Frumkin, 2e Part 5: The Practice of Environmental Health. Chapter 29: Risk Assessment

Transcription

1 Frumkin, 2e Part 5: The Practice of Environmental Health Chapter 29: Risk Assessment

2 Risk Assessment Risk assessment is the process of identifying and evaluating adverse events that could occur in defined scenarios. In the environmental health setting, risk assessors focus on health impacts that might result from being exposed to a particular agent or from working in, living in, or visiting a particular environment. Environmental health risk assessment can be viewed as a quantitative framework for evaluating and combining evidence from toxicology, epidemiology, and other disciplines, with the goal of providing a basis for decision making.

3 Environmental Health Risk Assessment In 1983, the conceptual framework for environmental health risk assessment was formalized in a National Research Council (NRC) report, Risk Assessment in the Federal Government, commonly referred to as the Red Book. The Red Book divides risk assessment into four elements: hazard identification, dose response assessment, exposure assessment, and risk characterization.

4 Hazard Identification Hazard identification is the process of identifying and selecting environmental agent(s) and health effect(s) for assessment. This process includes causal inference for particular health outcomes, based on the strength of the toxicological and epidemiological evidence for causation. For example, The International Agency for Research on Cancer (IARC) has worked in hazard identification, publishing over ninety monographs, each evaluating whether or not the weight of evidence suggests that human exposure to an agent or to a group of related agents causes cancer.

5 Dose-Response Assessment Dose-response assessment attempts to describe the quantitative relationship between exposure and disease. Dose-response assessments frequently rely on mathematical models in order to estimate responses for exposures that fall between experimental dose groups, or for observational data for which doses are typically continuous with few or no repetitions. One well-known dose-response model for cancer is the linearized multistage model. This model, like many other cancer dose-response models, assumes that every molecule of exposure adds more risk of cancer. In contrast, threshold models assume that nobody exposed at a level below a critical threshold dose will develop cancer as the result of exposure.

6 Exposure Assessment Exposure assessment includes the estimation or measurement of the magnitude, duration, and timing of human exposures to the agent of concern. Although ideal exposure assessment would produce a full profile of each individual s exposures over time, in practice most exposure assessments are limited to estimating summary values, such as time-averaged exposure rates. In addition, many exposure assessments rely on default assumptions about media contact rates, such as water and soil ingestion rates, rather than attempting to estimate specific exposure factors for every individual or population of interest.

7 Risk Characterization final step of risk assessment It consists of combining the information from the other three steps in order to estimate the level of response for the identified health effects(s) at the specific level of exposure to the agent(s) of interest in the defined population. Mathematically, the approach consists of substituting the specific dose amount into the dose-response equation and computing the response level. The risk that is contributed by the exposure itself is often of more interest than the overall probability of response, so analysts often summarize the result in terms of the relative risk; the additional risk, also known as the attributable risk; or the excess risk.

8 Risk Characterization The Red Book and subsequent reports emphasize that uncertainties associated with risk estimation should be assessed and discussed as part of the risk characterization step.

9 Risk Management Risk managers are faced with the challenge of judging the significance of risks, comparing risks and costs for different risk management strategies, discussing these assessments with stakeholders, and finally, making appropriate decisions or recommendations. There are many different philosophies on risk management.

10 De Minimis Risk Many risk managers and regulatory policies rely on the concept of de minimis risk, the idea that some risks are so small that they are acceptable or insignificant from a societal perspective. However, activities or exposures that cause risks above the threshold are not necessarily unacceptable under this paradigm, which is sometimes used to screen out extremely small risks so that more attention can be paid to activities that pose larger risks.

11 Risk Benefit Analysis Some risks result from activities that are otherwise beneficial. When conducting risk-benefit analysis, it is important to examine the risks and benefits in a balanced manner.

12 Cost-Benefit Analysis Although the financial costs of risk abatement have always been of concern to affected businesses, they are increasingly considered by risk managers. In fact cost-benefit analysis is legally required in promulgating certain types of environmental health regulations.

13 Decision Analysis or Alternative Analysis It has been argued that the best decisions are made after considering all the relevant potential consequences of a variety of options, rather than focusing only on particular consequences of a single option. This approach differs from the first three approaches discussed here in that the focus is on the comparison of options, rather than on a single activity or exposure. Decision analysis may contain elements of the other approaches, such as cost-benefit analysis, and may be done qualitatively or quantitatively.

14 Precautionary Principle The precautionary principle is the idea that serious risks should be avoided or mitigated when possible, even when those risks are unlikely or uncertain.

15 Risk Assessment vs. Safety Assessment Although closely related, risk assessment and safety assessment differ in several important ways. Safety assessment (also referred to as regulatory risk assessment or regulatory toxicology) is commonly used by regulatory agencies to select reasonably safe exposure limits or concentration limits in food, water, air, or other parts of the environment. Although safety assessment often relies on hazard identification and dose-response modeling, its aim is to answer the question, What dose or concentration is safe? rather than to assess the likelihood of adverse health effects at a given dose or concentration.

16 Dose-Response Modeling Dose-response models play an important role in environmental health risk assessment, as they determine interpolations between tested doses. Although the multistage linearized model is a commonly used dose-response model for carcinogenesis, there are many competing models. The multistage model is biologically motivated; however, the linearized version does not use quantitative information on specific toxic events.

17 Dose-Response Modeling In contrast, mechanistic dose-response models, biologically based dose-response models, and biologically motivated dose-response models attempt to model the dose to disease process in more detail, in the hope of providing a better understanding of the true shape of the dose-response curve and a better basis for extrapolating results from laboratory animals to humans. Examples include toxicodynamic dose-response models that predict the influence of toxicants on critical events, such as stochastic cell proliferation models for carcinogenesis or neurodevelopment; toxicokinetic dose-response models that model the transport, metabolism, and disposition of toxicants that enter the body; and models that include both toxicokinetic and toxicodynamic processes.

18 Dose-Response Modeling All monotonic dose-response models can be expressed using a simple concept known as the tolerance distribution, which assumes that each individual has a specific tolerance for a toxicant. The model expressing the probability that the individual s tolerance is less than or equal to her dose is known as the probit model. Threshold dose-response models assume that there is no change in response at low exposures to an agent. Hormesis is a phenomenon in which a toxicant reduces the probability of an adverse response at low doses compared to the probability of the same response at a zero dose but increases the probability of an adverse response at higher doses, creating a dose-response curve that is not monotonic.

19 Dose-Response Modeling Nonparametric regression (also called semiparametric regression) is an appealing but entirely different approach to dose-response modeling. Rather than relying on a mathematical function or functions to describe the shape of the dose-response curve, nonparametric regression relies heavily on the dose response data set itself to determine the shape of the curve. The primary advantage of nonparametric regression is its ability to show such features of data sets without imposing many a priori restrictions on the shape of the model.

20 Uncertainty Analysis At its simplest, uncertainty analysis consists of a qualitative description of the sources of uncertainty in a risk assessment and their potential impacts. However, uncertainty analysis is often quantitative, providing a range or distribution of reasonable risk estimates. Approaches to quantitative uncertainty analysis include probabilistic analysis and interval analysis.

21 Uncertainty Analysis In interval analysis, one estimates the risk twice, once using best-case parameter estimates and then again using worst-case parameter estimates. Although they are seldom identified as such by risk analysts, statistical confidence intervals and prediction intervals are examples of another type of interval analysis. Probabilistic risk analysis describes risk using one or more probability distributions indicating the plausibility of an entire range of risk estimates. The most popular method among environmental health risk assessors is an informal approach called Monte Carlo simulation.

22 Uncertainty Analysis The Monte Carlo approach to probabilistic risk assessment requires additional steps after the initial quantitative risk assessment. First, probability distributions are selected to represent uncertainty, or variability, in the model parameters. Next, plausible sets of parameter values are randomly and repeatedly selected according to the specified probability distributions and correlation structure. Finally, after risk estimates have been calculated for many sets of parameters, the collection of risk estimates approximates the distribution of uncertainty regarding the risk.

23 An Assessment of Risk Assessment Risk assessment has been widely criticized over the years. Some of the objections to risk assessment have been scientific. For example, some question the use of assumptions when data are not available, others question the generalizability from animal studies, and others have alleged that focusing on single chemicals is inappropriate. Other objections to risk assessment have turned on policy decisions, such as the selection of margins of safety and the possible legitimization of some risks.

24 An Assessment of Risk Assessment In 2008, the National Research Council s Committee on Improving Risk Analysis Approaches Used by the U.S. EPA issued a landmark report, Science and Decisions: Advancing Risk Assessment (called the Grey Book). The report provided a comprehensive look at risk assessment since its inception in the 1980s, identified a series of problems, and recommended solutions.