Quality control in integrated human exposure monitoring of lead and cadmium

Abstract

The Human Exposure Assessment Location, HEAL, Project is implemented by WHO/UNEP in close collaboration with national agencies in different countries. In a first phase methods for exposure monitoring of Pb, Cd, DDT, HCB and NO2 will be tested in China, Japan, Sweden, USA and Yugoslavia, possibly also Brazil and India. The Pb/Cd study, which involves measurements of lead and cadmium in blood, duplicate diets, feces and inhaled air, collected by a group of non-smoking women in each participating country, is coordinated by a Technical Coordinating Centre (TCC) in Sweden. In order to assure accuracy and comparability of data an extensive quality assurance programme has been developed. Quality control (QC) samples for lead and cadmium in blood, feces, air filters, dust and diets have been prepared. Sets of 4–6 External Quality Control (EQC) samples, the metal concentrations of which are unknown to the laboratories, and 1–2 Internal Quality Control (IQC) samples with stated Pb/Cd-levels, have been distributed. The analytical performance evaluation is based on linear regression analysis of reported results (y) versus reference values (x). Criteria for how much the regression line may deviate (Maximum Accepted Deviation, MAD) from the ideal line (y=x) have been developed. A power of 90% is employed, which means that the actual acceptance interval for the regression line is slightly narrower than the MAD-interval. This procedure gives an estimate of the maximum systematic errors involved in the analysis. The MAD criteria for the evaluation of QC results are based on the data quality requirements as well as the feasibility of the analytical techniques. The results of the first QC rounds show that good analytical performance on one QC sample is no guarantee for good results at higher or lower concentrations or good results with other types of samples. Furthermore, analytical performance may vary with time.