The use of Chernobyl data to test model predictions for interindividual variability of 137Cs concentrations in humans

Data sets assembled in the aftermath of the Chernobyl accident as a part of the International Atomic Energy Agencyʹs model testing program (VAMP) have provided a rare opportunity for ‘blind-testing’ predictions made with exposure assessment models. Measurements of Chernobyl-derived 137Cs in Central Bohemia (Czech Republic) and southern Finland were used to test model predictions for a number of endpoints, including the distribution of whole-body concentrations of 137Cs in adults in these regions at specified time points. This test endpoint required separation of uncertainty due to stochastic variability (aleatoric uncertainty) and uncertainty due to lack of knowledge about fixed but unknown values (epistemic uncertainty). Predictions of the distribution of whole-body 137Cs concentrations were made by a minority of the participants in these model-testing exercises. Major reasons for misprediction included bias in the bioavailability of 137Cs in soil and misestimation of the total intake of 137Cs in the diet. Overestimation of the amount of interindividual variability often resulted from confusion of uncertainty with variability. The spreads of the distributions for parameters describing interindividual variability were frequently increased to compensate for lack of knowledge about the uptake and metabolism of 137Cs in the population. Accurate results produced by participants are attributable both to a participantʹs access to additional site-specific data or choice of appropriate site-specific assumptions and to the effects of compensatory errors.