How difficult is it to achieve compliance with the quality control requirements of the Clinical and Laboratory Standards Institute's guideline M42-A?

Data accumulated in a multi-laboratory study of disc diffusion susceptibility data for Escherichia coli ATCC 25922 was analysed with respect to the acceptable ranges for this strain specified in the Clinical and Laboratory Standards Institute's M42-A guidelines. Seven laboratories each generated 60 zone size measurements for discs containing five agents for which acceptable ranges had been published. Of the 35 data sets analysed 37% contained > 10% of their measurements outside the acceptable ranges. For each agent, between one and four of the seven data sets obtained contained > 10% of the zone sizes outside the acceptable ranges. Only one laboratory obtained data sets that for all agents, contained < 10% of the zone sizes outside the acceptable. For the other six, non-compliant data sets were obtained for between one and four of the agents. These data suggest that the adherence to a specified test protocol cannot be assumed to result in the achievement of compliance with the quality control requirements of that protocol. They would further suggest that the need to undertake corrective actions, as specified in M42-A, would be frequently required before a laboratory could claim to be using this protocol and could legitimately apply any cut-off values or breakpoints that are or that might be, associated with it use. As laboratories involved in susceptibility testing of bacteria associated with aquaculture frequently handle only a small number of strains per year, the achievement of compliance with the quality control requirements of a standard protocol might require a disproportionate and unacceptable increase in their work load. The data obtained in this retrospective study also demonstrated that, even if all data sets containing > 10% non-compliant measurements were eliminated; significant inter-laboratory variation in mean zone sizes would remain.Laboratory-independent cut-off values will be of value only if inter-laboratory variation in disc diffusion data can be reduced to a minimum. It is argued that the data presented here provides strong grounds for questioning whether, particularly in the case of laboratories with small strain throughputs, inter-laboratory variation can be appropriately resolved by rigorous specifications of test protocols and quality control criteria. Thus, they would rather support the development of cut-off values by the application of normalised resistance interpretation. As the cut-off values generated by this approach are laboratory-specific, they are not influenced by inter-laboratory variation.