Applications of evidence synthesis in chemical health risk assessment

Abstract

This thesis aimed to combine all available evidence within and across evidence streams, i.e. evidence synthesis, to strengthen chemical risk assessment. The evidence considered was published data of the following evidence streams: epidemiologic studies, molecular epidemiology studies, animal studies and mechanistic assays. For the purpose of evidence synthesis, advanced statistical and biological informed tools and approaches were used including meta-analysis, meta-regression, text mining, systematic reviews, systems biology, physiological based kinetic (PBK) models, and Bayesian inference. Most work performed in this thesis is on two exemplary data rich chemicals: benzene and diisocyanates. The case-studies of this thesis illustrate that human biomonitoring studies can be a valuable source of information for both enhancing exposure assessment and hazard characterization. Tooling such as text mining and network visualization tools can help to make evidence identification for hazard identification more efficient. And evidence from human (epidemiology and biomonitoring) and animal studies can also be combined together, but this does require a large number of assumptions. The chapters in this thesis demonstrate that a wide range of interdisciplinary approaches are required to be able to integrate data, for it to become optimally useful for chemical risk assessment. The methods and tooling in this thesis are mostly applied to chemicals for which a large amount of data is known. In the Discussion chapter of this thesis, suggestions are given for approaches that can be used for chemicals for which only little information is known, possibly in conjunction with the new scientific approaches described in the other Chapters of this Thesis.