1. RESPONSE TO THE CRITIQUE 1.1. Bounding Analysis is Not an Alternative to Standard Risk Assessment

• Main Authors: Elizabeth A. Casman, Minh Ha-duong, M. Granger Morgan
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Subjects: Greenland
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.410.7829; http://halshs.archives-ouvertes.fr/docs/00/02/82/55/PDF/Response_to_Commentary.pdf

Greenland misses the point of our method. He suggests that we have proposed “bounding analysis as an alternative to probabilistic risk assessment, or even as a standard to judge the latter. ” This is not correct. Our opening paragraphs and our previously published qualitative arguments on this same subject (1) clearly explain that we are only proposing this approach to deal with situations in which the existing data are insufficient to support standard methods of risk assessment. The idea is to use the parts of the problem that can be characterized using conventional probabilistic risk analysis to back out an upper bound on the contribution made by the causes for which there are scanty data. The application of conventional probabilistic methods to such poorly understood causes can yield broad probability distributions, which, when added to the better defined estimates of cases due to the well-known causes, yield an estimate of the sum that includes many more cases than the number that is known to have occurred. In such circumstances we argue that it should be possible to bound the estimate of the number of deaths from poorly understood causes. When adequate science is available to characterize all the exposures that contribute to a specific health