Abstract: The problem of low-dose extrapolation is common in risk evaluation of carcinogens. To set safety standards, regulatory agencies often employ the benchmark dose (BMD) method with a default model of low-dose-linearity. They claim this approach is inherently conservative, leading to dose levels that are considered protective of the publics health. These dose levels have been historically referred to as Virtually Safe Doses (VSD) and they correspond to doses for which the upper bound on the projected lifetime incremental risk is, for example, 1 in 1,000,000. However, for carcinogens that are directly or indirectly beneficial, these VSD may be unpractical and/or excessively protective of the public's health.
This paper extends the framework in Fygenson (2008) to address the question of just how conservative is the current BMD method and provides, for the first time, a lower bound on the projected lifetime incremental risk from the so called VSD. The proposed lower bound complements the upper bound derived by the BMD method and can lead to more productive risk/benefit analyses.
Key words and phrases: Benchmark dose, low-dose extrapolation, model uncertainties, pessimistic distributions.