Iterative Cochran's C test as a multivariate method to detect higher than expected variability: a microbiological inter-laboratory ring trial as a case-study

Abstract

IntroductionAn interlaboratory calibration analysis was carried out to validate a methodology for a European standard for domestic laundry disinfection, using different doses of disinfectant and microorganisms. ISO 5725-2 and ISO 13528 form the basis of interlaboratory validations of quantitative methods, but there is a need for a simple graphical method to detect differences in laboratory behavior in terms of accuracy and variability. ObjectiveA novel multivariate method based on the classical Cochrans C test, as well as PCA and bootstrapping, which allows the inclusion of different correlated variables, was applied to identify higher variability than expected in factor (e.g. laboratories) levels, and the detection of multivariate outliers in a reduced space. MethodsThe proposed method is based on resampling, using the same sample many times but removing cases at random and performing Cochrans C test for all the variables together in a reduced space. ResultsThe method was tested by checking 7 laboratories for high variability in different parameters (logarithmic reduction (LR), cross contamination (RI), and wash water (WW)). After applying the proposed statistical analyses, no reasons were found to reject any of the participating laboratories. Multiple applications of the method are possible and we describe a case study in which the multivariant iterative Cochrans C test was used: variability detection with multiple microbiological parameters (with high variability) during an interlaboratory ring trial.