An assessment of the use of fixed limits to characterize intralaboratory performance by proficiency testing.

Abstract

Currently, proficiency-testing (PT) programs are expanding the use of fixed limits to evaluate interlaboratory performance. These limits are an attempt to relate total allowable intralaboratory analytical error and performance in a PT program. Fixed limits are a means of counteracting the effects of overly stringent performance requirements derived from the interlaboratory group mean and standard deviations (SD) achieved by today's very precise analytical systems. Our previously described computer model of a PT program is used to delineate the quantitative relationship between the magnitude of intralaboratory coefficient of variation (CV) and bias that is compatible with fixed interlaboratory limits of 5, 10, 15, and 20% relative error. However, fixed limits alone do not fully characterize intralaboratory performance. The efficiency, or the percentage of times a laboratory's PT results are required to fall within the fixed limit, also must be considered when criteria for PT performance are being set. For example, a 10% fixed limit and a 80% efficiency will allow laboratories having CV-bias combinations from 8%-0%, 7%-4%, etc., to pass PT. The use of a 95% efficiency will allow CV-bias combinations from 4.8%-0%, 3.5%-4%, etc., to pass PT. We give a figure that depicts all possible combinations for four fixed limits and six efficiencies.