Comparison of 3 options for choosing control limits in biochemistry testing.

Abstract

BACKGROUND: The purpose of statistical quality control (QC) is to provide peace of mind with regard to the production of results that are suitable for analytically sound clinical interpretation and making reliable decisions about patient diagnosis, monitoring, and prognosis. OBJECTIVES: In this study, we compared 3 options for choosing control limits for biochemistry testing. They focus on the probability of error detection (Ped) and probability of false rejection (Pfr) achievable for a veterinary biochemical analyzer using the following 3 combinations: the quality control material (QCM) manufacturer's acceptable ranges; a standard 1rule customized for the instrument's observed performance; and candidate rules selected for the instrument's observed performance using a computerized program (EZrules). METHODS: For assessing customized QC, we used mean, SD, CV, bias, total error, and sigma metrics calculated from 3 months of control measurements on a laboratory biochemical analyzer, for 24 commonly used analytes, on 2 QCM levels. RESULTS: Given the desirable combination of high Ped (> 90%) and low Pfr (≤ 5%), the candidate rules selected by the computerized program-related EZrules provided the best performance combinations. CONCLUSIONS: The present work shows acceptable QC performance basing the QC on customization of the acceptable ranges of results from the achievable performance of an individual instrument. The QC performance is maximized by the application of candidate rules based on customized ranges obtained from a computerized QC tool, providing the ability to achieve the highest Ped and acceptably low Pfr values.