Epidemiology and Risk Assessment: Examples From the US EPA Integrated Risk Information System

Abstract

ISEE-184 Objective: The Integrated Risk Information System (IRIS) is a public use electronic database (www.epa.gov/iris) developed by the US Environmental Protection Agency. It currently receives more than 250,000 Web-page hits per month from public and private sector users in over 100 countries. This searchable database of 545 chemicals includes qualitative and quantitative health effects information pertaining to the hazard identification and dose-response components of the risk assessment process. Our objective was to identify assessments that used human data for qualitative and quantitative risk estimation. Material and Methods: Two reviewers independently abstracted information from the IRIS database pertaining to studies used to derive the reference dose and reference concentrations for oral and inhalation exposures for noncancer effects, cancer weight of evidence descriptors, and cancer inhalation unit risks and oral slope factors. Results: Data from human studies were used to derive a reference dose for 27 chemicals and a reference concentration for 12 chemicals. Twelve assessments (arsenic, acrylonitrile, asbestos, benzene, benzidine, beryllium, 1,3-butadiene, cadmium, chromium (VI), coke oven emissions, nickel refinery dust, and nickel subsulfide) used human data to estimate a cancer inhalation unit risk or oral cancer slope factor. The number of studies available pertaining to a specific chemical, the extent to which exposures to other carcinogens was able to be characterized, the available data on levels of exposure, and aspects of the data presentation were some of the key issues affecting use of the data from studies in humans. Conclusions: Epidemiology data have played an important role in the quantification of risk estimates in approximately 10% of the IRIS assessments to date. The role of epidemiology in risk assessment could increase, as specific issues regarding the design and presentation of data are incorporated into studies.