How clean is clean? A comparative analysis of EPA's reliance on risk assessment in its contaminated site remediation programs

Over the past 12 years, Congress has repeatedly attempted to mandate, legislatively, that the U.S. Environmental Protection Agency (EPA) use best science-based risk assessments. Risk assessment is an important tool to assist in the determination of how clean is clean at sites contaminated with hazardous waste and substances. The results of a risk assessment are fed into the decisionmaking process referred to as risk management. EPA has three site remediation programs: the Superfund Remedial Response Action Program, the RCRA Corrective Action Program, and the Underground Storage Tank (UST) Corrective Action Program. Each program was created within a few years of each other, resides in the same overall program office, has the same statutory cleanup goal of protection of human health and the environment, adopted an original cleanup goal of remediation to background levels, requires the same carcinogenic risk protection level—10 −4 to 10−6, requires the same non-carcinogenic risk protection level—Hazard Index of 1, addresses some similar contaminants (benzene), addresses the same environmental media, uses the same toxicological data, and uses the same default exposure assumptions. However, the three programs' reliance on risk assessment in determining how clean is clean have become different. This difference occurred in spite of EPA's agency-wide efforts to harmonize its use of risk assessment. Previous studies suggest that various social, technical, political, organizational, and economic factors can influence the reliance on risk assessment. The reliance on risk assessment by EPA's three site remediation programs has not been compared. Through a comparative case study analysis based on interviews, internal EPA documentation, and external documentation, this study examines the impact of various social, technical, political, organizational, and economic factors on each program's reliance on risk assessment. The study found that program resources, degree of federal control, the remediation funding source, the adoption of Action Levels, and public risk perception universally all had a major impact. However, these major factors had different outcomes. For example, all the major factors decreased reliance on risk assessment for both the RCRA and UST Corrective Action Programs. In contrast, these same factors increased reliance for the Superfund program. A primary cause for this inverse phenomenon is the public's risk perception, which is very high for Superfund and low for the RCRA and UST programs. The study found that interest group influence, congressional influence, and problem size had some impact on programmatic reliance. Scientific uncertainty was found to have no appreciable effect. The study's conclusions are that (1) risk assessment has become primarily an economic decision and (2) risk assessment is vulnerable to organizational biases. The policy recommendations are (1) do not mandate risk assessments, (2) increase the development of standardized, numerical health-based cleanup goals/levels, and (3) combine all site remediation programs into one program office.