Planning Statistical Quality Control to Minimize Patient Risk: It's About Time.

Abstract

The purpose of statistical quality control (SQC)2 in the clinical laboratory is to assure that reported patient results are fit for their intended use, not only when a measurement procedure is operating in its stable incontrol state, but also when out-of-control conditions occur. The value of quality-control principles and practices in the laboratory has been well recognized and appreciated for many decades. The Clinical and Laboratory Standards Institute (CLSI; then known as the NCCLS) published its first approved guideline on statistical quality-control principles and definitions for quantitative measurement procedures in 1991 (1). The fourth edition of the guideline appeared last year (2). For many years SQC design primarily involved choosing how many quality control (QC) samples to measure and what QC rules to apply to the QC results. This approach originated in an era when batch testing was common. QC samples were placed in the batch along with patient specimens. The QC sample results were used to decide if the patient results in the batch were acceptable. The goal was for the QC rule to have a low probability of rejection when the batch was in control (probability of false rejection, Pfr) and a high probability of rejection when the batch was out-of-control (probability of error detection, Ped) (3). When continuous-production analyzers became prevalent in the laboratory, a new QC planning question arose: When should QC samples be measured? In batch testing, the answer was to measure QC samples with each batch. However, with continuous-production analyzers a link between QC results and patient results within a batch no longer exists. Instead, QC results simply reflect the current state of the measurement procedure at the time they are measured. Unfortunately, the traditional QC performance measures, …