The 27th Norwegian intercomparison study for analysis of fresh water, designated 19-27, was organized in January – April 2019 with 44 participants. It comprised analysis of three sample sets of four samples (A-D, E-H, I-L), and two sample sets of two samples (M-N and O-P), made by adding known amounts of stoichiometric material to water from Borstadtjern og Langsjøen in Aurskog-Høland municipality, which had been filtered through membrane filter with pore size 0,45 μm. The program included 32 different parameters: pH, conductivity, turbidity, colour, UV-absorption, sodium, potassium, calcium, magnesium, hardness, alkalinity, chloride, sulfate, fluoride, total organic carbon, chemical oxygen demand (CODMn), phosphate, total phosphorus, ammonium, nitrate, total nitrogen, aluminium, lead, iron, cadmium, copper, chromium, manganese, nickel, zinc, antimony and arsenic. The analysis was largely carried out according to Norwegian standards or equivalent methods (Table B1).
The median of the participants' results, after outliers have been omitted, is selected as the "true" value. The acceptance limit is normally set to ± 20 % of the average true value for the two samples that form a pair. The results are presented graphically in a Youden diagram, where a circle with the acceptance limit as radius is drawn. The result pairs within the circle are affected by a total error less than the limit and are hence considered acceptable (Appendix A).
In total 84 % of participants' results were acceptable, a result which is comparable to the latest intercomparison studies (Table 1). 21 parameters had acceptable results above 80 %. 11 parameters had 70-79 % and only one parameter had acceptable results below 70 %.
Some laboratories' results are dominated by large systematic or random deviations. As in previous intercomparison studies, some results have been reported in the wrong unit of measurement. This shows how all parts of the chain of analysis needs to have quality control if reliable data are to be achieved. Systematic deviations are particularly prominent in some instrumental analyses. In these cases, the goal of the troubleshooting should be to clarify whether the error is constant and/or dependent on the concentration, to obtain an indication of what the cause of the error could be (Appendix A). In some cases, the analysis methods haven't been sufficiently sensitive for the concentration level of the analytes. When considering their results, laboratories must consider the low level of concentration of analytes in some of the samples. Internal quality control [Hovind et al. 2006] is necessary for the running evaluation of the laboratorie' own methods and routines. If possible, the accuracy of the results should be controlled by certified reference materials (CRM's), or alternatively by reanalyzing samples from intercomparison studies they have previously participated in.
Vis fullstendig beskrivelse