process such as the four-step process described in Science and Judgment in Risk Assessment (NAS/NRC, 1994):
Hazard Identification: Identification of the potentially hazardous agent and a description of the specific forms of toxicity that may be expressed in exposed populations.
Dose-Response Assessment: Evaluation of the conditions under which an agent may be toxic and an evaluation of the quantitative relationship between the dose and the toxic response.
Exposure Assessment: Specification of the population that might be exposed, identification of the routes by which exposure can occur, and estimates of the magnitude and duration of exposure that people are likely to receive.
Risk Characterization: Development of a qualitative or quantitative estimate of the hazards associated with the agent that will be realized in exposed people. This includes a full discussion of the uncertainties associated with the estimates of risk.
In the case of risks of exposure to radiation at low levels (less than 0.1 gray (Gy) effective dose equivalent delivered instantaneously or less than 0.2 Gy delivered at a low dose rate), the scientific uncertainty about negative side effects, such as long-term cancer, is considerable. Radiogenic cancer (i.e., neoplasms caused by exposure to ionizing radiation) has not been observed in humans at low levels of exposure, and it is unknown whether such negative sequelae to diagnosis actually exist. If they exist, it is beyond the range of current science and medicine to observe and measure them.
Because of this scientific uncertainty, a public policy decision has had to be made regarding what approach to use in setting standards for protection of individuals against potential problems secondary to low-level exposures to radiation. U.S. regulatory agencies currently use a model that describes radiation injury as a linear function of radiation dose that has no lower threshold; this is called the linear, no-threshold model. Scientific consensus groups, including the International Commission on Radiological Protection (ICRP), the National Council on Radiation Protection and Measurements (NCRP), and a National Research Council Committee on the Biological Effects of Ionizing Radiation (BEIR), have also endorsed the use of this model for risk assessment (see NCRP, 1987; NAS/NRC, 1990; ICRP, 1991). This approach reflects an understandable tendency to be conservative in choosing analytical models that emphasize public safety.
In the linear, no-threshold model, data from high levels of exposure (greater than 0.5 Gy effective dose equivalent), where some radiogenic cancers have been observed in humans, are extrapolated to low levels of exposure, where radiogenic cancer has not been observed. The model can then be used to predict the risks of cancer at various levels of exposure; it specifically predicts very