Deep frozen raw milk standards. The way from reference methods to reference systems

摘要

Routine raw milk analysis plays a basic role in all milk payment and DHI schemes world wide. Precise and right data are the key factor for fair payment of raw milk and expressive and correct milk production data as well as breeding values and selection data. The routine methods used are powerful tools to gain mass data in short time and with reasonably low costs. But routine methods have to be calibrated carefully and steadily to ensure, that the obtained data are "true". The truth is defined with socalled reference methods, which give a reference value for a defined sample. For every routine lab it is crucial, that such reference material is available all the time. Because raw milk has a limited shelf life (even if it is chemically preserved) thistask happens to be time and money consuming. Thus precision of calibration is sub optimal, because of the limitations to characterize many little batches of milk extensively with costly and partly imprecise reference methods. A solution to this problemwas achieved by MILCHPRUEFRING BAYERN e.V. and MUVA (Bavaria, Germany) developing a deep frozen raw milk standard. A special protocol for homogenous packing and sparing deep freezing ensures, that every single vial of a batch represents the given targetvalue within a very narrow range. Each vial is accompanied by a quality certificate, which includes all statistical data of the QA process of this batch. Because batches can be large, it is feasible to do a very broad characterization, which enhances reliability ranges to a so far unknown level. Besides that, this material can be used to validate the performance of any lab carrying out the reference methods. This (together with further development of reference methods procedures) might be used to createa reference system rather than the present one-way-street from reference material to routine method. Additionally, if data from the routine procedures are implemented into such a calibration system, the systematically immanent error of prediction of "the true value" can be reduced efficiently.