Lower confidence limits for critical systematic errors

Abstract

Objectives: The critical systematic error (ΔSE C) can be used to select a quality control scheme. It is just an estimate of the true ΔSE C, however, and its precision depends on the sample size. Thus, the lower confidence limit (LCL) of ΔSE C should be adopted instead to guard against false optimism. Design and methods: LCLs are calculated as a function of ΔSE C and sample size using a modified version of Bissell's formula. Results: The uncertainty in ΔSE C may be significant. The largest relative difference between ΔSE C and its LCL is 52.2%, for ΔSE C = 2 and n = 20. This difference decreases with increasing ΔSE C and sample size. Conclusions: It is shown that taking into account the uncertainty in ΔSE C can greatly influence quality control planning decisions. Caution should thus be exercised when interpreting control data and choosing an appropriate quality control strategy. The derived equation for LCL is simple, and can be immediately applied to practical solutions.