Oral Presentation Australian Society for Microbiology Annual Scientific Meeting 2019

The Mathematical and Scientific Basis for Developing and Setting Epidemiological Cut-offs Values (#183)

John Turnidge 1
  1. University of Adelaide, Adelaide, South Australia, Australia

Minimum inhibitory concentrations are an in vitro measure of antimicrobial activity. They are measured on an interval and not a continuous scale, and therefore have special mathematical and statistical properties. For instance, the MIC value is at the upper end on the interval, but because of standard plotting techniques this property is often incorrectly overlooked when analysing data. Conventionally, MICs are measured on a 2-fold dilution scale. This accident of history turns out to be correct because wild type MICs are distributed on a lognormal scale. Wild types, namely isolates without phenotypically expressed resistance mechanisms, are important to recognise and characterise. They serve as the foundation for establishing susceptibility testing breakpoints and the detection of emerging resistance. The epidemiological cutoff value (ECOFF) is the MIC that identifies the upper end of the wild type population and is the lowest possible breakpoint for a species. The bell-shaped curve of MICs in the wild-type population is a result on variation in two broad parameters: biological (strain-to-strain) variation and assay variation. Assay variation is in turn a composite of variation in reagent/material, intra-laboratory and inter-laboratory variation. Assay variation can be observed in repeated testing of the same strain, such as that which happens with quality control isolates. In setting ECOFFs, all these sources of variation ideally need to be included. The European Committee on Antimicrobial Susceptibility Testing has developed standard procedures ECOFF-setting that incorporate as many of these types of variation as possible. There are several methods for estimating ECOFFs. Visual interpretation has led the way, but more recently some of mathematical and statistical techniques have been developed. The most widely favoured at thpresent is the iterative statistical method which has been implemented in Excel as ‘ECOFFinder’, which is available on the EUCAST website as a free download.