Bactoscan-8000 cell count instruments from the Swiss Quality Testing Laboratories (MIBD): starting routines 2003

In 2003 the six Swiss quality testing laboratories for commercial milk delivered 1664 data sets with the results of the daily starting routines of their instruments for detecting TMC in raw milk to the Swiss federal central office MIBD [Dairy Industry Inspection and Advisory Service]. A detailed evaluation was carried out on the basis of eight instruments in total to document measuring accuracy, safeguarding of measuring level, compliance with official requirements and also technical measuring limits and possible improvements. 7 criteria were assessed from each data set. 93.6% of data sets fulfilled all criteria4.9% contained one or more occasions when tolerance limits were exceeded and 1.5% contained single or multiple missing or wrongly recorded data (outliers). The 99 non-conformities are not evenly distributed over the different criteria, laboratories, instruments and batches of standard milk. The criteria «high-precision», «blank value-maximum» and «blank value-total» are easily complied with: the frequency of non-compliance is clearly less than 1.0% in each case. In the case of «high-accuracy», the frequency of non-compliance is 1.2%, whereas it is 1.5% for «low-precision» and «low-accuracy » and finally 1.6% for “pilot samples-precision”. There are clear differences in frequency of non-compliance between the instruments. A maximum of 2.7% of the criteria were not fulfilled by one instrument, the average being 0.9%. Since we are only dealing with BactoScan 8000 instruments from the firm Foss-Electric, the differences are probably mainly due to operation of the instruments, handling of the samples, and perhaps to servicing and technical support. Repeatability and reproducibility are satisfactory to good overall. They depend on the cell count: RSDr (Relative Standard Deviation of Repeatability) is 4.7 and RSDR (Relative Standard Deviation of Reproducibility) 6.9 for standard «high»for standard «low», 6.1 and8.3 (all measured in imp/ml). In logarithmized units log(imp/ ml) the corresponding values are 0.370, 0.556, 0.538 and 0.711. All these parameters are clearly higher than last year. Similarly, regressions to the target value are somewhat weaker, the corresponding correlation coefficients being 0.845 for standard «high» and 0.855 for standard «low». A special booklet with quality control charts and instrumentspecific graphics will be compiled for each of the 8 instruments and given to participants. Although the accredited laboratories are responsible for checking and observance of the tolerance limits, subsequent comparative and detailed evaluation proves to be imperative. This report documents and certifies to the participating laboratories the operating quality of their instruments with regard to measuring level, repeatability and reproducibility.