Use of statistical modeling to reassess the performance standard for the AOAC use-dilution methods (955.15 and 964.02).

The AOAC Use-dilution methods (UDM) 955.15 (Staphylococcus aureus) and 964.02 (Pseudomonas aeruginosa) are laboratory methods used to substantiate antimicrobial efficacy claims for liquid disinfectants on inanimate surfaces. The UDM is accepted by the U.S. Environmental Protection Agency for the purpose of product registration. To attain a hospital-level claim, testing against S. aureus and P. aeruginosa is required, and the product must pass against both microbes. Currently, the UDM's performance standard for a single 60-carrier test is the same for both microbes, and allows up to one positive carrier for the product to be considered as a pass. In this paper, the performance standards for these methods are reassessed using data from a 2009 five-laboratory collaborative study and a recently published statistical model. The reassessment focuses on the pass-error rate for ineffective products and the fail-error rate for highly effective products. The calculations indicate that the pass-error rate is between 9 and 24% and the fail-error rate between 18 and 23% when the current performance standard is used for a single test. For product registration, a smaller pass-error rate (1%) historically has been maintained by requiring that a disinfectant pass three UDM tests for each of the two microbes; however, the calculations also indicate that the fail-error rate is between 42 and 45%. This large fail-error rate is a compelling reason to consider a new performance standard for the two UDM methods, 955.15 (S. aureus) and 964.02 (P. aeruginosa). One alternative performance standard allows no more than six positive carriers out of 60 as a pass when using P. aeruginosa and no more than three positive carriers out of 60 when using S. aureus. In addition, the new performance standard requires that three UDM tests be performed with each of the two microbes, and the disinfectant must pass all six tests to be considered efficacious. The statistical calculations for this alternative performance standard indicate that the pass-error rate is no more than 3%, while the fail-error rate is as small as 5%. Based on these error rate calculations, proposed revisions to the performance standards for AOAC Methods 955.15 and 964.02 are provided.