National Academy of Sciences. Biosolids Applied to Land:
Advancing Standards and Practices

Chapter 4: Advances in Risk Assessment Since the Establishment of the
Part 503 Rule [abridged]

The committee's review of the risk assessment used to support the Part 503 rule was carried out in the context of current and emerging practice in risk assessment. The committee determined that its review of the risk assessment should communicate the committee's interpretation of how the risk-assessment process has evolved from the time the Part 503 rule was issued until present. Of particular interest to the committee were documents from EPA and the National Research Council (NRC) that propose and encourage methods that differ substantially from the methods used in the Part 503 risk assessment. This chapter provides a foundation and context for the following chapters.

This chapter first describes new approaches and considerations in risk assessment since the Part 503 rule (Standards for Use or Disposal of Sewage Sludge) was established in 1993 (40 CFR Part 503). It focuses on the changing priorities of cancer versus noncancer end points, acute versus chronic end points, probabilistic risk-assessment approaches, and the need to address aggregate exposures and cumulative risk. A brief description is then given of the changes in risk-assessment approaches of EPA over this period.

THE RISK-ASSESSMENT PROCESS

Risk assessment is a process for identifying potential adverse consequences along with their severity and likelihood. In contrast to other tools used for environmental evaluation and policy, the principal objective of the risk assessment and risk management approach is not to eliminate all risk but to quantify the risk and provide risk managers with tools to balance the level of risk against the cost of risk reduction, against competing risks, or against risks that are generally accepted as trivial or acceptable. Controlling the exposure of human populations to environmental contaminants in biosolids using a risk-based approach requires a definition of both an appropriate metric for assessing the impacts of contaminants on human health and a defensible process for assigning value to the predicted impacts. The end product of a risk-based approach to environmental management is either to identify an acceptable level of exposure or to prescribe the technical controls or political process needed to attain acceptable risk. Intervention can be achieved through technical or political controls.

Components of the Risk-Analysis Process
The NRC (1982, 1994) has divided and continues to divide the practice of risk analysis into two substantially different processes—risk assessment and risk management. Along with these processes are concurrent efforts to communicate and evaluate risk (NRC 1989, 1996). This section explores the evolution of the risk-assessment process over the last decade by considering the component steps in the process.

Risk assessment is the process of selecting and quantifying the adverse consequences that result from an action, such as application of biosolids to soils, or from inaction. A risk assessment begins with efforts to identify the potential hazards associated with a chemical or microbial agent and its use or occurrence. Hazard identification addresses the potential for harm but not the likelihood of harm. Risk characterization establishes the significance of an identified hazard by quantifying the likelihood and severity of exposure scenarios linked to that hazard. As applied to toxic agents, risk characterization has five principal elements: (1) quantification of sources and environmental concentrations in exposure media; (2) quantification of exposure to the target population and distribution of the dose among the population; (3) characterization of a dose-response function for all potential toxic agents that have been identified; (4) estimates of the number of people affected and severity of consequences expected within the population at risk; and (5) an assessment of the magnitude and sources of uncertainty that limit the precision of the estimate of consequences.

Risk management is the process of weighing policy alternatives and selecting the appropriate societal or institutional response. Risk management is used to integrate the results of a risk characterization with social, economic, and political valuation to reach a decision. The goal of the risk-management process is to establish the significance of the estimated risk, compare the costs of reducing this risk with the benefits gained, compare the estimated risks with the societal benefits derived from incurring the risk, and carry out the political and institutional process of reducing risk.

Linking the risk-assessment and risk-management processes are the concurrent efforts to evaluate and communicate risk. Risk evaluation is the process by which the risk-characterization and risk-management processes are reconciled with individual and societal valuations of risk (NRC 1996). A key step in this link is effective risk communication. According to the NRC (1989), risk communication has become more difficult in recent decades and common misconceptions often hamper communication efforts. In considering these issues, the NRC (1989) emphasizes that solving the problems of risk communication is as much about improving procedures as improving the content of risk messages.

Figure 4–1 provides a view of how the risk-analysis process might proceed for assessing the health impacts of pollutants in biosolids. Each of the major steps in this process involves one or more actions that are listed to the right of each major step.

Confronting Uncertainty and Variability
An important and often ignored final step in the risk characterization process is the characterization of uncertainties. Important sources of uncertainty and variability in risk assessments involve the data and models used. With incomplete data and models used to characterize contaminant transport representing heterogeneous geographic and climate regions, the variability and uncertainty associated with the resulting risk estimates are large.

In evaluations of uncertainty in risk assessment, Morgan et al. (1990) and Finkel (1990) distinguish among parameter uncertainty, model uncertainty, decision-rule uncertainty, and natural variability in any of the parameters and call for separate treatment of the different types of uncertainty. Probabilistic methods such as Monte Carlo analysis are available to evaluate uncertainty in parameters. According to Finkel (1990), model uncertainty derives from a number of actions, including the use of simplifications that might exclude relevant variables from the analysis; the use of surrogate variables that might not be appropriate for the variable of interest; the appearance of abnormal conditions that might occur in nature but that might not be appropriate in the model; and the use of incorrect model forms. Morgan et al. (1990) noted that relatively little research has been done on uncertainty or disagreement about what form of model to use. Decision-rule uncertainty applies to risk management and arises whenever ambiguity or controversy exists about quantifying or comparing social objectives. According to Finkel (1990, p. 16), "to take any actions using the outputs of a risk assessment, including the decision not to take action, one must be prepared to make a series of potentially controversial value judgments."

An important source of uncertainty in risk characterization is the development and application of dose-response models. Among the many issues that complicate the process of establishing a dose-response function is the variation in human susceptibility. In large heterogeneous populations, there are large variations in susceptibility to toxic effects. Those variations are due in part to variations in genetic predisposition to certain disease states, variations in age, and large variations in physical stresses and other chemical or non-chemical exposures that might be extant in the system of interest.

FINDINGS AND RECOMMENDATIONS

Finding: The Part 503 rule risk assessments were carried out more than a decade ago. In this chapter, the committee considered the likely impact of changes in risk-assessment practice in general and in various EPA offices in particular on the risk-assessment process for biosolids. The committee found that the development of methods in the broader academic community and the evolution of risk-assessment methods within various EPA offices and programs provide important benchmarks for the committee's assessment of the relevance and reliability of the Part 503 rule risk assessments. Of particular note are updates to the risk-assessment framework recommended by the NRC, the Presidential/Congressional Commission on Risk Assessment, and various EPA offices.

The risk-assessment methods and policies practiced and advocated at EPA have changed significantly, although not at the pace recommended by the NRC and the risk commission. As a result, the Part 503 rule, which has not been modified to account for any new methods and policies, is now inconsistent with current NRC recommendations and EPA policies within various offices. Particularly relevant examples of the inconsistency are the absence of stakeholder participation and the lack of explicit treatment of uncertainty and variability.

Recommendation: Because of the significant changes in risk-assessment methods and policies over the last decade, EPA should revise and update the Part 503 rule risk assessments. Important developments include recognition of the need to include stakeholders throughout the risk-assessment process, improvements in measuring and predicting adverse health effects, advances in measuring and predicting exposure, explicit treatment of uncertainty and variability, and improvements in describing and communicating risk. EPA should consider how the updated risk assessments would change the risk-management process. A similar approach can be taken with the issue of biological agent risks.

Finding: In recent years, health-effects research has made use of large-scale studies of human health end points at multiple sites. Health-effects research has also focused on early indicators of outcome, making it possible to shorten the time between the exposure and the observation of an effect. In addition, more use has been made of meta-analysis, better modeling of dose-response relationships, and more sophisticated regression models. These improvements make possible more site-specific assessments of the impacts of biosolids land-application practices.

Managing exposure of human populations to environmental contaminants using a risk-based approach requires an accurate metric for the impacts of contaminants on human health and a reliable process for monitoring and recording the exposures within populations assumed to be at risk. Over the past decade, the practitioners of exposure assessment have made important improvements in methods to measure and model source-to-dose relationships. These improvements have been made through greater use of time-activity surveys, personal monitors, and biomarkers of exposure, and they have made it possible to confirm some of the exposures predicted in risk assessments.

Recommendation: Many of the measures of risk used in developing the Part 503 rule guidelines cannot be monitored. Because of that inability to monitor, the committee acknowledges that EPA must perform theoretical risk assessments. Nevertheless, there is a continuing need to provide some measures of performance that can be monitored (e.g., concentrations of selected chemicals in exposure media, such as indoor air, house dust, or tap water of residences near land-application sites; and exposure biomarkers in the blood or urine of nearby residents). Recent improvements in health surveillance and exposure monitoring provide new opportunities for EPA to develop more explicit and measurable metrics of performance for biosolids land-application practices.

Finding: Advancements in monitoring health outcomes and exposure have resulted in improvements in the description and communication of risk. In particular, improved exposure assessments have led to better exposure classification in health-effects studies. Better descriptions of risk are available, using benchmark dose and margin of exposure to communicate hazard and risk in place of risk of death, hazard quotients, or exposure-potency product relationships. There have also been improved methods for prioritizing compounds using measures of risk.

Recommendation: In making revisions to the Part 503 rule risk assessment, EPA must strike a balance between expending resources to carry out site-specific data collection and expending resources to model and assess risk using existing information. In light of improvements in exposure and health monitoring, the committee encourages EPA to consider options carefully for collecting new data in support of risk-assessment assumptions before resorting to another risk assessment that relies only on existing data, models, and default assumptions. Among the data that would be of value are data on proximity of receptors to land-application sites; surveys of activities that could increase direct and indirect exposures; and samples of biosolids, air, vegetation, runoff, groundwater, and soil in environments surrounding land-application sites. In addition, EPA should conduct site-specific surveys of performance (e.g., monitor the extent to which rates and depth of application are consistent with risk-assessment assumptions) and scientifically relevant studies of health complaints.

Finding: Risk assessments make use of a number of assumptions to define chemical loading in biosolids that pose no undue risk to surrounding populations. Implicit in this process is the premise that these assumptions and the associated demographic and operational conditions will persist. However, there are no guidelines to ensure that these conditions persist.

Recommendation: Because there are no guidelines to ensure that conditions assumed in the risk assessment actually transpire, the committee recommends that the Part 503 rule provide guidance for periodic reassessments that will be used to ensure that the demographic and operational conditions of biosolids land application are consistent with the assumptions of the applicable risk assessment.