Health & Environmental Research Online (HERO)


Print Feedback Export to File
632561 
Technical Report 
Hypothesis-based weight of evidence: An approach to hazard identification and to uncertainty analysis for quantitative risk assessment 
Rhomberg, LR 
2007 
Aerospace Industries Association of America, Inc 
Arlington, VA 
Human health risk assessment consists of bringing to bear a diverse body of in vitro and animal testing results, epidemiologic studies, and background toxicological knowledge on the question of whether occupational or environmental exposures to a substance should be regarded as capable of impairing the health of an exposed population. The challenge is that the body of scientific information is rarely definitive — information is typically indirect (animal studies are applied to humans, high-dose studies are extrapolated to low doses), it is often incomplete (few strains/species are tested, epidemiologic studies may lack good exposure information), and it frequently contains apparent contradictions (endpoints may be discordant among studies, genotoxicity tests may include positive and negative outcomes). As a result, conclusions about potential human risks from chemical exposure are not firm and certain deductions; rather they are statements about possibilities, the evidentiary basis for which may vary from very strong to tenuous depending on the case. The task, therefore, is not only to make reasoned inferences about the potential for human impact, but just as importantly, to gauge and effectively to communicate how compelling those uncertain inferences should be deemed to be, acknowledging and giving proper consideration to the existence of contrary data and alternative scientifically plausible interpretations. There are qualitative questions (whether the agent should be regarded as capable of causing particular
effects in exposed humans—i.e., uncertainty in hazard identification) and quantitative questions (how accurate and precise the projected dose-dependent magnitude or probability of effect should be understood to be).