You are here:
CREATING A DECISION CONTEXT FOR COMPARATIVE ANALYSIS AND CONSISTENT APPLICATION OF INHALATION DOSIMETRY MODELS IN CHILDREN'S RISK ASSESSMENT
Citation:
CRAWFORD-BROWN, D. J. AND A. M. JARABEK. CREATING A DECISION CONTEXT FOR COMPARATIVE ANALYSIS AND CONSISTENT APPLICATION OF INHALATION DOSIMETRY MODELS IN CHILDREN'S RISK ASSESSMENT . Presented at Society of Toxicology Annual Meeting, Charlotte, NC, March 25 - 29, 2007.
Description:
Estimation of risks to children from exposure to airborne pollutants is often complicated by the lack of reliable epidemiological data specific to this age group. As a result, risks are generally estimated from extrapolations based on data obtained in other human age groups (e.g., adults) or experimental animals. This presents several significant challenges when comparing predictions from different models, i.e., there is a need to account for (1) potential differences in the mode of action (MOA), due to either pharmacokinetic (PK) or pharmacodynamic (PD) differences across the different data sources; (2) use and integration of mechanistic data that may range in analysis scale from in vitro biochemical studies to physiological/ morphological data to clinical/epidemiological data; and (3) analysis of the quality and reliability, or epistemic status, of any resulting predictions of children¿s risk. A variety of approaches have been used to address these issues. The simplest is the application of default uncertainty factors (UF) for intrahuman and interspecies variability . More recent approaches have incorporated mechanistic dosimetric adjustment factors that address aspects of interspecies differences in PK (US EPA, 1994). Data are being developed to provide PK parameters and PD endpoints for different life stages in experimental animals and advances in biomonitoring technologies may provide measurements directly in humans. While these latter approaches offer great promise to bring the full range of MOA data to bear on critical extrapolations, evaluation of their merit typically leads to mismatched comparisons between various models. For example, one model may be empirically estimating age differences measured at the population level while another model is mechanistically predicting, to the extent that understanding and data allow, potential differences at the tissue level. This suggests the need for a framework to systematically analyze and compare model structures, evaluate aspects of data interpretation, integration, and reliability to describe a given process, and to rationally assess the quality of resulting model predictions. We propose a decision analytic framework to meet these challenges. Our framework calls first for a conceptual logic tree to describe the process of interest in the target context (e.g., reactive gas uptake in nasal tissue of children) and to identify key parameters leading from exposure to clinical effect. This conceptual model determines the relevance of any given body of data, parameter values, and extrapolation premises used to support a given computational model structure and predictions. Based on these considerations, the framework assesses the rationality of results from different models to describe the process in the target context, including evaluation of the strengths and weaknesses of alternative PK and PD parameters and model forms, the reliability of resulting risk estimates, and the key sources of residual uncertainty. Explicit presentation of reliability and uncertainty will aid the design of research programs targeted at increasing the accuracy of risk predictions for children. (This abstract does not reflect Agency policy.)