You are here:
Minefields Associated with Mining Data from Peer-reviewed Literature
Citation:
RUSSOM, C. L. Minefields Associated with Mining Data from Peer-reviewed Literature. Presented at Society of Environmental Toxicology and Chemistry, Boston, MA, November 13 - 17, 2011.
Impact/Purpose:
This presentation will cover limitations and information gaps related to published research results; how these information gaps may affect the inclusion of a study in products of research and in risk assessments (e.g. models, Agency benchmarks, etc.); and possible solutions such as open access data repositories.
Description:
The USEPA’s ECOTOX database is the largest compilation of ecotoxicity study results, providing information on the adverse effects of single chemical stressors to ecologically relevant aquatic and terrestrial species. The primary source of data included in the ECOTOX database is the peer-reviewed literature, with information gathering accomplished through manual data abstraction and data entry. ECOTOX is used by all sectors of the environmental toxicology community in risk assessments, screening and prioritization exercises, data analysis projects, and species and chemical extrapolation model development. Typically an evaluation of the relevance, adequacy, and reliability of a test result is conducted prior to using it in risk assessment and modeling efforts. Therefore, critical to the usability of this data is the complete reporting of experimental design and test results by authors. The ECOTOX database is designed to capture, in discrete data fields, all aspects of test methods and associated results, but assumptions about test protocols, and interpretations of the data are not made. Therefore many data fields remain unpopulated due to lack of detail provided in the publication. This presentation will cover limitations and information gaps related to published research results; how these information gaps may affect the inclusion of a study in products of research and in risk assessments (e.g. models, Agency benchmarks, etc.); and possible solutions such as open access data repositories.