TL47: Precision Study for a Qualitative Assay
*Paul B. Hshieh, CBER/FDA  *Tie-Hua Ng, CBER/FDA 


CLSI (Clinical Laboratory Standards Institute) guideline EP5-A2 defines precision as closeness of agreement between independent test/measurement results obtained under stipulated conditions. Although the statistical analysis of the precision study for a quantitative assay is well-developed, it is not clear how the precision study for a qualitative assay be analyzed. One approach commonly used for a qualitative assay is based on percent agreement of the test results as compared with the known result. A 100% agreement is the best possible outcome one could hope for and would imply 100% agreement between all the test results. However, the interpretation of the results becomes difficult when less than perfect agreement is observed because percent agreement with the known results does not provide individual precision components. In this discussion, a pair-wise agreement is proposed as a measure of precision. For simplicity, time-related components of precision (e.g., repeatability, precisions due to between-run, within-day, between-day, and within-laboratory) as discussed in EP5-A2 will be presented using hypothetical examples to illustrate the proposed measure of precision.