Trace, and especially ultra-trace, element analysis require special facilities and infrastructure, are time-consuming and have not yet been developed to the point of complete automation as required by completely computerized clinical laboratories. The accuracy of the results can be invalidated by the greatest threat, sample contamination, and to a far lesser extent by element losses and changes in sample composition. The important of reliable samplingmay be illustrated by the problem of determining Mo in human serum or plasma. A study gives values published in the early seventies, which allow and estimation of the maximal extraneous addition of Mo to blood samples collected with a disposable stainless steel needle, which contains approximately 0.05-0.45% Mo(5). The amounts of Mo transferred from the needle and eventually from other sources to the serum were up to 20 times higher than the intrinsic concentration of Mo in serum. Such contamination precluded the investigation of possible changes in Mo concentrations occuring in certain pathological conditions. From a pragmatic point of view, it can be said that samples for the determaination of elements in the mg/L (mg/Kg) rang (Zn, Cu, Se, Rb, ...) can be treated in an ordinary analytical or clinical laboratory, at least when some basic procedures are observed. When concentrations down to the ug/L (ug/kg) or below (e.g., Al, V, Cr, Mn, Co, Ni, As, Mo, Cd, Hg, ....) have to be detected, the situation changes completely and all normal contamination precautionpractises become grossly inadequate.

The general thought is that if losses and/or additions are made during sample preparation, similar shortcomings will show up when handling a reference material. For this reason, it is paramount importance that the appropriate RM is chosen. such a RM would represent a good matrix match (identical is best) and one that contains a similar concentration of the analyte to the sample to be analyzed and in the same chemical form.

Many techniques require that the samples be introduced in liquid form. In the ideal situation a simple dilution of the biological fluid to reduce its viscosity and salt concentrationis sufficient for the subsequent detection with high selectivity, specificity and sensitivity. The usual procedure is acid digestion (nitric acid, sulphuric acid, perchloric acid, possibly supplemented with hydrogen peroxide). A revolution in sample digestion has occurred with the introduction of the microwave oven. For many biological matrices it is found to be faster, better controlled, more elegant and more amenable to automation than conventional open-beaker, reflux and closed-vessel pressurised techniques. However, one should remain cautious in case the microwave technique only partially breaks down the organic compounds, depending on the species present, the power, the medium and the duration.

When due attention is paid to the calibrants and the method of calibration, more often than not complete accordance with other existing data is produced when the exercise is repeated. One of the difficulties is that the classic solutions for calibration lack the matrix which plays such an important role in, e.g, the atomization characteristics of the analyte in a graphite furnace. The matrix may also play a major role in the signal-to-noise ratio. This may be exemplified with the Pb results in a fish tissue , with mean values produced by 12 laboratories varying between 0.057 and 0.118 mg/kg, using 6 different techniques. A fact ignored by some laboratories performing graphite furance atomic absorption spectrometry was the very poor signal-to-background ratio because the very low Pb content in the solution was measured in the highly acidic solutions obtained after digestion of the samples. To improve the sensitivity it would have been desirable to evaporate the acids first and then to dilute the final solution to about 0.1 M HNO3.

The need for reference materials certified for trace element species is evident. They are the best tool for harmonizing the results and providing quality assurance. The best method is to use a certified reference material (CRM) from external sources. They have the distinct advantage of wide acceptance, and they form the basis for inter-comparison of measurement systems and for testing data produced under diverse conditions and by various laboratories .

Quality assurance is a system of activities whose purpose is to provide to the producer or user of a product or a service the assurance that it meets defined standards of quality with a stated level of confidence(6).

Quality control is the overal system of activities whose purpose is to control the quality of a product or service so that it meets the needs of users. The aim is to provide quality that is satisfactory, adequate dependable, and economic(6).

Quality assessment is the overall system of activities whose purpose is to provide assurance that the overall quality control job is being done effectively. It involves a continuing evaluation of the products and of the performer of the production system(6).

Quality assurance refers to systems of external requirements placed upon laboratories by governmental agencies or private accreditation organisations. For analyses of trace elements in biological materials, these requirements may be regulatory in nature or voluntary, depending upon the matrix, the legal context and other factors(7).

Quality control refers to internal activities undertaken by laboratories to assure that their results are reliable. Quality control procedures may or may not be required by quality assurance systems. CRMs and RMs play a key role in evaluating these qualifications and allow the control of laboratory performance. Good laboratories appear to have developed an analytical methodology ensuring quality assurance through years of experience.

Two very important operational terms for describing the quality of the data are precision and accuracy. Whereas precision describes the variability of the individual results of replicate measurements, accuracy denotes the closeness of a measured value to the true value. The actual error of the analysis result is usually unknown whereas the error of the determination is inferred from its precision. Nearly all measurements will carry a certain degree of bias, i.e. the mean differs from the true or accepted value of the property measured due to some systematic error inherent to the method. CRMs are very useful for controlling both short-term and long-term repeatabilty, but this may also be done with internal RMs. The data can then be plotted sequentially in a Shewhart control chart on which warning limits and action limits are defiend. The results are considered to be out of control if: 1. The action line is exceeded. 2. The same warning line is exeeded twice in succession. 3. Successive measurements are on the same side of the line representing the mean value(8).

The evaluation of the data is relative to their end use. A measurement procedure capable of producing the same value within 10% or even 20% would be considered to be percise in some trace and surely in all ultratrace analyses. The requested accuracy and the precision of the analytical results must be sufficiently high to revealcritical differences.

When a laborattory obtains a result which is close enough to that of the certified value, then accuracy and traceability are established. When there is disagreement, then the user of the CRM is warned about the bias. Needless to say that it would be very unwise to apply a correction factor(9).

Quality assurance of sampling implies contrlling of the analytical blank. This "blank" not only contains the contamination with the analyte, but also the background-to-nosie signal during the measurement. An empirical rule in the case of trace element analysis is that the blank should never exceed one-third of the concentration level expected in the sample, in order to prevent minor errors in the two measured quantities from introducing large errors in the residual analytical signal. There is need for a better understanding of the contamination, for discovering ways to minimize the run-to-run variation in analytical practice, and for improving the signal-to-noise and sample-to-blank ratios. Quality control must therefore also cover this specific phase of the procedure.

1. Cornelis, R. (1991) A journey through the hazards of possible errors in the analysis of trace elements in body fluids and tissues. Microchim. Acta II, 37.

2. Vesterberg, O., G. Nordberg & D. Brune (1988) Database for trace element concentrations in bilogical samples. Fresenius Z. Anal. Chem. 332, 556.

3. Heydron, K., E. Damsgaard, N.A. Larsen & B. Nielsen (1979) Soureces of variability of trace element concentrations in human serum, in Nuclear Activation Techniques in the Life Sciences. International Atomic Energy Agency, Vienna, P.129.

4. Heydron, K. (1970) Environmental variation of arsenic levels in human blood determined by neutron activation analysis. Clin. Chim. Acta 28, 349.

5. BCR reference materials (community bureau of Reference) (1990) commission of the European Communities, Directorate General for Science, Research and development, B-1049 Brussels.

6. Taylor, J.K. (1987) Quality Assurance of Chemical Measurements. Lewis Publishers Inc., Michigan.

7. Brown, S.S. (1991) Quality assurance: achievements, problems, Prospects. In: Trace elements in health and disease (A. Aitio, A. Aro, J. Jarvisalo and H. Vainio, eds) The royal Socity of Chemistry, Cambridge, P. 37.

8. Griepink, B. (1990) The role of CRMs in measurements system. Fres. J. Anal. Chem. 338, 360.

9. Marchandise, H (1985) New Reference materials. Improvement of methods of measurements. Commission of the European Communities, Community Bureau of Reference, EUR 9921 EN. Luxembourg.