An evaluation of the ability of proficiency testing programs to determine intralaboratory performance. Peer group statistics vs clinical usefulness limits.

Proficiency testing programs use two basic approaches for evaluating participant performance: (1) state-of-the-art data based on the interlaboratory group mean and group SD, or the 1(2)s rule; and (2) fixed limits or assigned intervals around the true value. Our previously described computer model emulates the state-of-the-art approach by simulating the performance of 401 laboratories in interlaboratory proficiency testing programs using the 1(2)s rule. The use of fixed limits to evaluate laboratory performance can be simulated by simple probability calculations. We compare the 1(2)s rule and fixed limits on the basis of their ability to identify actual intralaboratory performance correctly using efficiency. In both cases, the maximum efficiency never exceeds 91% and depends on population SD or fixed limit, specific requirements for medical usefulness, and the prevalence (10%) of poorly performing laboratories in the proficiency testing population. This calculation gives some insight into why neither of these approaches adequately differentiates between good and bad intralaboratory performance.