscore to a hierarchy. This hierarchy, modified 5.12.1 100 per cent. contaminant in water (contributing less than 5 per cent of WoE analysis, Weed (2005) noted there to include some additional sources, is Level 1 sources the threshold TDI), the allocation to the is no clear or universally accepted reproduced in this section (5.12.1) RL = toxicity reference value (often a •• National Health and Medical Research There are some Australian precedents for specific exposure pathway of interest can definition of what constitutes a WoE NOAEL) Council (NHMRC) documents assuming default values for attributable approach. Weed pointed out that WoE can The sources are grouped into ‘levels’ in be 100 per cent of the threshold TDI. and documents from other joint background exposures. For example, be taken to include processes where the which the hierarchy generally signifies an BM = body mass, often 70 kg for an adult Commonwealth, state and territory in setting water quality standards, the A complication arises when the approach is: order of preference. For example, it may AF = the proportion of total exposure health organisations. These may be a Australian drinking water guidelines background exposure exceeds the TDI be assumed that Australian guidance attributable to drinking water •• metaphorical, where it is implied that source of Australian guidance values (ADWG) (NHMRC-NRMMC 2004) because, in theory, a health-based GV merely a collection of data or studies values accorded Level 1 status should •• ADI list from the relevant office of the allocates 10–80 per cent to drinking ED = exposure duration (if exposure is that is sufficiently protective of human has been taken into consideration, take precedence over other sources, Australian Government Health portfolio water as the source for setting ADWG less than continuous) health cannot be calculated. A few but there is no clear indication of what provided they are reasonably current. (currently the Office of Chemical Safety GVs. The Australian guidelines for water approaches are available to address type of weighting has been applied, or Other Level 1 sources may be more useful IR = ingestion rate, often taken as – OCS) – regularly updated at
78 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 79 UNEP International Programme first published by Toxicology •• Alliance for Risk Assessment (ARA) – With justification, and acceptance by the exposure to trace levels of chemicals Renwick (2004; 2005) has described the on Chemical Safety (IPCS), which Excellence for Risk Assessment a consortium of organisations formed local jurisdiction, they may be suitable to represents very low risks if chemicals act JECFA application of the TTC concept to produces environmental health criteria (TERA), a non-profit US organisation. to promote collaboration on risk use if no GVs are available. via more generalised modes of action. the safety evaluation of flavouring agents. monographs and concise international ITER is now also available via the assessment issues – maintains the TTCs have been developed for classes Since 1996 some 1,200 compounds have chemical assessment documents TOXNET network Risk Information Exchange (RiskIE) 5.12.4 of substances with a systemic mode of been assessed using the TTC concept. (CICADs). Documents detailing •• A special issue of Toxicology Applied database, which catalogues reports Selecting data sources toxicological action and with exposure international acceptable daily intakes Toxicology (Vol 233, Nov 2008) on risk assessment documents from by ingestion. The TTC concept has also been adapted The reliability of the data used in a (ADIs), tolerable daily intakes (TDI) or included several papers on databases various governmental and non- for: risk assessment should not be based tolerable weekly intakes (TWI) may be useful in risk assessment (see Waters governmental sources The TTC approach was for many years solely on the position of its source in the •• deriving criteria for soil risk found in evaluations by the WHO/FAO & Jackson 2008; Wexler 2008; •• Scientific societies and organisations put forward as a pragmatic solution for above hierarchies. The most important management for chemicals of Joint Meeting on Pesticide Residues Woodall 2008; Woodall & Goldberg that specialise in toxicology and risk addressing low concentrations of additives consideration is that the data is supported unknown toxicological hazard or (JMPR) and by the Joint FAO/WHO 2008; Wullenweber et al. 2008) assessment issues – the US Society in food (Frawley 1967; Munro 1990; by sound science and contemporary potency at contaminated sites Expert Committee on Food Additives of Toxicology (SOT) has long been a Munro et al.1996). While it was first •• Dutch National Institute for Public risk assessment methodologies, and (Wilson et al. 2000) (JECFA). The WHO has also published leader in developing and evaluating applied in a regulatory sense by the FDA Health & the Environment (RIVM) that a WoE approach has been used •• judging whether ingredients at guideline documents on ambient and toxicological and risk assessment (FDA 1995; CFR 2001) and was later reports and databases. to assess their worth. This means that low concentration in personal and indoor air quality (WHO 2000b, 2010) methodologies, and maintains used by the European Commission (EC) risk assessors should not simply rely on household-care products require •• enHealth Council documents specialist sections that undertake (2003) to address chemicals migrating 5.12.2 ‘looking up numbers’ but should gain an toxicological testing (Blackburn projects to advance risk assessment from plastic packaging into food, it is •• National Environmental Health Level 2 sources appreciation of the currency of the source et al. 2005) methodology. The Society for Risk now applied in a number of regulatory Forum documents distributed by the and the underlying science from which Peer-reviewed journals: These may Analysis (SRA) is another society activities (see Table 11). Commonwealth Department of Health the numbers have been derived. and Ageing provide opinions that do not meet that has played a leading role in general scientific agreement. and may advancing risk assessment and risk •• International Agency for Research on be a suitable source where no GVs communication methodologies Table 11: Current uses of the TTC approach in chemicals regulation Cancer (IARC) monographs 5.13 have been established by competent •• Unpublished industry reports •• WHO/FAO Joint Meeting on Pesticide authorities. submitted for regulatory purposes – CHEMICALS WITH NO Regulatory agency Use References Residues (JMPR) monographs these may have restricted availability AVAILABLE TOXICITY DATA United States Food and De minimis level (i.e. level of FDA (1993ab, 2006) •• NICNAS Priority Existing Chemical Industry or commercially based but information may be available in Drug Administration minimum importance) for regulation (PEC) reports publications: With justification, and evaluation reports from regulatory When no suitable toxicological data is (FDA) of minor contaminants (i.e. chemicals acceptance by the local jurisdiction, •• US Agency for Toxic Substances and agencies that have reviewed available for a COPC, it will be difficult in food packaging materials that can the following may be suitable for use: migrate). Disease Registry (ATSDR) documents individual reports to establish an appropriate toxicological for general toxicological reviews and •• European Centre for Ecotoxicology and •• Occupational health and safety sources reference value (ADI or TDI), and this Threshold of toxicological concern reference doses Toxicology of Chemicals (ECETOC) – these may be a useful source for will impact on the ability to derive an (TTC) is applied as a threshold of •• National Toxicology Program monographs, JACC reports and toxicological data and reviews, but appropriate health-based GV. In the regulation for indirect food additives. The US FDA has dealt with 183 (NTP) carcinogenicity appraisals, technical reports occupational exposure criteria must absence of a suitable GV, there may applications under this regulation and not be used in a general public health be grounds for assuming a level of which report in detail the results of •• Chemical Industry Institute of issued 78 exemptions using the TTC exposure to a COPC that would be of no carcinogenicity tests on a wide range Toxicology (CIIT) reports (now the CIIT context without appropriate adjustment concept (Barlow 2005). of chemicals Centers for Health Research, part for the different durations of exposure, toxicological consequence, and hence the Joint FAO/WHO Expert Evaluation of flavouring substances. JECFA (1993; 1995; 1999), •• OECD Standard Information Data Sets of the Hamner Institutes of Health the inclusion of susceptible sub- COPC could be discarded or discounted Committee on Food TTCs for different structural classes Munro et al. (1999), Renwick (SIDS) and SIDS initial assessment Sciences at Research Triangle Park, population in the general community in an EHRA. This concept has been (e.g. children) and the methodological variously described as the threshold of Additives (JECFA) have been used for the safety (2004; 2005) reports (SIAR) North Carolina evaluation of over 1,200 flavouring differences in the setting of criteria. toxicological concern (TTC) approach substances. •• EPA reference doses, cancer slope •• International Life Sciences Institute or the concentration of no toxicological factors and unit risk factors (ILSI) and ILSI – HESI (Health & concern (CoNTC) approach (Drew & European Food Safety Evaluation of flavouring substances EFSA (2004) Environmental Sciences Institute) 5.12.3 •• TOXNET – sets of toxicology databases Level 3 sources Frangos 2007). Authority (EFSA) maintained by the US National Library workshop and technical/project European Medicines Assessment of genotoxic impurities in CHMP (2004) committee reports These are sources not covered in of Medicine (includes TOXLINE, The thresholds determined using the Agency (EMEA) pharmaceutical preparations. CCRIS, GENETOX, IRIS) •• Toxicology Excellence for Risk Levels 1 and 2. Using this information TTC concept are intakes of chemicals See also Dolan et al. (2005) •• The International Toxicity Estimates for Assessment (TERA) – a non-profit requires justification that no other below which a given compound of Risk Assessment (ITER) database – a organisation that publishes reports on sources are available and an appraisal known structure is not expected to European Commission, The TTC principle has been endorsed EC JRC (2005) compilation of risk data from multiple methodologies and selected issues/ of the methodology detailing the present a toxicological concern. On the Joint Research Centre as a mechanism for the regulation of chemicals under draft chemical sources (including some respected chemicals and supports training in level of conservatism and range of basis of classical pharmacological and legislation reforms being considered risk assessment uncertainties inherent in the approach. toxicological concepts of dose–response, national risk assessment databases), by the European Union.
80 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 81 •• establishing ADIs for chemicals The 5th percentile NOEL (no observed in appreciable health risk, even if the in Europe, as well as the increasing chemical. Read across is primarily useful •• complex causality for many of the likely to be present as impurities in effect level) of each of the three substance was later found to be a number of high production volume (HPV) for hazard prediction. It has limited health conditions addressed in pharmaceutical manufacturing (Dolan Cramer classes was divided by an carcinogen. According to Munro (1990), chemicals requiring assessment in OECD capabilities for predicting quantitative the EHRA et al. 2005) uncertainty (safety) factor of 100 to a daily intake at the TTC of 0.02 µg/kg bw programs, should provide impetus for the dose–response behaviour. It relies on •• confidentiality of health and •• use as a screening tool for risk yield TTC values that are somewhat corresponds to a 96 per cent probability further development and use of SAR and there being a high-quality toxicological commercial information preventing assessment of air toxics (Drew & higher than those created by the that the lifetime risk of cancer would be QSAR techniques. dataset for the reference compound. full disclosure FDA for carcinogens. No formal less than one in a million (1 × 10–6). Frangos 2007) •• the atmosphere of fear, antagonism stratification of toxicological endpoints SARs already utilised in regulatory •• (more recently in Australia) setting and distrust being so charged that it was used in establishing NOAELs for The TTC that is protective of cancer toxicology include: 5.15 concentrations that would be unlikely inhibits meaningful dialogue between the three Cramer chemical classes. endpoints is termed a ‘generic TTC’, to to pose a human health risk for •• recognition of structural elements UNCERTAINTY AND the stakeholders. The NOAELs are: differentiate it from the TTC developed chemicals likely to be present in for non-cancer endpoints and using the that act as alerts for particular types recycled water (NRMMC, EPHC, •• Class I: 3 mg/kg/day (equates to a SENSITIVITY ANALYSIS Cramer classification. The TTC estimate of toxicological behaviour (e.g. In formulating an EHRA report it is crucial NHMRC 2008). TTC of 30 µg/kg bw/day) epoxides or other reactive metabolic that all uncertainties and knowledge of 0.02 µg/kg bw/day is conservative, 5.15.1 •• Class II: 0.9 mg/kg/day (equates to intermediates which that confer DNA- gaps be acknowledged and guide the erring on the side of safety, because of General The following description of the TTC a TTC of 9 µg/kg bw/day) the numerous compounding conservative and protein-interactive capabilities) development of risk management options approach is taken from the Australian •• Class III: P 0.15 mg/kg/day (equates assumptions used to derive the low-dose •• recognition of common structural At the completion of a risk assessment, (see Chapter 7). It is also important that guidelines for water recycling (NRMMC, to a TTC of 1.5 µg/kg bw/day). cancer risk estimates (Barlow et al. 2001; elements in chemical classes that it may become apparent that there are these uncertainties be managed in a EPHC, NHMRC 2008). Kroes et al. 2004). are consistent with known patterns inherent limitations to the outcomes, consistent and scientifically defensible In applying the TTCs to derivation of toxicity (e.g. organophosphonate such as: way, and that there is a clear explanation In establishing TTCs for chemicals of drinking-water guidelines, a more groupings that enable of how ‘scientific judgement’ may have that are not carcinogens, an conservative approach has been •• information gaps (e.g. effects of 5.14 phosphorylation of the active site been applied to the management of evaluation of toxicological databases applied to reflect the use of safety mixtures, low-level and variable on acetylcholinesterases) these uncertainties. This may include undertaken for non-carcinogenic factors used in the ADWG (NHMRC, QSAR AND READ ACROSS exposures over time, relative endpoints is used (Munro et al.1996; careful description of definitions of default NRMMC 2004). These guidelines •• grouping of chemicals based on contributions of lifestyle factors versus 1999; Kroes et al. 2000; 2004). In TECHNIQUES parameter inputs or using more complex apply a safety factor of 1,500 to recognisable structural features other environmental hazards, variations these evaluations, some 900 non- probabilistic approaches to defining organic chemicals (95th percentile). that lead to common toxicological in sensitivity) carcinogenic organic chemicals were Better understanding of structure–activity bounding values, or intervals within To achieve this, an additional safety properties (e.g. dioxin-like chemicals assigned to three ‘classes’ based on relationships (SAR), especially when •• poor exposure information (e.g. which the risk assessor expects the best factor of 15 has been applied in and others that interact with aryl their chemical structure, presence of combined with quantitative information complex mixtures of hazards estimates of risk to lay. converting TTCs (which already hydrocarbon (Ah) or peroxisome structural alerts for toxicity and known (quantitative structure–activity relationship with complex behaviours in the include a safety factor of 100) to proliferator (PPAR) receptors) metabolic pathways, according to the – QSAR) may facilitate prediction of environment, limited knowledge about It may be important to carry out proper drinking-water guidelines. •• computational systems that use classification scheme of Cramer et toxicological properties of chemicals the actual or potential population and sensitivity and uncertainty analyses so the a combination of features of the al. (1978). The Cramer classification without testing, or where no testing has sensitive sub-populations geographic, level of effort expended in an EHRA can molecule (electronic, physico- scheme divides chemicals into three 5.13.1 been done to establish the toxicological variations in exposure) be appropriately matched to the precision chemical, size, hydrophobicity, etc.) classes according to their predicted Can the TTC approach be applied profile of a new chemical. There will also •• limitations of toxicological and of the desired outcomes (NRC 2008). to predict properties (e.g. EPIWIN) toxicity as judged from structural alerts to carcinogens? be consequent benefits in terms of lower epidemiological research (e.g. If the outcomes or advice to the risk costs, shorter testing time frames and less •• knowledge-based or rule-based manager will not be materially affected and metabolism: A generic approach has been developed small populations, limited exposure use of animals. QSAR may also be useful systems that compare many by adopting more simplistic approaches, •• Class I: substances of simple for potentially genotoxic carcinogens information, multifactorial causes of in complementing the increasing use of in parameters of a dataset and enable it may be wasteful of scarce resources to chemical structure with known many diseases) using the TTC approach (NRMMC, EPHC, vitro and in silico technologies to provide predictions of the properties of metabolic pathways and innocuous •• ‘background noise’ affecting research use more sophisticated methodologies NHMRC 2008). insights into toxicological properties of chemicals that share common end products that suggest a low into common diseases or symptoms, (e.g. deterministic versus Monte Carlo chemicals without using live animals. structural features. One such assessment of exposures). Similarly, the order of toxicity The FDA (1995; CFR 2001) regulatory population heterogeneity, and the As well as facilitating chemical and drug commercially available system is sophistication of analytical techniques TTC is based on a carcinogenic fact that it is expensive and time •• Class II: chemical structures that development by industry, regulatory DEREK, a computer-based SAR used to measure environmental potency database of more than 500 consuming. are intermediate they are chemicals recognition of QSAR is also growing in program (Sanderson and Earnshaw, concentrations should be matched to the chemicals examined in more than 3,500 that are less innocuous they may importance. For example, it has been 1991), although its utility is mainly level of precision required in the EHRA. experiments. The FDA (1995; CFR 2001) Some of these limitations may be contain reactive functional groups suggested that up to 10 per cent of limited to predicting sensitisation but do not contain the structural and other leading researchers (Munro apparent before beginning the risk new chemical notifications in the UK and carcinogenic properties. 5.15.2 features suggestive of toxicity et al.1996; 1999) have concluded that, include QSAR data, and this proportion is assessment process. For example: if no toxicological data is available to Uncertainty analysis •• Class III: chemicals for which expected to grow over time. Another alternative approach when data •• the large number of combinations of derive a health-based guideline for a structural features or likely metabolic on a specific chemical is lacking is to hazards, exposures and health states Uncertainty in health risk assessment is chemical, intakes of 1.5 µg/person/ pathways permit no strong The implementation of the REACH use ‘read across’ techniques to make leading to complexity that cannot be the lack of knowledge about the correct day (0.02 µg/kg bw/day for a body presumption of safety, or may even (Registration, Evaluation, and readily resolved value such as a specific exposure weight of 70 kg) are unlikely to result informed predictions about the toxicity suggest significant toxicity. Authorization of Chemicals) program profile from a known, and closely related measure or estimate. Uncertainty is
82 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 83 distinguished from variability, which Uncertainty analysis is generally a in professional judgement, and •• The uncertainty analysis should seek as input parameters one at a time to The Monte Carlo method reduces the refers to true differences in attributes qualitative process; however, in some incomplete analysis. to communicate which uncertainties determine how such changes influence requirement for sensitivity analysis but due to diversity or heterogeneity; cases it can be semi-quantitative or •• Parameter uncertainty: uncertainty are most important to the conclusions the final output. Variables are changed may not eliminate it, depending on the variability cannot be reduced by further quantitative. affecting a particular parameter such of the risk assessment. within a defined range while leaving the model used. measurement or study, although it can be as measurement errors, sampling •• The level of detail in the uncertainty others constant and determining the better characterised (NRC 2008). The first step should be a consideration errors, variability, and use of generic or analysis should be commensurate with effect on the output – the risk estimate. 5.16 of the conceptual site model and what surrogate data. the scope of the risk assessment. The procedure involves fixing each Both uncertainty and variability contribute aspects of that model are uncertain uncertain quantity, one at a time, at its •• Model uncertainty: uncertainty in •• Uncertainty analysis should be INTERPLAY OF SCIENTIFIC to uncertainty in the estimation of risk and and how that uncertainty has been credible lower bound and then its upper scientific theory affecting the ability of expressed in terms that can be should be adequately assessed in a risk accounted for. bound (holding all other at their medians), JUDGEMENT AND SCIENCE a model to make predictions. understood by the risk manager and assessment. Such consideration needs to and then computing the outcomes for other stakeholders. POLICY be done transparently so that all users of The second most important part of the NRC (2008) provides a detailed each combination of values (US EPA the risk assessment can understand the uncertainty assessment is an evaluation •• Uncertainty and variability should be 1992). It can be used to test the effects evaluation of the techniques currently The interplay between these processes approach taken. of the uncertainty and variability in kept conceptually separate. of both uncertainty and variability in provided for in US EPA guidance and and the importance of providing the data available relating to the site, input values. The substitution of input concludes that although a number of appropriate explanation of assumptions, An analysis of the uncertainty in the situation or activity being assessed. Data The combination of uncertainty in parameters should be informed by usable methodologies are provided, the use of scientific judgement, and the risk assessment is important because of will always be limited. However, the risk the scientific data and assumptions knowledge of the upper and lower bounds it is unclear what level of detail is overlay of ‘science policy’ considerations the following: estimates based on even quite limited (the ‘inputs’) and inability to validate of the expected parameter distributions. required to capture and communicate is illustrated in the ebb and flow of data can be fit for purpose if the exposure assessment results directly or to isolate •• Information from different sources key uncertainties. A further comment regulatory actions and interpretations concentrations are a long way below (or and evaluate the impact of a resulting Sensitivity analyses can be used to carries different kinds of uncertainty is that quantitative methods suffer from surrounding the presence of chloroform above) toxicity reference values which decision (the ‘outputs’) creates a situation identify the most important input variables and knowledge of these differences the difficulty in sensibly quantifying all in drinking water and its surrogacy as indicate that the risks are very low (or in which decision makers, the scientific (or groups of variables) that are critical is important when uncertainties are uncertainties, and that the apparent an indicator of disinfection by-products very high). Decision making based on precision of quantitative analysis for some community, the public, industry and to the outcome of the risk assessment. combined for characterising risk. (Box 1). such uncertain but quite clear results is uncertainties may distract attention from other stakeholders have little choice but It follows that variation of some inputs •• The risk assessment process, with risk straightforward. Where risks are close other, possibly equally important but to rely on the overall quality of the many may have inconsequential effects. management input, involves decisions to or slightly above the relevant toxicity unquantifiable, uncertainties. processes used in the conduct of risk Sensitivity analysis can develop bounds regarding the collection of additional reference values or ‘target risk’ level (the assessment to provide some assurance on the distribution of exposure or risk. data (versus living with uncertainty). In ‘grey’ zone), the issue of the uncertainty In most health risk assessments, it is that the assessment is aligned with A sensitivity analysis can also estimate the risk characterisation, a discussion and variability in the data becomes much unlikely that quantitative uncertainty societal goals (NRC 2008). the range of exposures or risk that result of the uncertainties will help to more important and so the uncertainty analysis will provide value given the from combinations of minimum and identify where additional information/ assessment needs to be more detailed. effort required to undertake it. A clear 5.15.3 maximum values for some parameters data could contribute significantly qualitative analysis is considered Sensitivity analysis and mid-range values for others (US to reducing uncertainties in the risk When assessing risks, uncertainty sufficient in most cases to provide EPA 1989). Effort may then be directed Sensitivity analysis is an important final assessment. can arise from missing or incomplete the communication of the effects of to the collection of additional data for step in the risk characterisation process, •• A clear and explicit statement of information, be incorporated into the uncertainty that is necessary. these important variables; as additional especially where modelling has been used the strengths and limitations of a scientific theory behind the model used to data is collected, the uncertainty in the to determine important components of risk assessment requires a clear make predictions, and factors affecting a NRC (2008) and IPCS (2008) provide ‘true’ value is reduced, and it may be the EHRA (see Section 8.7.4). It provides and explicit statement of related particular parameter, for example, errors useful guidance on the principles to be possible to define a smaller range for a quantitative estimate of the effect of uncertainties (US EPA 1995b). in sampling. Such uncertainty has the adopted for uncertainty analysis; these a given parameter. The uncertainty in uncertainty and/or variability in the input potential to cumulatively overestimate or have been adapted for specific relevance the results of the risk assessment may •• Characterising uncertainty in parameters on the results of the risk underestimate risk during an assessment. to the enHealth document. therefore be reduced. a risk assessment informs the assessment and it should be undertaken An assessment of uncertainty is a part of stakeholders about the range of •• Risk assessments should provide when a risk assessment is conducted the health risk assessment process and All risk assessments where conclusions possible risks from an exposure. qualitative (as a minimum) or using a deterministic exposure model. Risk estimates may sometimes consequently must be addressed for each quantitative description of uncertainty are derived using modelling should step of the risk assessment and for its incorporate a sensitivity analysis and diverge widely (NRC 2008). and variability consistent with available While a single value must be entered for cumulative effect from all of the steps. describe the variability in the model •• Characterising the uncertainty in a data. The information required to each parameter in a deterministic model, outputs generated by plausible variation risk assessment associated with a conduct detailed uncertainty analysis it is unlikely that reasonable inputs for There are three broad types of uncertainty in the inputs. Note that some input given decision informs the decision may not be available in many each parameter can be limited to a single (US EPA 1992): variables may be connected and unable maker about the range of potential situations. value. This may be due to uncertainty to vary independently. Monte Carlo risks that may result from the decision •• Scenario uncertainty: uncertainty •• Sensitive sub-populations should be and/or variability. A range of reasonable models, where inputs are described by (NRC 2008). arising from missing or incomplete considered to the extent that they are values will be defined as appropriate for a probability distribution functions, provide information such as descriptive not covered by the selected toxicity given input parameter. Sensitivity analysis probability distribution function outputs. errors, aggregation errors, errors criteria (generally they will be). is the process of changing variables used
84 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 85 BOX 1: The initial extremely high dose bioassays results on The Chlorine Chemistry Council sought a court review of Chloroform – New evidence can eventually change regulatory perspective, albeit slowly chloroform (NCI 1976) provided the case that was widely the US EPA decision because the SDWA requires the US cited throughout the late 1970s and early 1980s that EPA to use the best available science in setting standards (Adapted from Hrudey 2009). chloroform and, by extension, THMs, were carcinogenic. and regulations. Although the US EPA acknowledged that The NCI (1976) results were obtained by a method (high the best available science called for raising the MCLG Chloroform and the related trihalomethanes (THMs) were delivering a slug dose once per day in a vehicle like corn oil dose of chloroform in corn oil) that was later found to be first identified as by-products of chlorine disinfection by provides a higher peak loading to the liver than consuming above zero, it had nevertheless decided to retain the zero much more toxic than the equivalent dosing of chloroform Johannes Rook, a Dutch water chemist (Rook 1974). Rook water with an equivalent dose of dissolved chloroform MCLG. On 31 March 2000, the US District Court ruled that in water (Bull et al. 1986). The comparison of corn oil had consistently identified chloroform in treated, but not spread out over a day. the US EPA had violated the SDWA by failing to use the raw water samples. He chose not to publish the identity versus water as a vehicle was undertaken to explain the best available science. The court found that the EPA action results from a study providing high concentrations of of the large chloroform peak until he had figured out what The results of this high dosing showed strong evidence of of setting the MCLG of chloroform at zero to be ‘arbitrary chloroform (up to 1,800,000 µg/L) dissolved in drinking was causing its formation but he was not troubled about liver tumours in mice (98% of males and 95% of females and capricious’ and in excess of statutory authority. The US water (Jorgenson et al. 1985) that produced no significant consumer health risk, noting (Symons 2001): ‘Our health at lifetime average doses of 277 mg/kg/d and 477 mg/kg/d, EPA withdrew the zero MCLG in May 2000, subsequently carcinogenic response. officer told us chloroform was a normal constituent of respectively, 36% of males and 80% of females at lifetime replacing it with a MCLG of 70 µg/L in 2003. cough syrups and was not known to be particularly toxic.’ average doses of 138 and 238 mg/kg-d respectively) in the mouse experiments. These high dose levels were from The impact of extremely high doses of chloroform in corn Meanwhile, WHO had changed its drinking water guideline There was also originally little health concern about oil to the liver was first noted as evidence of cytotoxicity 27 to 115% of published median lethal doses (LD50) for the for chloroform from 30 µg/L in its first edition ofGuidelines chloroform at US EPA (Symons 2001) because of the on liver cells. Larson et al. (1994, 1995) demonstrated by mouse (Hill et al. 1975), suggesting that the B6C3F1 strain for drinking-water quality (WHO 1984) to 200 µg/L in the widespread use of chloroform in consumer products, of mouse used in these cancer bioassays was unusually direct experimentation that the corn oil gavage delivery second edition (WHO 1993a). The rationale associated with but they confirmed finding higher levels of THMs with tolerant of chloroform. The rats dosed at up to 200 mg/kg/d of chloroform induced cytotoxicity and cell proliferation recognising that chloroform exhibited a threshold for acting increasing chlorine contact during disinfection (Bellar et in liver for mice and kidney and liver for rats. The failed to show a significant excess of liver tumours relative as a carcinogen was invoked to justify this change. al. 1974). The concern about chloroform only started to to controls. Rats did show a significant increase in kidney mouse experiments found this effect for the corn oil escalate when it was recognised that THMs were being gavage, but not for direct delivery of similar daily doses tumours, but mice did not. The initial NCI (1976) carcinogenic finding on chloroform, formed from the reaction of chlorine with natural organic orally by drinking water. These findings on a plausible taken together with the background expectation that material (NOM), a constituent that is ubiquitous in surface mechanism for chloroform carcinogenicity were supported Within four months of the publication of the NCI bioassay substantial numbers of human cancers could be explained water supplies. by extensive evidence showing virtually no mutagenic results, the US Food and Drug Administration banned by environmental contamination, resulted in more than activity for chloroform (Golden et al. 1997). The earlier the use of chloroform in cosmetics. This was a dramatic 65 epidemiology studies of varying quality from 1977 to Shortly after the growing body of evidence showing noted distinction in mechanism of tumour formation from change in relation to chloroform, which had been widely 2008, seeking to determine if some measure of chlorination chloroform appearing in chlorinated drinking water cytotoxicity rather than genotoxicity justifies a threshold used as an anaesthetic from the mid-1800s into the DBPs was associated with an increase in one or more supplies, the National Cancer Institute (NCI) published approach to risk assessment rather than a no-threshold early 1900s. types of cancer. The epidemiologic evidence regarding results of a rodent cancer bioassay on chloroform NCI approach for THMs (Fawell 2000). (1976). This bioassay was conducted in accordance with cancer has been reviewed at various times (IARC 1991, Health concerns associated with chloroform and THMs the practices of that day, i.e. it was designed to determine Mills et al. 1998, ICPS 2000, IARC 2004). Overall, the rapidly led to the adoption of drinking water guidelines and According to the US SDWA the maximum contaminant the potential for chemical substances to cause cancer in epidemiologic evidence has generally been found to be standards; Canada was first in 1978 to adopt a guideline level goal (MCLG) is the maximum level of a contaminant mammals and were designed to maximise the ability of the insufficient to declare chlorination DBPs to be carcinogenic maximum value for THM4 of 350 µg/L. Then in 1979, the in drinking water at which no known or anticipated adverse experiment to reveal any carcinogenic effect by using the in humans. The evidence for colon and rectal cancer has US adopted a regulatory standard for THM4 under the health effects would occur, and which allows an adequate maximum tolerated dose. been suggestive of a causal association while the evidence Safe Drinking Water Act of 100 µg/L as a running annual margin of safety. US EPA policy for carcinogens in drinking average of four quarterly samples. In 1984, WHO proposed water had required a MCLG to be zero, apparently ignoring for bladder cancer has been clearly the most consistent, Dosing in this experiment was done as a daily bolus dose providing the greatest likelihood of being causally of chloroform dissolved in corn oil. The initial high dose in a guideline for chloroform of 30 µg/L based on an estimate the possibility of a non-genotoxic carcinogen having a associated with chlorination DBPs (Mills et al. 1998). There female rats of 250 mg/kg(bw)-d had to be reduced to 180 that this would assure less than a 1 in 100,000 lifetime threshold. However, the foregoing toxicological evidence is now common understanding among experts on DBPs mg/kg(bw)/d after 22 weeks because of the frank toxic cancer risk assuming a linear extrapolation to zero dose. on the mode of action of chloroform resulted in a US EPA and health evidence that chloroform in particular and effects that were observed. Mice proved more tolerant The Australian drinking water guidelines established a expert review panel recommending the abandonment of to chloroform, so their initial doses of 200 and 400 mg/ guideline value of 250 µg/L in 1996, based on the NOAEL the MCLG of zero and replacement with a limit based on THMs in general are at best surrogates for some DBPs as kg(bw)-d were increased after 18 weeks to 300 and 500 for kidney toxicity in a 90d rat study, concluding that ‘In an estimated threshold. Thus in 1998, the US EPA (1998a) yet unidentified in chlorinated drinking water that may pose mg/kg(bw)-d. For comparison, a human dose of chloroform view of the safety factors used in the derivation of the proposed to raise the MCLG to 300 µg/L in accordance with a drinking water cancer risk. Despite the original focus on equivalent to the highest dose rate would correspond guideline value, it is unlikely that short-term consumption this expert advice. Because many intervenors protested this chloroform and THMs as carcinogens in drinking water, to more than 25,000 times the daily dose achieved by of water containing significantly higher concentrations of precedent-setting measure, the US EPA (1998b) Final Rule more than 30 years of evidence now verifies that these consuming 2 L per day of drinking water containing trihalomethanes would pose a health risk.’ This value was withdrew the proposal to change the MCLG for chloroform chemicals do not pose a cancer risk at realistic drinking 100 µg/L of chloroform daily for a lifetime. Furthermore, confirmed when reviewed in 2004 (ADWG 2004, 2010). from zero to 300 µg/L (Pontius 2000). water exposure levels (Hrudey et al. 2003).
86 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 87 Chapter 6: Community engagement
6.1 •• The Queensland Government has •• provide residents with knowledge so conflict and mistrust. High levels of stress, •• CRC-CARE has published guidance 6.2.2 published two guidance monographs they can participate more effectively concern or controversy are bound to on risk communication that The social context of risk perception INTRODUCTION on community engagement and in making decisions about how to make the already complex task of risk specifically addressed community risk communication (Queensland manage the risks. communication more difficult. engagement around contaminated The social amplification of risk framework (SARF) is a concept that explores how Engaging with stakeholders as part of Government 2001a, b). sites (Heath et al. 2010). media and other sources of information the EHRA risk process is a cornerstone It is crucial that consultation with This chapter does not set out to •• Two community consultation guides dissemination can provide a stimulus to effective risk management, since stakeholders occurs early in the EHRA be a complete guide to community 6.2.1 have been published by the WA to community engagement, but it can it provides a basis for developing process, and that it be done with an engagement and risk communication. Risk perception Government: one by the Department also lead to exacerbation of community sustainable risk management options of Environment and Conservation understanding of the factors likely to Rather, it attempts to summarise All parties, both expert and non-expert, concerns by maintaining an ongoing over which a concerned community can (2006) and another by the Department shape community perception of the risks. some key points from the expanding will have perceptions of risks. Experts and spreading of narrative relating feel a sense of ‘ownership’. This principle of Health (2006). While the first of literature on community engagement, and non-experts alike are influenced by to environmental incidents or issues is enunciated throughout Chapter 1 and these guidance documents is primarily It is also crucial that the risk assessor be risk perception and effective risk emotion, beliefs and their views of the (Pigeon et al. 2003). SARF can be both in all contemporary guidance on EHRA. aimed at managing consultations aware that ordinary people perceive risks communication. world (Thomas & Hrudey 1997). an integrative and a predictive model Effective community engagement can also in quite a different way to the scientific on contaminated sites, the second for analysing the social context of risk facilitate transfer of risk assessment and provides a framework for developers professional. Emotional factors often There are a number of books and reviews Heuristics is the psychological term used perception and risk communication. The risk management information, a process of new as well as existing proposals to dominate the way risks are perceived, that provide information relevant to risk to describe the process whereby people influence of media reports on maintaining referred to as risk communication. use with communities. Both include especially when the risks are thought to perception and risk communication, frame their perceptions of risk. Heuristics community concerns around a series of useful general advice on community be beyond the control of the individual which are important in the community are essentially ‘rules of thumb’ by which reports of cancer clusters in Australia It is important that risk communication engagement. and to affect themselves, family members engagement process. For example: we make judgements about everyday has been explored in the SARF context encompass a clear description of the or close relationships. Risk perception is •• enHealth has produced a national •• Paul Slovic, one of the leading occurrence. These rules are simple, (see Stebbing et al. 2008). There are also assumptions and underlying uncertainties discussed in Section 6.2.1. guidance document that includes ‘gurus’ on risk perception, has widely applicable and often reasonably instances where the media have raised inherent in the risk assessment pragmatic advice on how to written extensively on the topic. His accurate. However, they can lead people concerns about a particular or perceived processes. These points are further The potential mismatch in language and understand and respond to community monograph (Slovic 2000) addresses into making snap judgements about risk, and then ‘walked away’. In such emphasised in guidance on how to write understanding can be illustrated with concerns about environmental health heuristics (emotional factors) and other risks, when a more deliberate or logical instances, the media may engender comprehensive reports (see Chapter 7) a hypothetical description of the risks issues (enHealth 2006). cognitive theories of risk perception. community concern without awareness or associated with aflatoxin contamination analysis could lead to a different appraisal (Finucane 2004). acknowledgement of the consequences To be effective, community engagement of peanuts, as described by an ‘expert’, •• Melissa Finucane (Finucane 2004) and may use the argument when called must occur early in the process and be 6.2 juxtaposed against the real question has written a chapter on risk There are different types of heuristic to task, that ‘it is the community’s right sustained through a structured strategy. that a community member may be communication in an Australian approach: to know’. The strategy should provide for meetings RISK PERCEPTION AND concerned about. context in a textbook on Australian EHRA (Cromar et al. 2004). This with concerned groups as well as the •• An ‘availability heuristic’ describes RISK COMMUNICATION chapter also includes information on Critics of SARF have suggested that the dissemination of information in printed Expert: a probability estimate based on our the heuristic factors that shape or framework ignores the nature of perceived form (e.g. newsletters) or online. IN THE CONTEXT ‘A lifetime exposure to aflatoxin at a ability to imagine an outcome. It is influence risk perception. risk by implying that the risk must be concentration of 20 ppb in food, assuming influenced by the frequency with which shown to be ‘real’ before it is socially There are a number of sources of guidance OF COMMUNITY •• The US EPA has issued a guidance an average dietary pattern, yields an one can recall an event, or how vivid amplified or socially attenuated (Busby et on community engagement that have been document (workbook) on risk ENGAGEMENT estimated excess carcinogenic risk to the the memory is. Intense media coverage al. 2009). Even if no causal factor linked developed in an Australian context. communication that includes exposed population (at the 95 per cent can increase this form of heuristic by to the environment can be demonstrated, The goals of risk communication are to confidence level) of one case in a million.’ pragmatic advice on community making it more memorable. •• The Victorian Department of engagement (Reckelhoff-Dangel as in the above Australian case, the mere (Reckelhoff-Dangel & Petersen 2007): •• A ‘similarity heuristic’ is where Sustainability and Environment has & Petersen 2007). This document perception that a cancer cluster has Concerned parent: something is seen as representative, published a series of monographs •• help residents of affected communities complements an earlier US occurred is enough to create community ‘Will my children be safe if they eat an instance, a set or a group. It may on risk communication as part of a understand the processes of risk EPA guidance manual on risk anxiety and calls for action to be taken. peanut butter sandwiches every day?’ involve stereotyping or some other community engagement strategy along assessment and risk management communication (US EPA 2003b). Lay and media discussions of such events with consultation and community automatic process of categorisation are often based on the proposition that •• improve the quality of the risk involvement. The monograph series Providing an appropriate answer to •• Various authors have explored the based on preconceived ideas. there was a real, but undefined, hazard assessment by enabling community is: Effective engagement: Building the real question that concerns the significance of trust and credibility as out there somewhere that was causing the members to contribute relevant •• An ‘anchoring heuristic’ is where relationships with community and community can be quite challenging. factors influencing effective community risk event (Stebbing 2010). This confirms information, such as their observations people start with an original bias and other stakeholders, Books 1–3. See engagement (Peters et al.1997). that, real risk or not, consequences and local knowledge about the risks one ‘anchors’ their beliefs based on
88 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 89 or expanded beyond that expected by •• are related to an untrusted source that influence these perceptions. challenging because there is no quality communities is always an important The knowledge base of people involved experts, institutions and media and it •• are exotic or unfamiliar control on information on the internet. element of effective risk communication, in the consultation should never be indicates that risk cannot be studied or The findings of one such survey revealed there needs to be an awareness of the underestimated. It is good practice to •• affect children or pregnant women discussed in isolation from the social the following interesting findings (Starr, possibility of ‘consultation fatigue’. This at least conduct a ‘Google’ search to find context of engaged stakeholders. •• are ones that affect identifiable rather Langley & Taylor 2000): 6.3 can occur when engagement has been out what they have been reading about than anonymous people intense, covering a broad range of issues an issue. •• People are very concerned about RISK COMMUNICATION •• cause insidious and irreversible and over a long timescale. There must 6.2.3 chemical pollution. Differences in ‘real’ and ‘perceived’ damage – THINGS TO KNOW AND also be an awareness of what people The ultimate peril of poor risk •• More than 80 per cent of respondents risk •• cause dreaded health effects such expect as an outcome of the negotiations, communication is to lose control, tried to avoid chemicals in their daily THINGS TO AVOID as cancer and whether this might include an perhaps permanently, to another group Ideally ‘actual’, ‘estimated’ and ‘perceived’ life, but chemicals represent less of a expectation of payment, or other support, that earns a more trusted position, •• are poorly understood by science Risk communication may be erroneously risks should be closely aligned. This health hazard than lifestyle factors. for participation. and which may have opposing vested seen as a one-way process aimed at presents an immediate problem, as •• are subject to contradictory statements •• Two-thirds of respondents thought they interests. If possible, it is better to initiate rectifying incongruities between the actual risks are often unquantifiable and from responsible sources (or, even had some control over risks to their Communication must be always communication with people before a community’s perceptions and the unknowable. The aim of risk assessment worse, from the same source). health. be seen as ‘two-way’. Listening to crisis develops. It is much more difficult to opinions of regulators. However, it should should be to achieve the alignment of and respecting the views of other explain the complexities of a risk scenario •• Higher perceptions of health risks were be recognised that all parties will have actual and estimated risk and the aim of Concerns about risk will be lessened when: parties is just as important as clearly when people are panicking. It may be associated with personal involvement perceptions about a situation and the good risk communication should be the communicating a prepared risk message. more difficult to build the necessary trust •• the risks are voluntarily assumed (e.g. smoking, illegal drugs and ultimate aim is to draw these perceptions alignment of perceived and actual risk. The language of risk communication is if you are only beginning to reach out •• the risks have a natural origin sunbaking). about risk, the estimated levels of risks However, if the risk messages are not important. It must be kept simple and to people when a problem has reached •• Many people feel that ‘natural’ and the actual levels of risks as closely properly framed, or delivered in a context •• individuals or the community feel able questions must be answered directly. crisis point. chemicals are less risky than man- together as possible. of mistrust or limited engagement, the to exert some control over the risks Scientific language must be translated made chemicals. outcome may involve an exacerbation of •• there are clear benefits from the risks into language that is understandable by Since the initial public reaction is usually Risk communication should not be seen misunderstanding and conflict. •• 76 per cent of respondents believe an educated layperson. one of concern (perhaps strong concern •• the risks and benefits are fairly as a retrospective form of community there is more environmental or alarm), the risk communication distributed involvement and consultation. It is A simple numerical estimate of risk contamination than previously, but Effective communication requires mutual message must address the policy context •• the risks are associated with a trusted an interactive process involving the portrayed as the ‘real risk’ ignores the most thought their community is trust between all the parties. This trust within which it will be dealt and also source exchange among individuals, groups and subjectivity and multiple dimensions of becoming a healthier place to live. must be earned through a commitment address the demand for immediate institutions of information and expert risks (Thomas & Hrudey 1997). People •• the risks are familiar •• Respondents sought information on to open and honest interaction. Any action. Don’t try to convince a worried opinions about the nature, severity and see risk as multidimensional and not •• the risks only affect adults environmental risks from a wide range factual information presented should or sceptical public that ‘there is nothing represented by a numerical value, and will acceptability of risks and the decisions •• the risks are understood of sources. be correct. Exposure of an error of fact to fear’. Good risk communication is judge it according to its characteristics and taken to combat them. •• 60 per cent of respondents had at during a consultation, whether deliberate not simply a public relations exercise. context. For example, trauma or death as •• the process of how the risks are least a moderate degree of confidence or inadvertent, will undermine credibility, Delivering a ‘feel-good’ message about the result of an involuntary catastrophic determined is understood. Good risk communication and in news media reports. destroy confidence and potentially how well your organisation is handling an reaction is likely to be dreaded more consultation should result in an breach trust. Relevant uncertainty must issue is likely to be counterproductive, or •• Doctors are seen to be credible than the situation where the adverse These concepts about how concerns outcome where there is a high level be disclosed and be quantitatively or even offensive. sources of information on consequences are the result of a situation about risks may be heightened or of agreement between the affected qualitatively expressed. environmental health risks. parties. It also entails knowing how to where the risk is assumed voluntarily and lessened are drawn from the work of The aim of the communication respond to public concern, and is a the person feels some degree of control Sandman (1993), and they are essentially There must be a commitment to process need not always be to reduce Some of these findings may appear genuine process conducted with the (e.g. motor vehicle accidents). the same as the factors he describes that maintaining communication, including concern about risks. Many public paradoxical such as: the apparent community’s interest in mind. Good affect ‘outrage’. While knowledge of these providing specific and relevant health interventions are intended to mismatch between perceptions that risk communication and community Concerns about risk will be heightened by factors may not necessarily be helpful in information when requested or promised. increase public concerns about risks environmental pollution is becoming involvement will enable government risks that (DOH 1998): planning a risk communication strategy, It is important to establish realistic risk such as smoking or excessive alcohol worse; general concerns about man- and industry to better understand an awareness of the factors that influence communication objectives and ensure consumption. In communities with •• are involuntary or imposed on the made chemicals; and perceptions that public perceptions and to more readily community perception of risk could help if that openness is maintained throughout regional lead contamination (e.g. Port community communities are somehow ‘healthier’. anticipate community responses. It the planned strategy begins to fail. the processes. Pirie or Broken Hill), public health •• are man-made rather than natural Also interesting is the way in which will increase the effectiveness of risk activities have been designed to increase people source information from the media management decisions and reduce •• are inescapable 6.2.4 Key risk communication messages concerns about what are often subtle and the trust placed in such sources. unwarranted tension. It will explain •• are controlled by parties outside the Environmental risk perception – should be shared with other stakeholders effects, and to provide information about This survey was undertaken before the risks more effectively and constructively community an Australian context so that there is no apparent discord specific activities that can be undertaken widespread adoption of broadband inform communities. between parties attempting to deliver a to protect children. •• are of little or no benefit to the There is relatively little published internet access. It is likely that the internet consistent message. community information on how Australians perceive would feature as a primary information While engagement with affected •• are unfairly distributed environmental health risk and the factors source today. This will be increasingly
90 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 91 Some of the key principles of effective risk •• economic activity (e.g. jobs, property cites a number of guidance manuals government will listen to concerns 6.5.1 communication are (adapted from US values) versus conservation and health relating to risk communication in and how information about concerns Illustrative example of community EPA 1988; DOH 1998): protection emergency situations. will be sought. If the community is not engagement listened to, it will cease to listen. •• accepting and involving the public as a •• personal experiences and perceptions Chess and Hance (1994) outlined There are a number of examples of partner and stakeholder versus so-called ‘objective’ evidence •• Do not seek more feedback than a number of issues that need to be community engagement that illustrate the •• quality of life and aesthetics versus can be used because this will lead to •• carefully planning and evaluating addressed when designing community application of principles outlined in this defined disease problems community disillusion and loss of trust. the nature and content of the risk consultation and risk communication chapter. Equally, there may be some that •• Seeking grudging approval from the communication undertaken so that it is •• local control and involvement versus programs: illustrate how things can go horribly wrong relevant and understandable external control structures community will be far less productive if scant attention is paid to the principles. •• What is the purpose of the than genuinely seeking feedback that •• listening to the public’s specific •• local concerns versus national/ consultation? Is it to gain information, will be used, asking for comments in a concerns – trust, credibility, statewide/regional concerns The following example (Box 2) shows that ideas and options? Is it to build situation where plans can be changed. competence, fairness and empathy are a good outcome can be achieved where •• monitoring and health data versus credibility? Is it to meet regulatory often as important to the community community engagement is pursued using personal experience requirements? Is it to provide A simple, effective and underutilised as statistics and scientific details (trust appropriate consultative techniques. •• personal experience versus scientific maximum opportunity for public approach for identifying communication and credibility are very difficult to literature in making causal inferences involvement? challenges is to engage non-technical regain if lost; experts do not command family members or office staff in a automatic trust) •• broad community concerns versus •• Who is the audience? Those who narrow interest groups perceive themselves to be affected discussion of the issue to gauge how •• being honest, realistic and open those who are not driven by the science •• urgency versus priority determination should be able to participate in the •• appreciating that intentional process. ‘The community’ is diverse, and policy of the risk assessment will •• political activism versus incremental communication is often only a minor with different groups regarding risk in react to the messages. scientific analysis part of the message actually conveyed different ways. They may need a range (the manner of delivery and its tone may •• voluntary exposure hazards versus of messages and styles of delivery. It is important to avoid problems by be more important than its content) involuntary exposure hazards. anticipating issues such as: •• How will industry be involved? What •• ensuring that information is accurate, responsibility will be taken by industry •• lack of communication skills (by any of Communication and consultation are consistent between agencies and not versus the regulator? the parties) speculative important so that these conflicts are resolved. •• What does the community want to •• limited resources and time and staffing •• effectively communicating with the know? Local community leaders, (by any of the parties) media environmental groups and •• confusion between the ‘risk •• acknowledging the public concerns 6.5 environmental health officers may assessment’ and ‘risk management’ and the effects on the community often be able to provide broader phases PLANNING RISK information about particular concerns. •• focusing on issues and processes •• cultural differences rather than people and behaviours. COMMUNICATION •• Have appropriate persons been •• legal considerations identified who can represent the views Information about risks needs to take of potentially vulnerable groups within •• external politics, hidden agendas and 6.4 into account their complexities and the community, such as children and political pressures RISK COMMUNICATION uncertainties, and be constructed so it the elderly? •• conflicting interests within the varying can result in meaningful interpretation •• How will communication occur? parties concerned – UNDERSTANDING by all parties. People’s responses to risk Smaller, informal meetings are often •• impacts from the media will be strongly influenced by their wider CONFLICTS more effective than large impersonal •• evaluation of the consultation. values, so isolated facts about risks may meetings. At large meetings, some have limited impact on risk acceptability Even with good community consultation members of the community may feel Evaluation is a continuous process (DOH 1991), especially when risks are apprehensive about asking questions and risk communication there may be perceived to have little benefit. designed to avoid mid-course corrections disagreement between parties. or expressing opinions. There is a and repeating failures. Evaluation may need to avoid partisan chairpeople for cover: whether the communication was Planning for an appropriate risk meetings, and written materials may timely; whether the communication A consideration of potential conflicts will communication strategy is particularly have more credibility than the spoken was sufficient; whether the public assist in providing a context for effective important for managing a public health word. Materials need to be pre-tested was empowered; and whether the risk assessment, risk management, emergency or crisis situation, since the before they are printed and distributed, credibility and trust of the organisation risk communication and community level of community concern is bound to and evaluated afterwards. There is a was enhanced (adapted from Chess & consultation. Examples of these be heightened (Glik 2007). This review conflicts are: need to determine how industry and Hance 1994).
92 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 93 Chapter 7: Reviewing and appraising an environmental health risk assessment report
The credibility of site-specific health stakeholders and engagement are critical. chronic, developmental or reproductive risk assessments in the successful In reviewing or assessing an EHRA, the toxicity. The report must identify the BOX 2: management of environmental risks answers to the following questions should critical studies and the way these have Aluminium smelting and the community depends upon coherent and logically be clearly evident in relation to problem been used in the EHRA in a transparent, developed reports. Such reports may formulation and scope of the EHRA: accountable and defensible manner be prepared by consultants for industry (see Section 7.2.2). •• Has the problem been clearly defined? Public consultation and commitment to an independent evaluated before the application to increase sulfur dioxide clients, or to meet regulatory requirements. •• How will an EHRA assist with resolving study resulted in a successful resolution of public health emissions was granted. The purpose of this chapter is to provide this problem? 7.2.1 concerns in the Portland community. some guidance on how to review or Study identification The health study involved a literature review and a health appraise a report on an EHRA, and what •• Have the interests and concerns of In 1994 Portland Aluminium sought approval to increase survey. To determine whether there was an increase constitutes good practice. It does not affected parties been reflected in The toxicity studies (or reviews or sulfur dioxide emissions by nearly 30 per cent so that it in asthma and itchy eyes in Portland, the consultants purport to outline the specific reporting formulating the problem? monographs) on which the hazard could increase production at its aluminium smelter in surveyed residents of Portland and Warrnambool (a similar requirements that may be required by •• If existing environmental conditions identification and assessment are based Portland. population) using a questionnaire that covered a range of individual state, territory or Commonwealth appear to pose a threat to human should be clearly identified in the risk health symptoms. jurisdictions. However, information in this health, have options that exist for assessment report. This information is Members of the community were opposed to any increase chapter may be useful to state or territory altering these conditions been important for identifying the basic data in emissions, with the central issue being the effect of The study also reviewed the measurements of ground- regulators in determining whether reports described? (or reviews or monographs) on which the level concentrations of sulfur dioxide that resulted from risk assessment is based. sulfur dioxide on health. There was a widespread belief that submitted for their review contain relevant •• Has the scope of the risk assessment emissions from the older low stacks and the new tall stacks information that has been presented and asthma levels were high in the Portland area. There were been defined? Is it clear what elements after they were built. evaluated in a cogent and logical way, also similar concerns about the levels of sore and itchy eyes will or will not be considered? Such 7.2.2 with appropriate attention to underlying and skin irritations, as well as odours and acid smells. elements may include stressors, Checklist for hazard identification The literature review found there was no evidence that scientific principles. sulfur dioxide caused people to become asthmatic, but sources, exposure pathways, exposure The following checklist is adapted with Portland Aluminium stated that, with increased emissions, it did cause symptoms such as wheeze to occur more routes, populations, and effects or slight modification from US EPA (1995a). the use of taller stacks would improve air quality at ground The relevance and application of the key often. The survey showed that other health symptoms exposure endpoints to be evaluated. A summarised version can be used level by allowing sulfur dioxide to disperse higher into the review aspects identified in this section such as itchy eyes, cough, stuffy noise, sore throat and if tolerable intake data from WHO or atmosphere. will depend on the complexity of the •• Have the assessments that are skin rash were more common in Portland, but there was EHRA. It is not intended that they are required to characterise the risks of NHMRC are used. no significant difference in the proportion of people with Many residents had concerns about the reliability of air rigidly applied to all EHRAs. Some EHRAs existing conditions and the effects asthma and wheeze in Warrnambool, although both cities 1. Toxicological studies (see Chapters 9 monitoring within the Portland area and believed they were will be quite simple and will not need to on risk of the proposed options been had high rates. and 11 for more detail) not given complete information about the potential health address all aspects listed below. described? •• Where multiple stressors or COPCs •• What are the key toxicological effects associated with aluminium production. Monitoring data for 1995, 1996 and 1997 showed that have been identified, is the approach studies that provide the basis for the one-hour ‘acceptable level’ for sulfur dioxide at ground 7.1 taken consistent with guidance health concerns? How good is the In response to these concerns, the Victorian Department level was exceeded four times over this period. However, suggested in Chapter 12 , or in key study? of Human Services established the Health Professionals monitoring of emissions from the new tall stacks showed PROBLEM FORMULATION guidance contained in the Framework Advisory Committee, which included local health much lower levels. •• Is all relevant information presented professionals, a respiratory physician and department AND SCOPE for cumulative risk assessment (US and reviewed? EPA 2003a) or the recommendations representatives. The role of the committee was to organise The monitoring results were used to predict the ground- •• Does the report highlight critical There is increased emphasis on problem of Science and decisions advancing and oversee an independent health study to assess the level concentrations of sulfur dioxide that would occur with aspects of data quality? potential for any adverse health effects from the proposed formulation and scoping, including risk assessment (NRC 2008). the proposed 30 per cent increase in smelter emissions. •• Is the data from laboratory or field increase in sulfur dioxide emissions from the smelter. The levels in Portland and surrounding areas were upfront consideration of potential studies? Is the data for single predicted to be well below the standard. management options and the utility of the species or multiple species? A proactive program of community consultation was EHRA in helping discriminate between 7.2 established and local residents were interviewed and The results of the study were discussed with the these options. It is important that these HAZARD IDENTIFICATION •• If the hazard is carcinogenic, given the opportunity to raise key areas of concern. The community at a public meeting and a report of the study issues are addressed in the EHRA report. has comment been included on committee then ensured these concerns were addressed in was circulated. The study concluded there was no evidence The purpose of hazard identification is issues such as: observation of the study’s terms of reference. that the proposed increase in sulfur dioxide emissions from An EHRA can be initiated for a variety to identify what adverse human health single or multiple tumour sites; the taller stacks would be detrimental to health. of reasons. Regulatory requirements, effects are associated with the agent or occurrence of benign or malignant community concern over an issue, and The Victorian EPA then granted Portland Aluminium agents of concern. tumours; certain tumour types not approval to replace the low stacks at the smelter with six The report was well received by the community. Portland responding to accidents and emergencies linked to carcinogenicity use of are a few of the contexts that may form tall stacks and to monitor their emissions for 12 weeks. The Aluminium was given EPA approval to increase sulfur The hazard assessment component is the maximum tolerated dose; and the basis of the need for an EHRA. findings of the health study and the results of monitoring dioxide emissions from the smelter and ongoing monitoring likely to be based on a number of studies, whether a mode of action (MoA) For a risk assessor to understand the emissions from the old and new stacks were to be of air pollutants was included as a condition of the licence. conducted in different species within can be identified and supported problem and to define the scope of each toxicology study type, such as acute, by peer-reviewed studies? the assessment, early identification of
94 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 95 •• Has there been a weight-of- 3. Assessing the relationship between a •• Was a meta-analysis performed to Some key elements of the dose–response •• Where relevant, have differences The NRC has developed such a graphical evidence (WoE) approach in possible cause and an outcome combine the epidemiological studies? component of the report that should be between (e.g. WHO and US EPA method for representing the variability in presenting a judgement as to the What approach was used? Were present are as follows: toxicological assessments been toxicological endpoints. An example is •• How much is known about the likelihood of human carcinogenic studies excluded? Why? appraised and discussed?) shown in Figure 25, extracted from the biological mechanism by which the •• Valid datasets and plausible models hazard, and does this include a NRC review of the draft IRIS assessment agent produces adverse effects? for high- to low-dose and inter-species •• Has the dose–response relationship clear articulation of the rationale for 2. What model was used to develop the for agents of potential concern been of tetrachloroethylene (NRC 2010 p. 94). •• Were relevant studies on extrapolation are presented in dose– the position taken? dose–response curve? What rationale appraised and discussed? mechanisms of action discussed, response modelling. •• If the hazard is other than supports this choice? Is chemical- An alternative tabular method of including the possible impacts of •• The report offers an explicit rationale •• Has the data been presented in a form carcinogenic, what endpoints were specific information available to presentation could be adapted from the species differences in metabolism? for any preferred dataset(s) and amenable to efficient interpretation observed, and what is the basis for support this approach? approach used to summarise toxicological model(s) used in dose–response and review? the critical effect? •• To what extent does this information endpoints in JMPR pesticide evaluations help to interpret the toxicity data? For non-carcinogenic hazards: evaluation strengths and weaknesses (IPCS 2000 – see Appendix G for an •• Have other studies that support this of the preferred datasets are example). In such a table, the critical finding been described? Are there •• What are the implications for •• How was the tolerable intake (or the 7.4 discussed, and scientific consensus endpoints are listed, along with the any valid studies that conflict with potential health effects? acceptable range) estimated? or lack thereof is indicated for critical SELECTION OF GUIDELINE NOAELs and LOAELs, indicating the this finding? •• How were any negative or equivocal •• What assumptions or uncertainty issues or assumptions. lowest values that were used to drive the findings in animals or humans VALUES •• Has the report identified research factors were used? •• The report reveals how dose–response ADI development. addressed, and to what extent was that would reduce uncertainty in the •• What is the confidence in the estimates? relationships change with alternate this data considered in the hazard 1. What source has been used for EHRA and increase confidence in datasets, assumptions and models. 7.5.2 its outcomes? identification? guideline values? For carcinogenic hazards: Presentation of epidemiological data •• Besides the health effect observed Some other specific ideal considerations •• Is it a reputable source? 4. Summarise the hazard identification •• What dose–response model was used? The burgeoning number of epidemiological in the key study, are there other are: •• Are the guidelines relevant for the and discuss the significance of: What is the basis for selecting the studies and the need to integrate health endpoints of concern? situation being assessed? particular dose–response model used? •• Have all relevant toxicological facts toxicological and other data creates a •• Are there any significant data •• confidence in conclusions Are there other models that could have been checked for accuracy and •• Have they been transcribed challenge for selecting data, analysing gaps, and how have these been •• alternative conclusions that are also been used with equal plausibility and currency? correctly? the data and summarising the results. addressed? supported by the data scientific validity? •• Has the adequacy of the available The lucid presentation of the results in •• significant data gaps •• What is the basis for selecting the toxicological database been appraised? text, tables and figures is important for 2. Epidemiological studies 7.5 •• major assumptions. model used in this instance? communicating these processes. (see Chapter 10) •• Have the effects on each significant body system (e.g. renal, hepatic, DATA PRESENTATION •• What types of epidemiological 3. To what extent were the route and level cardiovascular) and the types of Although aimed at the assessment of studies were used (i.e. ecologic, 7.3 of exposure observed in the toxicology effects (e.g. allergy, genotoxicity and Data presentation is a critical part of interventions, the Cochrane handbook case-control or cohort)? DOSE–RESPONSE or epidemiology studies consistent with carcinogenicity, reproductive and any report outlining an EHRA. There for systematic reviews of interventions •• What was the size of the study the expected human exposures in the developmental) been appraised are a number of ways that data can be (Higgins and Green, 2011 has a range of population? CHECKLIST situation under appraisal? and summarised for the relevant summarised (e.g. in tables and diagrams) useful advice covering numerous topics, so that it makes it easier for a reader to including selecting data, incorporating •• How large are the confidence exposure routes? 1. What data was used to develop the •• Is the available data from the same grasp the essential information on which economic evidence and assessing risk of intervals on the observed measures •• Has the critical toxic effects and organ/ dose–response curve? Would the result route of exposure as the expected the outcomes of the EHRA have been bias. There is a chapter about presenting of risk? body system been identified? have been significantly different if human exposures? If not, is derived and to understand basis for results in text, tables and figures (e.g. •• Were exposures adequately based on a different dataset? pharmacokinetic data available to •• Have known toxicity modifying factors the main outcomes or conclusions from using forest plots that display effect described or categorised in the extrapolate across route of exposure? (e.g. synergistic and antagonistic effects the study. estimates and confidence intervals for both epidemiological studies? If animal data was used: •• Are there any potential anomalies resulting from exposure to multiple individual studies and meta-analyses). •• Was the degree to which in the toxicity data? (e.g. Was bolus contaminants) been considered? •• What species were used? The most 7.5.1 other causal factors could be dosing with a carrier like corn oil •• Have toxicologically sensitive sub- The key challenge is to find a way to sensitive, average of all species, or Presentation of toxicological data excluded well described in the used?) populations been identified? simplify the presentation of data from other? It would not be unusual for there to be epidemiological studies? •• What is the degree of extrapolation •• Has the toxicological basis of the multiple epidemiological studies and to •• Were any studies excluded? Why? a range of studies from which different •• How relevant is the available guidance value or potency factor, summarise the main findings. A common from the observed data in the toxicological endpoints can be identified. epidemiological evidence to the toxicological or epidemiological studies where applicable, been discussed and approach is to summarise the outcomes of If epidemiological data was used: These studies may indicate that the issue addressed in the EHRA to the expected human exposures in the uncertainties noted? individual studies in a consolidated table critical doses (NOAEL and/or LOAEL) that includes some brief information on •• Which studies were used? Only positive the situation under appraisal? One •• Have NHMRC (where applicable) for each of these effects can vary over the study design, the range of variation in studies, all studies, or some other to two orders of magnitude? Multiple or WHO toxicological assessments a wide range. It may be useful to utilise derived parameters (such as the adjusted combination? orders of magnitude? What is the been considered as the primary a graphical or tabular representation of odds ratios and their confidence limits). •• Were any studies excluded? Why? impact of such an extrapolation? toxicological resource? this variability.
96 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 97 Figure 25: Graphical depiction of the range of variability in toxicological endpoints that provides an explanation for any pathways Figure 26: Graphical depiction of the range of variability in odds ratios describing the effects of water disinfection may be used in an EHRA not considered, or where they have been by-products on birth defects considered not relevant in the analysis. This section should also address quality aspects of the sampling plans, measures to preserve sample integrity, and analytical data quality controls.
7.6.1 Interpretation of sampling data General issues to be considered in sampling and data interpretation are discussed in Section 8.5. This section summarises some key elements of collecting and interpreting data relating to exposure assessment.
An appraisal of data must show an understanding of: •• the context of the risk assessment •• the topography of the area affected •• the demography of the population •• environmental factors such as stratification of water bodies, movement of plumes in air or groundwater, soil structure (e.g. Each bar represents a single study, with the upper or right end indicating a possible point of departure presence of clay or fill, and the depths for risk assessment, after conversion of dose to ‘human equivalence’ or adjustment for continuous of individual strata), meteorological exposure in animal studies. Solid horizontal lines indicate the studies considered most applicable to factors, groundwater flows a risk assessment. Shadings indicate the application of various uncertainty factors (black for inter- •• the relevant current or future human species extrapolation; white for extrapolation from sub-chronic to chronic exposure; grey for intra- activities. species variation to account for sensitive individuals; light blue if the study indicated a LOAEL, but not a NOAEL; dark blue for uncertainty in the study database). The left end of the bar represents the toxicological reference dose (in this case RfC), which could be derived from this database. Too often numerical data is considered in isolation from other key parameters Reprinted with permission from NRC 2010 by the National Academy of Sciences, courtesy of the such as: National Academies Press, Washington, DC. •• levels of detection (and reporting) •• quality assurance for the data The IARC commonly uses such a tabulated for the effects of chlorinated disinfection •• uncertainty about the data approach to summarise a large number of by-products in water on birth defects •• geographical relationship of one epidemiological studies of different types. clustered around the value of unity, sample to another The summary of epidemiological data on designating no effect (Figure 26). formaldehyde is a good example of this •• current or potential human activities. approach (IARC 2006). 7.6 Other key failings in numerical data An alternative approach is to use a EXPOSURE ASSESSMENT analysis include: diagram that illustrates the relationships •• ignoring negative or unexceptional between the odds ratios describing the Exposure assessment is one of the more results by focusing on unusual or outcomes of individual studies. Such an critical aspects of an EHRA report. It is elevated results (the dataset needs to approach was used by Hrudey (2009) to important that the report describes all of be considered in its entirety) illustrate the extent to which odds ratios the exposure pathways considered, and Reprinted from Hrudey, 2009 © 2009, with permission from Elsevier
98 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 99 •• inadequately managing censored •• If models are used, their bases are in summary statements, along with •• Is the presentation of results consistent •• Is it possible to determine whether •• Does the report include or enable data (e.g. by assigning a zero value described, along with their validation quantitative estimates of risk, to give a with the units for which the standards an absence of results indicates an ecological risk assessment, if required to results below the level of detection status. combined and integrated view of the are written (e.g. using milligrams/ absence of testing or that results by regulatory authorities? 3 3 or reporting) •• Potential sources, pathways and routes evidence. m instead of nanograms/m when were non-detects?Where composite •• If toxicological data and the exposure •• accepting relatively high levels of of human exposure are identified and •• The report clearly identifies key sampling ambient air)? samples have been used, have these scenario led to the conclusion that detection or reporting so that the quantified; the reasons why any are assumptions, their rationale and the •• Have SI units been used correctly been identified? Is there an explanation a high concentration of agent is value of much data is obscured. This not included in the assessment are extent of scientific consensus; the throughout the report (Thompson & of compositing techniques? (Note: permissible, does the result violate may have the consequence of failing presented. uncertainties thus accepted; and Taylor 2008)? compositing is specified practice for ecological, aesthetic, land use or some types of food sampling). to reveal gradients that will help to •• Central estimates and, if possible, the effect of reasonable alternative •• Is there a map of testing sites that physical principles? highlight the presence and location of upper and lower bounds on assumptions on conclusions and enables ready identification of •• If there is likely to be significant •• Has a risk management decision been environmental ‘hot spots’. Examples exposures for the full population, estimates. sampling sites in relation to relevant heterogeneity in the individual made during the course of the risk have been seen where environmental and the distribution of exposures are •• The report outlines specific ongoing or environmental sources? components and n is the number of assessment and, if so, how might it health criteria levels have been treated described any preferred estimates samples, are there any reasons why potential research projects that would •• Does the presentation format enable have influenced the calculation of risk? as the level of reporting. are noted, together with supporting guideline values should not be divided probably clarify significantly the extent easy cross-referencing of results to •• What has been the involvement of documentation. by n in the assessment process to of uncertainty in the risk of estimation. maps and between different parts of a the public? The very existence of levels of detection •• Uncertainties in the estimates are help identify high concentrations in •• The report provides a sense of report? •• How has information been and reporting results in the need to described, and the relative importance one or two individual components of perspective about the risk through the •• Are results presented in linear communicated to the public? censor data. Censoring of data can be of key assumptions and data is a composite? use of appropriate analogy. geographic sequence (e.g. either going particularly important when the maximum highlighted. •• Where censored data has been used, •• What processes of community downstream or upstream, or towards permitted criterion is close to the level has the censoring been fully explained? consultation have taken place? •• Research or data necessary to improve or away from a point source)? of detection (e.g. with potable drinking- the exposure assessment is described. 7.8 water standards). Data censoring must •• Has the report been presented in Clarity of how the data has been In relation to the last two points about be addressed in an appropriate way (see Specific considerations are: CONFIRMATION OF UTILITY a suitable format to enable ‘track used, assumptions made and relevant community engagement and consultation, Sections 4.6.4 and 8.7). changes’, comments and cutting and it should be noted that a specific •• Does the report address the issues justification •• Has the potentially exposed population pasting into a Word document? PDF requirement for such consultation may described in the problem formulation Given two similar results, the result been identified? formats hinder rapid transcription •• Has information from previous reports be incorporated into EHRA guidance and scope? that can be explained (e.g. by history •• Have potentially exposed, unusually and can create the potential for on the situation been appropriately and/or regulations in some Australian or similarities with results from similar susceptible sub-populations been •• Does the report provide guidance as transcription errors. selected and incorporated into this jurisdictions. Further advice on strata) will tend to be of less concern identified? to the relative effectiveness of risk report? community engagement is presented in than the result that cannot be explained •• Have the estimates of chemical management options? Issues of data quality •• Has irrelevant information from other Chapter 6. (Langley 1993b). •• Has the report been peer reviewed or situations been excluded from the exposure for each significant exposure •• Has the nature of the analyte been is it being audited? report? Presentation of the report should include route and for each chemical of specified? This may be important potential concern been adequately attention to having an orderly structure 7.6.2 when the valency (e.g. chromium) or •• Have all assumptions and default data quantified and tabulated? and clear delineation of its component Exposure assessment checklist chemical form (e.g. organic arsenic been identified and justified? 7.9 parts. Figure 27 is presented as one •• In cases of presumed insignificant versus inorganic arsenic in fish; haem •• Has the analysis been based on an The general components in an acceptable example of a report structure that could exposure, has the exposure been GENERAL ASSESSMENT iron versus non-haem iron in animal up-to-date literature appraisal? exposure component to a risk assessment be used for an EHRA report. It has demonstrated to be small? samples) may be relevant for health report as follows (AIHC 1989): AND REPORT •• Has information been presented been developed primarily for assessing •• Has the relative significance of each risk assessment. coherently and in an appropriate •• The purpose and scope of the emissions from an industrial facility, but exposure pathway, based on the risk PRESENTATION •• Has the level of reporting (LOR) been sequence, to enable efficient appraisal exposure assessment and the its structure is designed to achieve the analysis, been discussed? specified for each analyte and for each of the report? underlying methodologies are clearly desired objectives of clarity, structure and A checklist of points that should be batch of results? described. •• Has the rationale for the sampling comprehension. Various state and territory considered in the overall presentation •• Have the laboratories involved in the program and selection of analytes jurisdictions may have different legislative •• The specific populations and sub- 7.7 of the report and an assessment of its assays been identified, along with their (including sampling objectives) been requirements for an EHRA report populations that are the subjects of the RISK CHARACTERISATION quality is provided below. QA/QC procedures? addressed? structure and content, and the example in assessment are clearly identified, and •• When there have been several •• Have environmental factors relevant Figure 27 is not meant to be prescriptive. the reasons for their selections and any The general components in an Report structure and data presentation sampling periods, have the results to the choice of analytes been noted, exclusions are given. acceptable risk characterisation •• Have all tables and figures been referred been collated into a single table to along with a rationale if unusual •• The available data is considered and component to a risk assessment report to correctly in the text of the report? enable efficient appraisal of results and analytes are included or common critically evaluated, and the degree are as follows (AIHC 1989): trends to be readily detected? analytes are excluded? of confidence in the data expressed •• The major components of risk (hazard (reasons for any data exclusion are •• Have all conclusions been justified? identification, dose–response and presented). exposure assessment) are presented
100 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 101 Part 2:
Figure 27: Example of a structured EHRA report for air emissions from an industrial facility Additional guidance on selected issues
102 ENVIRONMENTAL HEALTH RISK MANAGEMENT Chapter 8: Data collection and analysis
Collecting and analysing environmental environmental compartments and the sampling data is critical to the risk effects of environmental partitioning 8.2 assessment process. Quality assurance is necessary for developing sampling ENVIRONMENTAL (QA) of the collected data is also vital, plans for chemical agents and the and QA processes should be well process of exposure assessment. This SAMPLING AND ANALYSIS documented. Data may be available from is especially important for volatile and a preliminary or detailed site investigation gaseous substances, which can move Data collection involves acquiring and (either or both) and is usually available from sources such as contaminated analysing information about hazards on prior to commencing the risk assessment. soil or groundwater into dwellings and a site that may affect human health and In such cases, the assessor must open spaces where human ‘receptors’ which will be the focus for the particular determine whether the data quality may be exposed. Transport pathways risk assessment (US EPA 1989). objectives of any previous investigations for gases and vapours derived from are compatible with the objectives of soil or groundwater contamination may Sampling is often carried out to more the risk assessment and whether the be dependent on soil characteristics clearly define detected or suspected original data quality objectives have been such as soil type, porosity, depth of contamination and, if remediation occurs, to verify that contaminated material has satisfactorily met (NEPC 2010). the contamination and the presence of been removed and that any contamination organic carbon. These characteristics remaining does not constitute a health or The sampling plan may have been should be factored into the modelling that environmental risk. informed by a conceptual site model may be used to quantify the pathways. (see Section 4.4), which will assist Best practice modelling approaches The greatest concern in collecting samples with understanding the source and would, by definition, require model is to ensure the samples taken adequately medium (air, water, soil) from which parameterisation using site-specific represent potential exposures for the parameters in addition to soil vapour the samples have been obtained and situation. Consequently it is essential to information on the site history, which and indoor air data collection for be fully apprised of the context of the could assist in assessing the types and validation of model predictions. Transfer risk assessment, the objectives of the mobility of the contaminants and their characteristics are also important for task, the environmental conditions at the likely location. More specific guidance dusts and particulates, which can be site locations and what analytes will be on preparation of sampling plans and transported over significant distances tested in each sample, before sampling sample density requirements is available from source. commences (Lock 1996). (e.g. contaminated sites NEPM schedule B(2) NEPC 2010). If the sampling data Partitioning will reflect the fact Inappropriate sample collection is incomplete, or the sampling plan that substances will move to the procedures yield samples that are not inadequate, there should be specific environmental compartment for which representative of the population of interest comment in the risk assessment report. they have the most affinity (Calamari are of little use, seriously compromise the 1993; 1999). Transformation may occur purpose of sampling, and contribute to The purpose of this chapter is to in any environmental compartment. the uncertainty of the analytical results provide some guidance on sampling the (Keith 1990). environment to collect data suitable for 8.1.1 exposure assessment. It addresses some Meteorological data Laboratory errors can and do occur and issues of analytical sensitivity and quality if an aberrant or an unexpected result is assurance, but it does not purport to be a Meteorological data will be particularly provided, and the potential for laboratory comprehensive review of such topics. important when evaluating both error should always be considered. If a point source and generalised air laboratory error is suspected, it may be pollution and potential exposures of worthwhile requesting relevant QA/QC 8.1 populations. When environmental information. ENVIRONMENTAL monitoring is being undertaken, there is a need to have concurrent An important aspect of environmental DISTRIBUTION meteorological data and this may be sampling and analysis is that the available at minimal or no cost from environment is not static and sampling It is unlikely that contaminants will be the Australian Bureau of Meteorology results can vary over time. The uniformly distributed in the environment depending on the locality. methodology used for collecting samples under consideration. An understanding and interpreting environmental sampling of how chemical agents move between data should take this into account.
ENVIRONMENTAL HEALTH RISK ASSESSMENT 105 8.2.1 Further information on approaches to soil and contaminants; and whether identified and adequately remedied to •• not too complex: A procedure so guidelines for speciated metals and Strategies for sampling the environmental sampling and analysis are there is a renewable source or whether protect human health. complex as to be only useable by a few metalloids is limited so this factor should environment available in schedules B(2) and B(3) of the contamination will dissipate over highly trained persons will probably be be considered when deciding on the the contaminated sites NEPM, available time. This field is developing rapidly, and A considerable amount of expert of limited practical value; usefulness of speciated analysis. In sampling, the statistical considerations online at
106 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 107 8.3.3 Consideration will need to made as to AS 4482.1-1997 Appendix H provides a According to these definitions: Recovery check portions should be •• may be separately and independently whether routinely collected historical chain of custody form. fortified at concentrations that are quantified by virtue of (e.g. Detection limits •• Adequate QA is achieved when the data will be as appropriate to use as data easily quantified but within the range of chromatographic separation or results of QC demonstrate that agreed Ideally, the detection limit of the analytical collected de novo for the risk assessment. concentrations expected for real samples. production of different mass ions in 8.4.4 Quality assurance objectives such as freedom from method used should be lower than the a GC/MS system). contamination, method accuracy and level at which the contaminant might The information in Sections 8.4.4 to The method used to correct the reported 8.4.2 precision can be reliably achieved. An become a health concern. A health- 8.4.11 is adapted from Good (1993). data for recovering the analyte(s) from Surrogate compounds may be alkylated Sample handling, storage and important decision then is the correct based GV (e.g. HIL or Tier 1 screening the media must be explicitly stated. or halogenated analogues or structural transport level of QC. level) usually provides a benchmark All of the actions, procedures, checks Failure to do so may render the reported isomers of analytes of interest, or against which the detection limit of the Sample handling and transport should and decisions undertaken to ensure •• As a general rule, the level of required data meaningless, and significantly where GC or LC/MS is used, deuterated analytical methodology can be measured. be done according to relevant regulatory the representativeness and integrity of QC is that which adequately measures compromise the exposure assessment standards are an excellent choice. If the methodology cannot achieve such documents or Australian Standards. samples and accuracy and reliability of the effects of all possible influences and the derived risk assessment. a detection limit, there will inevitably analysis results must be recorded. upon sample integrity, accuracy and The purpose of surrogate spikes, which be some difficulty, bias or imprecision Some key issues are (Keith 1990): precision, and is capable of predicting 8.4.8 are added immediately before the in assessing risk, and efforts should In the field, this includes selecting their variation with a high degree sample extraction step, is to provide a •• Contamination may arise from Reference material analysis be made to find an alternative, more appropriate sampling methods, of confidence. QC is more often check for every analysis that no gross substances in the sampling devices sensitive analytical method. documentation and sample storage, performed inadequately than very well. This is analysis of a sample similar in processing errors have occurred that and storage containers. PVC and cleaning tools before sampling and matrix type to the samples, with accurately could have led to significant analyte plastics other than teflon tend to between samples, cleaning containers, 8.4.6 known concentration of the analyte(s) of losses or faulty calculation. 8.4 absorb organics and leach plasticisers and maintaining sample environment interest. Results of recovery checks and and other chemicals used in their Blanks to minimise sample contamination and reference material analyses for each batch Another term (used in schedule B(3) of QUALITY ASSURANCE manufacture. Some pesticides, analyte losses. A reagent blank (or preferably two for should be recorded so that the bias of a the contaminated sites NEPM) is that of a halogenated compounds and metals OF DATA USED IN SITE- very low-level analysis) is prepared by method may be estimated, and the day-to- ‘matrix spike’. The purpose of the matrix strongly adsorb to glass. In the laboratory, quality assurance processing reagents only in exactly the day method efficiency monitored. spike is to monitor the performance of SPECIFIC HEALTH RISK •• The loss of volatile analytes or reduced (QA) involves proper sample control, manner used for each sample. The aim the analytical methods used, and to ASSESSMENT concentrations from irreversible data transfer, instrument calibration, of the blank determination is to establish 8.4.9 determine whether matrix interferences the magnitude of that component of the absorption on the walls of sampling selecting properly trained staff and Surrogate spikes and internal exist. Matrix spikes should be performed analytical signal that can be ascribed to 8.4.1 containers can be a significant problem. suitable equipment, reagents and when validating a method by adding it contributions from reagents, glassware, standards Data quality objectives •• Sample preservation can be of analytical methods. to the analysis portion before extraction considerable importance. If incorrectly etc. The contribution established should Wherever appropriate, especially for or digestion. Data quality objectives ‘provide critical stored, materials can have accelerated 8.4.5 be subtracted from the gross analytical chromatographic analysis of organics, definitions of the confidence that must be breakdown, chemicals may be lost Quality control signal for each analysis before calculating using surrogate spikes and internal Immediately prior to instrumental analysis, inherent in the conclusions drawn from by volatilisation, and proliferation sample analyte concentration. standards is highly recommended. each sample, blank and recovery or Those parts of QA that serve to monitor the data produced by the whole project’ or diminution of microbiological Including this in methods requires little reference material extract is fortified with and measure the effectiveness of and determine the degree of uncertainty organisms can occur. The nature of 8.4.7 additional effort and greatly enhances a set amount of one or more compounds other QA procedures by comparison or error that can be tolerated in the data the storage container, its seal, and the Replicate analysis confidence in qualitative and quantitative to be used as internal standards. These with previously decided objectives. In (Keith 1990 p. 611). degree of refrigeration needed should results obtained. compounds should: the field, this may include checking Repeat analysis of at least one sample always be considered and addressed. •• not be found in real samples Data quality objectives that clearly specify sampling equipment cleanliness by or 10 per cent of the batch of samples. Surrogate spikes should be added to each keeping rinses for analysis, cross-checking The variation between replicate analyses •• not interfere with quantifying the the amount, nature and quality of the 8.4.3 sample, blank and recovery/reference data to be collected should be detailed. sample identities, duplicating sampling should be recorded for each batch to sample. They should be similar to the analytes of interest Chain of custody provide an estimate of the precision of Data quality objectives will be situation- sites and performance of ‘field blanks’ analytes of interest in terms of: •• be separately and independently the method. specific. More detail is available from the The consultant’s report must provide the and ‘field spikes’. In the laboratory, quantified. •• extractability website of the US EPA Quality System following chain of custody information quality control (QC) procedures involve for Environmental Data and Technology (EPA NSW 1997): measuring the quality of reagents, the 8.4.7 •• recovery through clean-up procedures The purpose of internal standards in at
108 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 109 8.4.10 results or more than a handful of results. of most datasets, a range of summary 8.5.3 Figure 28: Contour modelling of ground-level concentration (GLC) data Control charts At its worst, sample identification statistics needs to be presented, as the Contouring numbers, sampling points, technical logs mix of summary statistics will be more Nadkarni (1991) claims that the heart and results for each analyte will be on useful than a single summary statistic. Contour modelling (Figure 28) can of a QA/QC program is a control chart. separate pages. provide a useful way of representing Good (1993) agrees, explaining it is a As much sampling is judgemental rather dispersion data (e.g. ground-level ‘numerical picture (a plot) of the variation There is a constant tension between than random, caution needs to be concentrations of airborne contaminants of measured QC parameter (e.g. blank consultants who wish to maintain taken with using conventional statistical emitted from a point source, such as and recovery values). Data is plotted in individuality to their reports and methods that usually assume the random an industrial facility). While graphical the sequence in which it was obtained, government agencies that seek uniform collection of data and the use of normally representations of contours can provide and reviewed frequently in order to detect reports. A uniform approach to the distributed data. Much risk assessment useful information about situations any problem with minimal delay. The use location and presentation of data makes data is log-normally distributed or has such as the distribution and ‘trends’ of these charts is highly recommended’. for more rapid and accurate assessments skewed distributions and this will require of environmental hazards, poorly of reports. different statistical methods for analysis. constructed contour maps are often 8.4.11 More detail is given on the US EPA based on extrapolations from an Safety plans The major problems that can occur with website, accessible at
110 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 111 The Agency for Toxic Substances •• Use colour and shading only when It needs to be absolutely clear when 8.7 8.7.2 relevant to censored dataset users is and Disease Registry (ATSDR) in the necessary, and then only very carefully. logarithmic rather than arithmetic scales Diminishing levels of reporting also available from the US EPA website United States has used GIS to identify •• When possible, accompany graphs are being used on the axes of graphs. CENSORED DATA at
112 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 113 Chapter 9: Evaluation of toxicity data
Toxicology studies have been designed eye irritancy and sensitisation tests, and and Test guideline 416, Two-generation genotoxicity assays, and integration of to determine the toxic effects associated 9.1 some of these have now been incorporated reproduction toxicity study: (OECD 2009). their results, is also available in a UK 9.3 with exposure to chemical hazards. TOXICITY TESTING – into the OECD Test guidelines series For guidance on evaluating reproductive Department of Health document (see EVALUATING THE Such studies can provide information (e.g. TGs 429–435, and 437–438). toxicity studies, refer to IPCS EHC 225 COM 2000). Interpretation of the results relating to toxic effects and potential MAJOR IN VIVO STUDY Principles for evaluating health risks to of in vitro genotoxicity tests for the WEIGHT OF EVIDENCE health hazards likely to arise from single TYPES Sub-chronic toxicity studies are short-term reproduction associated with exposure to purposes of identifying potential human AND CONSIDERING THE or repeated exposures, in terms of repeat-dose studies. A short-term study chemicals (WHO 2001). genotoxins and, by inference, potential predicting potentially important toxicity human carcinogens, needs to be done Hazard identification mostly relies on has been defined (WHO 1990) as ‘having TOXICOLOGY DATABASE endpoints and identifying potential target within a well-defined science policy the results of in vivo toxicity studies a duration lasting up to 10 per cent of the Developmental toxicity studies are studies organs or systems. context (Thybaud et al. 2007). IN TOTO conducted according to standard animal’s life span, 90 days in rats and that examine the spectrum of possible mice, or 1 year in dogs’, although the US in utero outcomes for the conceptus, protocols. Guidance on the conduct of The essential purpose of toxicity studies This chapter on toxicological evaluation toxicity tests has been promulgated by EPA considers a 1-year dog study to be a including death, malformations, functional Other tests: The OECD Test guideline focuses on chemical hazards assessed chronic study. The main purpose of sub- deficits and developmental delays (Tyl & series now includes special tests for is detecting valid biological evidence of the OECD (OECD 2009). There have the hazard potential of the substance using traditional toxicity testing methods been 53 OECD Test guidelines published chronic testing is to identify any target Marr 1997). Exposure during sensitive such endpoints as neurotoxicity (TG424) and, in particular, on some of the organs and to establish dose levels for periods may alter normal development and developmental neurotoxicity being investigated. Evaluation of the since they were first promulgated in 3 problems and pitfalls that may arise chronic exposure studies. resulting in immediate effects, or may (TG426). It has also addressed animal weight of evidence (WoE) produced 1981, and many of these have been by toxicity studies is the process that during an assessment of possible periodically updated. subsequently compromise normal welfare issues through the development compound-related changes in the Chronic toxicity studies, or long-term physiological or behavioural functioning of a range of validated short-term in considers the cumulative data pertinent parameters measured in toxicity studies. later in life. Since some developmental vivo tests and in vitro tests, which may to arriving at a level of concern about the The following types of studies are defined. studies, are defined as studies lasting for It is intended to provide guidance on the greater part of the life span of the test processes can occur perinatally or complement, or possibly substitute for, potential adverse effects of a substance. the process of hazard identification animals, usually 18 months in mice and postnatally, protocols for developmental the conventional animal tests that have It is composed of a series of judgements Acute toxicity studies are studies that and assessment. 2 years in rats (WHO 1987; 1990). The studies are being modified and extended been used for many years. These include concerning the adequacy, validity, and investigate the effects of single doses of OECD protocols for these studies may to address developmental toxicity during tests for skin absorption (TG428) and appropriateness of the methods used a substance. The LD test, or medium The basic assumption is that traditional 50 cover the investigation of chronic toxicity the period covering major organogenesis tests for endocrine-related endpoints to produce the database, and those lethal dose test (OECD Test guideline methods of toxicity assessment will (TG452) or carcinogenicity (TG451), or as well as covering the perinatal and (in vivo tests TG440, 441 and in vitro judgements that bring into causal, TG401), which records gross toxicity and continue to be the mainstay of EHRA both (TG453). All three of the OECD Test early postnatal period. This could include test TG455). complementary, parallel or reciprocal mortality data over a 14-day post-dosing for some time. guidelines were updated in 2009 to better delayed toxicity associated with epigenetic relationships, all the data considered. period, has been commonly employed reflect developments in animal welfare effects during sensitive phases of foetal Because our knowledge about and may still be included in many data and to improve dose selection. development. Such attention to the critical 9.2 mechanisms of toxicity is still developing, It is also important to note that, over time, packages. However, TG401 was formally timing of exposure also accords with a because good epidemiological evidence the scientific community is gaining a withdrawn by the OECD in 2002 in GUIDANCE ON EVALUATING growing emphasis on understanding is seldom available, and because animal better understanding of the mechanisms response to animal welfare concerns. Reproductive toxicity studies are studies early-life susceptibility to carcinogenesis studies are not always conclusive, the of toxicity, and this is leading to changes Newer tests (‘limit’ tests and ‘up-and- designed to provide general information AND INTERPRETING in both methodology and interpretation about the effects of a test substance (see Section 5.5.2). information available at a given time down’ dosing methods) are now favoured TOXICITY TESTS may provide only ‘persuasive’ rather of toxicity data. It is inevitable that new as they reduce the numbers of animals on reproductive performance in both paradigms will be introduced as science male and female animals, such as Genotoxicity studies are designed to than ‘hard’ evidence of a defensible required and reduce the suffering seen Supplementary guidance on advances (see Section 9.4). effects on mating behaviour, gonadal determine whether test chemicals can presumption (one way or the other) in the classical LD50 test. OECD TG420 the evaluation and interpretation function, oestrous cycling, conception, perturb genetic material to cause gene or about the potential health effects of a covers acute oral toxicity determination by is provided in more detail in substance under given conditions of It is therefore important to acknowledge implantation, parturition, lactation, chromosomal mutations. A large number the ‘fixed-dose method’, TG 423 by the Appendix 1. This guidance is aimed exposure. Therefore, it is necessary that the analysis and evaluation of toxicity weaning and neonatal mortality. of assay systems, especially in vitro ‘acute toxic class method’, and TG 425 primarily at experienced toxicologists to succinctly discuss the rationale for studies reflects scientific consensus at These studies may also provide some systems, have been devised to detect the by the ‘up-and-down procedure’. who may be asked to provide a weight- judgements and conclusions contained the time the data is reviewed. This means information about developmental or genotoxic or mutagenic potential of agents of-evidence (WoE) analysis of the extent in risk assessments together with any that the toxicity studies underpinning teratogenic effects of the test substance. (IARC 1999). Most authorities consider The standard acute toxicity studies to which toxicity tests are able to define associated uncertainties. This becomes many EHRAs may contain data generated The conduct of and the results from that a minimum set of data is required include tests for: acute oral, dermal the hazard identification component of an important when new data or new scientific during an era when toxicity testing and these studies are very important to to define a mutagen/non-mutagen. This and inhalational toxicity; eye irritation; EHRA, and to provide useful information knowledge requires re-evaluation of the the interpretation of results were less assess with care, since the reproductive data usually consists of gene mutations skin irritation; and skin sensitisation. on dose–response relationships. It may database or a change in a previous risk well advanced. process is critical for perpetuation of in bacteria and mammalian cells, and Such studies may serve as the basis also be of value to less experienced the species and factors or agents that in vitro and in vivo cytogenetics. Newer assessment or regulatory action. for classifying and labelling a particular readers seeking further detail on using alter or disrupt this process can have assays that could provide additional chemical or mixture, and serve as an initial conventional toxicity tests. devastating consequences, both in information include the comet assay, guide to possible toxic modes of action 3 Strength of evidence’ is commonly taken to mean fact and in public perception (Korach mutations in transgenic animals, and in establishing a dosing regimen in the degree of conviction regarding the outcome of 1998). For information on study design, fluorescentin situ hybridisation and cell sub-chronic toxicity studies. Substantial an experiment such as NTP’s ‘clear evidence’, ‘some refer to OECD Test guideline 415, One- transformation. Guidance on the conduct evidence’, ‘equivocal evidence’ and ‘no evidence’ of work has been done to develop alternative generation reproduction toxicity study and interpretation of in vivo and in vitro carcinogenicity. ‘Weight of evidence’ involves integra- tests (mainly in vitro) to replace skin/ tion of all available data, not just one study.
114 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 115 There is no acceptable substitute for the principles upon which it is based conventional toxicity testing. The program to toxicological knowledge, bypassing informed judgement (based on sound are broad, and should enable its is a collaboration between the US EPA, some of the costs and time constraints of scientific principles) in analysing, use in analysing modes of action of National Institutes of Environmental conventional toxicity testing. evaluating, interpreting and weighing non-neoplastic effects of chemicals. Health Sciences/National Toxicology biological and toxicological data derived Irrespective of the nature of the disease Program, National Institutes of Health/ Another review (Creton et al. 2010) from animal toxicity studies conducted process, characterising the MoA will National Human Genome Research has focused on using acute toxicity according to currently available protocols. facilitate subsequent judgements about Institute, and the Food and Drug testing, and the possibility of replacing the human relevance of the toxicological Administration. < http://www.epa.gov/ in vivo tests with a range of alternative In addition to identifying toxic effects and findings, the possible need for further ncct/Tox21>. approaches. While recognising that the doses at which these effects do or data, risk quantification, and setting acute oral toxicity tests are important for do not occur, toxicity studies may yield appropriate regulatory standards for The project aims to: toxicity categorisation, Creton et al. (2010) insight into the mode (or mechanism) the chemical. point out that requiring additional studies •• research, develop, validate and of action (MoA) of a chemical toxicant. of dermal toxicity rarely adds anything to translate innovative chemical testing Assessing MoA is becoming a the process. Furthermore, while in vivo methods that characterise toxicity fundamental component of carcinogenic 9.4 tests for sensitisation potential probably pathways risk assessment, especially when it DEVELOPMENTS IN remain important for categorisation comes to classifying carcinogens (see •• research ways to use new tools to purposes, developing predictive in vitro Section 11.3), and making judgements TOXICITY TESTING identify chemical induced biological tests for skin/eye irritancy means that about whether a threshold or non- activity mechanisms conventional in vivo Draize-type testing is threshold approach to dose–response Environmental health risk assessment has •• prioritise which chemicals need more no longer justifiable. assessment is warranted (see Chapters traditionally been based on evaluating extensive toxicological evaluation 3 and 5). The evaluator may be toxicity data generated according to the •• develop models that can be used to While it is clear that development of able to combine information from a above principles, the knowledge base more effectively predict how chemicals alternative tests has been a high priority number of studies within the database of which dates back several decades. will affect biological responses for work and has progressed well for (e.g. metabolic/toxicokinetic, acute, However, it has long been recognised that acute endpoints, validation of the •• identify chemicals, assays, informatic short-term repeat-dose, sub-chronic, interpreting classic toxicity studies may alternative test methods for assessing eye analyses, and targeted testing needed chronic/carcinogenicity, developmental, be compromised by the high doses used irritancy are not as advanced as those for for the innovative testing methods reproductive and genotoxicity studies), and the uncertainties in extrapolating skin irritancy. Satisfactory tests for skin studies using genetically modified effects to the lower doses relevant to •• be able to provide the data generated sensitisation have yet to be validated. test mice and quantitative structure– environmental exposures. Furthermore, from the innovative chemical testing activity relationship (QSAR) analysis to animal welfare issues are likely to place methods to risk assessors to use when Morisseau et al. (2009) evaluated the adduce information about the mode increasing restrictions on using live making decisions about protecting efficacy of a range of high-throughput or mechanism of toxic action of the animals in toxicity testing. human health and environment. in vitro enzyme activation and receptor substance. assays to predict toxicity potential. They Recognising these as important issues, Andersen and Krewski (2009) concluded that such screening assays It is at the point of overviewing the entire the US EPA and the US NIEHS asked emphasised the importance of relating have some useful potential for prioritising toxicology database that the WHO/ the NRC to develop new guidance on events leading to toxicity in the context chemicals for further testing. IPCS Conceptual framework for cancer toxicity testing, which would incorporate of perturbations in biological functions, risk assessment (see Section 11.5) is new in vitro and in silico technologies some of which may be reversible The potential for incorporating new intended to be applied. This ‘framework’ and computational systems biology. An or capable of adaptive change. In types of testing strategies for assessing is an analytical tool providing a logical, interim report of this project (Toxicity relation to dose–response assessment, genotoxicity has been reviewed by structured approach to assessing the testing in the 21st century: a vision and greater emphasis was placed on using Elespuru et al. (2009) in the context overall WoE for a postulated mode of a strategy) was published in 2006 (NRC computational techniques (e.g. PBPK of criticisms of the over-interpretation carcinogenic action. Using the framework 2006) and a final report in 2007 (NRC models) to determine human exposure of positive results from the standard should increase the transparency of 2007). This has been followed by a scenarios likely to result in target tissue batteries of in vitro tests. the analysis by ensuring that the facts review of some of the challenges yet to be concentrations associated with critical and reasoning have been documented overcome and some practical approaches events in the transition from normal clearly, including any inconsistencies and to implementing this visionary program to abnormal biological function (i.e. uncertainties in the available data. (Andersen & Krewski 2009). toxicity). The use of QSAR and ‘read across’ techniques based on the known Note that although the conceptual There is no doubt that this program, now toxicological profiles of reference framework has been developed to assist called Tox21, is an important driving compounds is also on the increase, in assessing carcinogenic endpoints, force for the review and refinement of and can make a welcome contribution
116 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 117 Chapter 10: Use of epidemiological data
when the problem is a common but populations substantial effects in small and compare these exposures with However, ecological studies are relatively 10.1 ill-defined condition ... caused by numbers of sensitive individuals may 10.2 exposures among a non-diseased cohort. inexpensive for linking available health INTRODUCTION trace amounts of a highly reactive be ‘swamped’ by a lack of effect in TYPES OF The selection of appropriate controls to datasets and environmental information environmental toxin …’ (Last 2010) the majority. avoid bias is a significant challenge with and are useful for hypothesis-generation Epidemiology and toxicology are EPIDEMIOLOGICAL STUDY case-control studies. However, among (Yassi et al. 2001). Examples of ecological complementary in risk assessment. In simple terms, epidemiological methods Epidemiological studies are often most their advantages, case-control studies are studies are the assessments of the Epidemiology is the direct human compare health outcomes or calculated unreliable when measuring subtle Broadly speaking, epidemiological activity relatively inexpensive, ideal for studying relationship between tobacco sales in evidence component and, if based risk estimates in an exposed population effects, although these effects may can be either ‘descriptive’ (reporting and rare diseases and useful for investigating different countries and lung cancer on sound epidemiological methods, or group, with those in a non-exposed be of public health significance when describing the distribution of exposure multiple, different exposures (Gregg 1996). rates, and fluoride in water supplies and can provide the most important population. The types of epidemiological large populations are exposed. Even the and effect) or ‘analytical’ (designed to dental caries. evidence in characterising risk. studies that may be useful in making most rigorously conducted studies are analyse and understand the degree of Cross-sectional studies measure the Epidemiology is the principal driver in such evaluations are discussed briefly in unreliable for detecting small increases association between exposure and effect). prevalence of disease and measure A subset of ecological studies, known microbiological risk assessment, but Section 10.2. in risk. Since it is impossible to prove an Descriptive studies include case reports, exposure and effect at the same time. as ‘time series studies’, is regarded it can assist the formal framework of absence of risk from any human study, case series and cross-sectional surveys. They are relatively easy and economical as very helpful in understanding the EHRA at all of its stages. Environmental epidemiology is considered it is often considered that the principal Cross-sectional surveys examine exposure to conduct and are particularly useful influence of short-term fluctuations in to be a subspecialty of epidemiology that value of epidemiology is to exclude and disease prevalence at the same point for measuring fixed characteristics of air pollutants on day-to-day changes in The term ‘risk’ tends to be used in a addresses the effects of environmental major health effects of an exposure in time (or over a short duration of time) individuals such as socioeconomic status population morbidity and mortality after subtly different and more specific way in exposures on health and disease in the (NHMRC 2006). and thus are unable to support causal (Bonita et al. 2007). controlling for factors such as season and epidemiology than in risk assessment. population (Baker & Nieuwenhuijsen inference because it is not possible to air temperature. However, disentangling In epidemiology, risk describes the 2008). The purpose of this chapter is to provide know if exposure pre-dated the onset of Cohort studies follow cohorts or groups the effects of individual pollutants as ‘frequency of occurrence of a disease a basis for understanding the strengths disease. Similarly, case reports and cross- of individuals, defined in terms of their measured in a mixture such as urban air in one population compared with Epidemiological methods can be used and weaknesses of epidemiology in sectional surveys are unable to support exposures, over time to see if there are pollution can be quite difficult. another, either as a difference in rates to investigate the cause of adverse supporting risk assessment. As Mundt causal inference. differences in the development of new (attributable risk) or as a ratio of rates health effects, the natural history of et al. (1998) noted that if the limitations cases of the disease of interest (or other To strengthen the design of ecological (relative risk)’ (ACDP 1996 p. 20). health conditions and the description of epidemiological studies are not In practical terms in environmental health outcomes) between the groups studies, Nurminen (1995) recommended The feature distinguishing the two of the health status of populations, and understood by the risk assessment team, epidemiology, there are four main with and without exposure. Such studies selecting areas with populations that: categories of analytical study (from populations by its presence or distribution to evaluate health-related interventions the validity of an assessment might be can be carried out by either reviewing •• are homogeneously exposed (to Moolgavkar et al. 1999): is referred to as a ‘risk factor’. The (Bonita et al. 2007). They also allow for compromised by including inappropriate, past records (retrospective) or by tracking minimise within-area exposure reliance on comparisons of disease rates the control for possible confounding possibly misleading, epidemiological •• case-control studies people into the future (prospective variation) between populations creates substantial factors that may influence the results. data. In discussing the potential for cohort). The essential feature of these •• cross-sectional studies •• represent different extremes of In the context of environmental health, epidemiological studies to throw up false longitudinal studies is that for each limitations for the sensitivity of relative exposure distribution (to maximise epidemiological methods may also negative outcomes, particularly in relation •• cohort (longitudinal) studies individual prior exposure information risk determination for common diseases between-area exposure variations) (Thomas & Hrudey 1997). be used to characterise population to associations with cancer, Boffetta et al. •• ecological studies (including a special can be related to subsequent disease exposures, investigate perceived clusters (2008) made a plea for epidemiologists to sub-group known as time-series experience (Breslow & Day 1987). •• are comparable with respect Epidemiology is ‘the study of the of disease, to develop health surveillance practise ‘epistemological modesty’ in the studies). to co-variate distributions (e.g. distribution and determinants of health- programs, to establish a baseline, and types of claims or conclusions drawn from Ecological studies involve investigating socioeconomic status, demography) related states or events in specified to monitor the consequences of risk their studies. Cohort, cross-sectional and case-control a group of people such as those living •• use the smallest possible sampling populations, and the application of the management activities. studies differ from ecological studies in within a geographical area (e.g. a region, units for ecological analysis. study to the control of health problems’ The systematic appraisal of that information on exposure and disease state or territory). For example, place and (Last 1988). A more recent description of At the same time, there are often epidemiological studies is intended to is available on an individual basis. With time of residence can provide aggregate The largest number of environmental epidemiology by the same author states: unrealistic expectations of what an answer the question: ‘Is there any other ecological studies this information is exposures, so may be used to create epidemiology studies found in the epidemiological study may be able way of explaining the set of facts before only available on a group basis, so the surrogate measures of the real exposure literature are of the ecological or cross- ‘Epidemiology connects the dots, to achieve. us [i.e. the study results, is there any community or region is the unit of analysis. of interest (Elliott et al. 1992). Rates of sectional type, because they are easier the isolated bits of information that other answer equally, or more, likely disease and average exposure levels to carry out and cost less (Thomas & begin to form a coherent pattern In epidemiological studies of the potential than cause and effect’ (WHO 2000a). In case-control studies, exposure and other to a particular agent are determined Hrudey 1997). However, as noted above when connected in the right way ... influence of environmental factors, Alternative explanations may result from attributes of cases of the disease under independently, and on a group basis. and discussed further below in relation to The dots can come from anywhere. the exposures encountered are rarely chance, bias and confounding. investigation are compared with those from This may give rise to spurious apparent assessing causality, such studies may be Identifying them demands a broader extreme, in contrast to the high levels of a suitable control or comparison group correlation, called the ‘ecological fallacy’. useful for identifying potential hazards or perspective, the ability to see the Big exposure that can be applied in animal- of persons unaffected by the disease, Because it is not ascertained whether hypothesis generation, but they cannot Picture ... Sometimes the way the dots based toxicity tests. Such epidemiological and analysed to yield effect estimates. individuals who have been exposed to determine cause and effect. are connected is instantly apparent. studies are often confounded (see Section The approach is to start with a diseased the agent are the same individuals who Sometimes painstaking investigation 10.3), and any associations found are group and look backwards (retrospective developed the disease, statements about Characteristics of the various study types and analysis are required, for instance commonly weak. Furthermore in large cohort) to their history of exposures, causal associations are inappropriate. are summarised in Table 13.
118 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 119 Table 13: Study designs in environmental epidemiology that use the individual as the unit of analysis Epidemiological studies are rarely Table 14: Applications of different observational study designs definitive, and a single epidemiological Study design Population Exposure Health effect Confounders Problems Advantages study cannot establish causality. Any Ecological Cross-sectional Case-control Cohort such study is necessarily a sample of the Case reports, case Community or Records of past Mortality and Difficult to sort out Hard to establish Cheap; useful Investigation of rare disease ++++ – +++++ – total population of interest, so there will series and other various sub- measurements morbidity statistics; exposure-effect to formulate Investigation of rare cause ++ – – +++++ descriptive studies populations case registers; relationships hypotheses always be questions about being able other reports to generalise from an individual study Testing multiple effects of cause + ++ – +++++ sample to the total population. Moreover, Study of multiple exposures and Cross-sectional Communities or Current Current Usually Current exposure Can be done ++ ++ ++++ +++ undetected methodological biases can only determinants study special groups; may be irrelevant quickly; can use be overcome by having numerous studies exposed versus to current disease large populations; Measurement of time by different investigators with different ++ – +a +++++ non-exposed can estimate relationships prevalence population samples. If consistency of outcome is demonstrated across many Direct measurement of incidence – – +b +++++ Case-control study Diseased (cases) Records or Known at start of If confounders can Difficult to Relatively cheap studies, the causal hypothesis becomes Investigation of long latent periods – – +++ +c/- versus non- interview study be identified and generalise; may and quick; more likely. A ‘weight of evidence’ diseased (controls) measured they incorporate biases; particularly useful (WoE) approach is generally required, Key: may be addressed cannot derive rates for studying rare involving the interpretation of all available diseases + to +++++ indicates the degree of suitability from least to most suitable; – indicates not suitable Recall bias is a information and consideration of the a if prospective; b if population-based; c if retrospective or historical cohort study. major problem strengths and weaknesses of each study.
Time-series study Large community Current (e.g. Current (e.g. Often difficult Many confounding Useful for studies Table 15: Advantages and disadvantages of different observational study designs (several million daily) changes in daily) variations in to sort out; e.g. factors; often on acute effects Unfortunately, experimental interventions people); exposure mortality effects of influenza difficult to measure such as randomised controlled trials are Ecological Cross-sectional Case-control Cohort susceptible groups rarely available to assist environmental such as asthmatics health risk assessment, as it is not Probability of: ethical to purposely expose populations Selection bias N/A Medium High Low Historical Special groups; Records of past Records of past or Often difficult Need to rely on Less expensive to hazardous risks. It is appropriate, (retrospective) workers, patients, measurement current diagnosis because of records that may and quicker than a Recall bias N/A High High Low however, to expose groups to a lesser level cohort study insured persons retrospective not be accurate prospective study; Loss to follow-up N/A N/A Low High nature; depends can be used to of risk via a clean-up intervention. An on availability of study exposures example of an experimental intervention Confounding High Medium Medium Low previously obtained that no longer exist is a randomised trial of lead abatement Time required Low Medium Medium High data procedures undertaken in Broken Hill Cost Low Medium Medium High (S. Corbett, personal communication). Prospective cohort Community or Defined at outset of To be determined Usually easy to Expensive and Can estimate study special groups; study (can change during study measure time consuming; incidence and Epidemiological studies, depending on Tables 14 and 15 adapted from: Bonita et al. 2007. exposed versus during study) exposure relative risk; their design, may serve two purposes: non-exposed categories can can study hypothesis generation or assessment of a causal relationship. Their ability to change; high many diseases region or scenario linked to a possible dropout rate in one study; evaluate a causal relationship may be 10.3 environmental exposure source (e.g. near possible can describe limited by a lack of control of potential a waste dump, or in an occupational associations that confounders or a lack of power (usually INVESTIGATION OF setting). The assessment is usually first suggest cause– the result of limited sample sizes) (Samet APPARENT CLUSTERS centred on whether the observed number effect relationships et al. 1998). of cases is consistent with that expected Experimental Community or Controlled or To be measured Can be controlled Expensive; ethical Well accepted The assessment of an apparent cluster of from the background incidence, or (clinical/ special groups already known during study by randomisation considerations; results; strong 10.2.1 non-communicable disease is a complex whether there is a sufficiently common intervention) study of subjects study subjects evidence for Observational studies and resource-intensive task. It commonly compliance causality or efficacy pattern to the nature of the cancers or required of intervention Different observational study designs have involves investigation of a number of other health effect. different applications, advantages and reported cases of cancer, or some other adverse health effect likely to be linked Specific guidance on cluster investigation Adapted from: WHO (1991). disadvantages (see Tables 14 and 15). These comparisons assume the different to an environmental exposure. It should has been developed by Queensland types of studies are equally well designed. be managed using a multidisciplinary Health (2009), including suggested Even so, design variations may affect their approach using standardised analytical criteria for decision making. The NHMRC performance and provide exceptions. tools. The trigger for a cluster investigation is working towards the development of Bonita et al. (2007) provides a more is often the ‘perception’ that the incidence guidelines that can be adopted nationally. detailed description. of the disease is unusually high in a
120 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 121 that independently affects the risk of Causation of adverse health effects is Figure 29: Assessing the relationship and Greenland (1997) note, apart the laboratory). Notwithstanding this, 10.4 developing the disease. affected by four types of factors: between a possible cause and an outcome from temporality (whereby a putative causality may be imputed by the when an association is observed cause must precede the effect), there strength and consistency of evidence. BIAS AND CONFOUNDING: •• predisposing factors (e.g. immune Three conditions are necessary for a are no necessary and sufficient criteria deficiencies, gender and previous KEY CONCEPTS IN factor to be categorised as a confounder: for determining whether an observed With environmental health in particular, illness) association is causal. Lucas & McMichael much decision making rests on a WoE ENVIRONMENTAL •• It must be a risk factor for the disease •• enabling factors (e.g. poor nutrition (2005) provide a useful summary of approach rather than definitive proof of in the absence of the exposure under and bad housing may favour the these issues and conclude that, as with cause, which is commonly not available EPIDEMIOLOGY study (it does not have to be an actual development of disease) statistical p tests, the criteria of causality – hence the concept, ‘Judging the cause but can be a marker of an There are many ways in which error •• precipitating factors (e.g. the exposure must be viewed as aids to judgement, evidence’ in Table 16, is particularly actual cause). can be introduced into epidemiological to a specific disease agent) not as arbiters of reality. Thus the term relevant in assessment of risk. •• It must be associated with the studies. Error may be random (due to •• reinforcing factors (e.g. repeated ‘guidelines’ is more appropriate than the exposure in the study population. chance alone, and potentially reduced exposure may aggravate an established slightly more absolute ‘criteria’, and there The guidelines are arranged in a logical by improving sample size) or systematic •• It must not be affected by the exposure disease or state). is not necessarily an easy epidemiological sequence for making judgements on (and not reduced by increasing sample or disease. (In particular, it cannot be roadmap to finally determine causation. causality. They are not weighted equally, size). While this section does not attempt an intermediate factor in the causal The term ‘risk factor’ is commonly used and their relative contribution to a final to deal with the subject of systematic error pathway between exposure and to describe factors that are positively The guidelines are similar in some judgement will vary from one situation in any depth, the two key concepts of bias disease. An intermediate factor is one associated with the risk of development respects to Koch’s postulates for to another (Thomas & Hrudey 1997). and confounding must be highlighted. that is caused by the exposure and, of a disease but that are not sufficient determining whether an organism Consistency can be demonstrated if The size of the statistical confidence in turn, causes the disease.) in themselves to cause the disease. is causal of a particular disease. He several studies give the same result, intervals will provide an indication of the A ‘sufficient’ cause is one that inevitably required that the putative organism was especially if a variety of designs is potential for random sampling error, but There are specific approaches for produces or initiates a disease, and a to be found in every case of the disease, used in different settings since this statistical confidence intervals do not controlling confounding that can be used ‘necessary’ cause is one for which a and that it could be isolated and grown reduces the likelihood that all studies represent uncertainty arising from bias in both the design and analysis of analytic disease cannot develop in its absence from cases of the disease. Further, the are making the same mistake. However, or confounding. studies, providing that the confounding (Bonita et al. 2007). In the biological isolate should produce a like disease in a other factors such as different exposure variables have been identified and sciences, there is often a constellation new host and in turn the organism should levels or study conditions may need Bias occurs when there is a systematic measured. Randomisation of exposure of components acting in concert for a be recovered from that case. Modern to be taken into account, and the best tendency by a study to produce results can control confounding, however this cause to create an effect, and many microbiology is now finding instances designed studies should be given the that diverge from the truth. There are option is not usually available because of the components of a ‘sufficient where these postulates cannot be fully greatest weight. It is important to note that many sources and varieties of bias, but exposure tends not to be planned by cause’ may be unknown (Rothman complied with (e.g. the organism can in environmental epidemiology, reliance the most important include selection bias the researcher. Matching groups can, & Greenland 1997). At the low levels of be detected using molecular methods, on a single pivotal study is the exception and measurement (or classification) bias. however, remove confounding effects, for exposure commonly encountered in the but cannot be isolated or grown in rather than the rule. Bonita et al. (2007) include a succinct example, by investigating people of similar environment and where there may be account of bias. It may be difficult to age groups or gender, or by comparing a range of contributory factors present, precisely estimate the effect bias has in smokers with smokers. it may be difficult or inappropriate to Table 16: Guidelines for assessing causation a study, but it is vital for risk assessors assign this nomenclature to an agent to look for and attempt to identify the 10.5 even though the agent is accepted Temporal relation Does the cause precede the effect? (essential) potential size and direction of bias in as causing a specific effect with Plausibility Is the association consistent with other knowledge? interpreting a study’s findings. Recall ASSESSING THE high exposures. (mechanism of action; evidence from experimental animals) bias is important because case-control Consistency Have similar results been shown in other studies? studies are much more common for RELATIONSHIP BETWEEN As with other scientific disciplines, environmental epidemiology than are A POSSIBLE CAUSE AND epidemiology has attempted to define a Strength What is the strength of the association between the cause and cohort studies. set of causal criteria to help distinguish the effect? AN OUTCOME causal from non-causal associations. In Dose–response relationship Is increased exposure to the possible cause associated with an Confounding is the distortion of the effect the first place, other explanations for a increased effect? of the agent of interest by an extraneous A cause is ‘an event, condition, potentially causal association must be If alternative explanations such as bias factor (Moolgavkar et al. 1999). This Reversibility Does removal of the possible cause lead to a reduction of characteristic or a combination of excluded (such as chance, selection or and confounding can be excluded, it may occur if another exposure exists in disease risk? these factors which plays an important measurement bias, or confounding). is then useful to systematically apply the study population that is associated role in producing the disease’ (Bonita Particularly rigorous scrutiny should be guidelines for assessing causation as Study design Is the evidence based on a strong study design? with both the disease (or outcome) and et al. 2007). given to studies giving a positive but not shown in Table 16. The concepts in the exposure being studied, such as Judging the evidence How many lines of evidence lead to the conclusion? statistically significant result. Figure 29 these guidelines derive from work by Hill a third factor (‘confounding variable’) illustrates this process. (1965) and others. However, as Rothman Adapted from: Bonita et al. 2007.
122 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 123 The technique of meta-analysis grew In well-conducted cohort studies, bias •• ensuring that all relevant studies are •• genetic diversity may be broad in Areas where epidemiology may help to out of the need to reduce random error is minimised. Case-control studies are 10.6 collected and collated humans compared with the more inform dose–response relationships from in clinical trials. Meta-analysis in the subject to several forms of bias and THE STRENGTHS •• assessment of the quality of the studies restrictive phenotypes of selectively animal studies may include: context of systematic reviews can be weaknesses related to time sequence bred animal strains used in •• evaluation of the weight of evidence •• reduced uncertainty about inter- used to pool the data from well-designed but, if well designed, may still provide AND LIMITATIONS toxicological studies (WoE) each brings to the conclusions; species variability in metabolism, life studies, each of which may deal with a useful evidence for the causal nature of OF OBSERVATIONAL whether the study can be categorised •• epidemiological studies may include span, and genetic diversity relatively small sample size, in order to an association. Cross-sectional studies are as providing ‘acceptable’ results from different groups (e.g. the young, obtain a better overall estimate of effect. weaker as they provide no direct evidence •• complex temporal patterns of EPIDEMIOLOGY VERSUS a WoE perspective, ‘supplemental’ old and susceptible) that may not Meta-analysis has pitfalls if poor quality on the time sequence of events. exposure and doses – this relates results that have limited use in a WoE be included in the usually relatively studies are included, and needs to be EXPERIMENTAL to situations where risk assessment approach, or ‘unacceptable’ results homogeneous groups used in applied with caution to observational Ecological studies are the least may be impossible to replicate (e.g. TOXICOLOGY from a WoE perspective toxicological studies studies (which are less able to control satisfactory because of the dangers of animal studies), whereas some •• assignment of a scale value to the •• effects on some aspects of mental epidemiological studies may be for confounding than randomised trials). incorrect extrapolation to individuals from Epidemiological studies are crucial for conclusions, that can be related to grid function or behaviour, and more more useful for understanding these Standard methods for conducting data derived from regions or countries. assessing effects directly in humans and values that rank the epidemiological subjective effects such as nausea or complex dose–response relationships and reporting systematic reviews have However, where certain exposures cannot estimating population attributable risks. evidence from ‘for’ to ‘against’ a causal headache, can be better assessed in been published (Greenhalgh 1997). normally be measured individually (e.g. However, their power of resolution is •• the ability to assess large numbers hypothesis, and the toxicological human studies. See NHMRC (2000) How to review the air pollution, pesticides residues in food, limited, mainly because of the difficulties of people exposed to low levels of an evidence: Systematic identification and fluoride in drinking water) evidence from evidence as providing ‘high’ or ‘low’ agent. The doses from exposure to a in estimating exposure precisely and in Differences in hazard identification based review of the scientific literature. ecological studies may be important in evidence for biological plausibility of hazardous agent in epidemiological controlling bias. Toxicological studies on epidemiological and toxicological environmental health decision making. the findings studies are often considerably less are necessary for elucidating causal data may be seen in the matter of ‘site The strongest evidence comes from Time series studies demonstrating health •• assignment of the conclusion to a than in toxicological studies. This mechanisms, which may be important concordance’. The epidemiological data well-designed and competently outcomes associated with fluctuating air position on a causal relationship may have the advantage of providing for determining dose–response relations may suggest lung cancer is of concern, conducted randomised controlled trials. pollutant levels may be one particularly grid; with the grid categorising the information about the exposure range and extrapolating to low doses in risk whereas the toxicological data may The NHMRC (1999b) places strongest useful example. evidence for a causal relationship into of interest although, if they are the assessment. On the other hand, direct suggest liver cancer. Similar conflicts emphasis on evidence obtained from quadrants, labelled ‘likely’, ‘unlikely’ result of (prolonged) adult occupational generalisations to humans based on animal can arise where there are suggestions systematic reviews of all relevant (and The above principles about strength of or ‘uncertain’. exposures, the exposures are likely data are often uncertain (Pershagen 1999). of a problem from epidemiological data well-conducted) randomised controlled evidence obtained from various study to be considerably more than those unsupported by toxicological evidence trials (‘Level I’).4 types are summarised in Table 17. An illustration of the application of this experienced by people in the general Epidemiological studies are often given (Samet et al. 1998). increased weighting because they framework to assess the strength of population. With appropriate tools However, there are relatively few such Table 17: Relative ability of different types come from humans but, compared with epidemiological and toxicological evidence small differences in relative risk in large trials available for environmental health linking the herbicide atrazine to cancer 10.6.2 populations may be able to be assessed of study to ‘prove’ causation toxicological studies of animals, may be Dose–response assessment hazards that could form the basis for a more costly and time consuming and causation in humans was presented in a (Roseman 1998; Samet et al. 1998). systematic review. Most apply to the effects Ability to ‘prove’ more likely to result in ambiguous findings companion paper (Simpkins et al. 2011). Controlled exposure studies offer the only of treatment or prevention campaigns. Type of study causation (Samet et al. 1998). However, substantive real possibility of accurately assessing However, it is more common that A rare example is the Melbourne Water findings have been obtained at times 10.6.1 dose–response relationships in human epidemiological studies are limited by Quality Study, which was a blinded study Randomised controlled Strong the amount of data available on dose trials through opportunistic study of highly Hazard identification studies. These may occasionally still be involving real and sham domestic reverse exposed groups – such as occupational done under rigorous ethical controls, with and tend to address exposure–response Since human risk is assessed directly, osmosis water filters and an assessment Cohort studies Moderate cohorts or communities that have been some types of chemicals having minimal relationships (i.e. they are based on epidemiology has a number of potential of acute gastrointestinal disease (Hellard inadvertently exposed to contaminants toxicity potential. Examples include air whether or not exposure occurred) Case-control studies Weak/moderate advantages over animal toxicology in the et al. 2001). (e.g. via food or water). These can be chamber studies with mildly irritant or rather than dose–response relationships. Cross-sectional studies Weak area of hazard identification: either observational epidemiological odoriferous gases and vapours, where Doses are usually discontinuous and In practice, most evidence comes from Ecological studies Weak studies or what Lilienfield and Lilienfield •• differences between humans and the objective is to assess thresholds for variable in epidemiological studies observational studies (e.g. nearly all the (1980) called ‘natural experiments’. animals in relation to absorption, the onset of effects or sensory perception compared with controlled toxicological evidence on the health effects of smoking). Adapted from: Bonita et al. 2007. metabolism, detoxification and and for self-limiting, non-lethal microbial experiments. An integrated measure of An appropriate means of integrating excretion no longer need adjustment pathogens like Cryptosporidium using exposure may need to be developed to 4 The NHMRC document referred to is oriented The ranking in this table assumes that information derived from epidemiological or consideration healthy volunteers. represent the non-uniform doses. towards clinical interventions and clinical practice studies are well designed and well and toxicological studies would be a •• sample sizes for most types of guideline development. At present there are Quantitative description of dose–response conducted in each case. Even the significant step forward in assisting epidemiology studies are large Epidemiological data may still assist in no comparable endorsed ‘levels of evidence’ to relationships may be hampered by guide assessment of epidemiological evidence for presence of a strong ability to ‘prove’ the processes of risk assessment. The in comparison to the number of assessing dose–response relationships, incomplete information on exposure environmental health practice, although in 2009 causation should be supplemented by development of a framework for such animals used in animal studies even when controls over exposure in (especially for biologically relevant the NHMRC published a broader approach to mechanistic knowledge to be confident integration has recently been proposed by (usually no more than 10–50 of human studies are unavailable. evidence grading for use in developing guidelines. time windows), by exposure or dose of causation. Adami et al. (2011). The key elements of each sex per dose) Available at
124 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 125 surrogate markers of exposure. Incorrect If epidemiologic data adequate for dose– heterogeneity in the human population etc.) should be used to refine Future studies should be designed must have taken into account the information about the exposure may response evaluation is not available, and should be accounted for. Whenever low-dose extrapolation procedures in such a way as to better capture appropriate research strategy. bias the description of the exposure– a risk assessment is being developed for feasible, human data on metabolic when such phenomena are such information. For instance, if demonstrating causal response relationship. If there are wide use in making an important regulatory biomarkers and other biological adequately understood. effects is required, a survey may be confidence intervals around the results decision, and if it is feasible to develop measures should be employed to inappropriate for satisfying the aims there can be substantially different policy new epidemiologic data, or to extract adjust the risk estimates for known 10.6.3 10.7 of the research. endpoints depending on whether the new data from existing studies, an effort differences between species and Exposure assessment CRITICAL EVALUATION OF •• Selecting appropriate variables. The upper bound, the lower bound or the should be made to develop and provide between high and low doses. If operational definition of the variables A lack of good exposure data is a midpoint has been chosen for policy good epidemiologic dose–response data possible, data on susceptibility should PUBLISHED RESEARCH being investigated calls for selecting common pitfall of environmental making (Samet et al. 1998). that can be used together with, or in be included. appropriate measurement strategies. epidemiological studies, to the extent preference to, high-dose animal data. •• Principle 5: When epidemiologic This section is reproduced, with minor If the selection of the variables is that such studies tend only to be as good Commonly, too, there is insufficient studies are selected for dose–response adaptation, from Introduction to research inappropriate to the constructs being as their exposure data. The association epidemiological data to discriminate The ‘London principles’ (Federal Focus assessment, higher-quality studies in the health sciences by Polgar and investigated, then the investigation will of particular health effects and specific between alternative models that could 1996) may be used to guide the choice of should be given preference, especially Thomas (1991), pp. 302–306 by not produce useful results. patterns of exposure, if in keeping describe the dose–response relationship. studies in this critical area: those with precise and accurate permission of the publisher Churchill This is particularly important at very low with knowledge of pathophysiology, Livingstone. Italicised questions are •• Principle 1: Dose–response exposure information. The availability 10.7.2 exposure levels, and this is where both can provide strong support for causal from Riegelman 1981, p. 73 and assessment should include a range of information with respect to timing Critical evaluation of the methods epidemiological and toxicological data interpretations (WHO 2000a). British Medical Journal 1988, p. 50 of reasonable dose measures and an of exposure and response (time/age of section is often limited. Surrogate measures (with minor amendments). explanation why any were rejected, first exposure, intensity of exposure, of outcome (e.g. nerve conduction or Illustrating this problem, Saunders et A well-documented methods section is a and provide a rationale if any particular time to tumour), adjustment for tremor) and a relationship between the al. (1997) reviewed the 14 best studies 10.7.1 necessary condition for understanding, dose metric is preferred. In evaluations confounding variables, and potential surrogate measures and health outcomes (judged for potential to support causal evaluating and perhaps replicating a of both human and animal data, interaction with other effect modifiers Critical evaluation of the introduction may need to be established in order inference) selected from a shortlist of the research project. In general, the critical several different measures of dose is particularly important. to interpret the significance of a study, 43 analytical studies assessing human The introduction of a paper should evaluation of this section will reveal the should be evaluated (if possible). •• Principle 6: A properly conducted although care needs to be taken that the health effects in relation to hazardous essentially reflect the planning of the overall internal and external validity of meta-analysis, or preferably an surrogate outcomes do relate to clinically •• Principle 2: In the selection of a dose– waste sites that were identified among research. Inadequacies in this section the investigation. analysis based on the raw data in the meaningful outcomes. response model, the greatest weight hundreds that were published. They might signal that the research project was should be given to models that fit the original studies, may be used in hazard found that poor exposure measurement erroneously conceived, or poorly planned. identification and dose–response 10.7.2.1 Issues of sample size and whether observed animal and human data and was a major factor in the overall lack of The following issues are essential for evaluation when such combination Subjects a threshold or non-threshold can be are consistent with the biologically convincing evidence of causation from evaluating this section: relevant modes of action (genotoxic, includes an evaluation of individual these studies. It is often the case that The section shows if the sample was demonstrated also need to be addressed. •• Adequacy of the literature review. non-genotoxic, unclassified). When studies and an assessment of only a broad indication of the level or representative of the target population If assessing whole populations, the The literature review must be mechanistic knowledge is uncertain heterogeneity. The combined results nature of exposure may be deduced from and the adequacy of the sampling dose–response relationship is going to be sufficiently complete so as to reflect or limited, several plausible dose– ought to provide, more than any epidemiological studies. model used. different than in a smaller sub-sample. the current state of knowledge in response models should be considered single study, precise risk estimates The population sample is likely to include the area. Key papers should not be •• Sampling model used: A number of and the most plausible ones, based over a wider range of doses. Before Experimental toxicological studies, on the vulnerable groups, while the smaller omitted, particularly when their results sampling models can be employed on available data and professional using these tools, the gains should be other hand, generally have the advantage sample may not. This becomes an issue could have direct consequences for to optimise the representativeness judgement, should generally be used judged sufficient to justify potential of control and accurate measurement when standards are chosen on the basis the research hypotheses or aims. of a sample. If the sampling model in dose–response evaluation. errors in inference resulting from of exposures. Nevertheless, at times of toxicity data that appears to show a Researchers must be unbiased is inappropriate, then the sample combining studies of dissimilar design environmental epidemiological studies NOAEL based on a small study population •• Principle 3: When extrapolating cancer in presenting evidence that is might be biased, raising questions and quality. may be the only way of determining the and/or a study population that only risk to exposure levels below the unfavourable to their points of view. concerning the external validity of the includes healthy adults or, for example, observable range, mechanistic data •• Principle 7: When epidemiological data distribution of ‘real-life’ exposures in •• Clearly defined aims or hypotheses. research findings. only adults with mild asthma, when an air should be used to characterise the is used in dose–response assessment, terms of: a quantitative sensitivity analysis The aims or hypotheses of an •• Sample size: Use of a small sample pollutant is being assessed. shape of the dose–response function. •• magnitude should be conducted to determine the investigation should be clearly and is not necessarily a refutation of •• Principle 4: When the available •• duration operationally stated. If this is lacking, an investigation, if the sample is The reviewer or risk assessor should potential effects on risk estimates of epidemiologic data is not adequate then how the evidence obtained in representative. However, given a highly answer the basic question of whether the confounders, measurement error, and •• temporal patterns to perform dose–response analyses, the investigation is to be used for variable, heterogeneous population, epidemiologic data, in an individual study other sources of uncontrolled bias in •• routes requiring low-dose estimates of risk conceptual advances in the area will a small sample will not be adequate or cumulatively, is adequate for use in study design. to be derived exclusively from animal •• size of exposed population be ambiguous. to ensure representativeness. Also, a dose–response evaluation. There is no data, every effort should still be made •• Principle 8: Scientific understanding •• nature of exposed population. •• Selecting an appropriate research small sample size could decrease the formula or quantitative weighting scheme to use the available human data in of differentials in human susceptibility strategy. In formulating the aims power of the statistical analysis. prescribed for making this judgement. assessing the validity of low-dose to disease (racial/ethnic/gender/genetic of the investigation, the researcher risk estimates. To the extent feasible, differences, genetic polymorphisms,
126 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 127 •• Description of the sample: Was evaluation of causal effects. A variety experimenters or observers are •• Were the results adjusted to take •• p < 0.05 is not sacred: It is an arbitrary •• Drawing correct inferences from the there a power-based assessment of of threats to internal validity must be aware of the aims and predicted into account the effect of possible cut-off value. data: The inferences from the data adequacy of sample size? A clear considered when critically evaluating outcomes of the investigation, then confounding variables? Common •• Correlation or association: Does not must take account of the limitations description of key sample variables an investigation. the validity of the investigation will confounders are age and sex, imply causation. of descriptive and inferential statistics. (for example, age, sex, type and •• Control groups: A specific way of be threatened through bias and regional differences, socioeconomic Correlations do not necessarily imply severity of condition) should be controlling for extraneous effects expectancy effects. differentials, smoking, occupation, 10.7.2.6 causation, or that a lack of significance provided. When necessary and is the use of control groups. If no •• Settings: The setting in which a ethnic differences. Critical evaluation of the results in the analysis could imply a Type 2 possible, demographic information control groups are employed, then study is carried out has implications •• Was a significance test properly error (see below). concerning the population should the internal validity of the investigation for external (ecological) validity. An performed to assess the probability The ways in which epidemiological data •• Logically correct interpretations of be provided. Was the population of might be questioned. Also, if placebo adequate description of the setting that the difference was due to chance? are properly presented and analysed the findings: Interpretation of the adequate composition to answer the goes beyond the scope of this enHealth or untreated groups are not present, is necessary for evaluating the •• Was a proper measure of the size of findings must follow from the statistical study questions? If not, the reader document in terms of complexity and the size of the effects due to the generalisability of the findings. the difference presented? inferences, without extraneous cannot judge the representativeness depth, and reference should be made treatments might be difficult to •• Times of exposures and observations: evidence being introduced. For of the sample. Also, the readers might •• Was a proper measure of the degree of to standard texts. However, the following estimate. The sequence of exposures and instance, if the investigation used not be able to decide if the findings overlap of the differences presented? general points can be made. •• Subject assignment: When using an observations must be clearly indicated an n = 1 design, the conclusions are applicable to the specific groups of •• Were the confidence intervals given for experimental design, care must be so that issues such as series effects •• The results: Results should represent should not claim that a procedure is patients being treated. the main results? taken when assigning subjects so as and confounding can be detected. a statistically correct summary and generally useful. to avoid significant initial differences Identifying variability in treatment analysis of the data. Inadequacies •• Protocol deviations: In interpreting the Motulsky (1995) provides a useful 10.7.2.2 between exposure groups. Even and observation times can influence in this section could indicate that data, the investigator must indicate checklist of common pitfalls to bear in Instruments/apparatus when quasi-experimental or natural the internal validity of experimental, inferences drawn by the investigator and take into account unexpected mind when reading research papers that comparison strategies are used, quasi-experimental or n = 1 designs, were erroneous. deviations from the intended design. The validity and reliability of observations include statistical analysis, which has care must be taken to establish the resulting in, for instance, internal For instance, a placebo/active and measurements are fundamental been adapted as follows: •• Tables and graphs: Data should be characteristics of good research. In this equivalence of the groups. validity problems. correctly tabulated or drawn and treatment code might be broken, section, the investigator must demonstrate –– Was there a satisfactory statement •• Look at the data: Summary statistics adequately labelled for interpretation. or ‘contamination’ between control may result in the loss of useful and experimental groups might be the adequacy of the equipment used for given of the source of subjects? 10.7.2.4 •• Complete summaries: Complete information. discovered. If such deviations are the data collection. Assessment of outcome summaries of all the relevant findings –– Was a satisfactory response rate discovered by investigators, they are •• Validity and reliability: The investigator achieved? •• Was the assessment of outcome •• Beware of very large and very small should be presented. samples: large samples may generate obliged to report these, so that the should use standardised apparatus, properly performed in the study and •• Selection of statistics. Both descriptive –– Was the assignment of people statistically significant but unimportant implications on the results might be or establish the validity and reliability the control groups? and inferential statistics must be to study and control groups findings small samples have little taken into account. of new apparatus used. The lack of •• Was the measure of outcome selected appropriately. Selecting appropriate? power to detect important differences. •• Generalisation from the findings: proven validity and reliability will raise appropriate to the study aims? inappropriate statistics could distort Strictly speaking, the data obtained questions about the adequacy of the –– Could selection bias have occurred? •• Beware of multiple comparisons: the findings and lead to inappropriate •• Was the measure of outcome precise? from a given sample are generalisable empirical findings. –– If the study was experimental, were When analysing random data, on inferences. •• Was the measure of outcome average 1 out of 20 comparisons will only to the population from which •• Description of instrumentation: random and blind assignment •• Calculation of statistics: Both complete? be statistically significant p( < 0.05) the sample was drawn. This point is Full description of the structure and maintained? descriptive and inferential statistics •• Did the process of observation affect by chance. sometimes ignored by investigators, use of novel instrumentation should be –– Regardless of the study type, must be correctly calculated. Using and the findings are generalised to the outcome? •• Don’t focus on averages alone: presented so that the instrument can were the study and control groups computers generally ensures this, subjects or situations that were not Variability may reflect real biological be replicated by independent parties. comparable with respect to although some attention must be considered in the original sampling. 10.7.2.5 diversity, and outliers may be more paid to gross errors when evaluating characteristics other than the study •• Statistical and practical significance: Critical evaluation of statistical important. the data. 10.7.2.3 factors? Statistical significance does not analysis •• ‘Garbage in, garbage out’: Statistical Procedures •• Exposure parameters: Was exposure necessarily imply that the results of an •• Was there a statement adequately tests do not tell whether the study was 10.7.2.7 adequately defined? It is important to investigation are applicable in practical Full description of how the investigation describing or referencing all statistical conducted properly. describe all the exposures experienced Critically evaluating the discussion terms. In deciding on practical was carried out is necessary for both procedures used? replication and for the evaluation of its by the different groups. If the •• Confidence limits: Are as informative In the discussion, the investigator draws significance, factors such as the size of •• Were the statistical analyses used internal and external validity. exposures differ in intensity or in the as p values (and may be more inferences from the data in relation to effect, side effects, cost-effectiveness appropriate? quality of the administering personnel, so, particularly when dealing with the initial aims or hypotheses of the and value judgements concerning •• Adequacy of the design: A good design then the internal validity of the project •• Was the presentation of statistical hazards). investigation. Unless the inferences are outcome must be considered. should limit alternative interpretations is threatened. material satisfactory? •• Statistical significance:Does not correctly made, the conclusion drawn •• Theoretical significance: It is of the data. A poor design will result in necessarily indicate biological might lead to useless or dangerous necessary to relate the results of uncontrolled influences by extraneous •• Rosenthal and Hawthorne effects: •• Did the analysis properly compare importance. treatments being offered to clients. an investigation to previous related variables, negating the unequivocal Whenever possible, studies should be the outcomes in the study and the double or single blind. If the subjects, control groups?
128 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 129 findings, as identified in the literature 10.7.3 10.7.4 •• Lack of statistical power not considered Table 18: Checklist for evaluating published research review. Unless the results are logically Evaluation of meta-analyses Common omissions and errors in •• No allowance made for multiple testing related to the literature, the theoretical Problem that might be identified in a relevant published research •• Misunderstanding and misinterpretation significance of the investigation Meta-analysis is the process of research article Possible implications of results from models remains unclear. undertaking a quantitative review of the Rushton (2000) provides a report on 1. Inadequate literature review Misrepresentation of the conceptual basis for the research literature, seeking consistent patterns some of the most serious omissions and Research might lack direction; interpretation of evidence among, and sources of discrepancies Table 18 summarises some potential 2. Vague aims or hypotheses The processes involved in comparing the errors in papers presented in recent might be ambiguous problems that might emerge in the findings of a set or related papers are between, studies. An assessment should years to the journal, Occupational and context evaluation of an investigation. 3. Inappropriate research strategy Findings might not be relevant to the problem being introduced in the next sub-section. consider the homogeneity of the studies Environmental Medicine. These are: examined and whether summary effects A point that must be kept in mind is that investigated •• Was a valid interpretation drawn estimates will be calculated and by what Design even where an investigation is flawed, 4. Inappropriate sampling method Measurements might not be related to concepts being from the comparisons made between some useful knowledge might be drawn investigated methods (WHO 2000a). •• Authors unclear about type of the study and control groups during from it. The aim of critical analysis is not epidemiological study 5. Inadequate sampling method Sample might be biased; investigation could lack external analysis? WHO (2000a) has recommended that the to discredit or tear down published work, validity •• Adequacy of sample size not •• Did the investigators properly reject or following features be considered when but to ensure the reader understands its considered 6. Inadequate sample size Sample might be biased; statistical analysis might lack fail to reject the null hypothesis? conducting, or assessing the findings of, implications and limitations. power a meta-analysis: •• Bias in selection of subjects execution •• Did the investigators consider the 7. Inadequate description of sample Application of findings to specific groups or individuals possibility of Type 1 and Type 2 •• establishing or noting a protocol •• Data collection problems and missing 10.8 might be difficult data not adequately reported errors in interpreting the meaning of specifying the objectives of the review 8. Instruments lack validity or Findings might represent measurement errors the significance test? (Type 1 errors and the methods to be employed •• Non-respondents not investigated UNDERTAKING HEALTH reliability are the result of chance and are the •• having inclusion criteria that are more •• Sample selection and exclusions STUDIES 9. Inadequate design Investigation might lack internal validity (i.e. outcomes rejection of the null hypothesis when inclusive than exclusive, so enabling inadequately justified might be due to uncontrolled extraneous variables) no true difference exists in the larger sensitivity analysis using different levels The material in the following sections is population. Type 2 errors result from Analysis 10. Lack of adequate control groups Investigation might lack internal validity; size of the effect of inclusion to be undertaken adapted from ATSDR (1996). difficult to estimate chance or too small a sample size •• Parametric tests carried out on and are the failure to reject the null •• avoiding a single quality score of 11. Biased subject assignment Investigation might lack internal validity obviously skewed data In some situations, there will be a need hypothesis when a true difference studies and presenting, instead, to undertake health studies as part of a 12. Variations or lack of control of Investigation might lack internal validity exists in the larger population.) an assessment of a range of •• Use of multiple paired tests treatment parameters characteristics risk assessment. A risk assessment may •• Inappropriate analysis of repeated •• Were the size of the differences and have been prompted by health studies 13. Observer bias not controlled Investigation might lack internal and external validity •• weighting according to the precision of measures or longitudinal data the degree of overlap taken into undertaken by the community. The design (Rosenthal effects) the study consideration in the conclusions •• Incorrect analysis of matched case- of health studies should be underpinned 14. Subject expectations not controlled Investigation might lack internal and external validity reached about the meaning of •• assessing and addressing the impact control studies by epidemiological principles. A range (Hawthorne effects) of publication bias observed differences? •• Modelling incorrect (e.g. inadequate of factors need to be considered before 15. Research carried out in Investigation might lack ecological validity •• In interpreting the meaning of any •• systematically quantifying the adjustment for confounders, embarking on a health study. inappropriate setting relationship, was the concept of heterogeneity of the studies that can interaction terms not included, only 16. Confounding of times at which Possible series effects; investigation might lack internal cause and effect (causation) properly enable the identification of sources significant variables from preliminary 10.8.1 observations and interventions are validity applied? of variability in the results of studies analyses included) Public health significance carried out from factors such as the choice of •• Were the extrapolations to individuals 17. Inadequate presentation of The nature of the empirical findings might not be Presentation Public health significance is a key factor not included in the study properly methodology and the inclusion of descriptive statistics comprehensible susceptible sub-groups or unusual in considering the merits of a proposed performed? •• Inadequate description of the exposure conditions health study. Issues for consideration 18. Inappropriate statistics used to Distortion of data; false inferences may be drawn •• Did the investigators stay within the methodology and statistical procedures include: the toxicity of the agent; the describe and/or analyse the data •• using sensitivity analyses of factors limits of the data when extrapolating •• Inappropriate summary statistics for pathways of human exposure; severity 19. Erroneous calculation of statistics False inferences may be drawn such as different analytic approaches, the results? non-normal data and biological plausibility of the health different methods of extracting results 20. Drawing incorrect inferences from False conclusions might be made concerning the outcome; need for new information •• If the investigators extrapolated from from the studies or the inclusion or •• No presentation of risk estimates (e.g. the data outcome of an investigation population data to individual data, was (beyond what is already known or exclusion of particular studies or types odds ratios – and confidence intervals) 21. Protocol deviations Investigation might lack internal and external validity this appropriate and correct? what has already been done); size and of studies 22. Over-generalisation of finding External validity might be threatened •• Did the researchers take into Interpretation susceptibility of the population affected; •• appraising methods used to obtain ability to prevent or mitigate exposure 23. Confusing statistical and clinical Treatments lacking clinical usefulness might be consideration differences between the •• Potential bias due to sample selection, qualitative and quantitative summary or health outcomes; and relevance to significance encouraged study population and the population to no or poor response, missing values, estimates from a collection of studies. other situations with similar agents and which they extrapolated their data? exclusions 24. Findings not logically related to Theoretical significance of the investigation remains exposure pathways. previous research findings doubtful
130 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 131 Adapted from: Polgar & Thomas 1991. studied; to allow new laboratory tests or conduct and completion of any hypotheses about exposure–outcome health and demographic information be collected as part of the survey. study methods to be used or evaluated; proposed health study. Issues for associations and address specific to assess the occurrence of specific 10.8.2 to generalise to other situations or consideration include: the availability exposures, community health concerns, health effects in defined geographic When a Type 1 health study is considered Community perspective and populations; and provide confirmation of qualified personnel and technical or specific information needs. Examples areas and determine if the rates appropriate, there are several attributes involvement or additional support to a preliminary support; an ability to obtain necessary of Type 1 health studies follow. are elevated compared to similar that are considered necessary in order to hypothesis or theory. The principal tenet data and health information; and the populations elsewhere. Available ensure the quality of the study effort: •• Cross-sectional studies. These are Community involvement is critical to the for public health is promoting health availability of appropriate project time information might include: death surveys of a sample of individuals •• a reasonable ability to document and success of any proposed health study. of the population, for all sectors, and lines and resources. certificates, birth certificates, and to obtain information about current characterise exposure in the target Various community involvement methods with special attention to protecting the census data; tumour or disease and past health or environmental area can be used for health studies. Issues for vulnerable. The precautionary principle registry data; and health surveillance 10.8.6 exposures, or both. These studies can consideration include: an ability to involve is now well known among the public or disease notification data. A health •• an adequate study size for the type of Authority and support include comparison populations with key community stakeholders; an health sector and community advocates. statistics review may be performed study recommended demographics similar to those of the understanding of community health A tension arises when advances that It is critically important that local, state, in response to a reported cluster of exposed (target) population. •• an ability to identify and locate subjects concerns; an understanding of the promise benefits to society also bring risks territory and Commonwealth health specific diseases or conditions. and records approach and limitations of proposed agencies be involved early in discussions •• Pilot investigations collect additional (real or perceived), and this can become •• Exposure investigations use •• appropriate comparisons for rates of activities; and community support for the about potential health studies. Issues information to assess the feasibility and especially acute when the full weight environmental or biological testing, occurrence or levels of exposure study being conducted. It must be for consideration include: the ability to value of conducting a full-scale health of evidence cannot be known for many or both, for the hazardous agent(s) recognised that interested community support or provide technical assistance study. The investigation might include •• an ability to control confounding years. The once unquestioned faith in of interest. The biological test might members tend to be a more homogeneous requested by the local, state or territory assessments of data completeness and factors and biases (when possible). science has transformed to an educated measure the level of the hazardous group than a sample drawn from the health agency; the ability of local and quality, the level of documentation of and cautious scepticism among large agent, a metabolite of a hazardous general population. Significant effort must state or territory health agencies to exposures or health outcomes methods 10.8.7.2 sectors of the community. substance, or another marker of be made to identify strategies to alert address the community problem and to identify and track individuals, study Type 2 health studies exposure in human body fluids individuals and interested groups of the health concerns; and the involvement of size, and statistical power issues and 10.8.4 or tissues. The purpose of this Type 2 health studies are specifically discussion paper, and to invite their appropriate agencies with legislative and the availability of a control population Ability to provide definitive results investigation is to assess individual designed to test scientific hypotheses considered contribution. Any input to regulatory backing. or comparison. exposure levels to a specific agent about the associations between process is most likely to be performed in Since health studies may and frequently •• Cluster investigations evaluate associated with the situation. The adverse health outcomes and exposure personal time, that is, out of work hours. It will end up with inconclusive findings, it 10.8.7 when the reported occurrence of a levels identified should be compared to hazardous substances in the may also require discussions with group is important to consider how definitive the specific disease or condition is above Nature of the health study with that of a relevant reference group environment. The following are some colleagues during internal processes that study might be in providing scientifically the expected number for a given or with a known standard reference examples of Type 2 health studies: have scheduled meetings. Many useful results related to specific When the decision to conduct a health geographic location and time period. level. Depending on the hazardous dissemination processes for community exposure–outcome relationships. Issues study is being considered, several criteria These investigations can be conducted •• Case-control studies are designed agent, the investigation can be used groups (e.g. newsletters, also require long for consideration include the ability are used to determine the type of health to confirm case reports, determine to collect information and compare to explore for evidence of past or lead times to reach their constituent to: obtain appropriate measurements study. These relate to whether the relevant an unusual disease occurrence and differences in exposures and other current exposure. members). Community consultation time of exposure and to document health research hypothesis requires: explore potential risk factors. risk factors in two groups of people: •• Disease and symptom prevalence lines must therefore factor in these delays outcomes and exposures; use adequate •• the characterisation of environmental •• Comprehensive case reviews are those with specific illnesses or surveys are used to measure and to ensure effective community control or comparison populations; contaminants by type, media, and medical or epidemiological evaluations conditions (cases) and those without compare the occurrence of self- engagement. Advance warning of obtain community support to ensure concentration levels of the medical status of one or more the illnesses or conditions (controls). reported diseases, in some instances document release can help to facilitate an adequate participation rate; state individuals through medical record The controls are selected to represent •• documented evidence of human using medical records or physical processes. However, a key factor to clearly the study objectives and specific reviews, interviews or biomedical the population from which the cases exposure at a level of concern examinations to validate adverse community engagement remains allowing hypothesis to be tested; have sufficient testing to determine additional were identified. Usually the cases and •• level of current knowledge about the health conditions. Addressing sufficient time for the dissemination, full statistical power to detect predicted information about their health status or controls are identified first, and then relationship between exposure and potential health concerns raised by review, drafting and reviewing of responses. effects if they exist, obtain data on potential for exposure. information is collected about past important potential confounders, and specific adverse health outcomes the community, the survey compares exposures and other risk factors. and/or •• Situation-specific surveillance is an exposed population (target area) 10.8.3 evaluate a dose–response relationship •• Cohort studies are designed to collect designed to assess the specific with an unexposed population (control Scientific importance or gradients of exposure. Extrapolation •• documented excess of an adverse information from a group of people occurrence of one or more defined area) with similar demographic of findings to vulnerable groups, such health outcome, when known. followed over a period of time, and Scientific importance is closely related health conditions among a specific characteristics. The purpose is to as children remains problematic, as information on the occurrence of to public health significance, but they population potentially exposed to determine the need for further health involvement in these groups in studies is The health studies can be grouped into specific illnesses or conditions is do not form a perfect match. Issues hazardous agents in the environment. studies in the target area, provided fraught with complexities. Type 1 and Type 2 studies. collected. Cohort studies can be for consideration include: the ability to Data collection might include using there are statistically significant prospective, meaning that individuals provide new knowledge or information existing records of health events or excesses that are clinically important. 10.8.5 10.8.7.1 involved in the study are followed about an exposure–outcome relationship; records from relevant health care Depending on the contaminants and Resources Type 1 health studies into the future, or cohorts can be to address specific exposures or providers. circumstances, biological testing of retrospective, meaning that the cohort outcomes that have not been adequately Resources are critical to the support, Type 1 health studies explore or generate •• Health statistics reviews use available exposure or effect, or both, might also
132 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 133 is reconstructed from historical records 10.8.8 All standard quality control and •• Methods for measurement of and then followed over a specified Ensuring the quality of a health study quality assurance procedures must be exposure time period. They are expensive, and followed and documented. •• Collection of biological specimens require long periods of time, and large To ensure a useful and appropriate •• Additional data collection or numbers of people must be followed outcome the following factors should 10.8.9 be met: sources for rarer outcomes to provide enough Contents of a health study protocol cases for analysis. •• The group conducting the health study •• Chain of custody and shipping The following components should be •• Nested case-control studies are must be capable and fully responsible •• Laboratory methods and quality considered in drafting a report. Ethics another approach that uses both of the for conducting the health study. control approval should be outlined. Protocols for study designs previously mentioned. •• Privacy protection •• Personnel conducting the health health studies might not need to contain The nested case-control study uses study must be identified and have all of the items within this outline. •• Findings of immediate significance cohort individuals who have developed appropriate training and experience. •• Follow-up of abnormal lab results a specific illness or condition (case) •• The facilities and resources must be The listing is more comprehensive and persons sampled from the cohort •• Data analysis appropriate for successfully completing in order to cover the wide variety of who have not developed the illness •• Data entry, editing and the health study. study approaches. or condition (control). The case- management •• Contractors for a health study must control method is then used to collect 1. Title and identification page •• Data transformation additional information and analyse the follow written and approved work 2. Introduction and overview differences between these two groups. plans and their work must be carefully 7. Data analysis plan and methods reviewed by the sponsoring group. 3. Background 8. Study time lines •• For complex studies, a detailed study There are several attributes of •• Situation description Type 2 health studies that are considered protocol should be written and undergo •• Key activities or milestones •• Demographics necessary in order to ensure valid scientific peer review. 9. Community involvement and scientific findings, including: •• Ethical issues relating to the protection •• Contaminants and pathways notification of human subjects, consent and data •• an ability to reasonably estimate or •• Community health concerns 10. Interpretation of results document individual exposures confidentiality procedures must be addressed. •• Literature review 11. Limitations of the study •• an ability to document or validate human health outcomes •• Reports of complex health studies 4. Purpose 12. References may need to undergo scientific peer 5. Study objectives •• an adequate study size and statistical review prior to any public release of 13. Tables and figures power information. 6. Methods 14. Attachments •• an ability to identify and locate subjects •• Community involvement and •• Rationale for study design •• Data collection forms and and records knowledge of the health study are •• Study description questionnaire •• availability of an appropriate control or necessary: the involvement process will comparison population assist in ensuring that the community •• Eligibility criteria •• Study letters of notifications and •• an ability to control confounding understands and supports the study •• Selection of target area and consent form factors and minimise biases focus and design, and its limitations. population 15. Specimen collection and shipping •• an ability to determine influence of •• Depending on the community •• Selection of comparison area and protocols environmental, behavioural or other involvement approach, public meetings population factors. might be held to present and discuss •• Sample size and statistical power the study methods and findings. estimates However, final study methods must be •• Participant selection and scientifically valid before proceeding. definitions •• All study reports, data files and related •• Subject recruitment procedures documentation should be kept in the official records. •• Locations of data and specimen collection •• Any environmental sampling or biological testing must follow existing •• Informed consent procedure standards for collection, handling, •• Questionnaire procedures chain of custody, storage, analysis, and •• Interviewer training and methods reporting by an approved laboratory.
134 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 135 Chapter 11: Assessment of carcinogens
Cancer is an especially dreaded group of However, it is rare that a single causative Such tests require exposure of test relationship from a point of departure The US National Toxicology Program something of a misnomer, since its diseases. It may affect different tissues factor can be identified for cancer in animals for the majority of their life span (POD) on the dose–response curve (NTP), a cooperative program of the assessments essentially categorise and cell types and it is multifactorial in an individual. The best that can usually (18–30 months in rodents). The doses towards zero and estimate a finite risk National Cancer Institute and the carcinogenic hazards, and do not attempt its development. An essential feature be done is to identify risk factors used are as high as possible (up to the level at specified doses or levels of National Institutes of Environmental to quantify risk. This is recognised in of cancer is that it represents a process that may contribute to an increased conceptual maximum tolerated dose – exposure. Non-threshold methodology Health, has had an extensive chemicals the preamble to the IARC monographs, where a single cell or groups of cells overall cancer rate by contributing MTD) in order to maximise the sensitivity is applied most rigorously to chemicals testing program for over 30 years with statements that IARC considers lose the mechanisms that control to the initiation or promotion of the of the test to detect a carcinogenic that have the potential to damage DNA aimed at providing standardised data that a cancer ‘hazard’ is an agent that is differentiation and growth. Cancer cells carcinogenic process. This may be via response. As with epidemiological studies, (genotoxic chemicals), although it may be on rodent carcinogenesis (Bucher & capable of causing cancer under some may not only exhibit uncontrolled growth, a genetic transformation of a single it is important that the incidence of applied also to putatively non-genotoxic Portier 2004). The program is extensively circumstances, and a cancer ‘risk’ is a but when a malignant tumour occurs, progenitor cell, with subsequent clonal cancer in the rodent studies consider chemicals, where some doubt remains used as a database for both qualitative probability estimate of a carcinogenic it can invade adjacent or distant tissues expansion and transformation through age- and species-specific effects, as well about the carcinogenic mode of action and quantitative carcinogenic risk effect occurring from a defined amount, (metastasis). the various stages of its progression to as inter-current mortality. In some species (MoA). For non-genotoxic chemicals, assessment. The NTP 2-year rodent frequency and duration of exposure to a an established cancer. and strains, the incidence of tumours in where evidence provides more confidence bioassay program is extensively used as carcinogenic agent. Existing methodologies have difficulties untreated (control) animals can increase about the putative MoA, there may be a basis for carcinogenic risk assessment. in assessing and conveying the broad Some commentators have put forward a towards 100 per cent at some sites as adequate scientific grounds to adopt a However, the NTP also recognises that The International Program on range of human health implications of view that the majority of human cancers the animals age, making discrimination of threshold approach to EHRA. the future lies in better understanding of Chemical Safety (IPCS) has been exposure to environmental carcinogens. have a purely environmental factor a treatment-related effect more difficult. mechanism-based biological observations, developing a special project on This, combined with a high ‘dread factor’ associated with their causation, and There is further discussion of the broader A distinction is sometimes made which can be integrated with better carcinogenic risk assessment as part of for cancer, has resulted in many cases some appear to be still convinced that range of issues (such as dose and species between the terms ‘mode of action’ predictive testing regimens. its broader program on harmonisation in a disproportionate regulatory, political environmental factors are at the heart of selection) that need to be taken into and ‘mechanism of action’. The latter of risk assessment methodologies (see and public focus on cancer as compared the human cancer epidemic (Belpomme consideration in evaluating rodent-based implies a more detailed understanding
136 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 137 Ward (2007) presented a damning is becoming increasingly important as cause cancer through a species- animals. In some cases, an agent inadequate in humans but sufficient in 11.3 criticism of the reliance on animal understanding of mechanistic pathways specific mechanism which does not (mixture) may be classified in this experimental animals may be placed CLASSIFICATION OF studies to identify carcinogens. While he improves great advances are being made, operate in humans should also be category when there is inadequate in this category when there is strong acknowledges that regulatory authorities especially with the advent of sophisticated taken into consideration. evidence of carcinogenicity in humans, evidence that the mechanism of CARCINOGENS are sometimes prepared to accept that laboratory molecular techniques. sufficient evidence of carcinogenicity carcinogenicity in experimental animals carcinogenic responses associated The preamble goes on to define the in experimental animals and strong does not operate in humans. Agents, Advances in biological knowledge with high-dose studies in animals Essentially four descriptive dimensions of terms ‘sufficient evidence’, ‘limited evidence that the carcinogenesis is mixtures and exposure circumstances are enabling mechanistic data, may not be relevant for human risk mechanistic data are used in the IARC evidence’ and ‘inadequate evidence’. mediated by a mechanism that also that do not fall into any other group are pharmacokinetic data and other relevant assessment, especially when there is process (Fitzgerald 1993): Understanding how these terms are used operates in humans. Exceptionally, also placed in this group. data to be increasingly taken into account evidence for a species-specific MoA, the in the final evaluation is critical, as is the an agent, mixture or exposure •• evidence of genotoxicity (i.e. structural in classifying and assessing the risks of science underlying these hypothetical understanding that the evaluations: circumstance may be classified in this Group 4 – The agent (mixture) is change at the level of the gene) carcinogens. mechanisms of carcinogenesis in rodents category solely on the basis of limited probably not carcinogenic to humans. •• consider only published information and humans remains unproven. Even •• evidence of effects on the expression evidence of carcinogenicity in humans. Australia does not have a formal the concept that genotoxicity is the key of relevant genes (i.e. functional •• consider the route of exposure in This category is used for agents or mechanism for classifying carcinogens, initiating step in carcinogenesis has changes at the intracellular level) animal studies, but not necessarily Group 2B – The agent (mixture) is mixtures for which there is evidence but most Commonwealth, state and been challenged, with observations that •• evidence of relevant effects on cell its relevance to the conditions under possibly carcinogenic to humans. suggesting lack of carcinogenicity territory jurisdictions give weight to epigenetic events and mechanisms are behaviour which human exposure may occur The exposure circumstance in humans and in experimental IARC and US EPA classifications when also important (Trosko & Upham 2005). •• evidence of time and dose •• do not evaluate carcinogenic potency. entails exposures that are possibly animals. In some instances, agents or considering risk assessment based on relationships of carcinogenic effects carcinogenic to humans. mixtures for which there is inadequate a carcinogenic response. The difficulty Whether or not the introduction and interactions between agents. The use of these terms in reaching a evidence of carcinogenicity in associated with reaching harmonisation of mice genetically engineered to final classification is embodied in the This category is used for agents, humans but evidence suggesting lack of various national and international increase susceptibility to some classes In each monograph, the ‘preamble’ sets rules, as outlined in the IARC monograph mixtures and exposure circumstances of carcinogenicity in experimental classifications schemes for carcinogens of carcinogens will enhance the out the rules and approaches used by preambles: for which there is limited evidence animals, consistently and strongly was explored by Di Marco et al. (1998). predictive science of animal studies IARC expert panels to classify substances of carcinogenicity in humans and supported by a broad range of other remains to be seen (MacDonald et or exposures as ‘carcinogenic to humans’. Group 1 – The agent (mixture) is less than sufficient evidence of relevant data, may be classified in 11.3.1 al. 2004). Similarly, the combination carcinogenic to humans. The exposure carcinogenicity in experimental this group. IARC approach of toxicogenomic approaches to The term ‘carcinogen’ is used in circumstance entails exposures that animals. It may also be used when understanding carcinogenic mechanisms are carcinogenic to humans. there is inadequate evidence of Between 1972 and 2008, the IARC The IARC developed the first system the IARC monographs to denote ‘an of action may also hold some promise carcinogenicity in humans but there is monograph series (Volumes 1–100) had for qualitatively categorising chemical exposure that is capable of increasing for future advances in the science sufficient evidenceof carcinogenicity evaluated 935 environmental agents carcinogens (IARC 1978). Initially, the the incidence of malignant neoplasms. This category is used when there is (Guyton et al. 2009). This may be further in experimental animals. In some and exposures. The breakdown of approach was to adopt a strength-of- The induction of benign neoplasms sufficient evidenceof carcinogenicity enhanced by combinations of classical instances, an agent, mixture or classifications was: evidence scheme to decide whether, may in some circumstances contribute in humans. Exceptionally, an in vitro genotoxicity assays with in silico exposure circumstance for which for humans and experimental animals to the judgement that the exposure is agent (mixture) may be placed in screening techniques for genotoxicity there is inadequate evidence of Group 1 separately, there was sufficient or carcinogenic. The terms ‘neoplasm’ and this category when evidence of and computational techniques for SAR carcinogenicity in humans but (carcinogenic to humans) 108 limited evidence of carcinogenicity ‘tumour’ are used interchangeably. carcinogenicity in humans is less analysis (Benfenati et al. 2009). than sufficient but there issufficient limited evidence of carcinogenicity in for a substance, mixture or exposure Group 2A experimental animals together with circumstance, or whether the database The IARC evaluations consider whether evidence of carcinogenicity in A later decision by IARC was to supporting evidence from other relevant (probably carcinogenic to humans) 63 was inadequate for classification (prior the agent may intervene at one of more experimental animals and strong incorporate information on the data may be placed in this group. IARC monographs essentially only of the multiple stages of carcinogenesis evidence in exposed humans that Group 2B mechanism of action of chemicals in the the agent (mixture) acts through a summarised existing tumorigenicity or mechanisms. However, experimental (possibly carcinogenic to humans) 248 evaluation process (Vainio & Wilbourn Group 3 – The agent (mixture or studies). Since then, the scheme has evidence still maintains a high status in relevant mechanism of carcinogenicity 1992). In practical terms, this means that exposure circumstance) is not evolved whereby now all data, including the assessment hierarchy. The statement (e.g. dioxins). Group 3 category Group 1 (sufficient evidence classifiable as to its carcinogenicity human, animal and in vitro studies in current IARC monograph preambles (not classifiable as to carcinogenicity) 515 for carcinogenicity in humans) ‘could be to humans. are assessed for an overall weight-of- notes that: Group 2A – The agent (mixture) is extended to include agents for which the Group 4 evidence (WoE) evaluation of human probably carcinogenic to humans. evidence of carcinogenicity in humans is ... in the absence of adequate data on This category is used most commonly carcinogenicity (Vainio & Wilbourn 1992). The exposure circumstance (probably not carcinogenic to humans) 1 less than sufficient but for which there is humans, it is biologically plausible and entails exposures that are probably for agents, mixtures and exposure sufficient evidence of carcinogenicity in prudent to regard agents and mixtures circumstances for which the evidence Thus considerable weight is given to the carcinogenic to humans. The fact that only one substance has experimental animals and strong evidence for which there is sufficient evidence of carcinogenicity is inadequate in animal cancer bioassays, though some been classified in Group 4 (caprolactam in in exposed humans that the agent of carcinogenicity in experimental humans and inadequate or limited in scientists are not convinced of the validity This category is used when there is 1998) is probably because entry into the acts through a relevant mechanism of animals as if they presented a experimental animals. Exceptionally, of this philosophy. limited evidence of carcinogenicity IARC program for evaluation is a selective carcinogenesis’ (Vainio & Wilbourn 1992). carcinogenic risk to humans. The in humans and sufficient evidence agents (mixtures) for which the process. There must be at least some This aspect of the evaluation process possibility that a given agent may of carcinogenicity in experimental evidence of carcinogenicity is plausible published evidence suggestive
138 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 139 of carcinogenicity for a substance to be •• inadequate information to assess •• Group III: Possibly carcinogenic to The extent to which non-genotoxic MoAs of a weakly genotoxic chemical whose selected for evaluation. There are a number carcinogenic potential humans – as per IARC Group 2B, but 11.4 contribute to the universe of chemicals carcinogenic response in the kidney of substances used as human therapeutic •• not likely to be carcinogenic to humans. sub-categorised into four sub-groups WHEN IS A that have been classified positive for and liver appears to be determined by agents that have been classified as where the relationships between carcinogenicity by IARC (Categories 1, the extent to which cytotoxicity, followed Group 1 (not just a range of cytotoxic The narrative would set out, within each inconclusive epidemiological data and CARCINOGENIC RESPONSE 2A and 2B) was reviewed by Hernandezet by tissue repair, has occurred in those anti-cancer drugs). This exemplifies of these descriptors, the conditions positive/equivocal animal data are RELEVANT TO HUMANS? al. (2009). They found that 12 per cent target tissues. the importance attached to human under which a carcinogenic response is expanded. (45/371) of these proven and likely epidemiological data, and the relative human carcinogens had an MoA that likely to occur, and it suggests that more •• Group IV: Unlikely to be carcinogenic In characterising risk associated with A very good example of how lack of importance given to the conditions included a non-genotoxic component, than one conclusion may be possible to humans – essentially like IARC carcinogens that have been identified differentiation of genotoxic and non- under which humans may be exposed. with 27 per cent of these (12/45) for an individual agent, for example, Group 2B, but where there are no using rodent studies, the carcinogenic genotoxic MoAs can impact on the posing a potential human hazard as when carcinogenesis is dose- or route- useful epidemiological studies, and mechanisms (preferably termed quantitative risk assessment process is assessed through a margin of exposure 11.3.2 dependent. This could occur if a chemical the four sub-groups are based on the MoA) should be taken into account. found in a published risk assessment of (MoE) approach based on exposure US EPA classification of carcinogens produces tumours at the point of contact relative strengths of association using For carcinogens that are unequivocally 1,4-dichlorobenzene (Butterworth et al. for one route of exposure, but there is animal studies, including whether assessment potential. 2007). See Box 3. Current US guidance on carcinogen genotoxic, conservative risk assessment adequate evidence that no tumours occur there is evidence or not of genotoxicity. methodology requires extrapolation assessment (US EPA 2005a) also In some cases, it is difficult to differentiate Butterworth (2006) provides a framework at other sites when a different route of •• Group V: Probably not carcinogenic to of the dose–response curve from a emphasises the importance of integrating the significance of genotoxic and for assessing when a genotoxic MoA may exposure occurs. humans – like IARC Group 4, where suitable point of departure (POD) to and weighing all the available evidence non-genotoxic MoAs for a specific be important in the carcinogenic process, adequate epidemiological studies are zero to derive a cancer slope factor (see in reaching a decision about classifying chemical. For example, formaldehyde and may be differentiated from cytotoxic The narrative could also point out if lacking, but where the data in animal Section 3.10.3). Genotoxicity is MoA that a chemical as carcinogenic to humans. causes nasopharyngeal cancer in rats and regenerative cell proliferative there is evidence that a key event in studies is more limited, and evidence is considered relevant for carcinogenesis This should include a consideration of the after inhalation exposures, and while responses that occur in high-dose rodent the carcinogenic process may not indicating a lack of genotoxicity is across species, including humans, so it conditions under which a carcinogenic formaldehyde has some genotoxic bioassays and which may confound the response might occur in humans, as well occur below a certain dose range. stronger. is rare for a genotoxic carcinogen to be This provides scope for evaluating potential, the initiation of nasal cancers in MoA assessment. Butterworth proposed as its likelihood. •• Group VI: Unclassifiable with respect to subjected to HRA by any other method. carcinogenic responses in animal rats appears only to follow strong irritancy (pp. 19–20) the following classification carcinogenicity in humans – essentially studies where high doses have resulted of the nasal mucosa, followed by tissue criteria for a carcinogen acting via a The US EPA policy position (US EPA like IARC Group 3, but including a Carcinogenesis is a multifactorial and in a precursor cellular proliferative repair involving a cellular proliferative genotoxic MoA: 2005a pp. 1–10) is that: sub-category where the evidence from multi-stage process, involving initiation, response. If the studies are further response. Chloroform is another example epidemiological and animal studies promotion and progression. The presence ... in the absence of sufficiently informed by toxicokinetic or mode of is conflicting. of multiple mutations in critical genes scientifically justifiable mode of action action data that indicates a carcinogenic in cancer cells is a strong indicator that information, the EPA generally take response in animals may not be relevant In the UK, carcinogens are defined cancer is associated with accumulative public health-protective, default to humans, these findings should be built and categorised in both the Control irreversible DNA damage. While DNA BOX 3: positions regarding the interpretation into the narrative. damage or genotoxicity is probably of toxicologic and epidemiologic data: of Substances Hazardous to Health Example of the application of MoA analysis (COSHH) regulations (see www.hse.gov. involved at one or more stages, there animal tumour findings are judged It is clear that a degree of consistency has 1,4-dichlorobenzene induced liver cancer in male and female mice after uk/coshh) and the Chemicals (Hazard may also be stages of the carcinogenesis to be relevant to humans and cancer emerged between the US EPA, IARC and either gavage or inhalation exposure. No liver cancers were observed in Information and Packaging for Supply) process where epigenetic events or risks are assumed to conform with low IPCS approaches to assessing the weight rats. The inhalation route is the primary source for likely human exposure. regulations (CHIP), although the term non-genotoxic mechanisms may act to dose linearity. of evidence that suggests a chemical The MoA for the mouse liver tumours appears to be via a non-genotoxic ‘carcinogen’ has a wider meaning in CHIP increase cancer incidence. For example, could be carcinogenic to humans. mitogenic stimulation, with a putative threshold established from the dose– than in COSHH. The classification system non-genotoxic chemicals may increase However, in a departure from previous response relationship. Kidney tumours seen in male rats were probably is relatively simple, being based on three cancer incidence by stimulating cellular guidance on carcinogen classification, 11.3.3 proliferation following a cytotoxic response acting via deposition of alpha-2μ-microglobulin and were not included in it recommends that a descriptive categories only. Other international carcinogen (tissue injury), or by promoting the the risk assessment. The gavage and inhalation data were used to establish WoE narrative replace the numerical a BMD (1 per cent extra risk) for both routes of exposure, as a POD for classification schemes •• Category 1: substances known to further carcinogenic transformation of 01 classification scheme. Such a narrative cause cancer on the basis of human cells which have already undergone further risk assessment. Using a conventional threshold approach, and should include a summary of the The Canadian system uses six evidence- experience genetic predisposition to a pre-cancerous dividing the POD by a 300x uncertainty factor, yields an air concentration information about the mode of action. based categories for carcinogens, based •• Category 2: substances which it is state. Non-genotoxic mechanisms of 0.1 ppm below which there should be no appreciable cancer risk to essentially on the IARC classifications assumed can cause cancer, on the can also include hormonal effects, humans. In contrast, if an assumption of a genotoxic MoA had led to the use The narrative should be written around system. basis of reliable animal evidence immunosuppression, receptor activation of a non-threshold approach to the risk assessment, the estimate of the air five standardised descriptors: concentration associated with a 10-6 lifetime excess cancer risk would have •• Group I: Carcinogenic to humans – •• Category 3: substances where there (e.g. dioxins/Ah receptor peroxisome been 0.00004 ppm, about 2,500-fold lower (and 1,875,000 times lower than •• carcinogenic to humans essentially the same as IARC Group 1. is only evidence in animals that is of proliferators/PPAR receptor). the actual inhalation NOAEL of 75 ppm). •• likely to be carcinogenic to humans •• Group II: Probably carcinogenic to doubtful relevance to human health •• suggestive evidence of carcinogenic humans – essentially the same as (i.e. the evidence is not good enough potential IARC Group 2A. for Category 1 or 2).
140 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 141 Evidence that a chemical can induce compound. Experimental evidence the target site, such as cytolethality best addressed in a single analysis. regenerative hyperplasia of the differing experimental designs mutagenic activity: for the likelihood that genotoxic and regenerative cell proliferation It should also be noted that some urothelium. increases that support since different activity was the predominant driving in a dose-dependent manner that modes of action will involve multiple designs may reduce unknown biases •• Clear evidence for mutagenic force in tumour formation, such as indicates it was likely responsible for contributing components. or confounding. activity: Genotoxic activity is Dose–response relationship the identification of DNA adducts at tumour formation. observed consistently in several the target tissue site. This section should detail the observed Consistency, which addresses well-validated cell culture assays Postulated mode of action While these suggested guidelines from a dose–response relationships and repeatability of key events in the with different endpoints. Chemically •• Some evidence that cancer (theory of the case) discuss whether the dose–response was induced by a mutagenic research scientist may provide a useful postulated mode of action for cancer in induced genotoxic activity is This section comprises a brief for the key events parallels the dose– mode of action: Some evidence framework for decision making on the different studies, is distinguished from expressed in whole animal assays, description of the sequence of response relationship for tumours. for mutagenic activity of the relevance of a proposed MoA, they are coherence, which addresses relation particularly at the tumour target site. events on the path to cancer for the Ideally, one should be able to correlate compound. There may be a not proposals that have been endorsed by of the postulated mode of action with •• Some evidence for mutagenic postulated mode of action of the test increases in incidence of a key question as to whether genotoxic any regulatory agency. However, they are observations in the broader database. activity: Genotoxic activity is substance. This explanation of the event with increases in incidence or activity was occurring in the target reasonably consistent with guidance that Pertinent observations: e.g. tumour observed in several well-validated sequence of events leads into the next severity (e.g. lesion progression) of tissue. Other biological activity was has been formulated in the IPCS and ILSI response and key events in same cell cell culture assays with different section which identifies the events other key events occurring later in evident that could also have driven programs (see Section 11.5). type, sites of action logically relate to endpoints. No activity, however, is considered ‘key’ (i.e. measurable) the process, and with the ultimate event(s), initiation-promotion studies, tumour formation. tumour incidence. Comparative tabular expressed in whole animal assays. Determination and evaluation of the given the database available for stop/recovery studies. •• Equivocal evidence that cancer presentation of incidence of key •• Equivocal evidence for mutagenic potential MoA is a key feature of the IPCS the analysis. was induced by a mutagenic mode events and tumours is often helpful in activity: Negative response in and US programs aimed at harmonising Biological plausibility and of action: Equivocal evidence examining dose–response. tests that include those assays risk assessment for carcinogens. Key events for mutagenic activity of the coherence which over time have proven to compound. Clear evidence that This section briefly describes the ‘key be most reliable and relevant. Temporal association The postulated mode of action and other biological activity was likely events’ – i.e. measurable events that No responses are seen in whole 11.5 the events that are part of it need to driving tumour formation. Such are critical to the induction of tumours This section should detail the observed animal genotoxicity assays. be based on current understanding of activity would include endpoints IPCS PROGRAM as hypothesised in the postulated temporal relationships or sequence Positive responses are seen only in the biology of cancer to be accepted, such as necrosis, cytolethality, mode of action. To support an of events and discuss whether the unvalidated or unproven assays or though the extent to which biological regenerative cell proliferation, ON CHEMICAL association, a body of experiments key events precede the tumour in assays where the experimental plausibility as a criterion against which peroxisomal proliferation, and needs to define and measure an response. One should see the key procedures used were questionable. CARCINOGENESIS weight of evidence is assessed is associated mitogenic and event consistently. events before tumour appearance necessarily limited, due to considerable •• Clear evidence for lack of mutagenic promotional activity in the rodent The IPCS program on harmonising risk this is essential in deciding whether gaps in our knowledge in this regard. activity: Negative responses in tests liver and receptor-mediated assessment for carcinogens began in Pertinent observations: e.g. tumour the data support the postulated mode One should consider whether the that include those assays which mitogenic or other activity at doses 1997. The structure of the framework response and key events in same of action. Observations of key events mode of action is consistent with what over time have proven to be most associated with tumour formation. and the key steps proposed for a cell type, sites of action logically at the same time as the tumours is known about carcinogenesis in reliable and relevant. No responses (e.g. at the end of a bioassay) do not •• Evidence that cancer was induced carcinogen evaluation were described relate to event(s), increased cell general (biological plausibility) and in in whole animal genotoxicity assays. contribute to temporal association, but by both mutagenic and non- in Appendix 5 of the original 2002 growth, specific biochemical events, relation to what is also known for the The weight of evidence indicates can contribute to analysis in the next mutagenic activity: In many enHealth EHRA guidance document, organ weight, histology, proliferation substance specifically (coherence). that the compound is not genotoxic. section. Most often, complete datasets cases, tumours are formed by as they stood at April 2001. The following assays, hormone or other protein For the former, likeness of the case to address the criterion of temporality •• Insufficient data to make a a combination of mutagenic text is reproduced from that appendix. perturbations, receptor–ligand to others for structural analogues are not available. conclusion regarding genotoxic and non-mutagenic activity. In changes, DNA or chromosome effects, may be informative (i.e. structure activity: No or minimal testing such instances, it is important Introduction cell cycle effects. activity analysis). Additionally, this has been conducted with the to approximate the relative Strength, consistency and section should consider whether the This section describes the cancer compound. Compounds that contributions of the two different For example, key events for tumours specificity of association of tumour database on the agent is internally endpoint or endpoints that have been have structural alert for potential activities in order to select a more hypothesised to be associated with response with key events consistent in supporting the purported genotoxicity, but for which there observed and identifies which of these prolonged regenerative proliferation mode of action, including that for realistic risk assessment. This section should discuss the weight is no experimental data may be is addressed in the analysis. (The might be cytotoxicity in as measured relevant non-cancer toxicities. Some •• Clear evidence that cancer was of evidence linking the key events, placed in this category. nature of the framework is such that histopathologically and an increase modes of action can be anticipated induced by a non-genotoxic mode only one mode of action is analysed in labelling index. Key events for precursor lesions and the tumour to evoke effects other than cancer, of action: Weight of evidence response. Stop/recovery studies Evidence that the chemical produced at a time hence, for example, tumour induction of urinary bladder tumours e.g. reproductive effects of certain indicates an equivocal or clear showing absence or reduction of cancer via a mutagenic mode of action: types associated with a different mode hypothesised to be due to formation hormonal disturbances that are lack of evidence for mutagenic of action, even if recorded in the of bladder stones composed primarily subsequent events or tumour when carcinogenic. Moreover, some modes •• Clear evidence that cancer activity of the compound. Other same animals, will require separate of calcium phosphate might include a key event is blocked or diminished of action are consistent with observed was induced by a mutagenic biological activity known to have the framework analyses.) However, elevated urinary calcium, phosphate are particularly important tests of the lack of genotoxicity. Coherence, which mode of action: Clear evidence potential to drive tumour formation where different tumours are induced and pH, and formation of bladder association. Consistent observations addresses relation of the postulated for mutagenic activity of the is associated with the compound at by related mode of action, they are stones followed by irritation and in a number of such studies, with mode of action with observations in
142 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 143 the broader database – for example, There have been a series of workshops The NHMRC mBMD methodology A key driver for the development of association of mode of action for that have evaluated the application of 11.6 11.7 has many similarities with the BMD the now-rescinded NHMRC mBMD tumours with that for other endpoints the framework using a selected range of QUANTITATIVE RISK AN ALTERNATIVE methodology used by the US EPA to methodology was its lack of dependence – needs to be distinguished from substances with differing modes of action. derive a POD for its carcinogen risk on defining an ‘acceptable’ or ‘target risk’ consistency, which addresses For example: ASSESSMENT OF APPROACH TO assessments, but there are also some level, and driving the risk assessment repeatability of key events in the important differences. The NHMRC towards an excess risk estimate (e.g. •• conazoles – cytotoxicity, cellular CARCINOGENS QUANTITATIVE postulated mode of action for cancer advocates estimation of a mean BMD 10–6), which may be either excessively proliferation, metabolic induction 05 in different studies. from all the curve-fitting data, after conservative or widely misunderstood in The risk assessment of carcinogens CARCINOGEN RISK •• d-limonene – cytotoxicity, cellular discarding those models which have the community, or perhaps both. presents special challenges in quantitative proliferation, metabolic pathway ASSESSMENT? an evidently poor fit to the data. The Other modes of action EHRA. In part, this is because the •• di-ethylhexylphthalate – cytotoxicity, US methodology derives an upper While the methodology retains an element This section discusses alternative methodology has been driven by the cellular proliferation, peroxisome While there is strong precedent for both 95th percentile estimate of the BMD of ‘expert judgement’ in relation to the modes of action that logically present concept that there is no threshold for 10 proliferation Australian and international regulatory as a POD for linearised extrapolation curve-fitting and selection/justification themselves in the case. If alternative carcinogenic risk. The consequent systems to expect application of a non- to estimate doses associated with of the MFs, it is largely insulated from modes of action are supported, they •• vinclozolin – cytotoxicity, cellular use of mathematical constructs of the threshold approach to risk assessment of the target risk used in US regulations the science policy decision of what need their own framework analysis. proliferation, hormonal perturbation dose–response relationship to provide carcinogens, in 1999 the National Health (usually 10–6; see Section 5.6). The constitutes an acceptable level of risk These should be distinguished from •• captan – cytotoxicity, cellular a platform for extrapolation from a and Medical Research Council (NHMRC), use of a linearised extrapolation to zero (see also discussion in Section 5.6). This additional components of a single proliferation, genotoxicity. POD is the mainstay of quantitative acting on advice from a specialist working dose using this method does not differ methodology is also not captive to the mode of action which likely contribute HRA of carcinogens. party, endorsed a set of guidelines for conceptually from the linearised multi- extrapolation through the origin, which to the observed effect, since these Since 2000 there has been substantial risk assessment of carcinogens on stage procedure because the slope factor risks the artefact of correlation between would be addressed in the analysis of cooperation and alignment between the There is an extensive discussion in contaminated sites, which was based on a derived is anchored and therefore strongly the CSF and the MTD (Krewski et al. the principal mode of action. carcinogenic framework programs of the Sections 3.10, 3.11, 3.12, 5.4.2, 5.7 modified threshold approach, or modified influenced by passing through the origin 1993, Hrudey 1998). IPCS, the US EPA and a similar program and 5.10 of the basis for the assumption benchmark dose (mBMD) methodology (zero dose – zero response). of the International Life Sciences Institute of no threshold for a carcinogenic Assessment of postulated mode (NHMRC 1999a). While the NHMRC The NHMRC method also represents (ILSI) (Meek et al. 2003). A special issue response, on the application of non- of action document has since been rescinded, The modifying factors (MF) applied to an alignment with the methodology of Critical Reviews in Toxicology (Boobis threshold dose–response assessment it did outline an approach that could the BMD to derive a GD include a used for risk assessment of virtually all This section should include a clear et al. 2006) addresses the integration methodologies and the reliance on 05 harmonise the HRA of cancer and non- mixture of those used to adjust a NOAEL other toxicological endpoints. There statement of the outcome with an of cancer assessment frameworks establishment of a policy-driven ‘target cancer toxicological endpoints by moving to derive an ADI, TDI or RfD (inter-and have been a number of commentators indication of the level of confidence developed by these three programs. The risk’ for the evaluation of the EHRA away from the need to calculate or intra-species variability adequacy of the in the scientific literature who have in the postulated mode of action e.g. overall approach of all these programs outcomes. While there has been estimate a finite risk level for carcinogens database and data quality) as well as argued that it is no longer justifiable to high, moderate or low. has been essentially the same, with some movement towards accepting a alone, and the need to establish a ‘target some which are based on the nature of differentiate between threshold and non- emphasis on evaluating the MoA and threshold approach for carcinogenic risk’ with which this could be compared. the carcinogenic response (genotoxic threshold approaches to carcinogenic Uncertainties, inconsistencies and human relevance of the carcinogenic responses associated with some versus non-genotoxic rare, benign or risk assessment (see Gaylor et al. 1999; data gaps response. The special issue of Critical non-genotoxic modes of action, the The essential features of this alternative malignant tumours produced). The Purchase & Auton 1995). Reviews in Toxicology provided further dominant method of dose–response Uncertainties should include those approach, deriving a POD for range of possible MF values is up to illustration of the framework application assessment in Australian EHRA practice related to both the biology of tumour carcinogenic risk assessment were: 50,000. It is noted that, application of using three additional case studies: and regulatory requirements remains development and those for the a MF of 50,000 to the BMD would the non-threshold approach taken in US •• use of a range of mathematical models 05 database on the compound of interest. •• 4-aminobiphenyl – DNA reactivity to explore the data ‘fit’ to a derived be equivalent to the derivation of a regulatory agencies. –6 Inconsistencies should be flagged and •• formaldehyde/glutaraldehyde – nasal dose–response relationship 10 excess risk estimate using a linear data gaps identified. For the identified extrapolation approach, thus providing a cytotoxicity The detailed critique of quantitative •• estimation of a BMD at the 5 per cent data gaps, there should be some similar level of conservatism where the •• thiazopyr – thyroid disruption. cancer risk assessment provided in excess risk level (BMD05) to serve as a indication of whether they are critical data warrants use of such a large MF Hrudey (1998) is quite informative ‘point of departure’ (POD) for further as support for the postulated mode value. Refer to the NHMRC document and still relevant. The continued use risk assessment of action or simply serve to increase The various projects have distilled down for further guidance on how to assess the of a non-threshold approach for most confidence therein. to focus on three key questions (Cohen et •• application of a range of modifying carcinogenic hazard and how to select al. 2004): carcinogenic risk assessments results factors to the BMD05 dose estimate, and apply the various MFs. in quite conservative, and potentially 1. Is the weight of evidence sufficient to in order to derive a ‘guideline dose’ 11.5.1 costly outcomes in many types of establish the MoA in animals? (GD), which should be protective of Recent updates to the IPCS framework EHRA processes. 2. Are key events in the animal MoA human health. The outcomes of the IPCS project were plausible in humans? summarised in a review article in 2006 (Boobis et al. 2006). 3. Taking into account kinetic and dynamic factors, are key events in the animal MoA plausible in humans?
144 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 145 Chapter 12: Assessing multiple routes and sources of exposure
The passing of the US Food Quality However, the key issue is whether change in composition over time, due these families of chemicals is initiated by Protection Act of 1996 put the onus on 12.1 interactions are likely to occur at the 12.2 to differential translocation, degradation binding to a common intracellular receptor the US EPA to develop scientifically sound RISK ASSESSMENT OF low doses commonly associated with ASSESSMENT OF or ‘weathering’ effects. Such problems (the aryl hydrocarbon receptor or AhR) with methodologies to incorporate ‘aggregate’ environmental exposure scenarios. It are not unique to the approach of subsequent downstream gene activation and ‘cumulative’ methodologies into CHEMICAL MIXTURES has been proposed that there is an REPRESENTATIVE testing ‘representative’ mixtures, since leading to a variety of cellular events (e.g. environmental health risk assessment ‘interaction threshold’ representing a MIXTURES interpretation of environmental data cellular proliferation and differentiation, programs, particularly with regard to The conventional approach to health risk point below which interaction effects in based on single chemical toxicity data growth factor and hormone regulation). pesticide residues in food. The program assessment often relies on evaluating mixtures may not be relevant (Hamm et also needs to take into account time- and For this class of AhR agonists, the relative Simple binary mixtures, or in some cases led to the development of model software toxicity data where chemicals have been al. 2005). pathway-dependent variables in exposure. potencies are related to 2,3,7,8-tetrachloro- more complex mixture matrices, may termed CARES – Cumulative and administered as single entities using fixed dibenzodioxin (TCDD), putatively the most have been directly tested in conventional Aggregate Risk Evaluation System. dose rates. However, in the real world, For most of the environmental scenarios toxic congener and the one for which exposures may occur simultaneously or in vivo or in vitro toxicity testing systems. 12.3 most toxicological data exists. Tables that have been evaluated, the ‘interaction These tests allow an assessment of In this US context, the meaning of these consecutively with multiple chemicals threshold’ appears to be higher than the summarising the relative potencies of the and doses varying over time. Assessment whether a fixed chemical mixture may THE TOXICITY dioxin-like compounds were first published terms was: dose or exposure where toxicity is seen, cause toxicity that is either different in of such complex situations presents a thus suggesting that interactions are EQUIVALENCE FACTOR during the mid-1980s and have since been •• Aggregate exposure: the analysis of potency or not predicted by studies using real challenge to toxicologists and risk unlikely to be relevant (Yang & Dennison updated as new information has come to exposure to a chemical by multiple the individual components. assessment professionals. 2007; Crofton et al. 2005). A possible APPROACH hand (Van den Berg et al. 2007; US EPA pathways and routes of exposure 2000a; 2009c; NTP 2006). exception is where the toxic effects are Since the mixtures may have undergone a •• Cumulative exposure: the combined One of the more important challenges is The basic assumption in this approach is mediated via disruption of the endocrine concentration step in order to administer risk estimate where exposure occurs to determine whether the combined risk is that the toxicity contribution of individual Following an extensive review of the dioxin system, a system potentially sensitive doses that adequately reveal the toxicity simultaneously, or consecutively, to independent of the individual components chemicals in a mixture can be summed congener TEFs proposed by Van den to very low doses of environmental profile, this approach presents the multiple chemicals that exert toxicity whether the risk is additive (i.e. no by expressing the concentration or Berg et al. (2006), the US EPA (2009c) chemicals (Kortenkamp et al. 2007). same high-to-low dose extrapolation through a common mechanism. interactions among components) more dose of each component in terms of a has recently issued draft guidance that issues for health risk assessment as than additive (i.e. synergistic) or less than standard or reference compound. The proposes endorsement of these TEFs as While interactions between chemicals in conventional single chemical tests. The A key factor in US ‘cumulative’ HRA additive (antagonistic). All are theoretically amount of each component is adjusted by the best currently available estimates. an environmental mixture may or may not concentration step may also increase methodologies is the determination of possible at high doses, although there multiplication with a toxicity equivalence be relevant, it is generally assumed that the likelihood of observing confounding whether the multiple chemicals act by have been relatively few opportunities factor (TEF) based on relative toxicity Another class of chemicals for which TEFs additivity is the more normal outcome interactions that may not be relevant at a common mechanism of action. In to test whether additive models or those potency. The derivation of TEFs is based have been developed for EHRA purposes (Lambert & Lipscomb 2007). This may be lower doses. Tests based on in vitro and the earliest determinations, this was based on independent actions are better on assessment of a relative potency factor are the polycyclic aromatic hydrocarbons expressed as either: in silico exposures may have used more relatively easy, since the methodologies predictors of mixtures toxicology. One (RPF) for the individual components (PAHs). PAH TEFs developed as part of relevant lower concentrations, but the focused on cumulative risk assessment such review (Cedergreen et al. 2008) •• Dose addition: Where the combined (Chen et al. 2003). the contaminated sites workshop series in interpretative problem then becomes of organophosphonate (OP) pesticides, found that the predictive capability of effects are assumed to be by the same the 1990s have conventionally been used one of extrapolating the dose–response which act via the common mechanism either concept was relatively poor when mode of action (MoA) and are based The HRA approach sums the doses of in EHRA in Australia (Fitzgerald 1993; and interactive relationships to whole of inhibiting acetylcholinesterase (AChE) 158 datasets relating to tests with binary on relative potency at a fixed level of individual components of a mixture using 1998). These TEFs for selected PAHs are organisms, including humans. (US EPA 2001b). Subsequently, the work mixtures were evaluated. On the other response, or the formula: summarised in Table 20, which includes was extended to include other pesticides hand, Pohl et al. (2009) analysed 380 •• Response addition: Where the comparisons with TEFs developed by US The obvious limitation in such an Total toxicity equivalents (TEQ) = (N-methyl carbamates, triazines and binary combinations from the ATSDR combined effects are assumed to be EPA and the UK (from Fitzgerald 1998). approach is the practical impossibility ∑ (component 1 amount/concentration Table 20 also contains estimates of RPFs chloroacetanilides), which also share database and found that only 156 (41 independent of the mode of action of testing more than a few possible × TEF1) for selected PAHs from an external review a common mechanism of action (see per cent) indicated possible additive and defined by dose–response curves n permutations and combinations of the draft (US EPA 2010). While these values
146 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 147 Table 19: Toxicity equivalence factors (TEFs) for dioxin congeners Table 20: TEFs values for PAHs
Congener TEF Congener TEF Fitzgerald UK US EPA CCME PAH 1998 US EPA 1993 19951 20102 20103 Chlorinated dibenzo-p-dioxins Chlorinated dibenzofurans Benzo(a)pyrene 1 1 1 1 2,3,7,8-TCDD 1 2,3,7,8-TCDF 0.1 Dibenz(a,h)anthracene 4 1 1 10 (1–40) h 1
1,2,3,7,8-PeCDD 1 1,2,3,7,8-PeCDF 0.03# Benz(a)anthracene 0.1 0.1 0.1 0.2 (0.02–0.04) m 0.1 Benzo(b)fluoranthene 0.1 0.1 0.1 0.8 (0.1–2) h 0.1 1,2,3,4,7,8 –HxCDD 0.1 2,3,4,7,8-PeCDF 0.3# Benzo(k)fluoranthene 0.1 0.01 0.1 0.03 m 0.1 1,2,3,6,7,8-HxCDD 0.1 1,2,3,4,7,8-PHxCDF 0.1 I-pyrene 0.1 0.1 0.1 0 m nd 1,2,3,7,8,9-HxCDD 0.1 1,2,3,6,7,8-HxCDF 0.1 Anthracene 0.001 nd 0.001 0 m nd
1,2,3,4,7,8-HpCDD 0.01 1,2,3,7,8,9-HxCDF 0.1 B-perylene 0.1 nd 0.1 nd nd Chrysene 0.1 0.001 0.1 0.1 (0.04–0.2) h 0.01 OCDD 0.0003* 2,3,4,6,7,8-HxCDF 0.1 Acenaphthene 0.001 nd nd nd nd 1,2,3,4,6,7,8-HpCDF 0.01 Acenaphthylene 0.001 nd nd nd nd 1,2,3,4,7,8,9-HpCDF 0.01 Fluoranthene 0.01 nd 0.001 0.08 (0.009–0.2) l nd
OCDF 0.0003* Fluorene 0.001 nd nd nd nd Naphthalene 0.001 nd nd nd nd Non-ortho-substituted PCBs Mono-ortho-substituted PCBs Phenanthrene 0.001 nd 0.001 0 h nd 3,3’,4,4’-tetraCB (PCB 77) 0.0001 2,3,3’,4,4’-pentaCB (PCB 105) 0.00003# Pyrene 0.001 nd 0.001 0 m nd 3,4,4’5-tetraCB (PCB 81) 0.0003* 2,3,4,4’,5-pentaCB (PCB 114) 0.00003# Adapted from: Table 5, Fitzgerald 1998. 3,3’,4,4’5-pentaCB (PCB 126) 0.1 2,3’4,4’,5-pentaCB (PCB 118) 0.00003# nd = not determined
3,3’,4,4’,5,5’-hexaCB (PCB 169) 0.03* 2’,3,4,4’,5-pentaCB (PCB 123) 0.00003# 1 Cited in CRBE (1995). 2 Cited in an external review draft; not yet endorsed by US EPA; figures quoted are RPFs based on a review of rodent tumour bioassays, 2,3,3’,4,4’,5-hexaCB (PCB 156) 0.00003# using different exposure routes; figures in parentheses indicate the range of estimated RPFs; the letters h, m and l denote high, medium and 2,3,3’,4,4’,55-hexaCB (PCB 157) 0.00003# low confidence in the reported figures. 3 A set of values used by Canadian authorities in assessing contaminated sites. 2,3’,4,4’,5,5’-hexaCB (PCB167) 0.00003*
2,3,3’,4,4’,5,5’-heptaCB (PCB 189) 0.00003# The Canadian report (CCME 2010) •• possible presence of other carcinogens activation of key endocrine receptors * Adjusted 3x upwards (approx half log10) from the 1998 WHO values in the 2005 review. includes comprehensive tables of TEF or promoters in the mixture that are not (e.g. the oestrogen receptor complex) and # adjusted downwards from the 1998 WHO values in the 2005 WHO review. values for PAHs derived by a broader measured or addressed in the relative it is possible to use the TEF approach range of authorities, along with the source potency estimate to construct aggregate risk assessment Source: Van den Berg et al. 2006, as cited in the 2005 WHO review. references. However, the report notes that •• uncertainty about the relevance of models based on relative potencies for more than a dozen sets of equivalency rodent studies in predicting human this class of chemicals. numbers have been proposed over the cancer risk past two decades and cautions that The TEF approach has some key there can only be limited confidence in At this time, no one set of PAH TEFs has limitations. Since it is assumed that the derived potency estimates. Adoption been recommended for use in Australia, chemicals in a TEF class have the same of any one set of values requires an although it is likely that the Canadian MoA, the toxicity is assumed to be dose- understanding of the factors that can set is becoming more widely used, additive and that there are no significant create uncertainty: based on the fact that it the most recent interactions (e.g. receptor antagonism or •• prediction of mixture effects based on compilation of such values. modifications to metabolic clearance). It single-substance studies is also assumed that the relative potencies are based on similar test designs and •• prediction of potency for one route of Relative toxic potencies of some endocrine disrupting chemicals (EDCs) endpoints. TEFs based on relatively exposure based on data from another simple or short-term endpoints may not route of exposure have also been determined towards
148 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 149 necessarily reflect longer-term toxicity (HQ). To derive an overall estimate of would need to significantly erode this implies a common MoA and outcome, recycled water (EPHC, NHMRC, NRMMC The main limitation on the biomarker where differences in tissue distribution, risk across all components of the mixture, 100–10,000-fold margin between the ADI/ where the risk contributions of individual 2008). The TTC is a concentration or approach to mixture risk assessment is an clearance and elimination may alter or the individual HQs are summed to TDI and the level where toxic effects begin carcinogens in a mixture may in fact amount (dose) derived by analysing the incomplete understanding of the extent to confound the relative potency estimates. derive a hazard index (HI) (Herzberg & to occur. When the overall HI is less than be independent. To overcome this distribution of known toxic potencies for which the biomarker reflects the ultimate In the case of dioxin-like compounds, the Teuschler 2002). 1, it is generally assumed that cumulative inherent conservatism, the cumulative an extensive chemical toxicity database, expression of toxicity. TEFs are actually based on a composite risk is within reasonable bounds and that carcinogenic risk is sometimes adjusted and selecting an arbitrary low percentile of endpoints from short- and medium- n n there is no need to undertake a more to a higher level than that applied to (e.g. 5 per cent) of that distribution such term exposures, including some relative E j refined risk assessment. Even when HI individual components. For example, that it is unlikely that the toxicity potential 12.7 HI = ∑ = ∑ HQj toxicity endpoints based on in vitro tests is greater than 1, it does not imply that where the ‘target risk’ for an individual of any unknown chemical would be j = 1 j = 1 INTERNATIONAL systems, not all of which are strictly Rf Dj risks are unacceptable, although there is component may be set at 1 × 10–6, the greater that the TTC. Where components relevant for assessing toxicity associated clearly some erosion of the conservatism target risk for a mixture may be adjusted of a complex mixture can be shown to be PROGRAMS AND REVIEWS –5 with lifetime exposures (e.g. cancer). Ej = exposure level of chemical j built into each of the processes of to 1×10 (see Section 5.10). below the TTC, they may be disregarded In a recent 2-year rat bioassay study determining components of the HQ in the risk assessment. International risk assessment programs RfD = RfD of chemical j designed to assess the predictive powers j calculation (exposure and TDI). When have been quite active in addressing of TEFs for dioxins and dioxin-like PCBs the HI is greater than 10 there is more 12.5 the complex problems of mixture risk HQj = HQ for chemical j (dimensionless) (Walker et al. 2005; NTP 2006), ternary reason to undertake further investigation COMPONENT ELIMINATION 12.6 assessment. These programs include mixtures of TCDD, PeCDF and PCB of the risks, including an assessment of BIOMARKER APPROACH the following. 126 were administered at dose levels RfD is US terminology equivalent to whether addition of HQs is justified or OR SIMPLIFICATION the ADI or TDI whether the risk contribution of some of •• US EPA (2000; 2003a) – guidance representing the four TEF-equivalent The above approaches to risk assessment the components is independent. For mixtures where a common MoA for conducting health risk assessment doses used for TCDD alone in previous all require detailed knowledge of the or other determinants of risk additivity of chemical mixtures, which has NTP studies. While the dioxin mixtures components of a mixture, including While this pragmatic approach may be Nevertheless, it is a common approach cannot be assumed, the default approach interleaved with the development of produced a range of cellular proliferative their relative concentrations or a useful screening tool, is used quite in the professional practice of EHRA in becomes an assumption of toxicity programs for assessment of cumulative and carcinogenic responses comparable exposures/doses and their toxicological extensively in HRA practice and is Australia (especially when dealing with independence for components of the and aggregate risks. A further review of to that seen with TCDD alone, the characteristics. An alternative approach, generally accepted, or even required, government agencies) to not only require mixture. In such a case, the risk estimate the US EPA methodology for assessing dose–response relationships were not which may overcome some of the gaps by Australian regulatory authorities, it is the calculation of the HI for mixtures, but (e.g. HQ or assessed risk level) associated multiple chemical exposures and identical. Therefore, while this NTP in this knowledge database, requires based on a potentially flawed premise. As to undertake further refinement of the with the most toxic components or those effects using dosimetry-adjustment study provided some support for the TEF identification of a suitable biomarker noted above, additive risk estimates imply risk estimates when the HI exceeds 1. with the greatest exposure potential are using PBPK modelling finalised in concept, it did show that the numerical of effect (Ryan et al. 2007). Such a a common MoA, or at least a common There may be situations where additional assessed independently and may drive 2006–07 (US EPA 2006b; 2007c). values for TEFs may not be applicable biomarker would represent a common toxicological outcome. Where the toxicity information will be available that shows the risk assessment. This approach •• ATSDR guidance for the preparation across the full range of toxicological event in the progression between of individual components does not satisfy that the HI approach will overestimate makes the assumption that overall risk of an interaction profile (2001), with outcomes (Gray et al. 2006). On the exposure and a toxicological outcome. such an assumption, the contributions to the risk and scientific argument could be is no greater than that of the riskiest the subsequent publication of mixtures other hand, Smialowicz et al. (2008) were In such a case, the biomarker becomes risk are theoretically independent. advanced to show this approach to be component and that no additive or profiles for a number of pesticides able to validate the US EPA dioxin-like the measure that aggregates the effects of inappropriate. interactive effects alter the risk. and environmental contaminants, a TEFs to predict immunotoxic endpoints, all components of a mixture that operate based on their CYP inductive effects. The HI equation may be modified to take formal guidance manual for assessing Where the health risk assessment via this pathway. A more comprehensive review of the into account the potential for chemicals in In order to simplify the process by mixtures (ATSDR 2004), and an a mixture to interact and refinement of the approach makes a finite assessment eliminating those components of a overview of the ATSDR program by methodologies and databases used to Some of the areas in which such a exposure assessment using target tissue of the level of risk and compares it to mixture that make little or no effective De Rosa (2004). establish dioxin TEFs has been published biomarker approach may be useful in dose estimates (Herzberg & Teuschler some form of expressing an acceptable contribution to risk, there is an increasing by Haws et al. (2006). or ‘target risk’, as is the case for the risk assessment could include: •• Health Council of the Netherlands 2002; Pohl & Abadin 2008). Such use of the concept that a toxicological non-threshold risk associated with program on exposure to combinations modifications require a more extensive threshold of concern (TTC) may be •• using DNA adducts or other cytological carcinogens, a similar approach may be of substances (Feron et al. 2004). 12.4 database and use a weight-of-evidence derived for risk assessment purposes markers of DNA damage following used to assess the combined risk of a •• Danish Veterinary and Food (WoE) approach to evaluate such (see Section 5.13). This concept may exposure to a complex mixture of SUMMATION OF RISK mixture. In this case, the risk estimates Administration (2003) program for interactions and to modify terms in the alternatively be called the concentration polycyclic aromatic hydrocarbons for individual components derived assessing toxicological effects of above equation. of no toxicological concern (CoNTC) and ESTIMATES from the carcinogenic slope factor and (PAHs) exposures to mixtures of industrial and it is not really a new concept, since it has exposure estimates are simply summed to •• determining the extent of environmental chemicals. The HI approach is essentially quite been applied to various areas of drug and This approach is the one most commonly provide an overall risk estimate. cholinesterase inhibition following conservative in providing an estimate of •• Bjarnason (2004): Toxicology of used in health risk assessments. The food regulation in the US and Europe exposure to organophosphonate and/or cumulative risk, since ‘safety factors of since 1993 (Drew & Frangos 2007). chemical mixtures: a review of contribution of an individual mixture In some cases, the ‘target risk’ estimate carbamate pesticides. between 100 and 10,000 are commonly In an environmental context, the TTC mixtures assessment, for the Canadian component is calculated as a ratio of for the mixture may be set at the same used to adjust the estimated no observed concept has been embodied in recent Defence R & D as part of a NATO predicted exposure compared with a level as that for individual components adverse effect level (NOAEL) to derive the project NATO HFM-057/RTG-009. health-based benchmark (ADI, TDI). This (e.g. 1 × 10–6). This is particularly Australian guidance for assessment of ADI or TDI estimates. This means that derived ratio is termed a hazard quotient conservative, since the addition of risks the effects of the chemical combinations
150 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 151 Chapter 13: Exposure modelling
•• Project IRSST (2005): Impact of •• Which of the two assessment and ‘All models are wrong, some models Because direct measurement of •• plant and animal uptake models, toxicological interactions on the prediction concepts – dose addition 12.8 are useful.’ contaminant concentrations in potential which track the transfer of management of exposure to multiple or independent action – should be OTHER REVIEWS OF George EP Box (1979) exposure media (air, water, soil) may contaminants from sources of contaminants: a Canadian program utilised in practice? not available, or because direct sampling agricultural production (pastures, directed towards occupational •• Which chemicals should be subjected CHEMICAL MIXTURES Modelling is used in exposure assessment and analysis may not be practical, the gardens and crop sites) into food exposures. See
152 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 153 13.1 paucity of good distribution data needed probability of occurrence. The estimated Figure 30: Principles of the Monte Carlo method 13.2.1 for probabilistic techniques such as Monte risk values are combined to provide Weaknesses with the Monte Carlo USING POINT ESTIMATES Carlo. More recently, development of UK a frequency distribution of the output technique soil guidance values (SGVs) using the parameter, such as the consolidated AND PROBABILITY contaminated land exposure assessment estimate of exposure (from Langley Limitations of the Monte Carlo technique DISTRIBUTIONS (CLEA) methodology has reverted to using et al. 1998). include the following: single value (deterministic) data inputs 1. Complexity: While the Monte Carlo rather than probabilistic techniques Point estimates are most commonly used Figure 30 demonstrates the process of method has a very general applicability, (NEPC 2010). in Australia for exposure assessments. A using the Monte Carlo method to estimate changing one variable may mean large point estimate is a single value chosen the probability distribution of exposures in amounts of recalculation because to represent a population such as 70 kg ‘While methods using probability a population. of the extent of the iterative process as the weight of an adult. Point estimates distributions are ‘more informative and when using this model. The complexity are usually typical or average values for a inherently more representative’ [Ruffle et Monte Carlo analysis may add value to a reduces the ‘transparency’ of the population or an estimate of an upper end al. 1994 p. 403] than point estimates, if risk assessment (US EPA 1997b) when: method. This may create difficulties applied appropriately point estimates still of the population’s value such as 70 years •• exposures and risks are likely to in community consultation and risk have a major role in exposure assessment as the duration of residence on a property. approach or be above levels of concern communication it obscures errors, An upper-end value may be chosen for as they are readily understood and and creates difficulties for checking by •• screening assessments using reasons of conservatism or to provide a applied, and may incorporate safety both the modellers and administering conservative point estimates fall above ‘worse case’ scenario. factors that could be lost with Monte authorities. Carlo-type exposure assessments’ levels of concern (Langley & Sabordo 1996). 2. Loss of factor distinctions: The method Where a risk assessment uses a series of •• it is necessary to disclose the degree of does not indicate ‘which variables upper end estimates, the result can be a bias associated with point estimates of are the most important contributors worse than ‘worst case’ scenario due to Hattis and Silver (1994) propose that exposure to output uncertainty’ (US EPA 1992 the compounding effects of the estimates there will be greater uncertainties in •• exposures and exposure pathways p. 22928). (e.g. the person with the upper-end value estimates for the variability of a parameter need to be ranked value (i.e. the standard deviation) than for weight is unlikely to also have the •• there is a need to appraise the relative 3. Unrealistic probability assessments: the estimate of the parameter value upper-end value for water consumption, values of collecting different types US EPA (1992) notes that simulations (i.e. the mean). the upper-end value for contamination, of further information (Cullen & Frey such as that found with the Monte the upper-end value for duration of 1999) Carlo model often ‘include low In standard-setting, this is important if residence and the upper-end value for probability estimates at the upper end one is using point estimates, as a point •• the costs of action are likely to be high soil ingestion. that are higher than those actually estimate of a mean will be more certain and the gains are likely to be marginal experienced in a given population, than a point estimate of the level intended •• the outcomes of action affect different A point estimate of a median or a due to improbability of finding these to represent the 95th or 99th percentile. exposure pathways and the benefits Adapted from: Ferguson 1994. mean is inherently more certain than a exposures or doses in a specific need to be ranked (Cullen & Frey point estimate of the level intended to population of limited size, or due 1999) represent the 95th or 99th percentile 13.2 •• safety is an urgent concern and action For small-scale situations, the use of to non-obvious correlations among because there is more data involved •• sensitivity analysis is needed to apprise must be taken rapidly Monte Carlo methods is likely to be too parameters at the high ends of their in determining the estimates of central MONTE CARLO ANALYSIS the impact of default values and key ranges’. This results in overestimation •• probability distributions are so complex or costly, and it may be more points than for estimating extremes. This pathways of exposure dose or risk. The Science uncertain or incomplete that appropriate to use direct measurements will present problems if there are limited Monte Carlo-type exposure assessments Advisory Board of the US EPA •• the consequences of simplistic detailed probabilistic judgements are of exposure. The exposures of high- data for using point estimates if the point rely on the use of probability distribution has noted: ‘For large populations, exposure assessments are likely to be unreasonable end exposure ‘outliers’ must always be estimate is intended to be, for example, functions. This may be described as a unacceptable. acknowledged in risk assessments, and simulated exposures, doses and risks •• there is little variability or uncertainty in the 95th percentile. For the same reason, ‘distribution of possible values for each ways of identifying and accommodating above the 99.9 percentile may not be the data. similar problems will arise if the tails of a of the parameters [is] described along Monte Carlo analysis may not add value them must be considered. This is meaningful when unbounded log- probability distribution are to be estimated. with the probability of occurrence of to a risk assessment (Cullen & Frey 1999) particularly important in the assessment normal distributions are used as a If a Monte Carlo assessment is performed each value’ (Alsop et al. 1993 p. 407). when: of an existing situation (e.g. a default’ (US EPA 1992 p. 22922). the methodology must be ‘transparent’ Increasing attention has been paid to Using standard mathematical formulae, contaminated site where housing has •• exposures and risks are likely to be or problems will arise in community 4. Assessment endpoints: With Monte the use of Monte Carlo-type exposure several thousand iterations of the output already been developed), rather than negligible consultation. As with any form of risk Carlo-type assessments there is assessments and such methods have parameter are performed. a forecast exposure scenario, as the assessment, the basic principles of the still a need to determine what is an been acknowledged by the US EPA and •• the costs of reducing the exposure presence of an ‘outlier’ will severely test method must be able to be understood by acceptable level of exposure. Smith the UK Department of the Environment For each iteration, values for each and risk are smaller than the cost of the credibility of any risk communication the affected community. (1994 p. 48) considers that ‘the level (US EPA 1992; Ferguson 1994). However, parameter are selected randomly from probabilistic analysis exercise (Langley & Sabordo 1996). this trend may be reversing because of the each distribution based upon the of exposure exceeded by 1 in 20
154 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 155 exposed persons would seem to be Even the use of complex models frequent combinations of model exposure •• Short-term and long-term variation: 13.2.4 Finley et al. (1994) recommend the an appropriate reasonable maximum’. still provides a static picture of a scenarios. Latin hypercube techniques Interpersonal variability will be Selecting appropriate datasets following criteria for assessing data: This would allow 5 per cent of the dynamic world albeit a more elaborate use random sampling within equiprobable decreased if the length of time •• consistency with other studies population not to be included in the representation of reality (McKone intervals of the distribution so that there over which a factor is measured is When establishing probability exposure assessment. 1994) and such a picture must be will not be clustered sampling near increased. Short-term data tends to distributions, the distributions should be •• relevance of the survey population placed within a sound theoretical the mode. It also ‘maintains complete overestimate inter-individual variation determined, where possible, from relevant to the general population or the 5. Variability–uncertainty confusion: framework. independence of the variables’ (AIHC (Finley et al. 1994). For example, the datasets. If there is a need to estimate population being appraised as part Smith (1994) highlights the need 1994 p. 3.3). but this also means if 95th percentile of dietary intakes from a probability distribution, it should be of a risk assessment to distinguish between ‘variability’ 8. Misleading precision: The use of more correlations are intended between studies taken over one- to three-day appreciated that that many environmental •• minimisation of confounding variables (measurable factors that differ across complex models does not necessarily health factors are likely to be log-normally variables appropriate mathematical periods will be significantly higher •• whether there is sufficient data to populations such as height) and increase precision (McKone 1994). distributed rather than symmetrically actions must be taken. than for studies taken over longer adequately characterise variability and ‘uncertainty’ (unknown, difficult to The costs of collecting and analysing distributed. Examples of risk variables that periods such as 1 month to 1 year. the extremes of the distribution. measure factors such as frequency data, and constructing new models 13.2.3 This has been seen in the studies have been characterised by log-normal of trespassing on a site). Currently must be balanced by the value of the distributions are (Murphy 1998): Estimating distributions for exposure of tap water consumption and fish Haimes et al. (1994) propose several available software packages do not information obtained. There is a need factors consumption (Finley et al. 1994). It •• body weight (each sex) approaches to developing distributions distinguish between variability and to appraise the value of information can be particularly marked for rare when objective data is missing or scarce uncertainty. An administrator reviewing along with its uncertainties in defining Factors affect the choice of distributions •• bioaccumulation exposures (e.g. rarely eaten food such or not quite relevant: a Monte Carlo risk assessment will, the capabilities and limits of exposure (Finley et al. 1994) include: as shellfish). •• breathing rate however, need to appreciate the models (Langley & Sabordo 1996). •• cancer potency factors •• When data is sparse but relevant, differences between variability and •• Variability and uncertainty: Variability, Studies of shellfish consumption •• concentrations in air, soil, tissue, water expert judgement can be used to uncertainty and the nature and extent 9. Characterisation of extreme values: as an inherent characteristic of a taken over short periods of time may propose percentiles using available of both (Smith 1994). More recent The 50th percentile can always be population, will not be reduced with suggest only a very small proportion •• drinking-water rate data as ‘collaborators’ of the expert developments of Monte Carlo analysis, estimated with less uncertainty than additional data but will be more of the population consumes the •• exposed skin judgement. the 99th percentile (Finley et al. 1994). accurately characterised. Uncertainty, foodstuff and, if the common practice such as two-dimensional simulations, •• fish consumption •• Where data is not quite relevant to Problems in estimating the extreme however, will be reduced with of excluding all non-consumers is may assist with differentiating propose a distribution for a parameter, percentiles can come from limitations additional data. undertaken, there will be a poor •• lifetime uncertainty and variability in the input expert judgement again can be used in the measurement techniques (e.g. characterisation of the variability in the •• residence time parameters (Simon, 1999). Uncertainty may arise from factors collaborating the judgement with incorrect and implausible estimates of general population (Finley et al. 1994). intrinsic to the available data (e.g. •• shower duration analogous data. 6. Limited exposure data: Limited dietary consumption may be accepted limitations of study design and Short-term data tends to overestimate •• shower water use •• Where there is an absence of data the information is available about into the survey) the duration over analytical techniques) or from the inter-individual variation. many variables for the exposure which exposure data was collected •• soil ingestion rate formal elicitation of expert judgement application of data to non-sampled •• Parametric versus non-parametric assessments. As a consequence (see short-term and long-term •• surface area/body weight to construct a distribution can be used. populations (e.g. extrapolating distribution characterisation: For data of this, many input variables are variation below) and whether there Scandinavian data to an Australian to meet parametric distributions (e.g. •• total water use described as triangular distributions. are sub-populations that may have If there are a variety of studies then the population) (Finley et al. 1994). normal or log-normal), appropriate •• toxic susceptibility. Smith (1994) stresses the need to unusual exposures (e.g. vegetarians, purposes, designs and methodologies that statistical tests must be met. collect and verify distributions from subsistence fishers) (Finley et al. The characterisation of uncertainty are similar maybe able to be combined Theoretically normal or log-normal Much environmental data is log-normally many currently undescribed input 1994). Estimating extreme percentiles related to exposure factors has been (Finley et al. 1994). density distributions do not have an rather than normally distributed. Table 21 assumptions to improve accuracy. The can be a very time-consuming process. developed further than two other areas upper-bound limit yet for many factors gives some examples of output variables use of Monte Carlo methods may be of uncertainty that may in fact be more Haimes et al. (1994) highlight the (e.g. height, weight, fluid consumption) that can be represented by probability inappropriate where the predictions significant: the relationship between need to examine the tails of probability 13.2.2 there are obviously physiological distributions. of exposure are so dominated by the absorbed dose and the ultimate distribution functions and submit them Monte Carlo versus Latin hypercube limitations to the factors. Some of the uncertainties. McKone (1994) gives delivered dose to a target organ and to a ‘reality check’ and examine the currently available software enables the example of benzo(a)pyrene, Monte Carlo uses ‘random (or pseudo- the uncertainty about the response to For describing a probability distribution, combination of factors that resulted in such factors to be set within the model. where information on benzo(a)pyrene random) numbers to sample from the the dose (Finley et al. 1994). the relevant studies and the quality of the the extreme values. They highlight the input distribution ... [so that] ... samples data produced may vary considerably. extreme sensitivity of these tail values to exposure is ‘not readily available’ so •• Factor interdependence: Some factors, •• Shapes of distributions: Triangular are more likely to be drawn from values Unless datasets are rigorously scrutinised assumptions and reinforce the need to that the use of Monte Carlo methods such as body weight and skin surface shape distributions are often used that have higher probabilities (e.g. near the resulting uncertainty in the range assess the sensitivity of the tails to the to assess variability in population area, are interdependent, and this in Monte Carlo-type assessments the mode)’ (AIHC 1994 p. 3.3). This of risk estimates could be greater than assumptions. The assumptions need exposures is somewhat redundant. needs to be considered. Age-specific but may be viewed as conservative could be important if there is concern obtained using point estimates (Finley to be examined as to whether they are data should be used, as the factor characterisations of truncated 7. Simplification of complex situations: about exposures represented by the et al. 1994). mutually consistent. may be strongly related to age (e.g. normal or log-normal distributions Exposure assessments are comprised tails of the distributions (e.g. 99th (Finley et al. 1994). of combinations of modelling, percentile exposures). Large numbers inhalation rates). sampling, and modelling/sampling of iterations are required to overcome combinations (McKone 1994). this. It is more likely to result in unduly
156 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 157 Table 21: Some key variables for which probability distributions might be needed 13.2.5 •• Use probabilistic techniques (which –– a graph of the variable with Monte Carlo exposure assessment may be demanding in terms of time, administratively set allowable Model component Output variable Independent parameter variable datasets money and other resources) only risk criteria as annotations and where exposure pathways are likely to point estimates of risk using the Transport Air concentration Chemical emission rate Increasing amounts of data are becoming be significant. administratively set point estimates Stack exit temperature available to enable the use of Monte •• Provide detailed information about of exposure Carlo-type assessments. Stack exit velocity input distributions. The minimum –– a table of the mean, the standard Mixing heights stated by Burmaster and Anderson is: deviation, the minimum (if one The Exposure factors sourcebook (AIHC exists), the 5th percentile, the Meteorological factors Wind speed Taskforce 1994) presents extensive –– a graph showing the full distribution median, the 95th percentile, and documentation of probability distributions and the location of the point value Wind direction the maximum (if one exists). from a variety of sources intended for used in the [point estimate] risk Deposition Deposition rate Dry-deposition velocity the US population. Descriptions of the assessment •• Provide records of sensitivity analyses Wet-deposition velocity probability distributions are provided to –– a table showing the mean, and their impact that will enable the Fraction of time with rain enable easy use of @ RISKR software. standard deviation, the minimum determination of the most important (if one exists), the 5th percentile, input variables (or groups of variables). Overland Surface-water load Fraction of chemical in overland runoff The probability distributions for soil the median, the 95th percentile •• Assess the numerical stability of the Water Surface-water concentration River discharge ingestion by children use the studies and the maximum (if one exists) central moments (mean, standard (p. 478). There needs to be a deviation, skewness and kurtosis) and Chemical decay coefficient in river of Calabrese and Stanek (1991) and Burmaster et al. (1991). These sufficient justification of the selected the tails of the output distributions. Mixing depth distributions vary markedly and the distribution, which should be The latter are particularly sensitive Groundwater Groundwater concentration Predictions of plumes updated values of Calabrese and based on adequately referenced to the nature of the tails of the input Stanek (1995) are not included. sources and the statistical, distributions and, as they stabilise very Soil Surface-soil concentration Surface-soil depth physical, chemical, and biological slowly, sufficient iterations are required Exposure duration 13.2.6 mechanisms relevant to the to demonstrate the numerical stability. Exposure period Principles for the use of Monte Carlo- distribution. Burmaster and Anderson suggest that Cation-exchange capacity type techniques •• Detail how the input distributions commonly more than 10,000 iterations are required. Software that enables Decay coefficient in soil capture and represent both the The purpose and scope of the risk variability and the uncertainty in the Latin hypercube sampling results in Food chain Concentration in animal products Soil ingestion rates assessment should be clearly articulated input variables (p. 478) so as to enable more rapid stability of these output in the issues identification section. Plant to animal bioconcentration factors both variability and uncertainty to be tails. Burmaster and Anderson state Burmaster and Anderson (1994) stress described and analysed separately. that the changes in the tails of only Plant concentration Plant interception fraction that any method of exposure assessment a few input distributions contribute •• Use measured data to test the Weathering elimination rate must have a clearly defined assessment strongly to changes in the upper tail of relevance of the input distribution to Crop density endpoint and provide all relevant the output distribution. information so that the assessment can be the population, place and time of the Soil-to-plant bioconcentration factor •• Detail the name and statistical quality reproduced and evaluated. Burmaster and exposure assessment. Further data may need to be gathered to supply of the random number generator Fish concentration Water-to-fish bioconcentration factor Anderson (1994) detail 1-14 principles for missing information or supplement used. Some generators are inadequate good practice in Monte Carlo assessments. because of short recurrence periods. Dose Inhalation dose Inhalation rate incomplete information. Body weight These are as follows: •• Describe the methods by which •• Interpret the results and detail the measured data was used to derive a limitations of the methodology such Ingestion dose Plant ingestion rate •• Detail all formulae. probability distribution. as the effects of biases not elsewhere interpreted. Soil ingestion rate •• Detail point estimates of exposure •• Detail any correlations between Body weight where these are demanded by data where there are relatively high Burmaster and Anderson state that ‘the regulatory agencies. correlations. Sensitivity analysis may Dermal absorption dose Exposed skin surface area principles are not mutually exclusive •• Detail sensitivity analyses to enable be necessary to determine the effects nor collectively exhaustive’ (Langley & Soil absorption factor the identification of relevant and of correlations between variables on Sabordo 1996 pp. 140–141). Exposure frequency important input variables. Those the exposure analysis. Body weight variables that will drive risk assessment •• Provide detailed information and must obviously be included in the graphs for each output distribution. Adapted from: NRC 1994, p. 169; Seigneur et al. 1992. Monte Carlo analysis but reasons for Burmaster and Anderson suggest the excluding insignificant variables must following as a minimum: also be detailed.
158 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 159 Chapter 14: Biomonitoring
13.2.7 considered appropriate. Most regulators 13.3 14.1 14.1.1 The use of biomonitoring data in Administrative requirements for the are likely to discourage the use of this Why biomonitoring? environmental risk assessment was use of Monte Carlo methods technique, in the main due to the INTEGRATION OF BIOMARKERS reviewed at an international biomonitoring difficulty in explaining it to the affected Biological monitoring is a measuring workshop in 2004, at which six case procedure whereby validated indicators The range of total acceptable exposures community and the lack of robust EXPOSURE WITH The term ‘biomarker’ has been studies illustrated the applications and of the uptake of contaminants, or their and risk will need to be defined on a probability distributions for parameters used in recent times to describe the utilities of this technique in environmental EHRA OUTCOMES metabolites, and people’s individual situation-specific basis after consultation of interest. measurements used in biological health surveillance (Albertini et al. 2006). responses are determined and with stakeholders. Depending on how it monitoring. The term refers broadly to As part of the NRC review of toxicity interpreted. Whereas environmental is applied, the Monte Carlo method may Since the outputs of Monte Carlo analyses almost any measurement reflecting an Further reviews of the application of testing developments for the 21st century monitoring measures the composition lose much of the conservatism usually are distributions of risk estimates and interaction between a biological system biomonitoring to risk assessment have (NRC 2007), Hubal (2009) commented of the external environment around a inherent in point estimates. other parameters, some guidance will be and an environmental agent, which may been presented by Doerrer (2007), on the role that developments in exposure person, biological monitoring measures needed on where to define the cut-offs for be chemical, physical or biological (WHO Angerer et al. (2006) and Swenberg sciences must play in developing new the amount of contaminant absorbed into Regulatory authorities in Australia risk assessment purposes. This is likely to 1993b, 2001). Three classes of biomarker et al. (2008). paradigms of EHRA. In particular, the the body. are likely to require the following of fall into the realm of policy settings to be are identified by WHO (2001): assessments using Monte Carlo methods: development of models that could be determined by government authorities. As used to define exposures at levels ranging One of the difficulties traditionally noted above, UK guidance on establishing •• Biomarker of exposure: an exogenous Biological monitoring may be direct associated with the interpretation of •• meeting the 14 principles of good from environmental to cellular (target substance or its metabolite or the (e.g. the measurement of lead in blood) practice detailed above guidance values (GVs) for contaminated tissue doses) would be important for biomonitoring data has been the absence land exposure assessment (CLEA) is product of an interaction between or indirect (e.g. the measurement of of validated values representing specific •• providing adequate information to integrating animal testing data with the already showing signs of ‘back-pedalling’ a xenobiotic agent and some target the breakdown product of nicotine and levels of exposure or linking to levels of the authority to enable review of new generation of scientific tools using on the use of probabilistic approaches, molecule or cell that is measured in a cotinine in urine). Biological monitoring effect. This issue was addressed by an the assessment – this may require genetic, in vitro and in silico techniques compartment within an organism may measure a biological effect, such as such as Monte Carlo analysis. for profiling chemical toxicity and international panel convened to develop a providing the software (and underlying •• Biomarker of effect: a measurable enzyme depression, or a physiological individual susceptibility. A depiction of series of biomonitoring equivalents (BEs) formulae) and data biochemical, physiological, behavioural effect, such as tremor. The monitoring the interrelationships in such a model is (Hays et al. 2008). This panel established •• a demonstration of the relevance of or other alteration within an organism may be used to identify whether exposure shown in Figure 31. some guidelines on what should be taken the exposure data to the site (data that, depending upon the magnitude, has occurred at all, or the amount of into consideration in establishing BEs, from other countries or cultural can be recognised as associated with exposure. including consideration of toxicokinetics backgrounds may not be relevant) Figure 31: Proposals for integrating exposure with outcomes of EHRA an established or possible health and internal dose metrics, integration of •• an explanation of the data and method impairment or disease If biological monitoring is practicable, it human and animal data, and the choice will be more valuable than environmental that will be able to be understood by •• Biomarker of susceptibility: an of suitable tissues and analytes. This monitoring in determining the level of risk the relevant community (usually the indicator of an inherent or acquired expert group devised a series of flow from an environment, as it will measure most difficult aspect) ability of an organism to respond to charts (Figure 32) illustrating how animal whether exposure is occurring and the the challenge of exposure to a specific and human data could be integrated, •• using data that accounts for age and level of exposure (Langley 1991b). It can gender differences and takes into xenobiotic substance. depending on the extent to which be useful in identifying highly exposed pharmacokinetic data in either species account susceptible populations. individuals or sub-populations. For many environmental pollutants, the are well understood. flow of events between exposure and On a large site divided into housing lots, The prerequisites for biological monitoring health effects is not well understood. Biological monitoring should not be the results for specific housing lots that (Aitio et al. 1988) are as follows: may be affected by atypically elevated Biomarkers help address this problem commenced before: concentrations should not be obscured by improving the sensitivity, specificity •• The substance and/or metabolites •• the objective of the biological by averaging or Monte Carlo techniques and predictive value of detection and need to be present in a tissue, body monitoring is clearly defined applied to the entire site. In many quantification of adverse effects at low fluid or excretion suitable for sampling. •• a reference range of results that is instances, Monte Carlo methods will only dose and early exposure (Fowle 1989; •• Valid, accurate and practicable applicable for the population under be relevant to large sites or sites where Fowle & Sexton 1992; NRC 1992). methods of sampling and analysis are study is established – this is often direct measurements of exposure are not Sensitive sub-populations can be better available. not available (or a control group is practicable. Before the use of Monte Carlo pinpointed by biomarkers that measure •• The results of testing can be not available to establish a reference is commenced for any situation being increased absorption rate or a more interpreted in a meaningful way for range); the relationship of body burden assessed, the assessor should check severe biological response to a given individuals and groups. levels and exposure (or risk) are with the relevant regulator or government environmental exposure (Fowle & Sexton unavailable for many substances authority about whether such use is Hubel 2009. Reproduced with permission of Oxford University Press. 1992; Hemminki 1992; Lauwerys 1984; •• An appropriate management strategy NRC 1992). has been devised for sampling, •• consideration has been given as to how analysis, collation of results, results are to be managed – significant interpretation of results, and follow-up. anxiety may be caused by factors such as delays in providing information and
160 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 161 an inability to explain the meaning of Figure 32: Flow charts for deriving BEs for chemicals pharmacokinetic data are available Several aspects must be considered: the persistence of the substances acts •• Transient periods of high exposure may measured levels or to take action if for both animals and humans along with toxicity data from both species (a) or only from as a long-term marker of exposure not be detected. •• A good biological monitoring test the person is distressed by elevated animals (b) result may not correlate well with •• it enables an individualised Having decided a test for a substance is levels, perceives that any measure of environmental levels (mainly because assessment taking into consideration appropriate, further questions arise: exposure is unsatisfactory or equates of human factors). age, sex, personal hygiene, exposure to a health effect may cause biotransformation and elimination. •• Which compound should be •• The number of substances that can be •• the correct timing of sampling has measured? The substance, a used reliably for biological monitoring been established – correct timing The disadvantages and difficulties are: metabolite or both? is still small. is critical for substances with short •• Which biological fluid or tissue is to be •• the relatively wide range of individual biological half-lives or a particular •• Irritative, locally or rapidly acting sampled? substances are usually unsuitable response to a substance and the wide exposure is of concern •• In relation to what period of exposure? as the systemic absorption may be ‘normal’ range that may have to be •• a process has been established minimal and/or irrelevant to the level considered •• How frequently should sampling be to enable consistent analysis and of local reaction (e.g. SO , ammonia, •• the lack of simple specific analytical done? epidemiological appraisal of results 2 direct skin exposure to PAHs causing methods of sufficient sensitivity (in •• the ethical and confidentiality aspects skin cancer). many instances) of collecting, maintaining and 14.2 •• The substance must be in some tissue •• difficulties in sample collection, for distributing information and results are or fluid suitable for sampling. example, 24-hour urine collections CHOICE OF TISSUE fully considered •• Accurate, valid and practical •• unsuspected exposure can be shown •• a centralised collection point for OR FLUID measuring methods must be available. but the source cannot be pinpointed – results has been established to enable this will require detailed environmental consistent analysis and epidemiological •• The result should be interpretable in The biological samples used for monitoring appraisal of results. terms of health risk. monitoring may be (Fao & Allesio 1983): •• inferences caused by occupational •• The results are likely to have more •• blood, urine, fat, saliva, sweat, faeces The reasons for biological monitoring value for a group than an individual. exposure, for example, lead exposure in •• hair, nails, teeth include to: battery makers and radiator repairers The advantages of biological monitoring •• there must be a clear relationship •• expired air. •• detect whether exposure has occurred are: (if only on a group basis) between the •• quantitate exposure chosen biological indicator and the Physiological response to the exposure •• the exposed person is his or her •• enable the risk of health effects to be health risks of the substance. may be estimated by determining own sampler, so that many ‘samples’ assessed changes in: are taken over a 100 per cent Some analyses require specialised •• determine changes in exposure sampling time •• the amount of a critical biochemical over time or to assess the effects laboratories: constituent •• the evaluation of absorption can be of interventions such as health performed over a prolonged period •• There may be laboratory inaccuracy. •• the activity of a specific enzyme education or soil remediation (if of time •• If the substance has a short •• a particular physiological function serial measurements are done) or biological half-life, rapidly changing such as lung function. to determine exposure pathways •• the sampling takes into account all concentrations in body samples and their relative importance, such the person’s movements within and complicate interpretation and the body Choice of biological tissue or fluid for a as occupational versus domestic outside the domestic environment, burden may be under-predicted or hypothetical substance is represented exposures ingestion of soil versus and accidental and illicit exposures over-predicted. in Figure 33. inhalation of dust (if combined with •• the amount absorbed by various environmental monitoring) routes is considered (not only via the respiratory route as is presumed •• determine segments of the population Figure 33: Choice of biological tissue or fluid for a hypothetical substance at greatest risk, such as particular age by monitoring of atmospheric groups or those living in particular concentrations), for example, oral locations or circumstances (if absorption of lead compounds or in conducted as part of widespread situations where skin absorption is population studies). important •• it may show exposures where Results should always be available to past environmental monitoring is participants in biological monitoring unavailable, for example, PCBs where combined with a meaningful explanation Reproduced from Hays et al. 2008 2008 with permission from Elsevier. of the results.
162 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 163 14.2.1 accurate measure of excretion via hair measurement at low levels and the Table 22: Substances likely to be suitable for biological monitoring Blood cannot be performed. Hair analysis may possibility of results being overwhelmingly be useful for assessing intake from purely influenced by other sources of exposure Substance Fluid or tissue Comments Depending on the biological half-life of a dietary sources when there is no general (e.g. the influence of cadmium in food, Lead Blood Urinary lead does not accurately reflect either recent exposures or burden. Substantial data available substance, blood analysis may provide environmental contamination. tobacco smoke and the occupational an indication of exposure from recent on level of risk for particular blood lead ranges. Numerous Australian studies provide comparison data. environment will generally be far levels of concern available for both general population and groups (for example, children). hours to several years. Levels are often 14.2.4 greater than the influence of cadmium transient if the half-life is not prolonged. Cadmium Urine or blood Urinary levels tend to reflect body burden; blood levels reflect recent exposures. Urinary levels need Breast milk contamination of soils). The process of blood-taking may be to be adjusted for changes in urinary flow rates (results often given asµ g Cd/g creatinine or µg Cd/24 hour). Laboratory inaccuracy has always been a major problem, particularly prior to 1980. Limited unacceptable for some people, including Collecting breast milk is usually easy and For many substances, biological children. acceptable to nursing mothers. Breast Australian studies to provide comparison data. Most international studies have concentrated on monitoring is impracticable because: occupational exposures. Very limited data on children, especially for those less than 5 years. WHO milk provides an integrated exposure for (cited in Mueller et al. 1989) has set levels of concern. General diet and smoking will tend to have a When the volume of distribution is high, •• analytical techniques are not available very lipid soluble compounds for time major influence on levels. concentrations in blood are often too low periods related to the biological half-life of or are inaccurate at low levels or in the to be measured. Samples may require the substance. Breast milk measurements tissues or fluids being tested Arsenic Urine Short biological half-life; study must be done during exposure (or at most within 1–2 days afterwards). careful procedures, such as plasma of PCBs, organochlorine pesticides and •• insufficient information is available on Considerable interference from organic sources of arsenic (for example, seafood). Dietary sources from the environment not under study need to be excluded and testing for inorganic arsenic undertaken. separation and freezing. Substances dioxins have been used for exposure inter- and intra-individual toxicokinetics Limited comparison data and no set levels of concern. measurable in the plasma may not be assessments. The concentrations must and thresholds of health effects to responsible for the toxic effect which, be standardised for fat content and enable risk assessment of results Mercury Blood or urine At equilibrium, the concentration of mercury in the blood reflects daily intake and is probably the best indicator of exposure. Total measured mercury will also include methyl mercury from fish, so that a instead, arises from a metabolite. may vary according to the period since •• insufficient epidemiological studies have fractionated analysis of mercury salts and alkylated mercury compounds may be required (Aitio et al. breastfeeding first commenced. been done to determine normal ranges. 14.2.2 1988). Methyl mercury exposure will not affect urinary mercury levels although urinary levels show significant diurnal variation. Some international comparison data is available. Urine 14.2.5 The correct choice of biological tissue Expired air Polychlorinated biphenyls Blood, adipose Long biological half-life so that historical exposures (i.e. body burden) may be able to be monitored. Only a limited number of substances or fluid is important. Rarely can the concentration in the critical organ (PCBs) tissue (fat), Different PCBs will have different behaviours in the body and different biological half-lives. Some can be measured in urine because of Expired air is used to determine breast milk comparison data is available. It is difficult to obtain adipose tissue samples and blood sampling is be measured and compared with degradation of the parent substance to exposures to ethanol (e.g. traffic usually preferred. breakdown products. Urine samples, breathalyser) and some solvents and can concentrations that give rise to effects. Organochlorine (OC) Blood, adipose Long biological half-life so that body burden can be assessed. Some comparison data is available, in general, provide a more integrated be correlated to blood concentrations Attempts have been made for such pesticides tissue (fat), especially for blood. It is difficult to obtain adipose tissue samples and blood sampling is usually assessment of exposure than blood for based on the Henry’s Law constant of the breast milk preferred. periods of recent hours or days. Twenty- substance being measured. direct measurement, for example, in vivo four-hour sample collections may be more neutron activation analysis can directly Organophosphonate (OP) Blood Plasma butyrylcholinesterase or erythrocyte acetylcholinesterase (AChE) may be monitored to assess appropriate than spot samples but many measure renal or liver concentrations pesticides recent exposures. Depressed AChE activity may better reflect a level where a physiological response people find these collections onerous. 14.3 of cadmium but requires specialised may occur. Wide range of values reflect ‘normality’, so individual baseline values assist interpretation. First morning urine samples have been equipment and provides a dose of CHOICE OF A TEST Adapted from: Langley (1991b). found to be effective for representing 24 ionising radiation to the subject. hour urine samples. (Froese et al. 2002, Optimally, a biological monitoring test Bader et al. 2004, Zhang et al. 2009). For biological monitoring based on There are a range of other substances oxide, styrene oxide, vinyl chloride, substances practicable (e.g. manganese, would give a result that reflected the urine analysis, simple measurements Urine samples require rapid processing exposure, the concentration of the for which biological monitoring may be aromatic amines, polycyclic aromatic radioactive isotopes). and cooling. of concentrations can provide sufficient available – the tests should be assessed hydrocarbons). substance in the target organ and the risks information on exposure, but in many of adverse effects (Friberg 1985). Few and used on their individual merits for a A good knowledge of the toxicokinetics 14.2.3 instances, measurements of elimination particular situation. Biological monitoring Besides the pesticides mentioned of a substance is required for the correct tests are available that approach this ideal rates provide more precise information. Hair and toenails has been applied to a range of situations: in Table 22 specialised tests may be choice of method and interpretation of (Langley 1991a). Furthermore, what is of Urinary concentrations related to most importance is ‘not so much tobacco use (polycyclic aromatic available from some laboratories for results. The duration of persistence of the Hair and toenails can provide an integrated creatinine, or urinary flow rates may hydrocarbons, aromatic amines and pesticides such as glyphosate. agent will be important as is the volume of measure of exposure over a more the choice of medical test as much provide more accurate information, as the way the testing program is specific nitrosamines), dietary exposures distribution (e.g. many very lipid soluble prolonged period than blood or urine. but creatinine has not been found to (e.g. aflatoxins, N-nitrosamines, Most organic contaminants are not substances with a very high volume of They are only useful for chemicals known organised, the way the results are be worthwhile in some evaluations evaluated and communicated, and the heterocyclic amines), medicinal amenable to biological monitoring distribution have such low blood levels to accumulate in those tissues and they (Zhang et al. 2009) exposures (e.g. cisplatin, alkylating in general environmental situations that they can’t be measured in blood are inappropriate tissues for biological way abnormalities are pursued’ (Silverstein 1990). agents, 8-methoxypsoralen, ultraviolet because of the low levels of exposure but can be identified in breast milk). monitoring on or near contaminated Substances for which biological photoproducts), trichloroacetic acid for and the lack of comparison data Individual results may be distorted if there environments. External contamination of monitoring of general environmental In Australia, exposures from contaminated chlorinated disinfection by-products compared with occupational situations. is not constant exposure or equilibrium the hair cannot be adequately removed exposures is practicable are detailed in drinking water and occupational Specialised studies may make within the body (Langley et al. 1998). during sample preparation and an soil for example will be generally in Table 22. low, creating problems in accurate exposures (e.g. benzene, ethylene biological monitoring for some inorganic
164 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 165 Cytogenetic testing may occasionally be Aberrant results may need to be repeated may need to be assessed to determine This is one of the reasons why monitoring Before health monitoring is undertaken, of value but is often difficult to interpret before being accepted as ‘high’. Choice whether there may also be a significant blood lead is a more common approach 14.7 the following issues should be considered: as only small numbers of cells are usually of a laboratory should be governed by the risk from cadmium exposure. to lead EHRA and risk management HEALTH MONITORING •• how to ensure all parties involved do examined so that there is the potential presence of stringent internal and external (see Section 14.8). not have unreasonable expectations for considerable confidence limits quality control measures. Health monitoring is the organised about the ability of health monitoring to around the results and because there 14.4 The physico-chemical properties of the medical assessment of individuals resolve issues of causation or to detect can rarely be a link made to specific Contamination during sample collection INFLUENCES ON contaminant will have a crucial influence and groups of people. The medical any subtle effect (the studies rarely agent (one exception is aflatoxin). Tests is likely to be a significant problem unless on the bioavailability of the contaminant assessment will consist of history taking provide such evidence because of their such as sister chromatid exchange specialised collection protocols are BIOLOGICAL MONITORING and hence biological monitoring results. and clinical examination, and, where size and biases) and micronuclei are non-specific tests. rigorously followed. One example is skin A further crucial influence will be the indicated, particular tests (e.g. lung RESULTS •• confidentiality of information There are problems with confounding, contamination affecting blood samples characteristics of the exposed population function testing where there is a concern distinguishing recent from historical (especially capillary prick samples). (e.g. age, behaviours). about the effect of air pollutant). The •• how and when information will be Factors apart from environmental exposures, quantifying exposures epidemiological aspects of health surveys made available to participants (the contamination to be considered in and dealing with a finite background Twenty-four-hour urinary collections are The physico-chemical properties of the are covered in Chapter 10. information must be released to interpreting biological monitoring incidence of chromosomal abnormalities. likely to be impracticable during general contaminant and the characteristics of participants) results include (American Conference community studies and present significant the exposed population usually will be In Australia, health effects are likely to •• access to information (by whom and of Governmental Industrial Hygienists – Under the National Model Regulations risks for contamination during collection. more important predictors of biological be found in only a limited number of through what mechanisms) ACGIH 1990): for the Control of Workplace Hazardous Inappropriate sample containers can monitoring results than a statement of situations of environmental contamination. •• interpretation of information (at an Substances (adopted by the states and be a significant source of inaccuracy •• changes induced by strenuous the concentration of the contaminant in Subtle effects may only be able to be individual and group level and on territories), health surveillance is required from leaching or contamination. Without physical activity the soil. determined on a group basis rather what evidentiary basis) for specified substances. Biological appropriate selection of containers and •• changes induced by environmental than on an individual basis (e.g. subtle •• release of findings (which should be at monitoring methods developed for some of storage conditions, some heavy metals conditions (including heat, diet and 14.6 neurodevelopmental effects determined a group rather than individual level for these methods are detailed in the NOHSC will adsorb to some container materials cigarette smoking) by sophisticated testing in groups of reasons of confidentiality if the results series Guidelines for health surveillance. giving falsely low readings. A single children with different lead exposures). •• changes induced by water intake ABNORMAL RESULTS are made public) laboratory is preferred for studies to Similar problems of causation relating •• how the information will be used to 14.3.1 minimise problems arising from inter- •• changes in physiological functions to individual findings rather than group If the accuracy of an abnormal result address the relevant environmental laboratory variations and to enable a induced by pregnancy, disease or findings arise if the putative effects are Accuracy can be confirmed (this may require health issues. single body of data. diurnal rhythms common in the general population (e.g. Laboratory accuracy has always been a repeat testing), the health risks should •• changes in metabolism induced by headache or fatigue). Health effects problem because of the low levels of the be assessed and medical assessment 14.3.2 congenital variations of metabolic are rarely as specific to an exposure as substance being tested and analytical may be required. The reason for the high 14.8 Indicator analytes pathways or induced by simultaneous result should be determined, that is, the chloracne with PCB or dioxin exposure. problems, including those caused by administration of another chemical BIOMONITORING AND the biological matrix and the risk of Where there are multiple contaminants relevant exposure pathways. (induction or inhibition of activity of Health monitoring for specific health contamination. Gross analytical errors uniformly distributed in the environment BLOOD LEAD a critical enzyme by medication or effects is warranted where environmental have occurred in the measurement and with similar environmental and There should be a clear understanding by pre-exposure or co-exposure to or biological monitoring has indicated a Since the absorption and retention of of blood lead, and blood and urinary biological behaviour, the measurement of of the basis of how the ‘normal’ range another chemical). significant risk of effects (e.g. specific lead from various environmental matrices cadmium (Elinder 1985; Vahter 1982). one contaminant (the indicator analyte) was derived. (e.g. What populations were tests of renal function if urinary cadmium can be variable, biomonitoring (blood Friberg (1985) reports that ‘normal’ may be a surrogate measure for other studied? Were they comparable to this levels above the levels of concern are lead levels) has become the method values for aluminium in plasma and contaminants. The indicator analyte may population?). If the range is derived from 14.5 detected in biological monitoring). of choice for data inputs into health serum ‘decreased’ during the 10 years be chosen for the ease (or accuracy) normally distributed results in a general risk assessments and for managing 1975–1985 ‘from several 100 µg/l to a of analysis or its toxicity relative to the EXPOSURE AND population survey and the range is two When health monitoring is done, it environmental health risks associated few micrograms the only reason for this other contaminants. For example, if lead standard deviations each side of the BIOLOGICAL MONITORING should rarely be done in isolation with lead, particularly in children. being improved analytical technique’. Aitio and cadmium are uniformly present, mean, 5 per cent of this population will from environmental and/or biological et al. (1988) provide a further example lead may be chosen for the ease and have ‘abnormal’ results. If results are RESULTS monitoring. Clearly defined health for the values regarded as normal average relative accuracy of analysis as well being compared with health standards, 14.8.1 effects should be sought with specific serum chromium concentrations for as the availability of levels of concern For toxicokinetic reasons, the relationship how were these standards set? Do the Adult lead exposures case-definition criteria. Records of other occupationally unexposed men. Papers and comparison data. Alternatively, between exposure and biological standards incorporate a safety factor and, symptoms and clinical findings should These may be estimated using the US EPA published between 1956 and 1984 lead may also be chosen because it is monitoring results is often not linear. if so, how large is that safety factor? also be kept to enable epidemiological adult lead model methodology (US EPA showed a decrease in ‘normal’ values the predominant contaminant. In such For example, with air lead levels there assessment of other potential health 2003c). This model focuses on adult from 3,600 mmol/l to 2.1 mmol/l; Aitio instances, if the blood lead results are appears to be a greater influence on the effects (Langley 1991a). women and incorporates lead exposure, et al. (1988) attributed the decline not elevated, elevated levels of cadmium rate of change of blood lead levels with uptake into the body and biokinetic transfer to better techniques that avoided would not be expected. If high blood lead changes at lower air lead levels than into the blood and developing foetus. chromium contamination. results are demonstrated, cadmium levels moderate air lead levels (Friberg 1985).
166 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 167 Chapter 15: Microbiological risk assessment
Microbiological risk assessment (MRA) There is much support for the application The International Life Sciences Institute – Pbintake × BKSF × EF is an especially important part of and development of MRA (ACDP 1996). Risk Science Institute (in conjunction with PbB = PbB + adult background EHRA associated with food and water To date, MRA has predominantly been the US EPA): AT contamination. Recent Australian applied to two exposure sources, food •• A process that evaluates the likelihood initiatives to make more efficient use of and water, and much of the conceptual PbBadult = total adult blood lead concentration that have site exposures to of human health effects occurring scarce water resources through water development of MRA has resulted from soil lead (µg/dL) after exposure to a pathogenic recycling and stormwater harvesting the application of MRA to these media. micro-organism or to a medium in AT = averaging time (days/year) have placed a particular emphasis on which pathogens exist (ILSI 2000). EF = exposure frequency (days/year) MRA (NRMMC, NHMRC, AHMC 2006; MRA concepts and methodologies NRMMC, EPHC, NHMRC 2008; NRMMC, (particularly QMRA) are somewhat PbBbackground = background adult blood lead concentration (µg/dL) The Advisory Committee on Dangerous EPHC, NHMRC 2009a, b). These less well developed than comparable Pathogens (UK): Pbintake = total lead uptake from all media (g/day) initiatives required the development of methodologies in chemical risk BKSF = biokinetic slope factor (µg/dL per µg/day) HRA methodologies that predict disease assessment. For example, with QMRA, •• A formal structured procedure impacts associated with the pathogens vast datasets need to be developed, for identifying and characterising likely to be encountered in such water modelling needs to be improved (e.g. microbiological hazard and This approach allows for protection of •• The uptake component models the sources, as well as informing the secondary transmission) and analytical determining the risk associated with the most sensitive receptor in the adult process by which the lead intake setting of microbiological standards for techniques need to be refined. Methods it (ACDP 1996). scenario, which is an unborn child is transferred to blood plasma. reclaimed water. The Phase 1 recycled for extrapolation from animals to humans carried by a pregnant mother, and is The amount of lead that is taken up is water guidelines (NRMMC, NHMRC, are still being developed, with different The WHO and FAO guidelines on MRA: the approach taken for determination of controlled by the bioavailability of the AHMC 2006) introduced the concepts approaches being proposed. QMRA •• A tool used in the management the HIL for soil lead in the contaminated lead, which can be specified separately of quantitative MRA (QMRA), while does have an advantage over chemical of the risks posed by food-borne sites NEPM review for the adult exposure for soil, water and food. subsequent documents outlined how risk assessment in some cases where pathogens, including the elaboration scenario (NEPC 2010). these principles could be adopted into human volunteers’ dose–response data •• The biokinetic component models the of standards for food in international water management. is available. This obviates the need for balance of lead in the body between trade. Quantitative MRA is recognised 14.8.2 Lead exposures in children animal-to-human extrapolation which uptake and excretion. A central as a more resource-intensive task The aim of this chapter is to provide a introduces considerable uncertainty in The predominant modelling system for estimate of blood lead concentration is (WHO 2009). output from this component. brief description of the principles of MRA most cases of chemical risk assessment. assessing lead exposures in children is and QMRA in the context of EHRA and the Integrated Exposure Uptake Biokinetic •• The variability component applies a QMRA has been defined as follows: to briefly outline the processes. A more The development of models for MRA Model for Lead in Children (IEUBK log-normal distribution to the output detailed description of MRA can be have generally been developed based on •• The application of principles of model, Version 1.1 Build9, released June of the biokinetic component using obtained from the cited references. conventional risk assessment frameworks risk assessment to the estimate of 2009. See
168 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 169 9. Wherever possible, risk estimates The tolerable risk adopted in the AGWR 15.3 should be reassessed over time by 15.5 DALYs = YLL (years of life lost) + YLD is 10–6 DALYs per person per year, which GENERAL PRINCIPLES comparison with independent human EXPRESSION OF (years lived with a disability/illness) is consistent with the WHO Guidelines illness data. for drinking-water quality (WHO 2006a). The Codex principles of MRA, as MICROBIOLOGICAL RISK The advantage of using DALYs as a This is approximately equivalent to an 10. An MRA may need re-evaluation as –3 applied to food, are listed below. method of expressing QMRA risk is that annual diarrhoeal risk of illness of 10 new relevant information becomes In common with other types of EHRA, These principles can also be DALYs include a measurement of the (i.e. one illness per 1,000 people). available. estimating microbiological risk may generalised to the other media: water, air, severity of impacts on human health In comparison, the reported rate be expressed in numerical notation soil and the surfaces of inanimate objects. arising out of infection and illness. of diarrhoeal illness in Australia is or qualitatively, using terms such as Most of the principles listed are similar to 15.4 They differentiate between relatively 0.8–0.92 cases per person per year low/medium/high. Risk may also be established risk assessment principles, mild impacts, such as diarrhoea, and (NRMMC, EPHC, AHMC 2006). characterised by a narrative description except for item 7, which is unique to MRA MICROBIOLOGICAL severe impacts, such as haemolytic of the risk, or whether it breaches (Codex 1999, 2003). uremic syndrome and even death. However, there are also problems RISK ASSESSMENT standards or guidelines. In practice, In terms of waterborne disease, the with using DALYs. The methodology is 1. MRA should be soundly based however, a continuum exists from a fully – PARADIGMS AND most commonly recognised illness is relatively complex, potentially costly, upon science and conducted quantitative through to a wholly narrative gastroenteritis (involving symptoms such and inherently conservative. It requires according to a structured approach FRAMEWORKS expression of risk. validated knowledge of infectious doses that includes hazard identification, as diarrhoea and vomiting) following ingestion of enteric pathogens. However, of selected reference pathogens, which hazard characterisation, exposure Micro-organisms are living entities An MRA cannot always practically a number of these pathogens can cause may or may not reflect the full range assessment and risk characterisation. and are very different to chemicals achieve numerical expression of more severe and long-lasting symptoms of pathogenic organisms likely to have and physical hazards by their nature. microbiological risk (ACDP 1996). 2. A MRA should clearly state the in a small percentage of infected people, health impacts in an EHRA. It requires MRA requires additional methods This can be due to, for example, lack purpose of the exercise, including for example: estimation of likely human exposure and terminology that are particular to the form of risk estimate that will be of dose–response data or a lack of doses of a microbial population that microbiological risks (e.g. methods of the output. understanding of the route of entry of a •• diabetes, associated with Coxsackie B4 may change rapidly over time. The estimating secondary transmission), and pathogen. Semi-quantitative or qualitative virus (Mena et al. 2003) calculation of YLD is based on application 3. The conduct of a MRA should be infective doses need to be developed. MRA can be applied in these situations. •• myocarditis, associated with echovirus of severity factors for various health transparent. However, the enHealth model can, in and Coxsackievirus (Mena et al. 2003) outcomes that are necessarily value- general, be applied to MRA. The Australian guidelines for water 4. Any constraints that impact on the risk •• reactive arthritis and Guillain- laden and may not reflect the values of recycling (AGWR) (NRMMC, EPHC, assessment such as cost, resources Barré syndrome, associated with affected stakeholders. Like the processes Haas et al. (1999) have produced AHMC 2006) include a detailed or time should be identified and their Campylobacter jejuni (Havelaar et al. of non-threshold chemical EHRA, it the most comprehensive attempt at discussion of risk assessment metrics possible consequences described. 2000; Nachamkin et al. 2001) requires the policy-driven establishment describing the methods used in QMRA using disability-adjusted life years (DALYs) of acceptable levels of risk. Again, like •• haemolytic uraemic syndrome, 5. The risk estimate should contain a and the particular needs of MRA. and the reference pathogens used chemical EHRA, such target risk levels associated with haemorrhagic description of uncertainty and where However, they have not developed in their derivation. The use of DALYs carry a perception of exactitude. Some Escherichia coli (Teunis et al. 2004) the uncertainty arose during the risk a modified framework that attempts and reference pathogens is based on of the problems associated with using assessment process. to encompass these different needs. the approach described in the World •• reactive arthritis, associated with DALYs are acknowledged and discussed Instead, their approach to MRA and Salmonella (Rudwaleit et al. 2001). 6. Data should be such that uncertainty Health Organization’s Guidelines for in most references, but not always. While QMRA loosely follows the National in the risk estimate can be drinking-water quality (WHO 2006a). there may not be a better tool available Academy of Sciences (NAS) framework determined. The DALY concept for microbiological Determining DALYs for individual currently, it would be useful if these proposed for chemical risk assessment risk assessment was not included in hazards includes considering acute shortcomings were discussed more widely 7. Data and data collection systems (NRC 1983). risk assessment advice in the Australian impacts (e.g. diarrhoeal disease or and appreciated by those applying the should, as far as possible, be of drinking water guidelines (NHMRC, even death) and chronic impacts standards and methodologies. sufficient quality and precision that By contrast, the International Life NRMMC 2004). However, it was adopted (e.g. reactive arthritis, haemolytic uncertainty in the risk estimate is Sciences Institute and the US EPA (ILSI for risk management of water in Australia syndrome). Calculation of DALYs includes minimised. 1996; 2000) have explicitly adapted with publication of the AGWR. consideration of each of the symptoms the NAS framework to suit the unique caused by a particular pathogen and 8. An MRA should explicitly consider challenges presented by MRA. Like the relative frequency of occurrence the dynamics of microbiological In brief, the DALY concept is a measure the QRMA process described by Haas (NRMMC, EPHC, AHMC 2006). growth, survival and death in foods of the years of life lost by premature et al. it essentially follows the standard and the complexity of the interaction mortality (YLL) and years of healthy life description provided by the NAS (including sequelae) between human lost in states of less than full health (years paradigm but uses synonymous terms for and agent following consumption as lived with a disability – YLD). The YLD each part of the process. well as the potential for further spread. parameter is weighted according to the severity of the disability.
170 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 171 Chapter 16: Guidance on route-specific EHRA
General guidance in this updated designed to be protective of the health guidance on site assessment methods assumptions are available when assessing a particular setting or activity, and are The issue of whether the negative impacts document is relevant to EHRA for all of the more sensitive or vulnerable in line with technological changes in ambient air pollution from diffuse and monitored on a needs basis. of odours that affect quality of life should sources and routes of environmental ‘receptors’ on or near the site, it is Australia and overseas (EHPC 2010). point-source regions or large areas, be classified as adverse health effects exposures. There are, however, some important to emphasise that they are not localised air pollution from point sources, Currently, the air toxics NEPM is controversial. While it is undoubtedly specific EHRA guidelines available in intended to represent clean-up levels The assessment of site contamination and indoor air pollution such as may provides monitoring advice and health- important for these matters to be Australia relating to specific environmental or targets for clean-up. Levels found NEPM can be accessed at the occur in the home or workplace. Where based air quality guidelines for selected considered in a standard-setting process, sources, such as contaminated sites, to be marginally in excess of the HILs Environmental Protection and large populations are involved, different volatile organic compounds (VOCs), it is debatable whether the appropriate air pollution, contaminants in food, and do not imply unacceptability or that Heritage Council (EHPC) website at epidemiological methodologies such as semi-volatile compounds, polycyclic place is within the scientifically rigorous contaminants in potable and recycled a significant health risk is likely to be
172 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 173 It also recognises that odour emissions Some of the risk factors associated with •• food additives, processing aids and A key source of information is the series designed to provide an authoritative may not be preventable from some 16.3 food are: packaging materials of Australian Market Basket Survey reference on what defines safe, good- activities, and that ‘no odour’ may not publications that are published biannually quality water, how it can be achieved FOOD CONTAMINANTS •• agricultural and veterinary chemical •• mycotoxins, plant and marine toxins be a realistic objective. It is a reasonable by FSANZ. These provide information and how it can be assured. They are residues •• novel foods and ingredients objective that may be exceedingly difficult Food-related risks can occur because of about certain substances in a range of concerned both with safety from a health to achieve in some cases. a range of factors, and in many cases the •• biological agents, including micro- •• radionuclides foods across Australia. point of view and with aesthetic quality. interdependence of these factors needs to organisms, viruses and parasites •• other environmental contaminants. be considered when assessing risk. •• cooking and process-related artefacts The ADWG are not mandatory standards. Some of these may occur as a result of 16.4 However, they provide a basis for commercial growing or processing of the WATER CONTAMINATION determining the quality of water to be food, some of them may be because of supplied to consumers in all parts of localised pollution at the point of planting, There are a wide range of water types, Australia. These determinations need BOX 4: harvesting or animal husbandry. Where water uses and possible routes of to consider the diverse array of regional Clean and Healthy Air for Gladstone (CHAG) project food is home-grown (e.g. in backyard transmission of waterborne hazards or local factors, and take into account vegetable patches), the contamination to humans. In undertaking health risk economic, political and cultural issues, may be associated with uptake from assessments, the characteristics and including customer expectations, and Gladstone is a city located approximately 550 km north via public meetings. The EHRA was supplemented with locally contaminated soil. potential uses of water bodies need to willingness and ability to pay. The of Brisbane. It is a centre for substantial industrial activity a questionnaire and hospital admission health survey, be determined. Water sources include ADWG are intended for use by the and development, including aluminium smelting, mineral undertaken in 2008 to identify any increased incidence Where the contamination occurs as fresh, estuarine, marine and waste Australian community and all agencies and gas processing, coal shipment from its port facilities, of specific disease patterns that could be attributed to air the result of some commercial activity waters. Water uses can include supply of with responsibilities associated with coal-fired power generation, chemicals manufacturing and pollution in the Gladstone region. that can be controlled by regulation potable water for drinking and bathing, the supply of drinking water, including projected expansion as a liquid natural gas transport hub. (e.g. specifying allowable food additives recreation, aquaculture and irrigation of catchment and water resource managers, Planning the EHRA included benchmarking the measured or migration from packaging material crops. Human exposure to waterborne drinking-water suppliers, water regulators Community concerns about air quality and possible air pollution measurements (appropriately averaged over specifying good agricultural practice contaminants can include: and health authorities (NHMRC & impacts on public health caused the Queensland one hour for irritants, or 24-hour–12-month averages for to ensure food residues do not exceed NRMMC 2004). •• direct exposure through ingestion, Government to expand its air quality monitoring programs other pollutants) against published air quality standards regulated maximum pesticide residues dermal contact, inhalation of aerosols in the Gladstone region and implement a more detailed from Australian NEPMs and various WHO, US or other – MRLs), these aspects of food 16.4.2 or sprays EHRA, in conjunction with the further allocation of six well- international guidelines. regulation fall within the jurisdiction of Australian guidelines for water equipped monitoring stations to sample an extended range the Australian Pesticides and Veterinary •• indirect exposure through foods recycling of air pollutants in the airshed. An interim report covering the health risk analysis of Medicines Authority (APVMA) and/or contaminated by irrigation water The Australian guidelines for water the first six months air monitoring data was released Food Standards Australia New Zealand or water used for aquaculture and (FSANZ). MRLs can be accessed at seafoods contaminated by waste water recycling are designed to provide an The range of air pollutants monitored in the program over in November 2009. Since many of the measured air authoritative reference that can be used a period commencing in January 2009 included the six pollutants produced consistently low values, often below
174 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 175 Part 3: of these guidelines is simply to provide 16.4.3 Regulatory the scientific basis for implementing Guidelines for managing risks in those decisions in a safe and sustainable recreational water manner. National water recycling guidelines were produced in two phases: Published by the National Health and Medical Research Council (NHMRC), context Phase 1: Australian guidelines for the primary aim of these guidelines is to water recycling: managing health and protect the health of humans from threats environmental risks (Natural Resource posed by the recreational use of coastal, Ministerial Management Council estuarine and fresh waters. Threats may (NRMMC), Environment Protection and include natural hazards such as surf, Heritage Council (EPHC), Australian rip currents and aquatic organisms, and Health Ministers’ Conference (AHMC) those with an artificial aspect, such as 2006). Phase 1 of the guidelines provides discharges of wastewater (NHMRC 2008). a generic ‘framework for management of recycled water quality and use’ that 16.4.4 applies to all combinations of recycled Australian and New Zealand guidelines water and end uses. It also provides for fresh and marine water quality specific guidance on the use of treated These guidelines are published by the sewage and greywater for purposes other Australian and New Zealand Environment than drinking and environmental flows. and Conservation Council (ANZECC) and the Agriculture and Resource Phase 2 (Module 1): Australian guidelines Management Council of Australia and for water recycling: augmentation of New Zealand (ARMCANZ). The purpose drinking water supplies (NRMMC–EPHC– of the guidelines is to provide an NHMRC). This current document, the authoritative guide for setting water quality first module of Phase 2 of the guidelines, objectives required to sustain current, extends the guidance given in Phase 1 or likely future, environmental values on the planned use of recycled water uses for natural and semi-natural water (treated sewage and stormwater) to resources in Australia and New Zealand. augment drinking water supplies. The document focuses on the source of water, initial treatment processes and blending of recycled water with drinking water sources.
Phase 2 (Modules 2 and 3): – Modules 2 and 3 cover use of stormwater for uses other than drinking water augmentation and managed aquifer recharge.
These documents are accessible on the EPHC website at
176 ENVIRONMENTAL HEALTH RISK MANAGEMENT Chapter 17: Australian regulation of chemical hazards
Chemicals assessment in Australia Working Group on Aboriginal and Torres is mainly the province of agencies Strait Islander Environmental Health. 17.1 associated with the Australian OVERVIEW OF THE Government.5 However, the Under the guidance of the AHPC and with implementation of chemicals regulatory reference to the National environmental CHEMICAL RISK programs, especially those aimed at health strategy 2007–2012, enHealth’s ASSESSMENT AGENCIES management of environmental health terms of reference are to: risks, is mainly the province of state, •• provide nationally agreed Several bodies are involved in the process territory and local governments. environmental health policy advice, for undertaking chemical hazard and based on the best available evidence risk assessments. National chemicals The Environmental Health Committee and expertise, to the Australian Health legislation and responsible authorities are (enHealth) is a Commonwealth–state Ministers’ Advisory Council (AHMAC) outlined in Table 23. coordinating body that provides policy through the AHPC advice in these areas. It is constituted Essentially, if a product or chemical is as a subcommittee of the Australian •• coordinate implementation of nationally not intended for agricultural or veterinary Health Protection Committee (AHPC) agreed environmental health policies use, nor for human therapeutic or food/ (see
ENVIRONMENTAL HEALTH RISK ASSESSMENT 179 Table 23: Chemicals regulation in Australia and epidemiological data, detailed respiratory protection, gloves). However, 17.2.2 exposure information for workers, the where mechanical ventilation is installed, New chemicals Agency National Industrial Chemicals Australian Pesticides and Therapeutic Goods Administration Food Standards Australia New public and the environment and risk this can be factored into the EASE model, Notification and Assessment Veterinary Medicines (TGA) Zealand (FSANZ) management initiatives. The toxicological should suitable monitoring data not be For new industrial chemicals, the data Scheme (NICNAS) Authority (APVMA) package includes available human, available (i.e. measured when ventilation requirements depend on the notification Portfolio Health and Ageing Agriculture, Fisheries and Health and Ageing Health and Ageing animal and in vitro data and ecotoxicity has been installed and is operational). category and are stipulated under the Forestry and biodegradability/fate data for the The quality of the monitoring data should ICNA Act. A standard dataset comprises environmental assessment. also be a factor considered in the risk information confirming the identity of the Scope Assessment and review, not Assessment, product Assessment and product registration Assessment and standard- chemical, the physico-chemical properties registration based registration and review setting characterisation and exposure standard- The dataset for a preliminary PEC may or setting processes . and use of the chemical; detailed Relevant Industrial Chemicals Agricultural and Veterinary Therapeutic Goods Act 1989 Australia New Zealand Food may not include a detailed toxicological exposure information about how workers; legislation (Notification and Assessment) Chemicals Code Act 1994 Authority Act 1994. package or detailed exposure data. Risk Where dermal exposure is an important the public and the environment are Act 1989 and Veterinary Chemicals Food Standards Code exposed to the chemical; and a standard Administration Act 1994 assessment, in terms of a formal risk route of exposure or where the characterisation for specific uses, is not toxicological database does not provide an toxicological package. The toxicological About the Industrial chemicals are Agricultural products, Therapeutic goods, including Food additives and carried out for preliminary assessments. inhalation study, internal (dose) exposure package includes animal and in vitro data chemicals varied and include dyes, including chemicals that prescription medicines and processing aids used in food may be estimated utilising the available for the human health assessment, and solvents, adhesives, plastics, destroy/repel pests or plants. medicines available over the counter to assist with preservation, The dataset for a secondary notification pharmacokinetic data and used in the risk ecotoxicity and biodegradability data for laboratory chemicals, paints Veterinary products are used (OTC). OTCs include complementary flavouring, colouring, or the environmental assessment. and coatings, chemicals to prevent, diagnose or treat medicines (herbal, vitamin and modifying its functions; assessment is determined in accordance characterisation process. used in cleaning products, disease in animals. homeopathic preparations); some permitted levels of some with a set of circumstances (criteria) as cosmetics including some sterilants and disinfectants and environmental contaminants set out in the ICNA Act. Should these Estimates of public exposure are carried 17.2.2.1 sunscreens. cosmetic-type products that make are also controlled via the circumstances require a re-evaluation out similarly for existing chemicals and Exposure assessment therapeutic claims. Food Standards Code. new chemicals (see 17.2.2.1). of the risks assessed in the original PEC The occupational exposure assessment report, then a formal risk characterisation is conducted by establishing the use is usually carried out. 17.2.1.3 pattern of the chemical and identifying What follows is a brief synopsis of the health and environmental impact – over the The risk assessment entails some or all of Toxicological data the sources of occupational exposure. roles and methods of operation of the two life cycle of the chemical. the following elements: 17.2.1.2 Exposure is then estimated by taking agencies listed in Table 23 that have main Toxicological and epidemiological/case •• hazard identification Exposure data into account the routes of exposure, responsibility for chemicals likely to be NICNAS was established in July 1990 study/clinical data is also provided as the frequency and duration of exposure, released into the environment (industrial under Australian Government legislation: •• hazard assessment, incorporating the Occupational and public exposure data statutory obligation (under the ICNA Act) and measured worker data (e.g. chemicals, AgVets and consumer product the Industrial Chemicals (Notification and dose–response relationship is provided as a statutory obligation from applicants (for assessment). This example, atmospheric and/or biological chemicals). Note that the agencies and Assessment) Act 1989 (ICNA Act) •• exposure assessment (under the ICNA Act) from applicants data is supplemented from literature monitoring results). Information is responsibilities are current at 2011, (for assessment). This information is review and international reports (e.g. •• risk characterisation, where the needed for each of the scenarios but may change according to future NICNAS’s activities include: hazard and exposure assessments supplemented from literature review, site OECD SIARs, IPCS, IARC, ECETOC). where workers are potentially exposed government reorganisations. are integrated. visits, international reports (e.g. OECD •• assessing industrial chemicals that are SIARs), and where data is lacking from Currently, available toxicological and to the chemical. new to Australia for their risk to human Based on the risk assessment findings, modelling. The model that has been epidemiological data is evaluated health and the environment, before For new industrial chemicals, the 17.2 recommendations are made to mitigate used to date is the UK HSE EASE model, in conjunction with available use or release into the environment occupational exposure assessment is NATIONAL INDUSTRIAL the risks. which provides estimates of airborne pharmacokinetic data, to estimate the •• assessing industrial chemicals that and dermal exposure for different critical no observed adverse effect levels usually qualitative, as measured data is unlikely to be available and there is CHEMICALS NOTIFICATION are already in use in Australia (known 17.2.1 occupational scenarios. (NOAEL) or, if not determined, the lowest as ‘existing chemicals’) in response to insufficient information available to predict Existing chemicals observed adverse effect level (LOAEL) AND ASSESSMENT health and environmental concerns Where exposure by inhalation is the for both acute and chronic exposures reliable quantitative estimates. Modelling (e.g. using EASE) is occasionally used. SCHEME •• making risk assessment and safety 17.2.1.1 major route of exposure, and the for each relevant route of exposure (i.e. information on chemicals and Data requirements toxicological database includes good oral, dermal and inhalation). The health Estimates of public exposure from NICNAS is a chemical entity-based their potential occupational health quality inhalation data (human or animal), hazards for each endpoint are classified For existing industrial chemicals, the the common practice is to use ‘external’ consumer use (e.g. for cosmetics) are notification and pre-market risk assessment and safety, public health and in accordance with the Safe Work data requirements depend on the exposure data (i.e. not to attempt to made using the expected type and scheme. Industrial chemicals are defined environmental risks widely available to Australia (SWA) NOHSC Approved Criteria assessment type. A standard dataset for extrapolate to ‘internal’ dose) in the risk frequency of use. The possibility of in the context of their use and by their workers, the public, industry and other for classifying hazardous substances. a full-priority existing chemical (PEC) characterisation process (see Section public exposure arising from release exclusion as therapeutic goods, food or government agencies comprises information confirming the 17.2.3). When ‘external’ exposure data The quality of the toxicological database into the environment during transport, food additives, pesticides and veterinary •• enabling the public, organisations identity of the chemical, the physico- is used/determined, no adjustment is should be a factor considered in the risk manufacturing or end-use is also taken medicines. The scope of the NICNAS and key stakeholders to have effective chemical properties and use of the made to account for reduced personal characterisation and exposure standard- into account. assessments comprises three elements input into decision-making processes chemical (including import/manufacture exposures resulting from the use of – occupational health and safety, public setting processes. regarding the safe use of chemicals. volumes), all available toxicological personal protective equipment (e.g.
180 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 181 17.2.2.2 and dermal exposure (where relevant), that to determine the risk of adverse health 17.3.1 space (e.g. fumigants). Therefore, where the database, nature and severity of the Toxicological assessment is, by using NOAELs derived specifically effects such as acute respiratory effects Data requirements the toxicological database includes dermal health effect and known variability in from each route of exposure. and skin irritation. For longer-term and dosing studies of appropriate quality and human metabolism of the chemical. In Both human and experimental animal repeated exposures, the health risk The Agricultural and Veterinary Chemicals duration, ‘external’ exposure data is used general, a tenfold factor is considered data is assessed in accordance with Where exposures may be significant by to workers is characterised by firstly Code Act makes provision for approving in the risk characterisation process. appropriate to account for inter-species international guidelines to identify the both routes, the combined estimated comparing exposure estimates with active constituents, registering products extrapolation and a similar factor (10x) critical health effects of the chemical internal dose may be used. In this case, NOAELs to give an MoE, and then containing those active constituents, and Pesticide exposure assessments also take for intra-species variability. and to determine the dose–response the oral NOAEL (for the critical effect) is deciding whether there is cause for reconsideration of existing chemicals that into consideration the protection afforded relationship, with NOAELs established usually considered the more appropriate concern. have been nominated for review. Data by label-specified protective equipment. Current exposure mitigation methods wherever possible. For new industrial NOAEL for deriving the MoE. required for the OHS assessment of AgVet Default protection factors are utilised in are evaluated quantitatively, where chemicals, human data is usually not Similarly, an estimate of risk to the public chemical products includes: the absence of specific data. possible. In the absence of data or available. The health hazards of the The resulting MoE is then evaluated is characterised through the hazard of the •• use pattern of the product models, qualitative assessments are chemical are classified in accordance (for each route), taking into account chemical and determining whether there 17.3.3 conducted, based on generalised with the approved criteria for classifying •• formulation composition of product the quality of the available database is any significant public exposure. The Toxicological data information about the use pattern and hazardous substances. Classification is (e.g. whether derived from human data, approach taken will vary with the nature of •• physico-chemical properties of the ‘scientific judgement’. Where current also determined according to the Global uncertainties in the database) and nature any hazard posed by the chemical and the active constituent and product Toxicological and epidemiological/case exposure assessment methods are found Harmonized System for Classification or severity of effect (e.g. carcinogen, extent of data on exposure and toxicology. •• toxicology of the active constituent and study data provided by applicants is to result in unacceptable risk, additional and Labelling of Chemicals (GHS) (see sensitiser). No specific values are assigned product evaluated by the Office of Chemical exposure and risk reduction methods
182 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 183 agricultural and veterinary chemicals. 17.4.2 17.4.3 the market place based on criteria •• helps address community concerns summaries and company-sponsored Evaluation reports Chemical review programs now recognised as outdated by today’s and general interest in agricultural ‘expert reports’. It is important that all The origins of this function can be traced regulatory standards. The ECRP involved and veterinary chemicals by providing toxicity data and the methods by which back to 1984 when the then Toxicology Reports are structured to allow for ready Until the mid-1990s, there was no formal cooperative arrangements between information to the public on the use they are obtained be subjected to critical Evaluation Section (TES) was created access to the main points arising from the program to review ‘old’ pesticides and the health portfolio (public health), of chemicals and their environmental, and independent scientific assessment. following a Senate inquiry into hazardous assessment. They are written to provide veterinary drugs in Australia. However, environment portfolio (environmental public health and OHS aspects chemicals, which noted the increasing sufficiently detailed information for the two programs provided an informal assessment), Safe Work Australia Hazard/risk assessment: Given the reader to form an independent conclusion mechanism for reviewing a number •• considers public nomination of use of chemicals in the environment. (formerly the National Occupational chemicals for review. complexity of biological data interpretation Increasing public concern and media and aim to obviate the need, during of aspects of chemicals safety. First, a Health and Safety Commission) and the need for professional judgement a subsequent review of the chemical review process for individual chemicals attention to chemical exposure demanded (occupational health and safety) and AgVet chemicals are selected for review and a flexible approach when assessing greater accountability from the chemical (or product), to refer back to the original occurred on the suspicion of human the APVMA (chemistry, efficacy and the public health risk of chemicals, it study data. health and/or environmental concerns, or on the basis of agreed criteria including industry and from government regulators. agricultural issues, residues and their potential health and environment has not been the policy to establish In addition, chemical residues in export following international regulatory action.. registration). The ECRP ran in parallel prescriptive methodologies for hazard Reports include the following In practice, this usually involved reviewing hazard(s), exposure potential, age and produce (e.g. beef, wheat, dairy goods) with the so-called ‘Special Review adequacy of the database, efficacy, and risk assessment, although several have important implications for trade components. limited extra data related to the particular Program’ that was in place to deal guidance documents for evaluators have issue of concern rather than conducting a international regulatory actions, and so Australia must be able to ensure •• Submission summary – briefly with issues relating to existing chemicals been drafted. In general, a qualitative comprehensive review. trade and other agricultural implications. standards of chemical regulation outlines the results from all studies that needed to be dealt with rapidly The chemical selection process also approach is used to assess chemical acceptable to international markets, accompanying the application/ (e.g. that a particular use practice risk. The approach taken to derive an Second, the Technical Grade Active incorporates a mechanism for public as well as to domestic consumers. In submission and includes a discussion was leading to MRL exceedances in acceptable daily intake (ADI) follows Constituent (TGAC) Scheme, introduced nominations of chemicals. acknowledgment of public health and of the important findings and exported food commodities). In about the principles outlined in Environmental in 1985, while designed primarily to trade concerns, chemicals regulation as appropriate recommendations. 2000, the ECRP and Special Review The public and occupational health health criteria monographs 104 and an area of public policy has developed, identify the source and ascertain the Program were combined into one 210 prepared by the WHO/UNEP/ILO •• Main body of the report – contains aspects of reviews are assessed by recognising that the numbers of quality of technical grade materials Chemical Review Program. International Programme on Chemical detailed outlines of the studies staff of the OCS, which provides chemicals requiring assessment and used for formulating end-use products Safety (IPCS). conducted with the chemical, toxicological and chemicals policy advice the complexity of the assessment (EUPs) used in Australia, had significant The goal of the Chemical Review including methodology, the extent of as required to appropriate committees process have increased. elements of a review program in Program is to ensure agricultural and monitoring for biological changes, all of Commonwealth, state and territory While the main focus of agricultural and that mammalian toxicology data and veterinary chemicals in use in Australia treatment-related effects and any other government agencies, and to the public. veterinary chemical assessments is a environmental data were collected can be used safely and effectively. Recommendations on individual observations or information that may consideration of human exposure to and reviewed. As a result of the TGAC The program operates according to chemicals form an important component be pertinent to the assessment of the The OCS also undertakes technical pesticides through ingesting residues scheme, toxicology data on more than the principles of openness, fairness of the decisions by the APVMA in the significance of the findings. policy development and provides health in food and/or drinking water, the direct registration of chemicals. OCS reports 400 pesticides were reviewed with respect and consistency with regard to public dermal or inhalational exposure of the Reports on agricultural and veterinary to public health considerations over a advice on international chemicals treaty are also used to assist the Advisory consultation, selection of chemicals for negotiations. It interacts with other public, as users of chemicals (in the Committee on Chemicals Scheduling chemicals generally assess the types of 6–7 year period. As a consequence of the review, and standards of assessment. home garden/domestic setting) or as studies set out in Chapter 9. Submitted above two programs, Australia was well government agencies on a range of to make recommendations to the All aspects of a chemical (public health, environmental health issues. bystanders to agricultural or licensed pest Departmental Scheduling Delegate on studies usually include acute studies placed to develop further more formal OHS, environmental, efficacy, and animal control operator (PCO) use, is also taken poisons scheduling. The OCS provides with the technical grade active review arrangements. and crop safety) are considered in a into account. constituent (TGAC), relevant product An important role of OCS is to encourage the secretariat for the scheduling advisory review. The review program: and, where practical, to extend committees. The recommendations of the formulations and, where relevant, Australia introduced a formal program In general, a classification system for •• works towards the goal that AgVet international harmonisation of chemicals delegate are formally incorporated into the studies on the toxicity of key impurities. to review existing agricultural and public health aspects is not used when chemicals remain safe and effective regulation, including toxicological reviews Standard for the uniform scheduling of •• Consolidated summary – contains the veterinary chemicals in 1994 under the regulating chemicals with potential when used according to label and re-registration programs. medicines and poisons (SUSMP) which integrated summaries of study results title of the Existing Chemicals Review carcinogenicity. The potential human instructions by specifically considering forms the basis for national uniformity in from previous submissions relating to Program (ECRP), managed by the then carcinogenicity of chemicals is assessed toxicity and exposure patterns in 17.4.4 drugs and poisons scheduling. the particular active ingredient, plus National Registration Authority (NRA). using a weight-of-evidence (WoE) relation to public health, OHS and Assessment processes within the OCS the newly evaluated data. The program carried out systematic approach, which takes into account reviews of existing agricultural and environmental control mechanisms The scope of OCS assessments may be •• Confidential business information – Toxicologists within the OCS assess epidemiological data, carcinogenic veterinary chemicals on a priority basis. known and potential environmental expanded to address toxicological hazards impurity profiles, product ingredients mammalian toxicology and toxicokinetic potency in animals, biological relevance This program was one of a number of impacts efficacy safety issues in associated with a range of natural and and information on additives in data and prepare written assessment and potential human exposure. Australia, initiatives arising from a 1990 Senate relation to target species (animal and synthetic chemicals. Examples of formulations, etc. are included in reports that carry sufficient detail in this regard, supports the general inquiry into aspects of the legislative, crop) management options to reduce these include chemicals in certain removable appendices at the end of of the studies and findings to allow approach outlined by IPCS. administrative and regulatory procedures identified risks consumer products, and environmental the report. an independent assessment of the contaminants. for agricultural and veterinary chemicals. •• helps maintain the protection of data. As the primary emphasis is on Exposure assessment: Two aspects The ECRP stemmed largely from the fact Australian trade and commerce in independent assessment, limited of exposure to AgVet chemicals are that many registered chemicals entered agricultural produce and livestock regulatory status is given to company evaluated. The OCS takes responsibility
184 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 185 for evaluating worker and bystander (or be withdrawn), or registrations for potential impacts on Australian flora ‘risk analysis’ to describe the process. risk management decisions and licence This list is not exhaustive, but a special exposure, and uses these assessments to particular products may not be granted and fauna. Coupled with this are study Risk analysis is defined as a systematic conditions. The RAF has been recently note is made of ACTRA, which was recommend appropriate safety directions. (or be withdrawn), in order to eliminate or requirements to enable assessment of use of available information to determine updated and released in May 2009. The established with support from enHealth reduce potential public exposure. environmental fate and persistence. how often specified events may occur and outcome of the risk assessment process is partners in 2006 to specifically address Possible dietary exposure to pesticide Manuals explaining how these risk the magnitude of their consequences. a Risk assessment and risk management professional development in the areas of residues is based on calculations of likely The use of registered AgVet products assessments are undertaken are plan (RARMP) produced in response to toxicology and risk assessment relevant daily intakes of pesticide residues – either on the market can be regulated available at the EPHC website. The original risk management standard licence applications proposing dealings to EHRA. national theoretical maximum daily through poisons scheduling and was followed by a further standard, involving intentional release (DIR) of intakes (NTMDIs) or national estimated appropriate labelling. The Delegate 17.5.1 Environmental risk management: GMOs into the Australian environment. 17.8.1 dietary intakes (NEDIs) – and these are of the Commonwealth Department of Data requirements principles and process, HB 203: 2000. The Australasian College of Toxicology the responsibility of the APVMA and Health and Ageing, acting on the advice This gave more extensive detail on The RARMP categorises risks (including and Risk Assessment FSANZ. The procedures used are based of its Advisory Committee on Chemicals The data requirements for environmental environmental risk management using potential risks to human health) into the Guidelines for predicting dietary Scheduling (ACCS) recommends assessment of agricultural chemicals the framework established in the previous broad categories (negligible, low, With funding support and intake of pesticide residues (1989) classification of drugs and poisons is set out in Section 7 of the APVMA generic Australian Standard. moderate, high, very high) rather than encouragement from most of the published by the UNEP/FAO/WHO Food which are published in the Standard for Agricultural (and veterinary) manual of making numerical estimates of risk. enHealth partners, the Australasian Contamination Monitoring Programme in the uniform scheduling of medicines requirements and guidelines (Ag-MORAG Both standards provide qualitative College of Toxicology and Risk collaboration with the Codex Committee and poisons (SUSMP); these federal and Vet-MORAG). The requirements measures of consequence and likelihood. Assessment (ACTRA) was established on Pesticide Residues. recommendations are adopted by include studies on toxicity to birds, Appendixes in HB203: 2000 detail 17.8 in 2006 (see
186 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 187 Chapter 18: International programs and guidance on EHRA
This chapter summarises some of a discussion on why US air and water •• The scientific basis for risk assessment The linearised multi-stage model •• RAGS Part C – guidance on the human children. It only allowed this additional the EHRA guidance that has been quality standards may have developed should be detailed along with default using upper-bound estimates initially health risk evaluations of remedial safety factor to be removed where reliable developed in selected US, European differently from those in other parts of options. It was intended that the underpinned US regulatory risk alternatives that are conducted data established that it was not necessary and international programs. It is not the world. guidelines should be flexible and allow assessment of carcinogens. It has been during the feasibility study, during in order to provide protection for infants intended to be a comprehensive review departures from the defaults if there labelled as ‘one of the most conservative selection and documentation of a and children. of these programs, but to provide some One important landmark was a was appropriate data to indicate that models used in QRA’ (IEH 1999b). In remedy, and during and after remedy brief insights into their key elements and Supreme Court decision in 1980, when the default option was not appropriate. 1996, the US EPA proposed changing implementation 18.1.4 how they may have contributed to the an occupational safety and health from the linearised multi-stage approach •• RAGS Part D – guidance on Current guidance development of an Australian approach administration (OSHA) standard for The NRC committee did not recommend to a benchmark dose approach as their standardised planning, reporting and The US EPA has published extensive to EHRA. exposure to benzene in the workplace a specific methodology for risk default model (US EPA 1996a), and this review of Superfund risk assessments was struck down. The policy had assessment but noted there should be change has been reflected in more recent guidance on EHRA in recent years. Most •• RAGS Part E – supplemental guidance been aimed at reducing exposure as opportunities for continuing review of the guidance (US EPA 2005a; NRC 2008). of these have been specifically cited to address human health risk related to 18.1 far as technologically possible but science underlying the guidelines and elsewhere in this enHealth document. dermal exposures UNITED STATES did not consider whether the existing of the associated default options (NRC The development of US EPA risk Listing of the key guidance documents concentration posed a significant risk 1994). The report acknowledged the assessment policies has also been •• RAGS Part F – supplemental guidance in chronological order (other than PROGRAMS to health. The majority of the court critical role of science policy judgements given an introspective flavour, with the to address human health risk related to the RAGS documents listed above) concluded that under the legislation, and that these must be distinguished publication of an internal assessment inhalation exposures. shows the development of guidance The United States has been particularly OSHA could only regulate if benzene from scientific facts. of EPA risk assessment policies and in the management of health risks active in developing guidance on toxicity posed a significant risk of harm. While practices by staff in the risk assessment 18.1.3 associated with pesticides, soil air and water contaminants, carcinogens, and assessment and EHRA methodology. ‘whose significant risk of harm’ was not The Office of Science and Technology task force (US EPA 2004a). Cumulative and aggregate risk reproductive and neurological toxicants: Reference has been made elsewhere in defined, the decision highlighted that Policy brought together scientists assessment guidance this enHealth document (see Sections some form of quantitative risk assessment from regulatory agencies, the National 18.1.2 The passage of the Food Quality 1983 Good laboratory practice 1.8, 3.8, 3.9) to the NRC 2008 proposals was required as the basis for deciding Institutes of Health and other federal Risk Assessment Guidance for Protection Act of 1996 required US standards: toxicology testing to update US approaches to EHRA, and whether the risk was large enough to agencies. This body reviewed the Superfund (RAGS) regulatory agencies to develop specific Pesticide programs: good to the Tox21 Toxicity Testing in the 21st warrant regulation (NRC 1994). scientific basis of risk assessment of risk assessment methodologies that laboratory practice standards Century program (see Section 9.9) which chemical carcinogens and adopted An important stimulus to the development could address exposures to chemicals Technical assistance document seeks to supplement and/or replace Following from this judgement, Congress the framework for risk assessment of EHRA guidance in the US was from multiple environmental sources for sampling and analysis of toxic traditional animal-based toxicity testing instructed FDA to have the National proposed by the NRC. Only the the passing of the Comprehensive (termed ‘aggregate’ risk assessment) organic compounds in ambient air with newer in vitro, QSAR and in silico Research Council (NRC) appraise federal Environment Protection Agency (US Environmental Response, Compensation, and to consider where simultaneous techniques. efforts to use risk assessment in 1981. EPA) adopted a specific set of guidelines and Liability Act of 1980 (CERCLA) 1988 Seven cardinal rules of risk or consecutive exposures to different for carcinogen risk assessment. The US that established the Superfund for communication chemicals in an environmental mixture Current approaches and challenges in Drawing on work done previously by EPA carcinogenic HRA guidelines were contaminated site clean-up. This could result in health risks additional 1989 First version of the Exposure factors US chemicals regulatory policies were the Environmental Protection Agency, initially issued in 1986, and updated in legislation addressed assessment and to, or compounded by, these multiple handbook (updated in 2009) recently outlined by the administrator of the Food and Drug Administration, 2005, after a decade of consultation. remediation of contaminated sites exposures (termed ‘cumulative’ risk the US EPA in legislative hearings on the the Occupation Safety and Health The agency has subsequently gone on to in the US. It established a series of 1990 Compendium of methods for the assessment). These concepts are US Toxic Substances Control Act of 1976, Administration, International Agency for issue guidelines for other adverse health guidance documents, called the RAGS determination of air pollutants in discussed in Chapters 5 and 12. see
188 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 189 1996 Draft revision to the guidelines for Supplemental guidance for health from chemicals and to promote •• the treatment of data gaps and on Health Risks from Chemicals (IGHRC). important milestone in the development carcinogenic risk assessment assessing susceptibility from improved risk assessment decision efficiency The Health and Safety Executive (HSE) of cooperative international programs early-life exposure to carcinogens making. The agencies involved covered has published a number of pamphlets on chemical safety. The conference 1997 Guiding principles for Monte Carlo •• the degree of protection for general a wide variety of risks including those and reports on chemical safety issues, developed a number of themes analysis 2006 A framework for assessing health population exposures compared with from food contaminants and additives, mainly directed at workplace risk for effective global management risks of environmental exposures occupational exposures 1998 National primary drinking water agricultural pesticides, biocides, management. The HSE approach to of chemicals, and established, to children •• the degree of conservatism built into regulations: interim enhanced veterinary products, occupational risk assessment and decision making is under its Agenda 21 Chapter 19 on worst-case exposure estimates (IEH surface water treatment Approaches for the application exposures, consumer products, air summarised in HSE (1999). ‘Environmentally sound management of 1999b) of physiologically based quality, water quality, land quality and toxic chemicals, including prevention of 1999 Compendium of methods for pharmacokinetic (PBPK) models human medicines. As a result of the •• the approaches to the assessment of To date, the IGHRC and its precursors illegal international traffic in toxic and the determination of inorganic and supporting data in risk genotoxic carcinogens. The UK has compounds in ambient air deliberations of the initiative, a four-stage has published 14 reports covering topics dangerous products’, a set of principles assessment process of risk assessment was proposed tended to use a qualitative weight of such as: for developing chemicals management evidence approach to the evaluation 2000 Supplementary guidance for 2007 Guidance for evaluating the oral consisting of: programs, including: conducting health risk assessment of carcinogenic risk and has tended •• general risk assessment methodology bioavailability of metals in soils of chemical mixtures 1. Identifying the properties of chemicals to avoid the use of mathematical •• exposure assessment in EHRA •• expanding and accelerating for use in human health risk that can lead to adverse (toxic) health approaches for quantitatively assessing international assessment of Benchmark dose technical assessment •• weight of evidence effects (hazard identification) risks from genotoxic carcinogens and chemical risks guidance document •• route-to-route extrapolation Users guide for the integrated they are rarely, if ever, carried out •• harmonisation of classification and Methodology for deriving ambient 2. Obtaining quantitative information exposure uptake biokinetic model by UK regulatory agencies. The UK •• uncertainty analysis labelling of chemicals water quality criteria for the about the hazard including, where for lead in children (IEUBK) Department of Health’s Committee •• PBPK modelling protection of human health possible, information on dose– •• information exchange on toxic Considerations for developing a response relationships (hazard on Carcinogenicity did not support •• carcinogenic risk assessment – chemicals and chemical risks 2001 General principles for performing dosimetry-based cumulative risk the routine use of quantitative risk characterisation) the IGHRC document discusses •• establishment of risk reduction aggregate exposure and risk assessment approach for mixtures assessment (QRA) for chemical frameworks for carcinogen 3. Assessing exposure to the chemical programs assessments of environmental contaminants carcinogens (IEH 1999b). assessment while other documents (exposure assessment) •• strengthening of national capabilities Preliminary cumulative on carcinogenicity and mutagenicity 2009 Exposure factors handbook: and capacities for management of risk assessment of the 4. Comparing exposure and hazard Strategies for dealing with variability are published by the UK Committees 2009 update chemicals organophosphorus pesticides information (risk characterisation). within the human population as a result on Toxicity, Carcinogenicity and Recommended toxicity of factors such as age, sex, pregnancy Mutagenicity of chemicals in •• prevention of illegal international traffic 2002 A review of the reference dose and equivalency factors (TEFs) for The initiative described the variety status, health status, lifestyle and food, consumer products and the in toxic and dangerous products. reference concentration processes human health risk assessment of genetic factors were important parts of Guidance on cumulative risk of risk assessment practices used in environment (COT, COC and COM). dioxin and dioxin-like compounds the process. The Intergovernmental Forum on assessment of pesticide chemicals different government departments as 2010 Development of a relative potency Chemical Safety (IFCS) (
190 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 191 •• cooperation with chemicals •• Harmonization Project information The REACH regulation gives greater In the latter category, Santillo and •• Chemical Industry Institute of The mission given to ECVAM by the management programs of the brochure (2nd edition) responsibility to industry to manage the Johnston (2006), writing before the Toxicology – CIIT (see
192 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 193 Appendix 1: Guidance on the evaluation of toxicity studies
The contributions of NICEATM and This appendix, in conjunction with •• produce severe toxic, pharmacological For dermal exposure the material, in a ICCVAM to risk assessment methodology, the general guidance in Chapter 9, is or physiological effects that might suitable vehicle, is applied to the clipped and the challenges still to be faced in intended to provide additional guidance shorten duration of the study or skin of rats, rabbits or guinea pigs. OECD the future, have been summarised by to toxicologists who need to evaluate otherwise compromise the study results Test guideline TG410 recommends even Birnbaum and Stokes (2010). a package of toxicity tests to inform •• alter survival in a carcinogenicity study application to an area representing about an EHRA. in a significant manner due to effects 10 per cent of the total body surface other than tumour production. area. The site is generally occluded 18.5 A1.1 with polyethylene sheeting and gauze INTEGRATION OF Study protocol and design The International Life Sciences Institute patches (or semi-occluded) to prevent (ILSI) Risk Sciences Working Group dislodgment of material and oral ingestion A1.1.1 HUMAN HEALTH AND on Dose Selection has published its by the animal, which could affect the Dosing regimen deliberations on the selection of doses in validity or usefulness of the study. ECOTOXICOLOGY For volatile or semi-volatile materials, The purpose of toxicity studies is to detect chronic rodent bioassays (Foran & ILSI application and covering procedures valid biological evidence for any toxic Risk Sciences Working Group on Dose The assessment of risks to environmental should minimise the possibility of or oncogenic potential of the substance Selection 1997). flora and fauna (ecological risk evaporation. Useful chapters or sections being investigated. Therefore, protocols assessment) is also a part of on dermal toxicity testing may be found should maximise the sensitivity of the test Although it has been argued that environmental health regulation, although in textbooks on toxicology (e.g. Derelanko without significantly altering the accuracy responses observed at doses far in not a primary focus of this enHealth & Hollinger 1995; Hayes 1994), and in and interpretability of the biological data excess of levels experienced under real or guidance on EHRA, which is directed more recent US EPA RAGS-E guidance obtained. The dose regimen has an potential exposure conditions legitimately towards human health risk assessment; (US EPA 2004b). however, there have been calls for better extremely important bearing on these two fall within the classical dose–response integration of ecological and human critical elements. concept, there are valid scientific concerns The surface area of the respiratory health risk assessment. that such doses introduce a further layer of membrane is estimated at approximately Since the determination of dose uncertainty into the already difficult task of 50–100 square metres in the normal responses for any observed effects is one evaluating animal dose responses and the Suter et al. (2005) have reviewed a adult compared with the estimated area of the objectives of repeat-dose studies, assessment of their relevance to human framework developed by the IPCS that of the small intestine at 250 square at least three dose levels are normally hazard identification and risk (Paynter integrates risks to human health and metres (Guyton 1991) and much more required, as well as controls. US EPA 1984). High doses that overwhelm normal risks to non-human organisms and air (about 5,000 times, by volume) is guidelines allow a limit dose of mechanisms for metabolism, detoxification ecosystems. They note that: inhaled each day than food or water is 1,000 mg/kg in chronic and sub-chronic or excretion, or produce severe tissue ingested (McClellan & Henderson 1989). ... WHO’s framework recognises studies. If this dose produces no observed damage (i.e. necrosis, demyelination) Thus, exposure to airborne material is that stakeholders and risk managers toxic effects, and if toxicity is not expected can make interpretation difficult or lead potentially greater than via dermal or have their own processes that are based on data on structurally related to inappropriate conclusions about the oral exposure. Airborne material can parallel to the scientific process of compounds, then a full study using three extent of the hazard. be gases or vapours, liquid droplets or risk assessment and may interact dose levels might not be considered solutions, aerosols (solid and vapour with the risk assessment at various necessary. Ideally, the dose selection With respect to selecting the low dose, components), or dry fibres or powders. points, depending on the context. should maximise the detection of potential it is commonly accepted that the lowest As a consequence, to conduct inhalation Integration of health and ecology dose–response relationships and facilitate dose should not produce any evidence toxicity studies, mechanisms needed to provides consistent expressions of the extrapolation of these to potential of toxicity (i.e. allows the establishment deliver chemicals to a test chamber (in assessment results, incorporates the hazards for other species including of a ‘no observed adverse effect level’ – a form that can be inhaled) are quite interdependence of humans and the humans. The largest administered NOAEL). complex, particularly when coupled environment, uses sentinel organisms, dose should not compromise biological with the need to include measuring and improves the efficiency and quality interpretability of the observed responses. A1.1.2 devices which can establish particle of assessments relative to independent For example, it is generally considered Dosing route size, concentration and form of the human health and ecological risk that the upper dose should not: assessments. The advantage of the For repeat-dose studies, the most material in the exposure chamber. framework to toxicologists lies in the •• cause a body weight decrement from convenient route of administration is by Furthermore, many factors can influence opportunity to use understanding of concurrent control values of greater dietary admixture. However, depending on the inhalation, deposition and retention toxicokinetics and toxicodynamics to than 10–12 per cent the possible route of exposure of the public of inhaled materials in the respiratory inform the integrated assessment of all •• exceed 5 per cent of the total diet in or occupationally exposed workers to a tract. Thus, the conduct of inhalation exposed species. a dietary study because of potential chemical or an environmental contaminant, studies is considerably more complex nutritional imbalances caused at it may need to be investigated by the than equivalent studies by the dietary or higher levels dermal or inhalation route. dermal routes.
194 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 195 Of critical importance, in both the Examples of such response are the usual a histopathological change will depend on 3. The competency and completeness of A1.3.2 ones. Any untreated animal or group conduct and assessment of such studies, respiratory and pulse rate increases the ability of the injured organ or tissue to the conduct and reporting of the study. Analysing and evaluating major may exhibit some signs of abnormality in the need to establish what portion of associated with increased physical repair/regenerate. For example, the liver or drift from the norm for that species 4. The effects of modifying factors study parameters the material delivered to the exposure activity, systemic changes associated with has a relatively great ability to regenerate or strain. Because of the possibility that may result in major inequalities chamber was in a respirable form. In normal pregnancy, and those associated and many types of injury to this organ Not all observed effects of test that statistically significant differences between control and test animals. addition to standard toxicology texts, with homeostatic mechanisms. These are reversible. By contrast, differentiated substances are toxic effects. Rather, between treated and control groups are some useful specific references on variable factors are not important toxicity cells of the central nervous system are not they may be adaptive (e.g. liver enzyme the result of abnormal values among the This qualitative consideration has more to inhalation toxicology include McClellan endpoints in sub-chronic and chronic replaced and many injuries to the CNS induction leading to some hepatic controls, such differences should usually do with the evaluation and interpretation and Henderson (1989), Mohr et al. exposure studies unless their fluctuations are irreversible. enlargement) or may be a manifestation be dose-related and should delineate a of data than with acceptability of (1988) and Salem (1987). are abnormally altered by a dose regimen. of a pharmacological effect (e.g. in an trend away from the norm for that stock documentation. It is placed here because If such alterations occur at a particular animal colony suffering from various of animals, if they are to be indicative of a A1.3 determining the factors that may have dose or are part of a dose–response low-grade infections, an antibiotic true compound-related effect. A1.2 Assessing the quality of the data a major influence on toxicological data relationship, they should be correlated will lower leucocyte counts in treated Study findings – physiological, characterising the hazard needs to be made prior to analysing the with other toxicity endpoints that may animals relative to controls obviously it Historical control data may be useful pharmacological or toxic? data. There are many factors influencing be present. The following considerations address the is not appropriate to describe this as a when evaluating the acceptability of In conducting a hazard assessment, the acceptability of experimental studies and the responses of experimental animals leukopaenic effect of the chemical). the ‘normal’ data obtained from control to experimental treatment; some of evaluator needs to determine whether Pharmacological responses are altered the documentation provided. groups (Haseman et al. 1984; Paynter these are discussed by Doull (1980). effects seen in studies are physiological, physiological functions arising from Concurrent control groups should always 1984; Sumi et al. 1976; Tarone 1982). 1. The adequacy of the experimental Some influences may be quite subtle, pharmacological or toxic. interaction of a substance with a cellular be used notwithstanding the value of Any departure from the norm by the design and other experimental as exemplified by studies performed by receptor site. They are reversible, and historical control ranges in carconogenicity control groups should be taken into parameters, including: the Thompson et al. (1982), in which it was Responses produced by chemicals in are usually of limited duration following studies. It is generally not appropriate consideration, especially during the appropriateness of the observational noted that the onset of acute pulmonary humans and experimental animals may removal of the stimulus. While some of to rely on statistical comparisons with conduct of any statistical analysis. The and experimental methods; oedema in rats being used in immune differ according to the quantity of the these responses may be undesirable historical controls since the incidence of finding of consistent departures from the frequency and duration of exposure; hypersensitivity studies was sudden and substance received and the duration under certain circumstances, they spontaneous lesions can vary significantly norm in control groups may necessitate appropriateness of the species, strain, seasonal. Circadian rhythms and seasonal over time (and even between concurrent and frequency of exposure, for example, are distinguished from toxic (adverse) investigation of the source of the animals. sex and age of the animals used; the physiological variations can subtly randomised control groups). Controls responses to acute exposures (a single responses by generally not causing injury. numbers of animals used per dosage influence experimental results. Such must be age-matched because some exposure or multiple exposures occurring An example of this type of response Ideally, historical control data should group; justification of dose, route factors influencing animal responses forms of toxicity represent no more than within 24 hours or less) may be different is the increased activity of the hepatic be taken from the same laboratory and frequency of dosing; and the can be troublesome when their effects acceleration and/or enhancement of age- from those produced by sub-chronic cytochrome P-450 containing mono- undertaking the study, collected using conditions under which the substance are confused with or misinterpreted as related changes. Examples of pathological and chronic exposures. Not all observed oxygenase systems (enzyme induction) the same strain, age and sex of animals was tested. toxic responses to treatment. For further changes in aged rats that may be affected responses within a study, irrespective caused by exposure to many pesticides, obtained from the same supplier, and discussion of environmental effects on by compound administration include of exposure duration or frequency, will industrial chemicals and drugs (noting, 2. There are many guidelines to the only include those studies conducted experimental parameters see Herrington chronic progressive glomerulonephropathy, represent toxicity per se. They may however, that while not a direct adverse generation of scientifically valid data within a 2–3-year span on either side of and Nelbach (1942). peripheral nerve degeneration, amyloidosis encompass a range of effects from effect, a cytochrome P-450 inducer can, that concern good experimental the study under review, for retrospective and various neoplasms. physiological through pharmacological for example, alter hormonal homeostasis design, laboratory practice and studies. Important information to The acceptability of reports and other and toxic manifestations. and effect tumour promotion, or increase reporting, such as OECD and US EPA consider includes identification of study technical information is primarily Using both non-treated and vehicle- an organism’s susceptibility to other guidelines, and accepted codes of methodology, which could have affected a scientific judgement. Therefore, control groups helps to assess effects due Although it sometimes may be difficult to chemical exposures). good laboratory practice (GLP) (OECD the results, such as pre-sampling to vehicle or excipients. When a vehicle make a clear distinction between these 1982; US EPA 1983). They can be the rationale for rejecting a hazard conditions (e.g. fasting or non-fasting, assessment study should be succinctly is used to deliver the doses of the agent responses, an attempt to do so should Toxic responses may be reversible useful as aids in determining report assay methodology for study parameters, stated in the evaluation document. being investigated (e.g. a lipophilic agent be made. If an evaluator is uncertain of or irreversible, but are distinguished and data acceptability. However, the histopathological criteria for lesion delivered in corn oil), the need for vehicle- the type or the biological significance of from other types of responses by being evaluator needs to make a judgement identification, time of terminal sacrifice). treated controls is paramount. Since some a response, they should not hesitate to injurious and therefore adverse and about how well the study, in its totality, A1.3.1 parameters can be affected by animal obtain competent advice for resolving the harmful to living organisms or tissues. facilitates the identification of potential Analysing and evaluating toxicity Weil and McCollister (1963) analysed handling (e.g. serum ALT was raised in uncertainty. It is essential that all relevant adverse effects, or lack thereof, of the studies toxicity endpoints, other than A chemical that causes a physiological mice that were grasped around the body toxicity endpoints (statistically and/or substance being evaluated, rather than oncogenicity, from short- and long-term or pharmacological effect may produce Useful guidance documents for compared with unhandled or tail-handled biologically significant) be identified for how precisely it fits a prescribed test tests and concluded that only a relatively a toxic response if the exposure is evaluating data and conducting mice, see Swaim et al. 1985), control consideration when evaluating data for the guideline or ‘recipe’. The experience small number of endpoints were effective prolonged or if the dose is increased assessments include the IPCS animals should be treated in the same presence or absence of non-toxic levels. of senior evaluators can be helpful in in delineating the lowest observed adverse beyond a certain level. Environmental health criteria (EHC) way as test animals. resolving concerns about acceptability monographs, namely EHC 6, 70, 104 and effect level (LOAEL) in such tests. Body Physiological responses vary within of study conduct and/or reporting. weight, liver weight, kidney weight and The reversibility or otherwise of such 141 (WHO 1978; 1987; 1990; 1992). Control animals must receive as much limits that are in accord with the liver pathology delineated the LOAEL responses may also depend on these two attention during the analysis and normal functioning of a living organism. in 92 per cent of test chemicals in factors. The reversibility or irreversibility of evaluation process as do the treated
196 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 197 sub-chronic studies and 100 per cent in An evaluation of mortality patterns within the expected average life span for an Useful information on gross behavioural This effect on body weight is often quick identification of any unusual or chronic studies. To reach 100 per cent each group may indicate that mortality animal species/strain. To date, regulatory observations in laboratory animals and evidenced during the first two or three sudden changes in gain or loss by any efficiency in sub-chronic studies, renal is clustered early or late in the course of authorities have not come to any decision abnormal behaviour patterns can be weeks of the study. Sometimes animals group. and testicular histopathology had to be the study, is intermittent and scattered on recommending restricted diets versus found in Bayne (1996). in the affected groups are able to included. Heywood (1981) surveyed the throughout the duration of the study, or ad libitum feeding in toxicity study accommodate to the diet and a gradual A1.3.6 toxicological profiles of 50 compounds has a higher incidence in one sex than guidelines. Some useful references on The revised OECD test guidelines for increase in group weight gain will occur Haematological, clinical chemistry in rodent and non-rodent species and in the other. The analysis of the cause of this topic include Keenan et al. (1998), 90-day oral toxicity studies in rodents (Nolen 1972). In sub-chronic studies, and urinary measurements confirmed these observations. individual deaths will aid in determining Klinger et al. (1996), Masoro (1992) and and non-rodents (TGs 408 and 409) the lag in group weight gain may persist, the toxicological significance of these Thurman et al. (1995). have placed additional emphasis on even though the individual animal Haematological, clinical chemistry A1.3.3 various patterns. Early deaths within neurological endpoints, that is, studies gains per gram of food consumed (food and urinary parameters are routinely Mortality/survival treated groups may just reflect deaths of A1.3.4 should allow for identifying chemicals utilisation efficiency) are favourable after measured in sub-chronic and chronic the more susceptible animals in the test Clinical observations with the potential to cause neurotoxic the accommodation, and produce a toxicity studies compliant with regulatory Reasonable efforts should be made to population. Alternatively, it may indicate effects, which may warrant further statistically significant difference between guidelines. determine the cause or likely cause of changes in compound intake per unit Adverse clinical signs (gross observations) in-depth investigation. The reader the affected group and the concurrent individual deaths. The evaluation of body weight, in those experiments in noted during the exposure period may is referred to the references cited controls that is not related to toxicity of Normal biological variation in inter-animal pathological lesions or morphological which the quantity of test substance in correlate with toxicity endpoints or in Test guideline 408 relating to the test substance (McLean & McLean values and their alteration in response to a changes in belatedly observed deaths the diet is kept constant. Relative to body disease processes. These can be used as neurotoxicity assessment, including 1969). Sometimes the addition of the variety of inputs means that evaluators will is frequently complicated by post- weight, young rats ingest more food than supportive evidence for dose–response sensory reactivity to stimuli of different test substance will interact with one have to contend with much ‘noise’ in this mortem autolysis. The separation of older rats and therefore ingest relatively relationships and may play a role in types (auditory, visual, proprioceptive), or more essential nutritional elements area, and will frequently be presented with deaths caused by factors unrelated to more of the test substance. Early deaths determining the NOEL/NOAEL. However, grip strength, and motor activity. in the diet, thereby producing weight scattered, statistically significant effects in exposure to the test agent (e.g. acute or may therefore be the result of the higher not all adverse clinical signs will correlate The OECD promulgated Test guideline gain decrements or alterations of toxic the absence of any evidence of clinically chronic infections, age or disease-related exposure, on a mg/kg/d basis, of young with pathological or morphological TG 426 in 2007 to address neurotoxic responses (Casterline & Williams 1969; significant relationships to specific toxicity degenerative processes, anatomical animals compared with older animals. changes in organs or tissues. Some will effects during the developmental period Conner & Newbern 1984; Rogers et endpoints. To deal with ‘noise’ there is a abnormalities, negligent handling or be caused by biochemical or physiological in newborn animals. al. 1974). This phenomenon may be need to examine whether the effect noted accident) from toxicity-induced deaths is effects, such as incoordination, muscle Deaths that are clustered at a specific encountered in sub-chronic studies and, is within the normal range of variation important. All data relating to moribund twitching, tremor or diarrhoea, which may time period may reflect a spontaneous, A1.3.5 when identified, can usually be overcome (concurrent and historical controls). Some or dead animals during their study life, indicate acetylcholinesterase inhibition epidemic disease situation of limited by acceptable means before a chronic of these parameters can vary significantly as well as the results of post-mortem without any morphological changes being Body weight changes, food and duration. High mortality associated with study is initiated. Infrequently, control with no clinical manifestations but others examinations, should be scrutinised in evident in nervous tissue. water consumption infectious agents in treated groups, in values for weight gain (at one or more (e.g. serum potassium) have a very narrow an attempt to make this distinction. Note the absence of such evidence in the Body weight changes (gains or losses) time points) can be low, causing the other normal clinical range and small differences that US EPA guidelines state that the Many of these qualitative signs can concurrent control group, could indicate for individual animals and groups of value to appear unusually high. can be important. highest dose used in sub-chronic studies be counted, scored for intensity and an immunosuppressive action of the animals when compared with concurrent with non-rodents should not produce an tabulated as incidences. However, chemical being tested. control changes during the course of a Diet composition, food and water Frequently this data shows apparently incidence of fatalities that would prevent statistical analysis is of limited value. study are a criterion of some importance consumption, and body weight gains per ‘random’ changes in individual groups meaningful evaluation. The evaluator must rely on the number The effect of dietary intake on mortality (Heywood 1981; Roubicek et al. 1964; se can also have an important influence or, less commonly, non-dose-related of individuals per group exhibiting signs of needs to be considered. A compound Weil & McCollister 1963). Such changes on many aspects of animal responses trends in changes across several groups. Analysis of mortality requires more than a particular type, as well as the intensity administered in the diet may make the are usually related to food intake, and including shifts in metabolic, hormonal If using historical control data as an aid a statistical treatment of incidence at of the responses, to gain an impression laboratory food more or less palatable, analysis of one without an analysis of and homeostatic mechanisms (Kennedy to evaluation, only values produced by termination of a study. Survival/mortality of a dose–response relationship. may have a pharmacological stimulant the other is of limited value. Weight 1969) as well as disease processes (Berg the identical methods from the same data can be influenced by factors other or depressant effect on appetite, or may decrement may not always be related & Simms, 1960; Paynter 1984; Ross laboratory are valid in such comparisons. than the test substance. Changes in the Clinical observations, such as palpable affect the partitioning of the nutrients to toxicity per se (Seefeld & Peterson & Bras 1965; Tannenbaum 1940) and Literature values for normal ranges that do protocol during the course of a study tumours or those that might be associated in the food. Likewise, decreased water 1984). Occasionally the incorporation maturation (Innami et al. 1973). These not specify the method by which they were can complicate the analysis, such as with neoplasia (e.g. haematuria, consumption (e.g. in the case of an of the test substance into the diet variables and should be considered obtained should be used with caution. alterations in dosage levels can produce a abdominal distension or impaired unpalatable compound administered will reduce the palatability of the diet when unusual effects are observed in confusing mortality pattern. respiration), may be useful in defining in the water) will lead to reduced to many individuals in all treatment the absence of any indication of injury to A good review of factors that can the time a tumour was first suspected food consumption. These effects may groups or to the majority of individuals organs and other vital systems. complicate the interpretation of findings Any unusual mortality pattern should as being present. Such signs might help significantly influence longevity since it in the higher dietary level groups. in a toxicity study may be found in be explained by the test laboratory on evaluate decreased tumour latency in has been clearly shown in animal species Food spillage needs to be considered Evaluating body weight changes and the Handbook of toxicologic pathology biological or toxicological grounds. If overall long-term rodent studies. They may that long-term dietary restriction very when evaluating food palatability attendant effects is significantly aided by (Haschek et al. 2002). mortality is high (i.e. significantly greater also aid in determining cause of death. and compound intake. The same than expected for the particular colony and significantly increases life span (e.g. A statement of the correlations, or the the graphical presentation of group mean considerations apply if the compound is To gain maximum information from strain) for any repeat-dose study, or for a Tucker 1979). Conversely, excessive ad lack thereof, between clinical signs and body weights and food consumption administered in drinking water. enzyme determinations it is important to particular group within a study, a credible libitum intake of highly nutritious diets specific toxicity endpoints should be versus compound consumption (on a consider the most appropriate enzymes. explanation should be provided. can reduce life span compared with made in the evaluation. mg/kg body weight basis). This allows
198 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 199 It is important that organ distribution mouse and rat, whereas in primates, the long-duration studies. Urinalysis should be conducted A common problem in interpreting A1.3.8 and location of the enzyme in the cell is ALT is present in heart, skeletal muscle For repeated-dose studies in non- at least once during a study. For study findings is the misinterpretation Post-mortem observation known. ALT (alanine aminotransferase) and liver. Plasma alkaline phosphatase rodent species, clinical pathology routine urinalysis, an overnight of relative organ weight changes. For is found in greatest concentration in measurement has been less useful in testing is recommended at study collection (approximately 16 hours) example, an increase in relative brain Although much progress has been the liver in rats, even more so in dogs. detecting liver cell necrosis in the dog, termination and at least once at an is recommended. The core tests weight in a toxicity study in which the made in standardising nomenclature, AP (alkaline phosphatase, or ALP) is sheep, cow and rat but may be indicative earlier interval. For studies of two to should include an assessment of urine chemical causes a significant body to minimise any difficulties in this area, virtually absent from the liver in these of other types of liver damage, particularly six weeks in duration in non-rodent appearance (colour and turbidity), weight loss or reduced body weight gain an experienced pathologist will describe two species, being mainly confined to the those of a cholestatic nature in a number species, testing is also recommended volume, specific gravity or osmolality, may not be indicative of a toxic effect on each significant lesion type, at least once, kidney, intestine and bone. CPK (creatine of species. It is evident that species within 7 days of initiation of dosing, pH, and either the quantitative or the brain as the brain would be spared in such detail that another competent phosphokinase, or CK) is mainly located differences are of great importance when unless it compromises the health of the semi-quantitative determination of total under conditions leading to reduced pathologist can perceive a mental picture in skeletal and heart muscle, while AST specific clinical chemistries are selected animals. If a study contains recovery protein and glucose. body weight. Similarly, the relative weight of the lesion and form a judgement as (aspartate aminotransferase) is found in for inclusion in toxicity studies. groups, clinical pathology testing at of other organs may change because to its relevance to the histopathology For carcinogenicity studies, only various concentrations in most organs. It study termination is recommended. of changes in body weight rather than induced by the chemical being tested. blood smears should be made from is clear that CPK is the most appropriate When analysis and evaluation of as a result of a specific compound The core haematology tests unscheduled sacrifices (decedents) enzyme to detect muscle damage, while clinical data indicate a dose–response effect. Tables of the relationship of To assist in the uniform description recommended are total leukocyte and at study termination, to aid in the changes in ALT would probably reflect relationship or a biologically important relative organ weights to various levels of pathologies, a series of articles on (white blood cell) count, absolute identification and differentiation of some liver necrosis. Although AST is not drift from concurrent control values, the of reduced bodyweights (produced pathology nomenclature have been differential leukocyte count, erythrocyte haematopoietic neoplasia. organ-specific, it serves to confirm organ effects observed should be correlated by dietary restriction) for rats may be published, under the title Standardized (red blood cell) count, evaluation of damage, especially for muscle and liver, with other manifestations of toxicity. found in Sharer (1977). Changes in system of nomenclature and diagnostic red blood cell morphology, platelet if its activity changes in parallel with other The evaluator should indicate whether or A1.3.7 organ weight/body weight ratios as a criteria guides for toxicologic pathology (thrombocyte) count, haemoglobin enzymes. In dogs, AP is a sensitive test not a correlation could be established. Absolute and relative organ weights physiological response of the organism to by the US Society of Toxicologic concentration, haematocrit (or packed for biliary function, but in the rat it is of decreased nutrient intake and markedly Pathologists (STP), in cooperation with cell volume), mean corpuscular It is generally considered that little diagnostic value since it is absent Standard veterinary (e.g. Bush 1991; reduced growth must be differentiated the Armed Forces Institute of Pathology volume, mean corpuscular histopathology is more sensitive for from the liver and principally derived Duncan et al. 1994; Evans 1996) and from organ weight changes resulting from (AFIP) and the American Registry of haemoglobin, and mean corpuscular establishing the lowest dose producing from the intestines. For hepatocellular human clinical manuals (e.g. Henry primary toxic effects of the compound Pathology (ARP). haemoglobin concentration. In the an effect than are organ or body weight evaluation, ALT, AST, SDH (sorbitol 1984; Tyson & Sawhney 1985; Walach being tested. absence of automated reticulocyte changes. Organ weights are usually dehydrogenase) and GLDH (glutamate 1996) should be consulted for information Age-associated, especially geriatric, counting capabilities, blood smears reported as absolute organ weights and dehydrogenase) are the most appropriate, about laboratory diagnostic tests and The interpretation of organ weight changes can have an extremely from each animal should be prepared as relative organ weights (relative to body while for hepatobiliary evaluation, AP, to assist in the evaluation of potential changes must not be made solely on the important effect on histopathology, as for reticulocyte counts. Bone marrow weight and/or brain weight). Relative 5’-nucleotidase, GGT (gamma-glutamyl correlations between clinical chemistry, determination of a statistically significant well as clinical chemistry, metabolic and cytology slides should be prepared organ weight comparisons are used transferase) and total bilirubin are the haematological, and urinary data with difference between the concurrent pharmacokinetic parameters (Grice & from each animal at termination. since body weights are often affected most appropriate measurements. It is adverse effects. control value and a treatment group Burek 1983; Mohr et al. 1992; 1994; Prothrombin time and activated partial by compound administration. important to understand that many of value. A proper evaluation will also 1996) and therefore, important overt, and thromboplastin time (or appropriate these types of serum enzyme tests and The deliberations of a joint international include consideration of any correlation frequently subtle, influences on observed alternatives) and platelet count are the Experimentally controllable and urinalysis fail to detect minor injury or committee, established to provide between organ weights (absolute and physiological, pharmacological and toxic minimum recommended laboratory uncontrollable factors, such as circadian may reflect only transient or reversible advice for clinical pathology testing relative), histopathological and metabolic/ responses during the latter part of any tests of haemostasis. The core clinical rhythms, food intake, dehydration, changes. Therefore, evaluation and of laboratory animal species used in pharmacodynamic data. long-term study. It is essential in all cases chemistry tests recommended are nature of the diet, age of animals, interpretation of the test results must regulated toxicity and safety studies, where spontaneous and/or age-associated glucose, urea nitrogen, creatinine, organ workload, stress, and method of be performed carefully and correlated has published its recommendations, Since the evaluation of organ weight lesions are present to differentiate total protein, albumin, calculated killing, have an influence on organ and with more specific, sensitive and reliable including those parameters that should changes is such an integral part of between such lesions and treatment globulin, calcium, sodium, potassium, body weights and the variability of such histopathological findings. be measured (Weingand et al. 1996). toxicity assessment, the US Society of induced lesions. Grice and Burek (1983) total cholesterol and appropriate data. A review of this subject by Weil While these recommendations have not Toxicologic Pathology (STP) has made and Benirschke et al. (1978) are very hepatocellular and hepatobiliary (1970) should be consulted. The most Sensitivity and specificity of the enzyme been formally incorporated into national recommendations regarding the protocols helpful in this respect, as is advice from a tests. For hepatocellular evaluation, important influencing factor appears to changes as diagnostic of organ pathology or international guidelines at this stage, for weighing organs and interpreting competent and experienced pathologist. measurement of a minimum of two be the method of killing and the timing are greatly influenced by the species they are noted, as follows: the findings in the light of the class of scientifically appropriate blood tests of necropsy. The killing method used not selected for testing (Tyson & Sawhney compound under test, the MoA for toxicity An overview of factors (physiological, For repeated-dose studies in rodent is recommended, such as ALT, AST, only affects the appearance of the tissue, 1985). For example, in mammalian and the combined dataset for the study environmental, etc.) that can complicate species, clinical pathology testing is SDH, GLGH or total bile acids. important in describing gross necropsy species, aspartate transaminase is (Sellers et al. 2007). the interpretation of morphological necessary at study termination. Interim observations, but also, in conjunction not specific to any tissue and thereby For hepatobiliary evaluation, measuring findings in a toxicity study may be found study testing may not be necessary with the timing of necropsies, may cause elevated plasma AST activity may suggest a minimum of two appropriate blood in the Handbook of toxicologic pathology in long-duration studies provided post-mortem shifts in organ weights damage to any one of many tissues. In tests is recommended, such as AP, (Haschek et al. 2002). that it has been done in short- (Boyd & Knight 1963). contrast, alanine transaminase is relatively GGT, 5’-nucleotidase, total bilirubin duration studies using dose levels not specific to the liver in the cat, dog, ferret, or total bile acids. substantially lower than those used in
200 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 201 A1.4 of pregnancies resulting in live litters) outlines the protocol for a standard in risk assessment (NRC 1994) A1.6 Data obtained from studies carried out Analysing and evaluating study viability index (percentage of pups that developmental or ‘teratology’ study. has useful sections on the impact General comments many years ago should not simply be parameters in acute, developmental, survive 4 or more days) and lactation of pharmacokinetic information in dismissed out of hand simply because index (percentage of pups alive at four A1.4.4 risk assessment. If possible, compound-related changes they do not meet today’s standards. Such reproductive and special toxicity in biochemical, haematological or studies days that survived to day 21 (i.e. weaning) Special studies studies may still provide some useful gross necropsy and histopathology on A1.5 urinalysis parameters should be linked information if an experienced toxicologist Different classes of chemicals may with organ weight, gross pathology or A1.4.1 some parents (sires and dams), with Statistical tests is able to evaluate the study using a Acute toxicity studies attention paid to the reproductive organs require special toxicology studies that histopathological changes. weight-of-evidence approach. This is a and gross necropsy on weanlings. are not part of the ‘standard’ package The objective of a toxicology study is matter for scientific interpretation and Important endpoints in acute toxicity Guidelines and procedures for assessing of studies. For example, it is common to to demonstrate responses of biological The following points also should be noted judgement on a case-by-case basis. studies are clinical signs, gross necropsy these endpoints are well documented test organophosphate (OP) pesticides importance. Where statistical analyses in evaluating repeat-dose toxicity data. signs and mortality incidence; each of (see e.g. Korach 1998). for their ability to cause delayed are used in the judgement process, an these endpoints are discussed in detail neuropathy by conducting tests in hens awareness of the validity of the test and Findings should be considered on the in Sections A1.3.4, A1.3.8 and A1.3.3 A1.4.3 (OECD TG 419), since this species the degree of certainty (confidence) basis of both statistical significance respectively. Since the purpose of acute is especially sensitive to inhibition of pertaining within the context of the study and likely biological significance. The Developmental toxicity studies toxicity studies has moved away from neuropathy target esterase (NTE) by should be demonstrated. variability of biological data must be establishing a strict, quantitative number The critical endpoints in developmental OPs. Further in vitro and in vivo studies remembered in assessing a statistically for the median lethal dose to an estimate toxicity studies relate to developmental may be conducted to resolve possible There are limitations associated with significant result. Conversely, a finding of the likely toxicity range, the emphasis effects in utero, including death, mechanisms for toxic effects seen in the using statistics in toxicology (Gad & that is not statistically significant may have is more on clinical signs and gross organ malformations, functional deficits and standard toxicology tests. Because of the Weil 1986): biological significance when considered pathology than on mortality. developmental delays in foetuses. wide range of types of studies that may in the light of the likely toxicological or •• Statistics cannot make poor data be classified in this category, it is not pharmacological action of the compound, better. A1.4.2 Parameters assessed include: number of possible to comment on the assessment or when combined with results from other Reproductive toxicity studies live litters; number of live foetuses/litter of particular endpoints, but the evaluator •• Statistical significance may not imply studies. Thus, evaluators should note (total and by sex); sex ratio of foetuses; should apply sound scientific judgement biological significance. trends or transient changes in parameters Section A1.3 describes the major study foetal body weights; litter weights; in reviewing these studies. •• An effect that may have biological if there is an indication that these may be parameters to be considered in repeat- number and percentage of foetuses with significance may not be statistically related to dosing with the compound in dose toxicity studies; these endpoints malformations; number and percentage A1.4.5 significant. some way. This information may be useful may also apply to the sires and dams in of litters with malformations; number and Toxicokinetic and metabolism data when comparing results across studies developmental toxicity studies. However, •• The lack of statistical significance does percentage of foetuses with variations; not prove safety. and in considering the overall significance the critical endpoints relate to potential number and percentage of litters with Toxicokinetic (absorption, distribution or relevance of an observed effect (i.e. in toxic effects on reproductive parameters, variations; number and percentage of and elimination) and metabolic data The importance and relevance of any one study an effect may be only a trend including effects on mating behaviour foetuses/litter with malformations; number on the handling of the substance in effect observed in a study must be while in another study it may be very (both sexes), on fertility (both sexes), the and percentage of foetuses/litter with the test species can be very useful in assessed within the limitations imposed clearly treatment-related). implantation of blastocysts, embryonic variations; and types of malformations evaluating and interpreting sub-chronic by the study design and the species and foetal development and survival, and variations. and chronic exposure study data, being studied. A difficult problem for evaluators is when parturition, lactation and postnatal as discussed by Paynter (1984) and studies have significant defects in design survival and development. Thus, a Guidelines and procedures for soft references cited therein. If statistical tests have not been used, or reporting, such that they should be plethora of reproductive parameters tissue and skeletal examination are well if inappropriate tests appear to have been considered seriously flawed. The use need to be assessed in one or more documented (e.g. Tyl & Marr 1997). References in this paper also discuss of a flawed study reporting either a generations, depending on whether the dose-dependent effects in the absorption used, or if tests not commonly employed have been used, then this should be positive or negative outcome may provide study is a one-generation (OECD TG Additional reproductive parameters process and in biotransformation noted and action taken such as data a false sense of security, especially if 415), two-generation (TG 416) or three- assessed include number of females interactions (Levy 1968), the potential re-analysis. there is pressure to consider it because generation test. Important endpoints to pregnant; number of corpora lutea/ difficulties presented by impurities, the no valid alternative study exists, and assess within each generation include: dam; number of implants/dam; and overloading of detoxification mechanisms A number of textbooks and papers on the the study cannot be replaced with a time after pairing to mating; mating number and percentage of pre- (Munro 1977) and other important application of statistics in experimental valid or unflawed study. It is a matter of behaviour; percentage of females implantation loss/litter. While the dosing experimental considerations (Dayton & toxicology and the life sciences are judgement as to how much weight should pregnant; number of pregnancies going to period in a standard teratology study Sanders 1983). available these include Dickens and be accorded to a flawed study, and this full term; litter size; number of live births; commences after mating, conception in turn requires an element of experience number of stillborns; pup viability and and commencement of implantation, it A number of toxicology textbooks Robinson (1996), Gad and Weil (1986), Gad and Weil (1989) and Lee (1993). in evaluating toxicity studies to make weight at parturition, and postnatal days is appropriate to check these parameters include chapters on pharmacokinetics a determination of what type of flaws 4, 7, 14 and 21 days of age the fertility to see that the study has not been and toxicology assessment (e.g. seriously compromise the study outcomes. index (percentage of matings resulting in compromised by factors other than the Sharma & Coulombe 1996). The pregnancy) gestation index (percentage compound under test. OECD TG 414 publication, Science and judgement
202 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 203 Appendix 2: Extracts from a 2010 NHMRC discussion paper on health-based targets for microbial safety of drinking water supplies
The Australian drinking water guidelines •• mechanisms for setting performance The DALY per case is then determined (ADWG) provides the authoritative targets for different types of source water by multiplying the severity by the reference and benchmark on drinking •• mechanisms for setting performance persistence and percentage occurrence BOX 5: water quality for Australian drinking targets for urban and non-urban of each outcome. As shown in Box 5, Disability-adjusted life years water suppliers, health authorities and supplies. the DALYs per case for illness caused by consumers. It is therefore essential Cryptosporidium is 10 times less than The basic principle of the DALY is to weight each health •• severe diarrhoea (severity rating of 0.23) lasting seven that the scientific content of the illness caused by Rotavirus. The benefit of such an approach is impact in terms of its severity within the range of 0 for good days in 2.5 per cent of cases guidelines is maintained and represents that it provides a process for identifying health to 1 for death. Severities for outcomes of microbial •• rare deaths of very young children in 0.015 per cent of best practice. The ADWG was at the appropriate barriers for producing safe A2.2 infection include: cases (a death at under 1 year of age means a loss of up forefront of introducing a comprehensive drinking water. The 2006 WHO GDWQ Microbial safety and performance risk-management-based approach uses a combination of quantitative (treatment) targets •• 0.02–0.12 for mild diarrhoea to 80 years of life). to assure drinking water quality. microbial risk assessment (QMRA) and •• 0.21 for reactive arthritis This process, which started in the A key advantage of applying a quantitative The DALY per case then = (0.1 × 3/365 × 0.975) + the metric of disability-adjusted life years •• 0.23 for severe diarrhoea late 1990s, culminated in publication (DALYs) to define microbial safety. metric for assigning public health impacts (0.23 × 7/365 × 0.025) + of the Framework for management is that it can then be used to define safety. •• 1 for death. (1 × 80 × 0.00015) of drinking water quality in the 2004 QMRA is applied to determine the In the GDWQ, WHO defines microbial = 0.0008 + 0.0001 + 0.012 The severity is then multiplied by duration of the effect and edition of the guidelines. likelihood of infection and illness safety as water that does not give rise to the relative frequency of occurrence in those who become = 0.013 occurring from exposure to specific more than 1 DALY per 1 million people per year (10–6 DALYs). This is equivalent ill. In the case of death, duration is regarded as the years However, the ADWG has not matched pathogens contained in water, while Infection with Cryptosporidium can cause watery diarrhoea to about one case of diarrhoea per 1,000 lost in relation to normal life expectancy. international developments in setting DALYs are used to convert the likelihood (severity weighting of 0.067) lasting for seven days with health-based performance targets for of illness into impacts or burdens people per year. This value is considered Hence, DALYs = YLL (years of life lost) + YLD (years lived extremely rare deaths in 0.0001 per cent of cases. This achieving microbial safety. These targets of disease. The advantage of using to be the upper level of tolerable risk. with a disability/illness). equates to a DALY per case of 0.0015. are required to provide measurable DALYs is that the metric recognises benchmarks for effective risk that not all pathogens cause the same Having established the benchmark, the In this context, ‘disability’ refers to conditions that detract Campylobacter can cause diarrhoea of varying severity, management systems. level or severity of disease. Some like GDWQ shows how to use it to identify from good health. In the context of water-related guidelines, Guillain-Barré syndrome of varying severity, reactive Cryptosporidium cause mild diarrhoea in levels of treatment (performance targets) it generally relates to illness but in other areas it can also arthritis and occasional deaths. The calculated DALY per A2.1 the general population while others such to produce safe drinking water from raw case is 0.0046. water. Setting performance targets is a relate to physical or mental impairment. Quantitative microbial risk assessment as E. coli 0157 and Rotavirus can result in death. In the Walkerton incident, seven four step process: Based on DALYs per case the impacts of the three Performance targets derived from Using an Australian example, infection with Rotavirus people died from an outbreak involving E. 1. Determining or estimating pathogens is Rotavirus > Campylobacter > Cryptosporidium quantitative definitions of microbial safety causes: coli 0157 and Campylobacter (Hrudey & concentrations of pathogens in the are a common theme in guidelines and Hrudey 2004). Severity of consequence is source water •• mild diarrhoea (severity rating of 0.1) lasting three days Adapted from: NRMMC, EPHC, AHMC 2006. DALYs per case standards developed by WHO, US EPA, an important consideration in developing in 97.5 per cent of cases based on Havelaar and Melse 2003 and WSAA 2004. Health Canada and New Zealand where risk management systems and the 2. Use of a quantitative microbial risk they are used to identify appropriate greatest attention should be paid to assessment to determine the extent barriers for ensuring safety of drinking hazards that can cause the largest harm. of infection and illness arising from water supplies. The Australian guidelines these pathogens •• waterborne transmission established There are a range of micro-organisms A2.3 for water recycling (NRMMC, EPHC, DALYs represent the sum of time with an 3. Translating the frequency of illnesses as a route of infection that could be selected as reference NHMRC 2006) including the module on illness (i.e. loss of time in good health) Calculating health-based targets to burdens of disease and DALYs pathogens including Campylobacter, Augmentation of drinking water supplies and years lost through premature death. •• sufficient data available to enable E. coli 0157, Salmonella, Shigella, Calculating health-based targets for (NRMMC, EPHC, NHMRC 2008) has DALYs take account of each of the 4. Calculating required reductions in a quantitative risk assessment to Rotavirus, Norovirus, Giardia and source water containing 1 organism per adopted the WHO definition of microbial possible outcomes associated with a pathogen concentrations to comply be performed including data on Cryptosporidium. Selecting reference litre of Cryptosporidium, Campylobacter safety to underpin the setting of microbial particular pathogen, for example: with the reference level of 10–6 DALYs occurrence in source waters, dose pathogens should take account of local or Rotavirus per litre is illustrated in Table targets for pathogenic bacteria, viruses per person per year. responses in humans and disease •• watery diarrhoea, bloody diarrhoea, conditions, including the prevalence of A1. The calculations are based on the and protozoa. burden haemolytic uraemic syndrome and waterborne transmission and source consumption of 2 litres of water per person It is not practical, nor is there sufficient •• a relatively high occurrence in source death from infection with E. coli 0157 water characteristics, and may vary per day. The probability of infection from This paper canvasses the inclusion of a data, to develop health-based targets waters between different countries and regions. single organisms is based on QMRA using similar approach in the ADWG. Issues •• gastroenteritis, reactive arthritis and for all microbial pathogens. The •• a relatively high survival in the In the GDWQ, Campylobacter, Rotavirus published data largely derived from human discussed are: Guillain-Barré syndrome from infection recommended approach is to use environment and Cryptosporidium are used as dosing experiments (Haas et al.1999; with Campylobacter. reference organisms representing the •• a definition of microbial safety •• a relatively high infectivity and severe reference pathogens to illustrate the Messner et al. 2001). DALYs per case major groups of pathogens (i.e. bacteria, were calculated as shown in Box 5. The •• the principle of setting health-based Each of the outcomes is assigned a disease burden method for calculating health-based viruses and protozoa). The selection required reductions range from 4.3 log for performance targets for producing safe severity ranging from 0.02 to 0.12 for targets (WHO 2006a). criteria for reference pathogens include: •• a relatively low sensitivity to removal or Campylobacter to 5.5 log for Rotavirus. drinking water supplies mild diarrhoea, and up to 1 for death. inactivation by treatment processes.
204 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 205 Table A1: Calculation of pathogen reductions for a source water containing 1 organism per litre of Cryptosporidium, Rotaviruses The ADWG has been at the forefront of The Australian guidelines for water and Campylobacter1 international guidelines and standards in recycling (NRMMC, EPHC, AHMC 2006), adopting a risk management framework including the module on Augmentation of Step Equation Cryptosporidium Campylobacter Rotavirus for assuring drinking water safety. drinking water supplies (NRMMC, EPHC, a Probability of infection per organism2 5.9 × 10–2 1.9 × 10–2 5.9 × 10–1 However, the ADWG does not provide a NHMRC 2008), use DALYs and the WHO quantitative definition for microbial safety. definition of tolerable risk as a basis for b Probability of infection per year a × 730 43.1 13.9 431 (730 litres = 730 organisms) It also does not provide benchmarks to setting performance targets for achieving measure the success of risk management. microbial safety. c Percentage of infection leading to illness 70% 30%t 88% In this important aspect, the ADWG lags d Probability of illness per year b × c 30.1 4.16 379 behind comparable guidance provided Adopting a quantitative definition of e DALYs per case 1.5 × 10–3 4.6 × 10–3 1.3 × 10–2 by the US EPA (2006a), Health Canada microbial safety and establishing (2004a and b), New Zealand (2008), health-based performance targets f Percentage of population susceptible to illness 100% 100% 6% WHO (2006a) and the Australian has practical implications, including g DALYs per person per year d × e × f 4.52 × 10–2 1.9 × 10–2 2.96 × 10–1 guidelines for water recycling (NRMMC, the need to categorise source water Required reduction to achieve 10–6 DALY per year (g – 10–6) 99.998% 99.995% 99.9997% EPHC, AHMC 2006). quality. The most direct method is to + g100 (4.7 log) (4.3 log) (5.5 log) undertake system-specific monitoring In the absence of microbial targets, based on source water characteristics.
1 The information in this table is taken from WHO (2006a) as adapted by the Australian guidelines for recycled water (NRMMC, EPHC, AHMC 2006). many water suppliers have adopted parts However, this is not always practically 2 Low-dose formulae used as described in Appendix 2 of the Australian guidelines for recycled water (NRMMC, EPHC, AHMC 2006) as doses in safe of the US EPA regulations to provide achievable. Alternative methods could be drinking water below 10–5. performance benchmarks for assuring the based on source water characteristics, safety of drinking water. sanitary surveys and monitoring for indicator bacteria. Working backwards, based on A2.4 from mild diarrhoea to reactive arthritis, The ‘do nothing’ option means that the consumption of 730 litres of water Include a definition of microbial safety haemolytic uraemic syndrome and ADWG does not meet basic requirements The benefit of incorporating a definition –6 per year, 10 DALYs is equivalent occasionally death. of water suppliers or regulators. Reliance of microbial safety and health-based –5 The two alternatives are: to 2.2 × 10 Cryptosporidium, on the production of water containing targets is that they provide a mechanism –5 –6 5.2 × 10 Campylobacter and 3.4 × 10 •• the definition that underpins the A review of acceptable risk published by 0 E. coli per 100 mL from unprotected for identifying appropriate barriers
Rotavirus, Cryptosporidium per litre of US EPA drinking water standards the American Academy of Microbiology water sources is not sustainable and (catchment protection and treatment) to water. These concentrations reinforce (1 infection per 10,000 people per year) (2006) indicated that while a goal of is difficult to defend given the well- ensure safety of drinking water supplies. the impracticality of direct pathogen or less than one infection per 10,000 established limitations of this indicator monitoring of drinking water. people per year is a useful benchmark •• DALYs as applied by the WHO (10–6 in reflecting the presence or absence of it has limitations and a more descriptive DALYs per person per year). enteric viruses and protozoa. Depending on the source water quality, endpoint, taking into account the log reductions required to produce measurements of injury, may be more The two approaches share a number Internationally two alternatives have been safe water can then be calculated from advantageous. The DALY metric fulfils of common features and, importantly, established for defining microbial safety. the formula: this requirement. both can be used to identify quantifiable The US EPA has performance targets set in drinking water standards that are Log reduction = log (organisms in raw and measurable performance targets for Application of DALYs requires data based on an upper limit of one infection water per L ÷ 2.2 × 10–5 producing safe drinking water. Typically on disease outcomes. As discussed or illness per 10,000 people per year Cryptosporidium, or these targets have been developed in Section 3.1, WHO has published a (US EPA 1989; 2002; 2006). WHO has 5.2 × 10–5 Campylobacter for reference pathogens such as scoring system for severity of outcomes adopted the metric of DALYs and defined or 3.4 × 10–6 Rotavirus) Cryptosporidium or groups of pathogens and estimates of DALYs per case for –6 such as protozoa, viruses and bacteria. tolerable risk as being less than 1 × 10 Cryptosporidium, Campylobacter, The GDWQ indicates that ideally the DALYs per person per year. As noted by E. coli 0157 and Rotavirus (Havelaar & concentration of pathogens in source The difference between the two metrics the American Academy of Microbiology Melse 2003). These estimates are based water should be determined on a system- is that application of DALYs includes (2006), while benchmarks based on on international data from developed specific basis and then used to determine an additional step. After calculating the infection or illness rates are useful, a more countries. In the AGWR, an adjustment the log reductions that will produce safe likelihood of infection, the additional step descriptive endpoint, taking into account was included for diarrhoea caused by drinking water. However, the GDWQ is to consider the likelihood and severity impacts, may be more advantageous. Rotavirus in Australia. Further refinements recognises that this is not always possible of resulting disease. In this way, DALYs based on Australian data would increase and suggests that default values could be recognise that not all pathogens cause the accuracy of applying DALYs to used as an alternative. A limited range of the same level or severity of disease. define safety of Australian drinking water defaults derived from international data Infections can be asymptomatic and supplies. A research proposal to provide are included in the guidelines. where symptoms occur they can range this information has been developed.
206 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 207 Abbreviations
ACCS Advisory Committee on Chemicals Scheduling COPC chemicals of potential concern AChE acetylcholinesterase COSHH control of substances hazardous to health ACHHRA Australian Centre for Human Health Risk Assessment COT Committee on Toxicity (UK) ACS American Chemical Society CPK creatine phosphokinase ACTRA Australasian College of Toxicology and Risk Assessment CRC-CARE Co-operative Research Centre for Contamination Assessment and Remediation of the Environment ADI acceptable daily intake CSAF chemical-specific adjustment factors ADS Australian Dietary Survey CSF cancer slope factor ADWG Australian drinking water guidelines CSIRO Commonwealth Scientific and Industrial Research Organisation AEF Australian Exposure Factor CSM conceptual site model AF attribution factors CUF combined uncertainty factor AFIP Armed Forces Institute of Pathology CYP cytochrome P450 AGWR Australian guidelines for water recycling DALY disability-adjusted life years AHMAC Australian Health Ministers’ Advisory Council DETR Department of the Environment, Transport and the Regions AHPC Australian Health Protection Committee DIR dealings involving intentional release AhR aryl hydrocarbon receptor DSEWPC Department of Sustainability, Environment, Water, Population and Communities AIHC American Industrial Health Council EASE estimation and assessment of substance exposure ALP alkaline phosphatase EC exposure concentration ALT alanine aminotransferase ECETOC European Centre for Ecotoxicology and Toxicology of Chemicals AST aspartate aminotransferase ECHA European Chemicals Agency AUC area under the curve ECRP Existing Chemicals Review Program ANZECC Australian and New Zealand Environment and Conservation Council ECVAM European Centre for the Validation of Alternative Methods AP alkaline phosphatase EDC endocrine disrupting chemicals APVMA Australian Pesticides and Veterinary Medicines Authority EHC environmental health criteria ARfD acute reference dose EHPC Environmental Protection and Heritage Council ARMCANZ Agriculture and Resource Management Council of Australia and New Zealand EHRA environmental health risk assessment ARP American Registry of Pathology EPA Environment Protection Authority AT averaging time EPHC Environment Protection and Heritage Council ATSDR Agency for Toxic Substances and Disease Registry EPHSC Environment Protection and Heritage Standing Committee BaP benzo(a)pyrene EUP end-use products BMD benchmark dose FDA Food and Drug Administration BMDL benchmark dose, lower confidence limit FQPA Food Quality Protection Act BMR benchmark risk FSANZ Food Standards Australia New Zealand CHIP Chemicals (Hazard Information and Packaging for Supply) regulation GAF gastrointestinal absorption factor (UK) GD guideline dose CICAD concise international chemical assessment documents GDWQ guidelines for drinking water quality CIIT Chemical Industry Institute of Toxicology GHS Globally Harmonised System for classification and labelling of chemicals CLEA contaminated land exposure assessment GIS geographic information systems CNS central nervous system GLC ground-level concentration COAG Council of Australian Governments GLDH glutamate dehydrogenase COC Committee on Carcinogenicity (UK) GMO genetically modified organism COM Committee on Mutagenicity (UK) GOF goodness of fit CoNTC concentration of no toxicological concern GV guideline value
208 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 209 HEC human-equivalent concentration NATO North Atlantic Treaty Organisation HED human-equivalent dose NCEA National Center for Environmental Assessment HESI Health and Environmental Sciences Institute NChEM National Framework for Chemicals Environmental Management HI hazard index NEPC National Environment Protection Council HIA health impact assessment NEPM National Environmental Protection Measure HIL health investigation level NGO non-governmental organisations HPV high production volume NHMRC National Health and Medical Research Council HQ hazard quotient NICEATM NTP Interagency Center for the Evaluation of Alternative Toxicological Methods HRA health risk assessment NICNAS National Industrial Chemicals Notification and Assessment Scheme HSE health and safety executive NIEHS National Institute of Environmental Health Sciences HSL health screening level NIST National Institute of Standards & Technology IARC International Agency for Research on Cancer NOAEL no observed adverse effect level ICCVAM Interagency Coordinating Committee on the Validation of Alternative Methods NOEL no observed effect level ICNA Industrial Chemicals Notification and Assessment (Act) NOHSC National Occupational Health and Safety Commission IEH Institute for Environmental Health NRA National Registration Authority for Agricultural and Veterinary Chemicals IEUBK Integrated Exposure Uptake Biokinetic Model for Lead in Children NRC National Research Council IFCS Intergovernmental Forum on Chemical Safety NRMMC Natural Resource Management Ministerial Council IGAE Inter Governmental Agreement on the Environment NTE neuropathy target esterase IGHRC Interdepartmental Group on Health Risks from Chemicals NTMDI national theoretical maximum daily intakes IGR ingestion rate NTP National Toxicology Program ILCR increased lifetime cancer risk OCS Office of Chemical Safety ILO International Labour Organization OECD Organisation for Economic Cooperation and Development ILSI International Life Sciences Institute OGTR Office of the Gene Technology Regulator IOMC Inter-Organization Programme for the Sound Management of Chemicals OHS occupational health and safety IPCS International Programme on Chemical Safety OSHA Occupational Safety and Health Administration IRIS Integrated Risk Information System OTC over the counter ITER International Toxicity Estimates for Risk PAD population-adjusted dose JECFA Joint FAO/WHO Expert Committee on Food Additives PAH polycyclic aromatic hydrocarbon JMPR Joint Meeting on Pesticide Residues PBPK physiologically based pharmacokinetic models LD Longfellow DG PCB polychlorinated biphenyls LOAEL lowest observed adverse effect level PCO pest control operator LOD limit of determination PEC priority existing chemical LOR level of reporting PEL permissible exposure limits MBS market basket survey PHED Pesticide Handlers Exposure Database MF modifying factor POD point of departure ML maximum levels POEM Predictive Operator Exposure Model MoA mode of action POP Persistent Organic Pollutant MoE margin of exposure PP precautionary principle MRA microbiological risk assessment PPAR peroxisome proliferator-activated receptor MRL maximum residue level PQRA preliminary quantitative risk assessment MTD maximum tolerated dose PRA probabilistic risk assessment NAS National Academy of Science PRG preliminary remediation goals
210 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 211 PTWI provisional tolerable weekly intake TRV toxicity reference value QA quality assurance TTC threshold of toxicological concern QC quality control TWI tolerable weekly intake QRA quantitative risk assessment UF uncertainty factor QSAR quantitative structure–activity relationship UNCED UN Conference on Environment and Development QMRA quantitative microbial risk assessment UNEP United Nations Environment Programme RAF risk analysis framework URF unit risk factor RAGS risk assessment guidance for superfund VOC volatile organic compounds RARMP risk assessment and risk management plan WHO World Health Organization REACH registration, evaluation, and authorisation of chemicals WoE weight of evidence RfC reference concentration XRF X-ray fluorescence RfD reference dose YLD years lived with a disability RI risk index YLL years of life lost RiskIE Risk Information Exchange RIVM National Institute for Public Health and the Environment (Netherlands) RPF relative potency factor RSC relative source contribution SA surface area SARF social amplification of risk framework SF safety factor SGV soil guidance values SIDS Standard Information Data Sets SOT Society of Toxicology SRA Society for Risk Analysis SRM standard reference materials SSRA site-specific risk assessment STEL short-term exposure limits STP Society of Toxicologic Pathology SUSMP Standard for the Uniform Scheduling of Medicines and Poisons SWA Safe Work Australia TCDD 2,3,7,8-Tetrachlorodibenzo-p-dioxin TDI tolerable daily intake TEF toxicity equivalence factor TEQ toxicity equivalent TES Toxicology Evaluation Section TGAC technical grade active constituent THM trihalomethane TI tolerable intake TLV threshold limit values TMI tolerable monthly intake TPH total petroleum hydrocarbons TOXNET toxicology databases maintained by the US National Library of Medicine
212 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 213 Glossary
Note: the terminology in this glossary has been largely based on that used in the original 2002 edition of the enHealth guidance Background level The amount (or concentration) of agent in a medium (e.g. water or soil) that is not attributed to the on EHRA, plus some terms copied from the glossary of the 2010 revision of the contaminated sites NEPM, and the 2006 NHMRC (or concentration) sources under investigation in an exposure assessment. Background levels can be naturally occurring guidance on ambient air quality standard-setting. or the result of human activities.
Absorbed dose The amount of chemical that, after contact with the exchange boundary (skin, lungs, gut), actually Benchmark dose The dose associated with a given incidence (the benchmark risk, e.g. 1 per cent, 5 per cent or 10 per penetrates the exchange boundary and enters the circulatory system. The amount may be the same (BMD) cent incidence) of effect, based on the best-fitting dose–response curve. or less than the applied dose. (See Table 6 for other types of doses used in health risk assessment.) Benchmark risk (BMR) A predetermined incidence of adverse response that determines the benchmark dose. Accuracy The degree to which a measurement represents the true value of the variable that is being measured Bias A process resulting in a tendency to produce results that differ in a systematic value from the true (NHMRC 2000); or the degree of agreement between the average predictions of a model or the values. Also known as systematic error. average of measurements and the true value of the quantity being predicted or measured (WHO 2003). Bioaccessibility The fraction of a contaminant in an exposure medium that is soluble in the relevant physiological milieu (usually the gastrointestinal tract) and available for absorption. Generically, it is the ability for Acceptable daily intake The daily intake of a chemical that, during a lifetime, appears to be without appreciable risk on the a chemical to come into contact with the absorbing surfaces in an organism. It is related to solubility (ADI) basis of all the facts known at the time. It is expressed in milligrams per kilogram of body weight per and dissolution, since absorption usually can only occur from a liquid or gaseous phase and not from day (mg/kg/day). For this purpose, ‘without appreciable risk’ is taken to mean that adverse effects will a solid phase. not result even after a lifetime of exposure. Furthermore, for a pesticide residue, the ADI is intended to give a guide to the maximum amount that can be taken daily in the food without appreciable risk Bioavailability A generic term defined as the fraction of a contaminant that is absorbed into the body following to the consumer. Accordingly, the figure is derived as far as possible from feeding studies in animals. dermal contact, ingestion or inhalation. It is expressed as the ratio (or percentage) of the absorbed (See also ‘tolerable daily intake’ and ‘reference dose’.) dose (systemic dose) to the administered dose. (See Table 6 for other measures of bioavailability.)
Acceptable risk This is a risk management term. The acceptability of risk depends on scientific data, social, Biological monitoring Measurement of a contaminant or metabolite in body tissue, fluid, blood or expired air. economic and political factors, and the perceived benefits arising from exposure to an agent. (See also ‘target risk’.) Biomarker Any measurement reflecting an interaction between a biological system and an environmental agent that may be chemical, physical or biological (WHO 1993). Often used to describe measurements used Acute exposure A contact between an agent and a target occurring over a short time, generally less than 14 days, in biological monitoring. with a single or repeated dose. (Other terms, such as ‘short-term exposure’ and ‘single-dose’ are also used.) Cancer or A disease of heritable, somatic mutations affecting cell growth and differentiation. That is, genetic carcinogenesis alterations incurred in the first damaged cells are acquired in subsequent cells after cell division within Adduct A chemical moiety that is covalently bound to a large molecule such as DNA or protein. the same individual. It encompasses the origin, causation and development of tumours and applies to all forms of tumours (e.g. benign and malignant). Adverse effect The change in the morphology, physiology, growth, development, reproduction or life span of an organism, system population or sub-population that results in an impairment of functional capacity, an Cancer slope factor The plausible upper-bound estimate of the probability of a response per unit of intake of an agent over impairment of the capacity to compensate for additional stress or an increase in susceptibility to other (CSF) a lifetime. influences. Some adaptive changes are not generally considered to be adverse (e.g. some changes in enzyme levels). Carcinogen Chemical, biological or physical cancer-causing agent. A distinction may be made based on the presumed mode of action (MoA) – see ‘genotoxicity’ and ‘non-genotoxic carcinogen’. Agent A chemical, physical or biological substance or factor (including social factor) being assessed in the context of an environmental health risk assessment. Carcinogenicity A property of an agent that enables it to produce tumours, whether benign or malignant.
Aggregate/cumulative Terminology derived from US legislation. The term ‘aggregate risk’ in this context, implies Causality The relating of causes to the effects they produce. Most of epidemiology concerns causality, risk consideration of all sources of exposure to determine a total (or aggregated) exposure estimate. The and several types of causes can be distinguished. However, epidemiological evidence by itself is term ‘cumulative’ risk implies that the risk associated with substances sharing a common mode of insufficient to establish causality, although it can provide powerful circumstantial evidence. action or toxicity outcome, are aggregated across the exposure estimates for all such substances. Chronic exposure A contact between an agent and a target occurring over a continuous or repeated basis for a duration Air pollution The presence of contaminants (air pollutants) in high enough concentrations in the air that could of three months or greater. (See also ‘sub-chronic exposure’ and ‘lifetime’.) interfere with human health or welfare, or produce other harmful environmental effects. Chronic toxicity An adverse effect that is generally induced by prolonged exposure to a chemical. It may also include Ambient air An unconfined portion of the atmosphere; also open air or surrounding air . an ability to produce an adverse effect that persists over a long period of time, whether or not it occurs immediately upon exposure to a chemical or is delayed. Applied dose Amount of an agent presented to an absorption barrier and available for absorption. The amount may be the same or more than the absorbed dose. (See Table 6 for other types of doses used in health risk Chemical of potential An agent that is potentially associated with the site or exposure medium under consideration and assessment.) concern (COPC) whose data is of sufficient quality to be judged as potentially causing an adverse health effect.
214 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 215 Cluster A greater than expected number of cases that occur within a group of people in a geographic area Disability-adjusted life For a given health condition, the sum of the years of life lost due to premature mortality in the over a period of time (Queensland Health 2009). years (DALYs) population and the years lost due to disability for incident cases. It is a term used more commonly in quantitative microbiological risk assessment (QMRA) rather than in HRA for chemicals. Cluster assessment A scientific process to determine if there is an increased number of cases of a specific disease or condition and to determine if there is a biologically plausible causal agent(s) for the disease Dose A stated quantity or concentration of a substance to which an organism, system, population or (Queensland Health 2009). sub-population is exposed over a continuous or intermittent duration of exposure. It is generally the total amount of a chemical administered, but there may be other expressions relating to the amounts Community Those individuals or groups residing in a locality where a site assessment is to be conducted and actually absorbed or taken up (see Table 6 for other types of doses used in health risk assessment). who may be affected by the assessment and/or possible site contamination physically (e.g. through Dose is most commonly expressed as the amount of test substance per unit weight of test species risks to health or the environment, loss of amenity) or non-physically such as via concern about (e.g. mg/kg body weight). possible contamination). The term ‘wider community’ may be applied to individuals and/or groups not necessarily residing in the locality of the site assessment who may have an interest in the assessment Dosage A general term comprising the dose, its frequency and the duration of dosing. Dosage is properly (NEPC 2010). applied to any rate or ratio involving a dose. Dosages often involve the dimension of time (e.g. mg/kg/ day), but the meaning is not restricted to this relationship. Conceptual site model A description of a site including the environmental setting, geological, hydrogeological and soil characteristics, together with nature and distribution of contaminants. Potentially exposed populations Dose–response Relationship between the amount of chemical administered to, taken up by, or absorbed by an and exposure pathways are identified. Presentation is usually graphical or tabular with accompanying organism, system or (sub)population and the change developed in that organism, system or (sub) explanatory text. population in reaction to the agent. It is the correlative association existing between the dose administered and the response (effect) or spectrum of responses that is obtained. The concept Confidence Weight assigned by the evaluator to the quality of the information available (high, medium or low expressed by this term is indispensable to the identification, evaluation and interpretation of most confidence) to indicate that a chemical possesses certain toxicological properties. pharmacological and toxicological responses to chemicals. The basic assumptions that underlie and support the concept are: (a) the observed response is a function of the concentration at a site; (b) Confidence limit A range of values determined by the degree of presumed random variability in a set of data, within the concentration at a site is a function of the dose; and (c) response and dose are causally related which the value of a parameter (e.g. the mean) lies with a specified level of confidence or probability (Eaton & Klaassen 1996). The existence of a dose–response relationship for a particular biological (e.g. 95 per cent). The upper and lower confidence limits refer to the values at opposite ends of the or toxicological response (effect) provides a defensible conclusion that the response is a result of specified range. exposure to a known substance.
Confounding factor A factor that distorts the apparent effect or magnitude of the effect of a study factor or risk. Dose–response curve Graphical representation of a dose–response relationship that is essential to any quantitative Confounding factors must be controlled for in order to obtain an undistorted estimate of the effect estimation of risk for a given exposure. under study. Endpoint An observable or measurable biological event used as an indicator of the effect of a chemical on a Conservatism A cautious or conservative approach to evaluating and managing the uncertainties inherent in a risk biological system (cell, organism, organ etc.). It may also be expressed as a ‘toxicological endpoint’. assessment, which reduces the probability of harm occurring. Environmental health Those aspects of human health determined by physical, chemical, biological and social factors in the Contaminant Any chemical existing in the environment above background levels and representing, or potentially environment. Environmental health practice covers the assessment, correction, control and prevention representing, an adverse health or environmental risk (may be synonymous with a pollutant). of environmental factors that can adversely affect health, as well as the enhancement of those aspects of the environment that can improve human health. Contamination The condition of land, water or food where any chemical substance or waste has been added or detected at above background level and represents, or potentially represents, an adverse health or Environmental The monitoring of the concentration of substances in the physical environment of air, water, soil environmental impact (NEPC 2010). monitoring and food.
Critical effect The adverse effect judged to be the most important for setting an acceptable human intake or Epidemiology The study of the distribution and determinants of health-related states or events in specified exposure. It is usually the most sensitive adverse effect, that is, that with the lowest effect level, or populations, and the application of the study to the control of health problems. sometimes a more severe effect, not necessarily having the lowest effect level. Expert An expert has: (1) training and experience in the subject area resulting in superior knowledge in the Data quality objectives The establishment of the amount, nature and quality of data required to complete a specific risk field; (2) access to relevant information; (3) an ability to process and effectively use the information; and (DQOs) assessment. (4) is recognised by his or her peers or those conducting the study as qualified to provide judgements about assumptions, models and model parameters at the level of detail required (NCRP 1996). Default value A pragmatic, fixed or standard value used in the absence of relevant data. Expert/professional Opinion of an authoritative person on a particular subject. Deterministic A deterministic approach uses single values or point estimates as input values in an exposure or risk judgement estimation model. These are intended to be ‘best estimates’ of the value of the input variables. (see also ‘probabilistic’).
216 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 217 Exposure Concentration or amount of a particular chemical that reaches a target organism, system, population Genotoxic carcinogen A chemical for which there is adequate evidence that the ability to induce tumours is via a mechanism or sub-population in a specific frequency for a defined duration. Exposure is usually quantified as the involving direct damage to DNA. concentration of the agent in the medium integrated over the time duration of contact. Guideline values (GVs) Values such as concentrations in air or water that are derived after appropriate allocation of tolerable Exposure assessment The estimation (qualitative or quantitative) of the magnitude, frequency, duration, route and intake (TI) among the possible different media of exposure. Combined exposure from all media at the extent (e.g. number of organisms) of exposure to one or more contaminated media for the general guidance values over a lifetime would be expected to be without appreciable health risk. The aim of a population, for different sub-groups of the population, or for individuals. guidance value is to provide quantitative information from risk assessment for risk managers to enable them to make decisions concerning the protection of human health. (WHO 1994a, p. 16). Exposure concentration The exposure mass divided by the contact volume or the exposure mass divided by the mass of contact volume, depending on the medium. Hazard Inherent property of a contaminant or situation having the potential to cause adverse effects when a population may be exposed to that contaminant. It is also described as the disposition of a thing, a Exposure duration The length of time over which continuous or intermittent contacts occur between a chemical and the condition or a situation to produce an adverse health or environmental effect; or an event, sequence exposed population. of events or combination of circumstances that could potentially have adverse consequences (adapted from ACDP 1996). Note the definition of risk to distinguish hazard from risk. Exposure event The occurrence of continuous contact between chemical and exposed population. Hazard identification The identification of the type and nature of adverse effects that a contaminant has an inherent Exposure frequency The number of exposure events within an exposure duration. capacity to cause harm to an exposed population.
Exposure route or The way a chemical enters an organism after contact (e.g. by ingestion, inhalation or dermal Hazard indices/index The sum(s) of at least two hazard quotients. pathway absorption). The pathway usually describes the course a chemical or physical agent takes from a (HI) source to an exposed organism. An exposure pathway describes a unique mechanism by which an individual or population is exposed to chemicals or physical agents at or originating from a site. Each Hazard quotient (HQ) The ratio of the mean daily intake to the reference dose or tolerable daily intake for threshold exposure pathway includes a source or release from a source, an exposure point and an exposure exposure. route. If the exposure point differs from the source, a transport/exposure medium (e.g. air) or media (in cases of inter-media transfer) is also indicated. Health Health is a state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity (WHO 1946). Exposure scenario A set of conditions or assumptions about sources, exposure pathways, concentration of contaminants involved and exposed population (e.g. numbers, characteristics, habits) used in the evaluation and Health investigation Screening criteria based on health risk, presented in Schedule B(7) of the contaminated sites NEPM. quantification of exposure(s) in a given situation. The exposure scenario may be expressed in terms of levels (HILs) May also be called health screening levels (HSLs) to emphasise the fact that they represent an a model, that is, a conceptual or mathematical representation of the exposure process. outcome of a Tier 1-type screening level risk assessment, and may require a more refined Tier 2–3 level process to better define the risk. Exposed population The people who may be exposed to the contaminant. Synonymous with ‘receptors’. Health risk assessment The process of estimating the potential impact of a chemical, biological, physical or social agent Extrapolation For dose–response curves, an estimate of the response at a point outside the range of the (HRA) on a specified human population system under a specific set of conditions and for a certain time experimental data most commonly extrapolated to low dose. Also refers to the estimation of a response frame. May also be described as a process intended to calculate or estimate the risk to a given target in different species or by different routes than that used in the experimental study of interest. organism, system or (sub)population, including the identification of attendant uncertainties following exposure to a particular contaminant, taking into account the inherent characteristics of the agent of Factor A single factor or product of several single factors used to derive an acceptable intake. These factors concern as well as the characteristics of the specific target system. account for adequacy of the study, inter-species extrapolation, inter-individual variability in humans, adequacy of the overall database, nature and extent of toxicity, public health regulatory concern and Heuristics A psychological term used to describe the process whereby people frame their perceptions of risk, scientific uncertainty. The terms safety factor (SF), uncertainty factor (UF) and modifying factor (MF) based on ‘rules of thumb’ and other emotional (affective) factors by which we make judgements about are examples of the terminology used in different jurisdictions to imply essentially the same process. everyday occurrences.
False negative A result that is erroneously negative leading to a determination that the factor under study is not Hormesis Demonstrated beneficial effects of an agent at low (but not homeopathic) doses but with toxicity present. In statistical inference, this is a Type 2 error. occurring at higher doses. Also used to describe ‘hockey-stick’ or other J-shaped non-monotonic dose–response relationships where biological effects may appear to become greater as the dose False positive A result that is erroneously positive leading to a determination that the factor under study is present decreases. when it is not. In statistical inference, this is a Type 1 error. Immunotoxicity The ability to produce an adverse effect on the functioning of organs and cells involved in immune Genotoxicity A broad term describing the ability to produce damage to the genetic material (DNA) of cells or competence (IEH 1999b). organisms. In vitro/in silico Describes tests undertaken in test tubes, culture dishes or other systems where a non-living organism Genotoxic chemical A chemical for which there is adequate evidence of the potential to interact with, and/or modify the is exposed to a test agent. In silico techniques refer to modern genomic methodologies where genes or function of genetic material. DNA arrays on microchips are the responsive agents.
218 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 219 Integrated Risk The database of the US EPA that provides the agency’s adopted hazard and dose–response No observed adverse The highest concentration or amount of a substance, found by experiment or observation, that causes Information System assessment for chemical and radiological agents. Used as guidance and to provide consistency in the effect level (NOAEL) no observable alterations of morphology, functional capacity, growth, development or life span of target (IRIS) agency’s regulatory decisions designed to reduce risk related to environmental exposures. organisms. The NOAEL is the next dose below the LOAEL in the series of doses tested in a study, (may also be cited as where no toxic (i.e. adverse) effects are observed. It may also be worded in more detail thus: The LD50 The quantity of a chemical compound that, when applied directly to test organisms via inhalation, ‘no observable adverse NOAEL is defined as the highest exposure at which there is no statistically or biologically significant oral or dermal exposure, is estimated to be fatal to 50 per cent of those organisms under the stated effect level’) increase in the frequency of an adverse effect when compared with a control group (National conditions of the test. Academy of Sciences, National Research Council 1994). The definition of NOEL is equivalent, but with the removal of the term, ‘adverse’. Often, the difficult issue in the use of the terms NOEL or Lowest observed effect The lowest concentration or amount of a substance found by experiment or observation that causes NOAEL is in deciding whether a compound-related effect noted in a particular study is necessarily an level (LOEL) alterations of morphology, functional capacity, growth, development or life span of target organisms. ‘adverse’ effect. Alterations of morphology, functional capacity, growth, development or life span of the WHO (1990) defines it as the lowest dose of a substance that causes changes distinguishable from target organism may be detected, which are judged not to be adverse. those observed in normal (control) animals. No observed effect The ‘highest dose of a substance administered to a group of experimental animals at which there Lowest observed The lowest concentration or amount of a substance found by experiment or observation that causes level (NOEL) is an absence of observable effects on morphology, functional capacity, growth, development or life adverse effect level adverse alterations of morphology, functional capacity, growth, development or life span of target span that are observed or measured at higher dose levels used in the study. Thus, dosing animals at (LOAEL) organisms. the NOEL should not produce any biologically significant differences between the group of chemically exposed animals and an unexposed control group of animals maintained under identical conditions. Level of detection The minimum concentration or mass of analyte that can be detected at a known confidence level. The NOEL is expressed in milligrams of chemical per kilogram of body weight per day (mg/kg bw/day) (LOD) or, in a feeding study, in ppm in food (converted to mg/kg bw of compound intake by measured or Level (limit) of The value calculated from the instrumentation detection limits and with appropriate scale-up factors estimated food intake over the period of the study). reporting (LOR) applied. The scale-up factors are affected by the procedures, methods and the size of the sample. The NOEL has been simply defined as the highest dose of a substance that causes no changes Lifestyle factors Behaviours or habits that are a matter of individual choice and that may impinge in the outcomes of a distinguishable from those observed in normal (control) animals (WHO 1990). risk assessment. Examples include smoking, poor diet and alcohol intake. Non-genotoxic An agent that induces tumours via a mechanism that does not involve direct damage to genetic Lifetime A figure used in exposure assessment and risk characterisation representing the average life span of carcinogen material (DNA); sometimes referred to as ‘epigenetic’. an organism. Seventy years has been conventionally used for humans, but newer demographic data suggests that human life spans are expanding. Physiologically based Modelling the dose or degree of exposure to a chemical at a target tissue, cell or receptor by pharmacokinetic integration of pharmacokinetic data with anatomical, physiological and biochemical data (IEH 1999b). Metabolite A substance that is the product of biochemical alteration of the parent compound in an organism. (PBPK) model
Mode of action (MoA) A description of observable key events or processes from interaction of an agent with a cell or tissue Particulate matter The fraction of particles passing an inlet with a 50 per cent cut-off efficiency at an aerodynamic
through operational and anatomical changes to the disease state (EPA 2005). (PM10, PM2.5) diameter of 10 µm (PM10) or 2.5 µm (PM2.5). May also be referred to as ultrafine particulate matter.
Model A mathematical representation of a biological system intended to mimic the behaviour of the real Pica A behaviour exhibited occasionally by young children characterised by the deliberate ingestion of non- system, allowing description about empirical data and predictions about untested states of the system. nutritive substances, such as soil.
Mutagenicity The ability to produce a permanent, heritable change in the amount or structure of genetic material of Point of departure A point on a dose–response curve that is defined by the available data and close to the range of cells or organisms (IEH 1999b) (see also ‘genotoxicity’). (POD) observed data points, from which extrapolation techniques (e.g. linearised extrapolation and/or application of safety/uncertainty factors) are used to estimate a toxicity reference value. National Environment National guidance on assessment and management of environmental pollution, established under Protection Measure the National Environment Protection Act. NEPMs are broad framework-setting statutory instruments Probabilistic A probabilistic approach uses frequency distributions of parameters from which input data is (NEPM) defined in the NEPC Act. They outline agreed national objectives for protecting or managing particular randomly selected for repeated calculations to generate a frequency distribution of the output aspects of the environment. Establishment, maintenance and review of NEPMs is the responsibility (exposure or risk). (See also ‘deterministic’) of the Environment Protection and Heritage Council (EPHC), which incorporates the National Environment Protection Council (NEPC), a statutory body under the NEPC Acts of the Commonwealth, Provisional tolerable The tolerable intake of a chemical expressed as a weekly amount. The term was established by WHO states and the territories. The EPHC addresses broad national policy issues relating to environmental weekly intake (PTWI) (1972) for several heavy metals which ‘are able to accumulate within the body at a rate and to an protection, particularly in regard to air, water and waste matters. extent determined by the level of intake and by the chemical form of the heavy metal present in food’ (WHO 1989). Neurotoxicity The ability to produce an adverse effect in the central or peripheral nervous system (IEH 1999b). Public health The science and art of preventing disease, prolonging life and promoting health through the organised efforts of society.
220 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 221 REACH program The Registration, Evaluation, Authorisation and Restriction of Chemical substances program (REACH), Sensitive groups Refers to populations with both susceptibility and vulnerability factors (see ‘susceptibility’ and ‘vulnerability’). established in 2006 as a new European Community program for regulating chemicals and their safe use. The process of changing one variable while leaving the others constant and determining the effect on the output. The procedure commonly involves fixing each uncertain quantity, one at a time, at Read across An extrapolation technique that may be applied when information on the toxicological properties of a Sensitivity analysis its credible lower bound and then its upper bound (holding all other at their medians), and then substance is missing or incomplete. It relies on extrapolating from the toxicological profile of a known, computing the outcomes for each combination of values (USEPA 1992). It can be used to test the and related, substance to the substance under consideration. effects of both uncertainty and variability in input values.
Reproductive toxicity The ability to produce an adverse effect on any aspect of reproductive capacity, function or Skin irritancy A local inflammatory reaction affecting the skin. outcome. It includes effects on the embryo, foetus, neonate and prepubertal organism and on adult reproductive and neuroendocrine systems (IEH 1999b). Stakeholder One who has an interest in a project or who may be affected by it.
Reference dose (RfD) An estimate (with uncertainty factors spanning perhaps an order of magnitude) of the daily exposure Stochastic A random probabilistic phenomenon. (mg/kg/day) to the general human population (including sensitive sub-groups) that is likely to be without an appreciable risk of deleterious effects during a lifetime of exposure. It is derived from the Structure–activity The relationship between the biological activity of a chemical or series of chemicals and their NOAEL or the LOAEL by application of uncertainty factors that reflect various types of data used to relationship (SAR) molecular structure. The relationships can be described qualitatively and quantitatively. estimate RfD and an additional modifying factor, which is based on professional judgement of the Sub-chronic exposure A contact between an agent and a target of intermediate duration between acute and chronic. entire database of the chemical (IRIS 1996). The RfD is equivalent in meaning to tolerable daily intake Different bodies vary on their definitions of the duration of ‘sub-chronic’ exposure, since it varies with (TDI) and acceptable daily intake (ADI). Usually doses less than the RfD are not likely to be associated species. US EPA uses up to 10 per cent of an organism’s lifetime; however, between three and six with adverse health risks, and are therefore less likely to be of regulatory concern. As the frequency months is often used when discussing sub-chronic exposure to people (see also ‘chronic exposure’). and/or magnitude of the exposures exceeding the RfD increase, the probability of adverse effects in a human population increases. However, all doses below the RfD are not assumed to be ‘acceptable’ Susceptibility Refers to intrinsic biological factors that can increase the health risk of an individual at a given (or risk-free) and nor are all doses that exceed the RfD necessarily ‘unacceptable’ (i.e. likely to result exposure level; examples of susceptibility factors include: genetic factors, late-age and early-life, and in adverse effects) (US EPA). The term acute reference dose (ARfD) is used to designate a level of prior or existing disease. exposure (using the same types of uncertainty and other qualifiers) that is likely to be without an appreciable risk or deleterious effect after a single dose or short period of exposure. Target risk The risk level assessed by extrapolation of a dose–response relationship, suggesting an exposure level where the risk could be considered to be ‘negligible, tolerable or acceptable’ to the risk manager. (See Risk The probability that, in a certain time frame, an adverse outcome will occur in a person, group of also ‘acceptable risk’). people, plants, animals and/or the ecology of a specified area that is exposed to a particular dose or concentration of a hazardous agent, that is, it depends on both the intrinsic toxicity of the agent and Teratogenicity The ability to produce a structural malformation or defect in an embryo or foetus (IEH 1999b). the level of exposure. Risk differs from hazard primarily because risk considers probability. Threshold The lowest dose or exposure level that will produce a toxic effect, and below which no toxicity is Risk characterisation The qualitative and, wherever possible, quantitative determination, including attendant uncertainties, observed (IEH 1999b). A non-threshold dose–response relationship implies that the response of the probability of occurrence of known and potential adverse effects of an agent in a given incidence is only zero at zero exposure, and that a finite level of risk may be determined (using organism, system or (sub)population under defined exposure conditions. extrapolation methodology) at any exposure level above zero. Linear extrapolation typically refers to extrapolation to the zero exposure or zero effect origin of a dose–response curve. Risk communication An interactive two-way process involving the exchange among individuals, groups and institutions of information and expert opinion about the nature, severity and acceptability of risks and the decisions Tolerable intake (TI) An estimate of the intake of a substance that over a lifetime is without appreciable health risk (WHO taken to combat them. It usually involves an interactive exchange of information about health and 1994a). Examples are the ADI, TDI and reference dose. environmental risks among risk assessors, managers, news media, interested groups and the general public (see also ‘stakeholders’). Tolerable daily intake An estimate of the daily intake of a substance that can occur over a lifetime without appreciable (TDI) health risk. It may have different units depending on the route of administration (WHO 1994a). The Risk management The process of evaluating alternative actions, selecting options and implementing them in response to term ‘acceptable daily intake’ is used for chemicals such as pesticides (herbicides, insecticides and risk assessments. The decision making will incorporate scientific, technological, social, economic and antifungals) that are deliberately used on food crops or food-producing animals and for which some political information. The process requires value judgements (e.g. on tolerability and reasonableness level of residues may be expected to occur in food. The term ‘tolerable daily intake’ is used when the of costs). chemical is a potential food or environmental contaminant. While exposure should not occur, a TDI is an established health limit below which lifetime exposure should not have any adverse health effects. Safety Practical certainty that adverse effects will not result from exposure to an agent under defined (See also ‘acceptable daily intake’ and ‘reference dose’.) circumstances. It is the reciprocal of risk. Safety does not demand zero risk and would be a meaningless term if it did. Tolerable weekly The tolerable intake (TI) expressed as a weekly or monthly amount. (monthly) intake Safety factor (SF) See ‘factor’. Composite (reductive) factor by which an observed or estimated no observed adverse (TWI/TMI) effect level (NOAEL) is divided to arrive at a criterion or standard that is considered safe or without appreciable risk. Toxicity Inherent property of a chemical to cause an adverse biological effect.
222 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 223 References
Abelson PR (1994). Risk assessments of Aitio A, Jarvisalo J, Riihimaki V and Ames BN (1987). Six common errors Toxicity equivalence A method of expressing the combined (assumed additive) toxicity of a group of like chemicals that low-level exposures (editorial). Science, Hernberg S (1988). Biologic monitoring. relating to environmental pollution. (TEQ) share a common mode of action. The TEQ is based on summing exposure estimates for individual 265:1507. In C. Zenz (Ed.), Occupational medicine: Regulat Toxicol Pharmacol 7:379–383. components of a mixture multiplied by an estimate of their toxic potency (toxicity equivalence factor Principles and practical applications. – TEF) relative to a reference substance. An alternative US terminology for the TEF is relative potency Adami H-O, Berry CL, Breckenridge CB, Chicago: Year Book Medical Publishers. Ames BN, Durston WE, Yamasaki factor (RPF). Smith LL, Swenberg JA, Trichopoulos E and Lee FD (1973). Carcinogens D, Weiss NS and Pastoor TP (2011). Albertini R, Bird, M, Doerrer, N, Needham are mutagens: A simple test system Toxicity reference value Measures of tolerable intake or acceptable risk, such as reference doses and cancer slope factors. Toxicology and epidemiology: Improving the L, Robinson S, Sheldon L and Zenick combining liver homogenates for activation (TRV) science with a framework for combining H (2006). The use of biomonitoring and bacteria for detection. Proceedings of Tumour A mass of abnormal, disorganised cells arising from pre-existing tissue that is characterised toxicological and epidemiological evidence data in exposure and human health risk the National Academy of Sciences of the by excessive and uncoordinated cell proliferation or growth and by abnormal differentiation to establish causal inference. Toxicol Sci assessments. Environ Health Perspect USA 70:2281–2285. 122:223–234. 114:1755–1762. (specialisation). There are two types of tumours: benign and malignant. Benign tumours Andersen ME and Dennison JE (2004). morphologically resemble their tissue of origin, grow slowly (may also stop growing) and form Advisory Committee on Dangerous Alsop WR, Ruffle B, Frohmberg EJ and Mechanistic approaches for mixture risk encapsulated masses; they do not infiltrate other tissues, they do not metastasise and are rarely fatal Pathogens (ACDP) (1996). Microbiological Anderson PD (1993). Development of risk- assessments: Present capabilities with unless they cause physical disruption of a critical body function (e.g. a brain tumour). Malignant risk assessment: An interim report. based clean-up standards using Monte simple mixtures and future directions. tumours (also called carcinomas) resemble their parent tissue less closely and are composed of London: HMSO. Carlo analysis. In Developing clean-up Environ Toxicol Pharmacol 16:1–11. increasingly abnormal cells genetically, morphologically and functionally. Most grow rapidly, spread standards for contaminated soil, sediment, progressively through adjacent tissues and metastasise to distant tissues. Agency for Toxic Substances and Disease and groundwater: How clean is clean? Andersen ME and Krewski D (2009). Registry (ATSDR) (1992) Agency for Washington, DC: ATSDR/US EPA. Toxicity testing in the 21st century: Tumour initiation The first step in carcinogenesis whereby a small number of cells (or one cell) are irreversibly changed Toxic Substances and Disease Registry. Bringing the vision to life. Toxicol Sci due to genetic damage. Public Health Assessment Guidance Altenburger R, Walter H and Grote M 107:324–330. Tumour progression The stage in carcinogenesis when tumours acquire the features of malignant growth. Manual Division of Health Studies, ATSDR. (2004) What contributes to the combined Washington. Available at
224 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 225 Bader, EL, Hrudey,SE and Froese, KL Benirschke K, Garner FM. and Jones Boekelheide K (2007). Toxicological Bucher JR and Portier C (2004). Human Calabrese EJ (2005). Toxicological Carlson L and Gelber K (2001). Ideas for (2004). Urinary excretion half life of TC (Eds) (1978). Pathology of laboratory highlights: Mixed messages. Toxicol Sci carcinogenic risk evaluation. Part V: The awakenings: The rebirth of hormesis as community consultation: A discussion trichloroacetic acid as a biomarker of animals. Vols I and II. New York: Springer 99:1–2. National Toxicology Program vision for a central pillar of toxicology. Toxicol Appl on principles and procedures for making exposure to chlorinated drinking water Verlag. assessing the human carcinogenic hazard Pharmacol 204:1–8. consultation work. NSW Department for disinfection by-products. Occup Environ Boffetta P, McLaughlin JK, La Vecchia of chemicals. Toxicol Sci 82:363–366. Urban Affairs and Planning. Calabrese EJ (2008). Hormesis: Why it is Med 61: 715–716. Benner TC (2004). Brief survey of EPA C, Tarone RE, Lipworth L and Blot WJ standard-setting and health assessment. (2008). False-positive results in cancer Bull RJ, Brown JM, Meierhenry EA, important to toxicology and toxicologists. Casterline JL Jr and Williams CH (1969). Baker D and Nieuwenhuijsen MJ (2008). Environ Sci Technol. 34:3457–3464. epidemiology: A plea for epistemological Jorgenson TA, Robinson M and Stober Environ Toxicol Chem 27:1451–1474. Effect of pesticide administration upon Environmental epidemiology: Study modesty. J Nat Cancer Inst 100:988–995. JA (1986). Enhancement of the esterase activities in serum and tissue of Calabrese EJ and Baldwin LA (1994). methods and application. New York: Berg BN and Simms HS (1960). Nutrition hepatoxicity of chloroform in B6C3F1 rats fed variable casein diets. Toxicol Appl Improved method of selection of the Oxford University Press. and longevity in the rat: II. Longevity and Bogen KT, Cullen AC, Frey HC and Price mice by corn oil: Implications for Pharmacol 14:266–275. onset of disease with different levels of PS (2009). Probabilistic exposure analysis NOAEL. Regulat Toxicol Pharmacol chloroform carcinogenesis. Environ Health 19:48–50. Barlow SM (2005) Threshold of food intake J Nutr 71:225–263. for chemical risk characterization. Tox Sci Perspectives 69:49–58. Cedergreen N, Christensen AM, Kamper toxicological concern: A tool for assessing 109:4–17 A, Kudsk P, Mathiassen SK and Streibig Biello D (2006). Mixing it up: Harmless Calabrese EJ and Gilbert (1993). Lack substances of unknown toxicity present at Budzt-Jorgensen E, Keiding N and of total independence of uncertainty JC and Sorensen H (2008). A review low levels in the diet. Belgium: ILSI Europe levels of chemicals prove toxic together. Bonita R, Beaglehole R and Kjellstrom Grandjean P (2004). Effects of exposure of independent action compared to Scientific American 294:18. T (2007). Basic epidemiology. 2nd edn. factors (UFs): Implication for size of the Concise Monograph Series. imprecision on estimation of the benchmark total uncertainty factor. Regulat Toxicol concentration addition as reference Geneva: World Health Organization. dose. Risk Analysis 24:1689–1696. models for mixtures of compounds with Barlow SM, Kozianowski G, Wurtzen Birnbaum LS and Stokes WS (2010). Pharmacol 17:44–51. Safety testing: Moving towards alternative Boobis AR (2007). The future of mixtures different molecular target sites. Env Toxicol G and Schlatter J (2001). Threshold Burmaster DE and Anderson PD (1994). Calabrese EJ and Stanek EJ (1995). methods. Environ Health Perspect toxicology. Toxicology 240:142. Chem 27:1621–1632. of toxicological concern for chemical Principles of good practice for the use of Resolving intertracer inconsistencies in 118:A12–13. substances present in the diet. Food Boobis AR, Cohen SM, Dellarco V, Monte Carlo techniques in human health soil ingestion estimation. Environ Health Centre for Research into the Built Chem Toxicol 39:893–905. Birnbaum LS, Harris MW, Stocking McGregor D, Meek ME, Vickers C, and ecological risk assessments. Risk Perspect.103:454–457. Environment (CRBE) (1995). Guideline LM, Clark AM and Morrissey Willcocks D and Farland W (2006). IPCS Analysis 14:477–481. values for polynuclear aromatic Barnett V and Lewis T (1994). Outliers in Calabrese, EJ and Stanek EJ (1991). RE (1989). Retinoic acid and framework for analyzing the relevance of hydrocarbon contamination in soils statistical data. 3rd edn. New York: John Busby JS Alcock RE and Macgillivray A guide to interpreting soil ingestion 2,3,7,8-tetrachlorobenzo-p-dioxin a cancer mode of action for humans. Crit expressed as benzo(a)pyrene toxic Wiley & Sons. BH (2009). Interrupting the social studies: II. Qualitative and quantitative equivalents. Draft prepared for the selectively enhances teratogenesis in Rev Toxicol 36:781–92. amplification of risk process: A case study evidence of soil ingestion. Regulat Toxicol Bayne K (1996). Revised guide for the c57bl/6n mice. Toxicol Appl Pharmacol Department of the Environment, EPG Borak J and Sirianni G (2005). Hormesis: in collective emissions reduction. Environ and Pharmacol 13:278–292. 1/6/18. Centre for Research into the Built care and use of laboratory animals. 98:487–500. Science & Policy, 12:297–308. Physiologist 39:199, 208–211. Implications for cancer risk assessment. Calamari D (1999). Assessment of Environment, Nottingham Trent University. Biosecurity Australia (2006). Final import Dose Response, 3:443–451. environmental distribution: The use Bush BM (1991). Interpretation of Chen J, Chen YJ, Teuschler L, Rice G, Bellar TA, Lichtenberg JJ and Kroner RC risk analysis report for apples from New of fugacity models for this purpose. In Borgert CJ, Quill TF, McCarty LS and laboratory results for small animal Kamernik K, Protzel A, and Kodel RL (1974). The occurrence of organohalides Zealand. Part B. Canberra: Biosecurity UNEP/IPCS, Chemical risk assessment. Mason AM (2004). Can mode of clinicians. London: Blackwell Scientific (2003). Cumulative risk assessment for in chlorinated drinking waters. Journal of Australia. Training Module No. 3. WHO/PCS/99.2. action predict mixture toxicity for risk Publications. quantitative response data. Environmetrics the American Water Works Association Geneva: WHO. 66:703–706. Bjarnason (2004). Toxicology of chemical assessment? Tox Appl Pharmacol Butterworth BE (2006). A classification 14:339–353. mixtures: A review of mixtures assessment 201:85–96. Calamari D (Ed.) (1993). Chemical framework and practical guidance for Chess C and Hance, BJ (1994) Belpomme D, Irigaray P, Hardell L, Clapp for the Canadian Defence R & D as part exposure prediction. Chelsea: Lewis Box GEP (1979). Robustness in the establishing a mode of action for chemical Communicating with the public: 10 R, Montagnier L, Epstein S and Sasco AJ of a NATO project. NATO HFM-057/RTG- Publishers. strategy of scientific model building. In carcinogens. Regulat Toxicol Pharmacol questions environmental managers (2007). The multitude and diversity of 009. 45:119–143. environmental carcinogens. Environ Res RL Launer and GN Wilkinson GN (Eds). Calder MC (2008). Contemporary risk should ask. Center for Environmental 105:414–429. Blackburn K, Stickney JA, Carlson-Lynch Robustness in statistics. New York: Butterworth BE, Aylwood LL and Hays assessment in safeguarding children. Lyme Communication, State University of New HL, McGinnis PM, Chappell L and Felter Academic Press. SM (2007). A mechanism-based cancer Regis, UK: Russell House Publishing. Jersey. Rutgers, New Jersey. Benfenati E, Benigni R, DeMarini SP (2005). Application of the threshold Boyd EM and Knight LM (1963). risk assessment for 1,4-dichlorobenzene. Callahan MA and Sexton K (2007). Cleveland WS (1994). The elements of DM, Helma C, Kirkland D, Martin TM, of toxicological concern approach to Regulat Toxicol Pharmacol 49:138–148. Mazzatorta P, Ouedraogo-Arras G, Richard ingredients in personal and household Postmortem shifts in the weight and If ‘cumulative risk assessment’ is the graphing data. Summit, NJ: Hobart Press. answer, what is the question? Environ AM, Schilter B, Schoonen WGEJ, Snyder care products. Regulat Tox Pharmacol water levels of body organs. Toxicol Appl Calabrese EJ (1985). Toxic susceptibility: Pharmacol 5:119–128. Health Perspect 115:799–806. Code of Federal Regulations (CFR) RD and Yang C (2009). Predictive models 43:249–259. Male/female differences. New York: John (2001). Threshold of regulation for for carcinogenicity and mutagenicity: Wiley & Sons. Canadian Council of Ministers of the Blando JD and Cohn P (2004). Exposure Breslow NE and Day NE (1987). Statistical substances used in food-contact Frameworks, state-of-the-art, and Environment (CCME) (2010). Canadian and health risk from swimming in outdoor methods in cancer research. Vol. II – The Calabrese EJ (2004). Hormesis: Basic, articles. 21 CFR 170.39, United States perspectives. J Environ Sci Hlth Part C soil quality guidelines: Carcinogenic and pools contaminated by trichloroethylene. design and analysis of cohort studies. generizable, central to toxicology and a Government Printing Office. 27:57–90. other polycyclic aromatic hydrocarbons Human and Ecological Risk Assessment. International Agency for Research on method to improve the risk-assessment (environmental and human health effects). Codex (1999). Principles and guidelines 10:717–731. Cancer, Scientific Publication 82, Lyon. process. Int J Occup Environ Health Scientific Criteria Document (revised) for the conduct of microbiological risk 10:466–467. PN 1445. assessment, Cac/Gl-30.
226 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 227 Codex Committee on Food Hygiene Crofton KM, Craft ES, Hedge JM, De Rosa CT, El-Masri HA, Pohl H, Cibulas Doerrer N (2007). Integration of human Elinder C-G (1985). Normal values for Fao V and Allesio L (1983). Biological (CCFH) (2003). Proposed draft principles Gennings C, Simmons JE, Carchman W and Mumtaz MM (2004). Implications biomonitoring exposure data into cadmium in human tissues, blood and monitoring. In L Parmeggiani (Ed.), and guidelines for the conduct of RA, Carter WH Jr and DeVito MJ (2005). of chemical mixtures in public health risk assessment: HESI initiatives and urine in different countries. In L Friberg Encyclopaedia of occupational health microbiological risk management. Codex Thyroid-hormone-disrupting chemicals: practice. J Toxicol Environ Hlth, Part B, perspectives. Int J Hyg Environ Health et al. (Eds), Cadmium and health: and safety. Geneva: International Labour Alimentarius Commission. Joint FAO/WHO Evidence for dose-dependent additivity 7:339–350. 210:247–251. A toxicological and epidemiological Organization. pp. 271–274. Food Standards Program. CCFH, 35th or synergism. Environ Health Perspect appraisal, Exposure, dose and Dolan DG, Naumann BD, Sargent session, Orlando, 27 Jan – 1 Feb 2003. 113:1549–1554. De Rosa CT, Hicks HE, Ashizawa AE, Pohl metabolism. Boca Raton, Florida: Faustman EM and Omenn GS (1996). HR and Mumtaz MM (2006). A regional EV, Maier A and Doursen M (2005). CRC Press Inc. Vol. 1. 81–102. Risk assessment. In Klaassen CD (Ed.), Cohen SM, Klaunig J, Meek ME, Hill Cromar N, Cameron S & Fallowfield H approach to assess the impact of living in Application of the threshold of toxicological Casarett and Doull’s toxicology: The basic RN, Pastoor T, Lehman-McKeeman (2004) Environmental health in Australia a chemical world. Ann NY Acad Sci 1076: concern concept to pharmaceutical Elliott P, Cuzick, English D, Stern R science of poisons. 5th edn. New York: L, Bucher J, Longfellow DG, Seed J, and New Zealand. Oxford University Press. 829–38. manufacturing operations. Regulat Toxicol (1992). Geographical and environmental McGraw Hill, pp. 5–88. Dellarco V, Fenner-Crisp P, and Patton D Pharmacol 43:1–9. epidemiology: Methods for small-area (2004). Evaluating the human relevance Crome P (1999). The elderly. In Institute Department of Health (DOH) (1991). studies. Oxford: World Health Organization. Fawell J (2000). Risk assessment case Doull J (1980). Factors influencing of chemically induced animal tumours. for Environment and Health (IEH). Risk Guidelines for the evaluation of chemicals study: Chloroform and related substances. toxicology. In Casarett and Doull‘s Ellison SLR, Gregory S and Hardcastle Toxicol Sci 78:181–186. assessment strategies in relation to for carcinogenicity. Committee On Food Chem Toxicol 38:S91–S95. population sub groups. Leicester: Institute Carcinogenicity of Chemicals in Food, Toxicology, 2nd edn, pp. 70–83, W (1998). Quantifying uncertainty Conner MW and Newbern PM (1984). for Environment and Health. Consumer Products and the Environment. Macmillan Pub. Co., New York. (Note: in qualititative analysis. The Analyst Federal Focus (1996). Principles for Drug-nutrient interactions and their Department of Health, Report RHSS 42, This reference is to an old edition of this 123:1155–1161. evaluating epidemiologic data in regulatory Crump K (1984). A new method for well-known textbook but the latest edition risk assessment: Recommendations for implications for safety evaluations. Fund London: HMSO. enHealth (2001). Health impact determining allowable daily intakes. Fund (at page 22) still refers to this particular implementing the ‘London principles’ and Appl Toxicol 4:5341–5356. assessment guidelines. enHealth, Appl Toxicol 4:854–891. Department of Health (DOH) (1998). chapter in the 2nd edition for detailed for risk assessment guidance. Developed Canberra: Commonwealth of Australia. Connolly RB and Lutz WK (2004). Communication about risks to public discussion on factors influencing animal by an Expert Panel at London Conference Crump K (1985). Mechanisms leading to Nonmonotonic dose–response health: Pointers to good practice. responses to toxic substances). enHealth (2003). Guidelines for October, 1995. Washington: Federal Focus. dose–response models. In P Ricci (Ed.), relationships: Mechanistic basis, kinetic London: HMSO. economic evaluation of environmental Principles of health risk assessment. Drew R & Frangos J (2007). The Ferguson CC (1994). The Contaminated modelling and implications for risk health planning and assessment. Englewood Cliffs, NJ: Prentice Hall. Department of the Environment, Transport Concentration of No Toxicological Concern Land Exposure Assessment Model assessment. Toxicol Sci. 77:151–157. Vol. 1 – The Guidelines Vol. 2 – Case and the Regions (DETR), the Environment (CoNTC): A risk assessment screening tool (CLEA): Technical basis and algorithms. studies. Canberra: Department of Health Cullen AC and Frey HC (1999). for air toxics. J Toxicol Env Hlth. Part A Co-operative Research Centre for Agency, Institute for Environment and Ageing and enHealth Council. Draft report submitted to Department Contamination Assessment and Probabilistic techniques in exposure and Health (2000). Guidelines for 70:1584–1593. of the Environment. Nottingham: The Remediation of the Environment Pty Ltd assessment: A handbook for dealing with environmental risk assessment and enHealth (2006) Responding to Nottingham Trent University Centre for Duncan JR, Prasse KW and Mahaffey EA (CRC-CARE) (2009). HSLs for petroleum uncertainty and variability in models and management.
228 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 229 Finley B et al. (1994). Recommended Fowles JR, Alexeeff GV and Dodge D uncertainty/safety factors. Regul Toxicol Gregg MB (Ed.) (1996). Field epidemiology. Haschek WM, Rousseaux CG and Wallig Lipscomb J, Kilpatrick ME, Krewski distributions for exposure factors (1999). The use of benchmark dose Pharmacol 29:151–157. New York: Oxford University Press. MA (Eds) (2002). Handbook of toxicologic D, Krishnan K, Nordberg M, Okino M, frequently used in health risk assessment. methodology with acute inhalation lethality pathology. Vol 1. Academic Press, San Tan Y-M, Viau C and Yager JW (2008). Gilbert RO (1987). Statistical methods for Risk Analysis 14:533–53. data. Regul Toxicol Pharmacol 29:262–278. Grice HC and Burek JD (1983). Age Diego, CA. Guidelines for the development of environmental pollution monitoring. New associated (geriatric) pathology: Its impact biomonitoring equivalents: Report from York: Van Nostrand Reinhold. Finucane, ML (2004). The psychology Frawley JP (1967). Scientific evidence on long term toxicity studies. In Current Haseman JK, Huff J and Boorman GA the Biomonitoring Equivalents Workshop. of risk judgments and decisions. In N. and common sense as a basis for food- Issues in Toxicology, New York: Springer (1984). Use of historical control data in Glik DC (2007). Risk communication Regul Toxicol Pharmacol 51:S4–S15. Cromar, S. Cameron, and H. Fallowfield packaging regulations. Food Cosmet Verlag. carcinogenicity studies in rodents. Toxicol for public health emergencies. Ann Rev (Eds), Environmental health in Australia Toxicol 5:293–308. Pathol 12:126–135. Health and Safety Executive (HSE) (1999). Public Hlth 28:33–59. and New Zealand (pp. 142–155). Groten JP, Feron VJ and Sühnel J Reducing risks, protecting people: HSE’s New York: Oxford University Press. Friberg LT (1985). The rationale of Golden RJ, Holm SE, Robinson D, (2001). Toxicology of simple and Hattis D and Silver K (1994). Human decision-making process.
230 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 231 Hellard ME, Sinclair MI, Forbes AB Higgins JPT, Green S (2011) (Eds). Hrudey S. and Leiss W (2003). Risk International Agency for Research on International Programme on Chemical Joint FAO/WHO Expert Committee and Fairley CK (2001). A randomized, Cochrane handbook for systematic management and precaution: Insights Cancer (IARC) (1978). Some N-Nitroso Safety (IPCS) (2005). Chemical specific on Food Additives (JECFA) (2002). blinded, controlled trial investigating the reviews of interventions Version 5.1.0 on the cautious use of evidence. Environ compounds. Vol. 17. IARC Monographs adjustment factors for interspecies Evaluation of certain food additives gastrointestinal health effects of drinking (updated March 2011). The Cochrane Health Perspect 111: 1577–1581. on the Evaluation of the Carcinogenic Risk differences and human variation: and contaminants. WHO Technical water quality. Environ Health Perspect Collaboration, 2011. Available from
232 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 233 Klinger MM, MacCarter GD and Boozer Langley AJ (1991a). Exposure pathways. Larson JL, Wolf DC, Morgan MK and Lord DW (1987). Appropriate site McKone TE (1993). CalTOX. A multimedia Messner MJ, Chappell CL and Okhuysen CN (1996). Body weight and composition In O El Saadi, AJ Langley (Eds), The Butterworth BE (1995). Toxicity and cell investigations. In T Cairney (Ed), total-exposure model for hazardous-wastes PC (2001). Risk assessment for in the Sprague-Dawley rat: Comparison health risk assessment and management proliferation in the liver, kidneys and nasal Reclaiming contaminated land. Glasgow: sites. Part 1: Executive Summary prepared Cryptosporidium: a hierarchical Bayesian of three outbred sources. Lab Animal Sci of contaminated sites. Adelaide: South passages of female F-344 rats, induced by Blackie. for the State of California. Department analysis of human response data. Water 46:67–69. Australian Health Commission. chloroform administered by gavage. Food of Toxic Substances Control UCRL- Research 35:3934–3940. Chem Toxicol 33:443–456. Lovell DP and Thomas G (1996). CR-111456 Pt 1. Livermore, CA: Lawrence Knafla A, Phillipps KA, Brecher RW, Langley AJ (1991b). Protecting the public: Quantitative risk assessment and the Livermore National Laboratory. Mills CJ, Bull RJ, Cantor KP, Reif J, Petrovic S and Richardson M (2006). Biological monitoring. In O El Saadi, AJ Last JM (1988). A dictionary of limitations of the linearised multistage Hrudey SE and Huston P (1998). Health Development of a dermal cancer Langley (Eds), The health risk assessment epidemiology. 2nd edn. Oxford: Oxford model. Human Exptl Toxicol 15:87–104. McKone TE (1994). Uncertainty in risks of drinking water chlorination by- slope factor for benzo[a]pyrene. Regul and management of contaminated University Press. variability in human exposures to soil products: Report of an expert working Pharmacol Toxicol 45:159–168. sites. Adelaide: South Australian Health Lu FC and Sielken Jr RL (1991). contaminants through home grown food: group. Chronic diseases in Canada Last JM (2010). Connecting the dots: Assessment of safety/risk of chemicals: A Monte Carlo assessment. Risk Analysis, Korach KS (Ed) (1998). Reproductive Commission. 19:91–102. The power of words and the diversity of Inception and evolution of the ADI and 14:449–463. and developmental toxicology. New York: Langley AJ (1993a). Refining exposure epidemiological information. J Epidemiol dose–response modelling procedures. Moerbeek M, Piersma AH and Wout Marcel Dekker. assessment. In AJ Langley, M Van Alphen Community Hlth 64:106–108. Toxicology Letters 59:5–40. McLean AEM and McLean EK (1969) Diet Slob W (2004). A comparison of three (Eds), The health risk assessment and and toxicity. Br Med Bull 25:278–28. methods for calculating confidence Kortenkamp A, Backhaus T and Faust M Lauwerys R (1984). Basic concepts management of contaminated sites. Lucas RM and McMichael AJ (2005). intervals for the benchmark dose. Risk (2009). State of the art report on mixture of human exposures. In A Berlin, M McMichael AJ (1991). Setting Adelaide: South Australian Health Association or causation: Evaluating links Analysis 24:31–40. toxicity. Report to the Commission of the Draper, K Hemminki and H Vainio (Eds), environmental exposure standards: Commission. between ‘environment and disease’. European Union (Directorate General for Monitoring exposure to carcinogenic Bulletin WHO 83:792–795. current concepts and controversies. Int J Mohr U, Dungworth D, Kimmerle G, the Environment). Langley AJ (1993b). The representation and mutagenic agents. IARC Scientific Envir Hlth Res 1:2–13. Lewkowski J, McClellan R and Stöber W McAuley C and Hrudey S (2006). Kortenkamp A, Faust M, Scholze M and of data and the appraisal of contaminated Publications No. 59. Lyon: IARC. (Eds) (1988). Inhalation toxicology: The Towards meaningful stakeholder Manahan SE (1993). Fundamentals of Backhaus T (2007). Low-level exposure land. In AJ Langley, M Van Alphen pp. 31–36. design and interpretation of inhalation comprehension of sour gas facility risk Environmental Chemistry. Boca Raton, to multiple chemicals: reason for human (Eds), The health risk assessment and studies and their use in risk assessment. Lead Safe (1997). A guide for health care assessments. J Environ Eng Sci 5:1–11. Florida: Lewis Publishers. health concerns? Environ Health Perspect management of contaminated sites. ILSI Monograph. New York: Springer- professionals. Lead Education Program. 115:Suppl 1 106–114. Adelaide: South Australian Health Mason AM, Borgert CJ, Bus JS, Mumtaz Verlag. NSW: NSW Lead Reference Centre. McCarty LS and Borgert CJ (2006a). Commission. Review of the toxicity of chemical MM, Simmons JE and Sipes IG (2007). Krewski D, Gaylor D, Soms, A and Improving the scientific foundation Mohr U, Dungworth DL and Capen CC Lee PN (1993). Statistics. In D Anderson mixtures containing at least one Szyskowicz, M (1993). An overview of the Langley A (2003) What does it mean when for mixtures joint toxicity and risk (Eds) (1992). Pathobiology of the Aging –5 and DM Conning (Eds), Experimental organochlorine. Regul Toxicol Pharmacol report: Correlation between carcinogenic the risk assessment says 4.73 × 10 ? assessment: Contributions from the SOT Rat. Vol. 1. Washington, DC: ILSI Press. toxicology: The basic issues. Cambridge, 45:104–118. potency and the maximum tolerated dose: NSW Public Hlth Bull 14:166–167. mixtures project. Toxicol Appl Pharmacol UK: Royal Society of Chemistry. Mohr U, Dungworth DL and Capen CC Implications for risk assessment. Risk Anal 223:99–103. Langley AJ, Sabordo L (1996). Exposure McCarty LS and Borgert CJ (2006b). (Eds) (1994) Pathobiology of the Aging 13: 383–398. Levy B and Wegman DH (1988). Review of the toxicity of chemical factors in risk assessment. In A Langley, Masoro EJ (1992). Aging and proliferative Rat. Vol. 2. Washington, DC: ILSI Press. Occupational health: Recognising and mixtures: theory, policy, and regulatory Kroes R, Galli C, Munro I, Schilter B, B Markey and H Hill (Eds), The health homeostasis: Modulation by food preventing work-related disease. 2nd edn. practice. Regul Toxicol Pharmacol Mohr U, Dungworth DL, Ward J, Capen Tran L-A, Walker R and Wurtzen G risk assessment and management of restriction in rodents. Lab Animal Sci Little Brown, Boston. 45:119–143. CC, Carlton W and Sundberg J (Eds) (2000). Threshold of toxicological concern contaminated sites. Adelaide: South 42:132–137. (1996). Pathobiology of the Aging for chemical substances in the diet: A Australian Health Commission. Levy G (1968). Dose dependent effects MacDonald J, French JE, Gerson Mouse. Vols 1 and 2. Washington, DC: practical tool for assessing the need in pharmacokinetics. In Important Meek ME (Bette), Bucher JR, Cohen SM, Langley AJ, Taylor A and Dal Grande E RJ, Goodman J, Inoue T, Jacobs A, Dellarco V, Hill RN, Lehmann-McKeeman ILSI Press. for toxicity testing. Food Chem Toxicol fundamental principles in drug Evaluation (1998). 1996 Australian exposure factors. Kasper P, Keller D, Lavin A, Long G, LD, Longfellow DG, Pastoor T, Seed J 38:255–312. pp. 141–172. In A Langley, P Imray, W Lock and H McCullough B, Sistare FD, Storer R and and Patton DE (2003). A framework for Monosson E (2005). Chemical mixtures: van der Laan JW (2004). The utility of considering the evolution of toxicology Kroes R, Renwick AG, Cheeseman M, Hill (Eds), The health risk assessment Lewis SC, Lynch JR and Nikiforov AI human relevance analysis of information genetically modified mouse assays for and chemical assessment. Env Health Kleiner J, Mangelsdorf I, Piersma AB. and management of contaminated (1990). A new approach to deriving on carcinogenic modes of action. Crit Rev identifying human carcinogens: A basic Perspect 113:383–390. Schilter B, Schlatter J, van Schothorst sites, Adelaide. South Australian Health community exposure guidelines for ‘no Toxicol. 33:591–653. F, Vos JG and G Wurtzen G (2004). understanding and path forward. Toxicol Commission. observed adverse effect levels’. Regul Structure-based thresholds of toxicological Sci 77:188–194. Mena KD, Gerba CP, Haas CN and Rose Moolgavkar SH, Moller H and Toxicol Pharmacol 3:17–34. concern (TTC): Guidance for application Larson JL, Wolf DC and Butterworth JB (2003). Risk assessment of waterborne Woodward A (1999). Principles of the Mackay D (1991). Multimedia epidemiological approach to QEP. In to substances present at low levels in the BE (1994). Induced cytotoxicity and Lilienfield AM and Lilienfield DE (1980). coxsachievirus. J Amer Water Works environmental models: the fugacity SH Moolgavkar, L Krewski, L Zeise et diet. Food Chem Toxicol 42:65–83. cell proliferation in the hepatotoxicity Foundations of epidemiology. 2nd edn. Assoc. 95:122–131. approach. Michigan: Lewis Publishers. al. (Eds), Quantitative estimation and of chloroform in female B6C3F1 mice: Oxford University Press. Lambert JC and Lipscomb JC (2007). Menzie CA, MacDonell MM and Mumtaz Comparison of administration by gavage prediction of human cancer risks. Lyon: Mode of action as a determining factor McClellan RO and Henderson RF M (2007). A phased approach for in corn oil vs. ad libitum in drinking water. Lock WH (1996). Composite sampling. International Agency for Research on in additivity models for chemical mixture (Eds) (1989). Concepts in inhalation assessing combined effects from multiple Fund Appl Toxicol 22:90–102. National Environmental Health Cancer, Scientific Publication 131. risk assessment. Regul Toxicol Pharmacol Monographs Soil Series No. 3. Adelaide: toxicology. Hemisphere Publishing Corp, stressors. Environ Health Perspect 49:183–194. South Australian Health Commission. New York. 115:807–816.
234 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 235 Morisseau C, Merzlikin O, Lin A, He G, Nadkarni RA (1991). The quest for quality National Health and Medical Research National Toxicology Program (NTP) (2006). Nolen G (1972). Effects of various Paustenbach DJ (2002). Human and Feng W, Padilla I, Denison MS, Pessah in the laboratory. Analytical Chemistry Council (NHMRC) (2008). Guidelines NTP toxicology and carcinogenesis studies restricted dietary regimens on the growth, ecological risk assessment: Theory and IN and Hammock BD (2009). Toxicology 63(13): 675A. for managing risk in recreational water. of a mixture of 2,3,7,8-Tetrachlorodibenzo- health and longevity of albino rats. J Nutr practice. New York: John Wiley & Sons. in the fast lane: Application of high- Canberra: Commonwealth of Australia. p-Dioxin (TCDD) (CAS No. 1746-01-6), 102:1477–1494. throughput bioassays to detect modulation National Academy of Science (NAS) 2,3,4,7,8-Pentachlorodibenzofuran Paynter OE (1984). Oncogenic potential: of key enzymes and receptors. Environ (1991). Human exposure assessment National Health and Medical Research (PeCDF) (CAS No. 57117-31-4), and New South Wales Department of Guidance for analysis and evaluation Health Perspect 117:1867–1872. for airborne pollutants. Washington, DC: Council and Natural Resource 3,3’,4,4’,5-Pentachlorobiphenyl (PCB 126) Environment and Conservation (NSW DEC) of long-term rodent studies. Evaluation National Academy Press. Management Ministerial Council (NHMRC, (CAS No. 57465-28-8) in Female Harlan (2006). Assessment and management of Procedure #1000.1 Office of Pesticide and Mumtaz MM, Ruiz P and De Rosa, NRMMC) (2004). Australian drinking odour from stationary sources in NSW. Toxic Substances. Washington, DC: EPA. National Cancer Institute (NCI) (1976). Sprague-Dawley Rats (Gavage Studies). CT (2007). Toxicity assessment of water guidelines. Canberra: NHMRC and Natl Toxicol Program Tech Rep Ser. NSW: Department of Environment and Carcinogenesis Bioassay of Chloroform. Perkins JL (1997). Modern Industrial unintentional exposure to multiple NRMMC. September 526:1–180. Conservation. chemicals. Toxicol Appl Pharmacol Bethesda, MD: National Cancer Institute. Hygiene. Vol. 1. Recognition and 223:104–113. National Research Council (NRC) Natural Resource Management Ministerial Nurminen M (1995). Linkage failures in evaluation of chemical agents. New York: National Environment Protection Council (1983). Risk assessment in the federal Council, Environment Protection and ecological studies. World Health Statistical Van Nostrand Reinhold. (NEPC) (2009). An Australian approach Mundt KA, Tritschler JP, Dell LD (1998). government: managing the process. Heritage Council, Australian Health Quarterly. 48:78–84. to setting air quality standards. Technical Pershagen G (1999). Research priorities Validity of epidemiological data in Washington, DC: National Academy Press. Minister’s Conference (NRMMC, EPHC, Support Document
236 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 237 Presidential/Congressional Commission on Renwick AG (2005). Structure-based Rook JJ (1974). Formation of haloforms Ryker SJ and Small MJ (2008). Combining Saunders LD, Chen W and Hrudey SE Sexton K and Hattis D (2006). Assessing Risk Assessment and Risk Management thresholds of toxicological concern: during chlorination of natural waters. occurrence and toxicity information to (1997). Studies of hazardous waste cumulative health risks from exposure (P/CCRARM) (1997). Risk assessment guidance for application to substances Water Treatment and Examination. identify priorities for drinking-water mixture sites and human health: evaluation of to environmental mixtures: Three and risk management in regulatory present at low levels in the diet. Toxicol Appl 23:234–243. research. Risk Analysis 28:653–666. internal validity. Environmental Review. fundamental questions. Environ Health decision making. P/CCRARM. Pharmacol 207 (Supplement 1): 585–591. 5:167–180. Perspect 115:825–832. Roseman JM (1998). What can be done to Salem H (Ed) (1987). Inhalation Purchase IFH and Auton TR (1995). Renwick AG and Walker R (1993). ensure the usefulness of epidemiological toxicology: Research methods, Schaafsma G, Kroese ED, Tielemans Sharer K (1977). The effect of chronic Thresholds in chemical carcinogenesis. An analysis of the risk of exceeding data for risk assessment purposes? applications and evaluation. New York and ELJP, van Sandt JJM and van Leeuwen underfeeding on organ weights in rats: Regul Toxicol Pharmacol 22:199–205. acceptable or tolerable daily intake. Regul Human and Ecological Risk Assessment. Basel: Marcel Dekker. CJ (2009). REACH, non-testing How to interpret organ weight changes Toxicol Pharmacol 18:463–480. 4:737–746. approaches and the urgent need for in cases of marked growth retardation in Queensland Government (2001a). A Samet JM, Schnatter R and Gibb H a change in mind set. Regul Toxicol toxicity tests? Toxicology 7:45–56. guide to effective community engagement. Rice G, MacDonell M, Hertzberg RC, (1998). Invited commentary: Epidemiology Ross MH and Bras G (1965). Tumour Pharmacol 53:70–80. Community Engagement Unit Strategic Teuschler L, Picel K, Butler J, Chang Y-S incidence patterns and nutrition in the rat. and risk assessment. American Journal of Sharma RP and Coulombe RA Jr (1996). and Executive Services.
238 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 239 Slovic P (2000). The perception of risk. Suk WA and Olden K (2005). Teunis P, Takumi K and Shinigawa K Travis KZ, Pate I, Zoe K and Welsh ZK United States Environmental Protection United States Environmental London: Earthscan Publications. Multidisciplinary research: Strategies for (2004). Escherischia coli O157:H7 from (2005). The role of the benchmark dose Agency (US EPA) (1984). Risk Protection Agency (US EPA) (1996a). assessing chemical mixtures to reduce outbreak data. Risk Analysis 24:401–407. in a regulatory context. Regul Toxicol assessment and management: Framework Draft Revision to the guidelines for Smialowicz RJ, DeVito MJ, Williams WC risk of exposure and disease. Human Pharmacol 43:280–291. for decision making. EPA 600/9-85-002. carcinogenic risk assessment. Office of and Birnbaum LS (2008). Relative potency Ecological Risk Assess. 11:141–151. Teuschler L, Klaunig J, Carney E, Health and Environmental Assessment, based on hepatic enzyme induction Chambers J et al. (2002). Support of Trivedi B (2007). Toxic cocktail. United States Environmental Protection Office of Research and Development, Sumi N, Stavrou D, Frohberg H and predicts immunosuppressive effects of science-based decisions concerning the New Scientist Sept 44–47. Agency (US EPA) (1988). 7 Cardinal rules United States Environmental Protection Jochmann G (1976). The incidence of a mixture of PCDDS/PCDFS and PCBS. evaluation of the toxicology of mixtures: of risk communication. OPA-87-020. Agency, Washington DC. Toxicol Appl Pharmacol 227:477–484. spontaneous tumours of the central nervous A new beginning. Regul Toxicol Pharmacol Trosko JE and Upham BL (2005). The Washington: US Environmental Protection system of Wistar rats. Arch Toxicol 35:1–13. 36:34–39. emperor wears no clothes in the field Agency. United States Environmental Protection Smith LL, Brent, Robert L, Cohen of carcinogen risk assessment: Ignored Agency (US EPA) (1996b). Guidance Suter G, Vermeire T, Munns W and SM, Doerrer NG, Goodman JI, Greim Teuschler LK (2007). Deciding which concepts in cancer risk assessment. United States Environmental Protection for data quality assessment. Practical Sekizawa J (2005). Framework for the H, Holsapple MP and Lightfoot RM chemical mixtures risk assessment Mutagenesis 20:81–92. Agency (US EPA) (1989). Risk methods for data analysis. EPA QA/G-9. integration of health and ecological risk (2008). Predicting future human and methods work best for what mixtures. assessment guidance for Superfund. EPA/600/R-96/064. environmental health challenges: The assessment. Toxicol Appl. Pharmacol. Toxicol Appl Pharmacol 223:139–147. Tucker MJ (1979). The effect of long-term Vol 1. Human health evaluation manual, health and environmental sciences 207: Suppl 1 611–616. food restriction on tumours in rodents. (Part A). Interim final. EPA/540/1-89/002. United States Environmental Protection Thayer KA, Melnick R, Huff J, Burns Int J Cancer, 23:803–807. institute’s scientific mapping exercise. Swaim LD, Taylor HW and Jersey GC Washington: US Environmental Protection Agency (US EPA) (1997a). Exposure Critical Rev Toxicol 38:817–845. K and Davis D (2006). Hormesis: A Agency, Office of Emergency and factor handbook. National Center for (1985). The effect of handling techniques new religion? Environ Health Perspect. Tufte ER (1983). The visual display on serum ALT activity in mice. J Appl of quantitative information. Cheshire, Remedial Response. Environmental Assessment. Office of Smith RL (1994). Monte Carlo simulation 114:A632–633. Research & Development. for human exposure assessment at a Toxicol 5:160–162. Connecticut: Graphics Press. United States Environmental Protection Superfund site. Risk Analysis, 14(4):433– Thomas SP and Hrudey SE (1997). Agency (US EPA) (1992). Guidelines United States Environmental Protection Swenberg JA, Fryar-Tita E, Jeong Y-C, Risk of death in Canada: What we Tufte ER (1990). Envisioning information. 439. Boysen G, Starr T, Walker VE and Albertini Cheshire, Connecticut: Graphics Press. for exposure assessment notice. Part Agency (US EPA) (1997b). Guiding know and how we know it. Edmonton: VI. Federal Register 29 May 1992 principles for Monte Carlo analysis, Snodgrass WR (1992). Physiological and RJ (2008). Biomarkers in toxicology and University of Alberta Press. risk assessment: Informing critical dose– Tuomisto J, Pekkanen J, Kirivanta H, 57(104):22888–22938. Risk Assessment Forum, Washington, biochemical differences between children Tukiainen E, Vartiainen T, Viluksela M and DC: US EPA. and adults as determinance of toxic response relationships. Chem Res. Toxicol. Thompson A and Taylor BN (2008). United States Environmental Protection 21:253–265. Tuomisto JT (2005). Dioxin cancer risk response to an environmental pollutant. Guide for the use of the International Agency (US EPA) (1993). Provisional United States Environmental Protection System of Units (SI). National Institute of – example of hormesis? Dose Response In PS Guzelian, CJ Henry, SS Olin (Eds), Symons J M (2001). The early history guidance for quantitative risk assessment Agency (US EPA) (1998a). National Standards & Technology (NIST) Special 3:332–341. Similarities and differences between of disinfection by-products: A personal of polycyclic aromatic hydrocarbons. Primary Drinking Water Regulations: Publication 811. children and adults: implications for risk chronicle (Part I). Environmental Engineer Turczynowicz L and Robinson NI (2007). EPA/600/R-93/089. Environment Criteria Disinfectants and Disinfection assessment. Washington: International Life 37:20–26. Thompson GE, Scheel LD and Lowe D Exposure assessment modeling for and Assessment Office, US Environmental Byproducts Notice of Data Availability: Sciences Institute. (1982). Seasonal alteration in response to volatiles – towards an Australian indoor Protection Agency, Cincinatti, Ohio. Proposed Rule, Federal Register. Tal A (1997). Assessing the environmental stress or physiological change. Drug and vapour intrusion model. J Toxicol Env Hlth 63:15673–15692. Standards Australia (2009). Australia/New movement’s attitudes towards risk United States Environmental Protection Chem. Toxicol 5:189–199. Part A 70:1619–1634. Zealand Standard. Risk management. assessment. Environmental Science and Agency (US EPA) (1994) Air emissions United States Environmental Protection Principles and guidelines AS/NZS ISO Technology, 31 (10):470A–476A. Thornton KW and Paulsen SG. (1998). Tyl RW and Marr MC (1997) In RD models for waste and wastewater U.S. Agency (US EPA) (1998b). 40 CFR 31000:2009. Can anything significant come out of Hood (Ed.), Handbook of developmental EPA contract no. 68d10118 November Parts 9, 141, and 142 – Disinfectants Tannenbaum A (1940). Relationship of toxicology. Boca Raton, Florida: CRC 1994 prepared for U.S. Environmental and Disinfection Byproducts; Final Starr G, Langley A and Taylor A (2000). body weight to cancer incidence. Arch monitoring? Human and Ecological Risk Assessment, 4(4):797–805. Press. Protection Agency Office of Air Quality Rule. National Primary Drinking Environmental health risk perception Pathol 30:509–517. Planning and Standards. EPA-453/R-94- Water Regulations, Federal Register. Tyson CA and Sawhney DS (1985). Organ in Australia: A research report to the Thurman JD, Moeller RB and Turturro A 080A. 63:69389–69476. Commonwealth Department of Health Tarone RE (1982). The use of historical function tests in toxicology evaluation. control information in testing for a trend in (1995). Proliferative lesions of the testis in and Aged Care. Adelaide: Centre for ad-libitum-fed and food-restricted Fischer Park Ridge, NJ: Noyes Publications. United States Environmental Protection United States Environmental Population Studies in Epidemiology, proportions. Biometrics 38:215–220. Agency (US EPA) (1995a) Guidance for Protection Agency (US EPA) (2000a) F-344 and FBNF1 rats. Lab Animal Sci United States Environmental Protection Department of Human Services. risk characterisation. US Environmental Supplementary guidance for conducting Taylor R and Langley A. (1998). Exposure 45:635–640. Agency (US EPA) (1983) Environmental scenarios and exposure settings. National Protection Agency Science Policy Council. health risk assessment of chemical Stebbing M (2010). Exploring risk: Protection Agency. Good laboratory Environmental Health Forum Monographs Thybaud V, Aardema M, Clements J, Washington. mixtures. Risk Assessment Forum. Current issues in health risk perception in practice standards: Toxicology testing. Soil Series No. 2. 2nd edn. Adelaide: Dearfirled K, Galloway S, Hayashi M, pp. 1–209. Australia. PhD thesis, Monash University. Fed Reg 48: page 53922, and: Pesticide United States Environmental Protection South Australian Health Commission. Jacobsen-Kram D, Kirkland D, MacGregor Programs: Good Laboratory Practice Agency (US EPA) (1995b). The use of United States Environmental Protection Stebbing M, Katz E, Priestly B and JT, Marzin D, Ohyama W, Schuler M, Ten Berge WF, Zwart A, Appelman LM. Standards. Final Rule. Fed Reg 46 the benchmark dose approach in health Agency (US EPA) (2000b) Benchmark Abramson M (2008). Cancer cluster Suzuki H and Zeiger E (2007). Strategy for 1986. Concentration-time mortality (No. 230): pp. 53946–53969, Tuesday risk assessment. Risk Assessment dose technical guidance document. in the news: Risk perception, risk genotoxicity testing: Hazard identification response relationship of irritant and 11/29/83. Forum, EPA/630/R-94-007, United EPA/630/R-00/001. Risk Assessment communication and the media. Australian and risk assessment in relation to in vitro systemically acting vapours and gases. States Environmental Protection Agency, Forum. Epidemiologist 15:19–34. testing. Mutation Res. 627:41–58. J Hazrd Mater 13:301–309. Washington DC.
240 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 241 United States Environmental Protection United States Environmental United States Environmental Protection Vahter M (Ed.) (1982). Assessment of Weed DL. (2005). Weight of evidence: A Wilson RM, Sigal EA, Bacigalupo CM, Agency (US EPA) (2000c) Methodology Protection Agency (US EPA) (2003c). Agency (US EPA) (2007b). Users guide human exposure to lead and cadmium review of concept and methods. Risk Anal Willes RF and Munro IC (2000). Derivation for deriving ambient water quality criteria Recommendations of the Technical for the integrated exposure uptake through biological monitoring, National 25:1545–1557. of risk management criteria for chemicals for the protection of human health. Review Workgroup for Lead: An approach biokinetic model for lead in children Swedish Institute of Environmental of unknown toxic potency at contaminated Technical support document: Vol. 1 risk to assessing risks associated with adult (IEUBK). EPA 9285.7-42. Medicine and Karolinska Institute, Weil CS (1970). Significance of organ- sites. Human and Ecological Risk assessment. Washington, DC, Office of exposure to lead in soil. Technical Working Stockholm. weight changes in food safety evaluation. Assessment 6:131–139. Water, Office of Science and Technology, United States Environmental In Metabolic aspects of food safety. Group for Lead. EPA-540-R-03-001. Vainio H and Wilbourn J (1992). US Environmental Protection Agency Protection Agency (US EPA) (2007c). pp. 419–454, New York: Academic Press. Woodall GM (2008). An exposure- Identification of carcinogens within EPA-822-B-00-005. United States Environmental Protection Considerations for developing a response database for detailed toxicity Agency (US EPA) (2004a). An dosimetry-based cumulative risk the IARC Monograph program. Scand. Weil SC and McCollister DD (1963). data. Toxicol Appl Pharmacol 233:14–16. United States Environmental Protection examination of EPA risk assessment assessment approach for mixtures of J Work Environ. Health, 18 (Suppl Relationship between short- and long-term Agency (US EPA) (2001a). General principles and practices. Office of the environmental contaminants. National 1):64–73. feeding studies in designing an effective Woodall GM and Goldberg RB (2008). toxicity test. Agr Food Chem 11:486–491. Summary of the workshop on the power principles for performing aggregate Science Advisor EPA/100/B-04/001. Center for Environmental Assessment. Van den Berg M, Birnbaum LS, of aggregated toxicity data. Toxicol Appl exposure and risk assessments. Office of Office of Research and Development. Denison M, De Vito M, Farland W, United States Environmental Protection Weingand K, Brown G, Hall R, Davies D, Pharmacol 233:71–75. Pesticide Programs, Office of Prevention, US Environmental Protection Agency. Feeley M, Fiedler H et al. (2006). The Agency (US EPA) (2004b). Risk Gossett K, Neptun D, Waner T, Matsuzawa Pesticides and Toxic Substances, Cincinatti. EPA/600/R-07/064. 2005 World Health Organization re- assessment guidance for Superfund T, Salemink P, Froelke W, Provost JP, World Health Organization (1999a). Washington, DC: US EPA. evaluation of human and mammalian Vol 1: Human health evaluation manual United States Environmental Protection Dal-Negro G, Batchelor J, Nomura M, Chemical risk assessment. Training module toxic equivalency factors for dioxins and United States Environmental Protection (Part E Supplemental guidance for dermal Agency (US EPA) (2009a). Risk Groetsch H, Boink A, Kimball J, Woodman No. 3. WHO/PCS/99.2. Geneva: WHO. Agency (US EPA) (2001b). Preliminary dioxin-like compounds. Regul Toxicol D, York M, Fabianson-Johnson E, Lupart risk assessment). EPA/540/R/99/005. assessment guidance for Superfund World Health Organization (1999b). cumulative risk assessment of the Pharmacol. 45:9–23. M and Melloni E (1996). Harmonization Vol 1: Human health evaluation manual Principles for the assessment of organophosphorus pesticides. Office of United States Environmental Protection Vose D (2008) Risk analysis. A of animal clinical pathology testing in (Part F Supplemental guidance for risks to human health of chemicals. Pesticide Programs, Office of Prevention, Agency (US EPA) (2005a). Guidelines for quantitative guide. 3rd edn. John Wiley toxicity and safety studies. Joint Scientific inhalation risk assessment). EPA- Environmental Health Criteria 210. Pesticides, and Toxic Substances, carcinogenic risk assessment. EPA/630/P- & Sons. Committee for International Harmonization 540-R-070-002. IPCS/WHO. Geneva: WHO. Washington, DC: US EPA. 03/001B. of Clinical Pathology Testing. Walach JB (1996). Interpretation of United States Environmental Protection World Health Organization (WHO) (1978). United States Environmental Protection United States Environmental Protection diagnostic tests. 6th edn. Little Brown, Western Australia Department of Agency (US EPA) (2009b). Exposure Principles and methods for evaluating Agency (US EPA) (2002a). Guidance on Agency (US EPA) (2005b). Guidance Boston. Environment and Conservation (2006). factors handbook: 2009 update. the toxicity of chemicals. Environmental cumulative risk assessment of pesticide on selecting age groups for monitoring Contaminated sites management series: EPA/600/R-09/052A. Wallace LA and Ott WR (1982). Personal Health Criteria 6. International chemicals that have a common mechanism and assessing childhood exposures to community consultation guideline. Land monitors: A state-of-the-art survey. J Air Programme on Chemical Safety (IPCS)/ of toxicity. Office of Pesticide Programs, environmental contaminants. EPA/630/P- United States Environmental and Water Quality Branch. Pollut Control Assoc. 32, 601. WHO. Geneva: WHO. Office of Prevention, Pesticides and Toxic 03/00F. Protection Agency (US EPA) (2009c). Walker NJ, Crockett PW, Nyska A, Brix Western Australia Department of Health. Substances, Washington, DC: US EPA. Recommended toxicity equivalency World Health Organization (WHO) (1987). United States Environmental Protection AE, Jokinen MP, Sells DM, Hailey JR, (2006) Public health consultation: A guide factors (TEFs) for human health risk Principles for the safety assessment of United States Environmental Protection Agency (US EPA) (2005c). Supplemental Easterling M, Haseman JK, Yin M, for developers,
242 ENVIRONMENTAL HEALTH RISK MANAGEMENT ENVIRONMENTAL HEALTH RISK ASSESSMENT 243 World Health Organization (WHO) World Health Organization (WHO) (1993a). Guidelines for drinking water (2006c). Dermal absorption. quality. Vol. 1 recommendations. 2nd edn. Environmental Health Criteria 235. IPCS/ WHO. Geneva: WHO. WHO. Geneva: WHO.
World Health Organization (WHO) World Health Organization (WHO) (1993b). Biomarkers and risk (2006d). Principles for evaluating health assessment: Concepts and principles. risk in children associated with exposure Environmental Health Criteria 155. IPCS/ to chemicals. Environmental Health WHO. Geneva: WHO. Criteria 237. IPCS/WHO. Geneva: WHO.
World Health Organization (WHO) World Health Organization (WHO) (2009). (1994a). Assessing human health risks Risk characterization of microbiological of chemicals: Derivation of guidance hazards in food. Guidelines. values for health-based exposure limits. Microbiological risk assessment series 17. Environmental Health Criteria 170. IPCS/ WHO Food and Agriculture Organization of WHO. Geneva: WHO. the United Nations. Geneva: WHO.
World Health Organization (WHO) World Health Organization (WHO) (2010). (1996). WHO/FAO/IASEA Report on trace WHO guidelines for indoor air quality: elements in human nutrition and human Selected pollutants. WHO Regional Office health. Geneva: WHO. for Europe.
World Health Organization (WHO) Wullenweber A, Kroner O, Kohrman M, (1999a). Chemical risk assessment. Maier A, Dourson M, Rak A, Wexler P Training Module No. 3 WHO/PCS/99.2. and Tomljanovic C (2008). Resources for Geneva: WHO. global risk assessment: The International Toxicity Estimates for Risk Assessment World Health Organization (WHO) (ITER) and Risk Information Exchange (2000a). Global burden of disease. (RiskIE) databases. Toxicol Appl. Geneva: WHO. Pharmacol. 233:45–53.
World Health Organization (WHO) Yang RSH and Dennison JE (2007). Initial (2000b). Air quality guidelines for Europe. analyses of the relationship between 2nd edn. WHO regional publications. ‘thresholds’ of toxicity for individual European series. No. 91. chemicals and ‘interaction thresholds’ for chemical mixtures. Toxicol Appl. World Health Organization (WHO) (2001). Pharmacol. 223:133–138. Biomarkers in risk assessment: Validity and validation. Environmental Health Yassi A, Kjellstrom T, de Kok T and Criteria 222. IPCS/WHO. Geneva: WHO. Guidotti T (2001). Basic environmental health. World Organization, United World Health Organization (WHO) Nations Environment Programme, United (2006a) (World Health Organization). Nations Education Science and Cultural Guidelines for drinking-water quality, Organization, Council of Rectors of 3rd edn, incorporating first addendum, European Universities. Geneva: WHO.
244 ENVIRONMENTAL HEALTH RISK MANAGEMENT