Source: BVL

Proficiency tests (PTs) or interlaboratory studies are an important means of quality control in chemical analysis. The EU as well as the National Reference Laboratories are legally bound to organise such tests. Laboratories accredited according to ISO 17025 – which is the case for all laboratories active in official residue control - are obliged to participate regularly in these tests. The successful participation in proficiency tests is a prerequisite for obtaining accreditation or for renewing an existing accreditation.

In the framework of these tests the individual laboratories can compare their own results obtained from the analysis of a certain sample with the results of a group of other laboratories, and thus gain information about the proficiency of their own test methods.

At the beginning of each proficiency test the sample material is produced, which must be identical for all participants. The concentrations of the substances contained in the samples (for example veterinary drugs in muscle) should lie within a range which can realistically be expected in residue control. Randomly selected sample portions from the material are checked to see whether the substances to be analysed are distributed homogeneously. Only if this is the case is a sensible evaluation of the study possible. Equally it must be proven by means of appropriate tests that the material does not change, at least not for the duration of the study.

The samples must be distributed to the participating laboratories in such a way as to exclude any changes to the material during shipment. Depending on the samples and substance group concerned this might mean shipment in isolated packages or even shipment on dry ice.

The laboratories analyse the samples with their routine test methods. The results must be submitted by the laboratories within a certain period of time, since timeliness may also be a criterion in the assessment of a laboratory’s quality.

For the evaluation of the study the following two parameters are of particular importance: the target value and the accepted dispersion of the results between the laboratories (“target standard deviation”). On this basis the tolerable deviation of a laboratory value from the target value is calculated.

There are different ways of establishing the target value. It may be the case, for example, that the concentrations of the substances to be analysed in the sample are known exactly, e.g. from the production process of the samples or from pre-analyses carried out by selected, particularly qualified laboratories. Another possibility is to calculate the mean on the basis of the results from all participants. This approach makes sense for example in an animal study when samples are obtained by treating animals in a given way. However, it presupposes that the laboratories participating in the study have comparable proficiency levels and that at least the majority are not complete "beginners" in this field of analysis.

In such cases it is also recommended to apply so-called “robust statistics” for calculating the target value. These are methods of calculation which produce meaningful results even if the data deviate from a symmetric, normal distribution (e.g. clearly showing “outliers”, which would mean that the results of particular laboratories would have a comparatively large influence on the calculation of the target value).

The target standard deviation, from which the limits of tolerance are calculated, can either be established on the basis of the actual dispersion of the data, or a fixed value is set. This may be the case when certain legal regulations are to be taken into consideration.

The requirements for carrying out proficiency tests are now described in an ISO standard (DIN EN ISO/IEC 17043).

The diagram below depicts the results of a proficiency test on the determination of benzo[a]pyrene in smoked fish.

It should be kept in mind that measurement values are always afflicted with a certain uncertainty, depending on the analytical method used. However, this does not represent a loss in quality; it is unavoidable in science.

Nevertheless the measurement uncertainty of a method must not be great enough to jeopardise the meaningfulness of the assessed results. Another important aspect in this context are the existing legal regulations about the minimum required precision of test methods, which are in force for many substance groups (i.e. the maximum acceptable reproducibility standard deviation).

Therefore the diagram also shows the uncertainty of the target value (green belt) as well as the measurement uncertainties determined by the respective laboratories for their methods (bars).

Moreover the tolerance limits are depicted, within which a deviation from the target value is regarded as being acceptable. One laboratory was not successful in this test, as the mean of its three submitted measurement values was below the tolerance limit. As a consequence, this laboratory is under the obligation to look into the reasons for this outcome and, if necessary, to optimise its method.