Estimation of exposure levels by measurements and models

Summary Results from environmental and biological monitoring programs are abundant, but it is a question to what extent such data can be used for exposure assessments. The quality of such data depends not only on sampling and analytical errors, but also, and probably more, on the sampling strategy used. Therefore, uncritically use of such results can be biased and associated with large variations. The size of bias between two studies carried out in the same industry was found to be a factor of 5–7. Samples that are not representative of the exposed population studied are misleading. An estimation of exposure to cobalt in the porcelain industry illustrates the distinction between biological measurement data and air cobalt measurement data done on selected individuals. Following improvement of the working environment during the period 1983–1990, the change of cobalt concentration in urine indicates a reduction of cobalt exposure with a factor of 10, but air monitoring data do not verify that figure. Furthermore, results must always be interpreted with caution, when used beyond the purpose for what they were originally made. Great care should be taken to secure that data are representative and a quality assurance-quality control program should be used to ensure data quality.