Statistical quality control: defining goals by measuring performance

We cannot detect if our process is failing if we do not know what it should look like when it is running well. Processes fail when predefined expectations of performance are not met. The expectations must be objectively defined, and clinically useful. Then, and only then, can we consider applying rules to monitor it. Those rules, when used, must relate to the inherent performance of the assay, otherwise we run the risk of allowing our assay to run out of control, and not be able to detect it. Although it may have been fine in the past, it is no longer acceptable to use a single approach, set of rules or to have the same expectation of performance for all assays. We must now define what standard we expect our assays to achieve.

Held in Milan in 2014, the 1st Strategic Conference of the European Federation of Clinical Chemistry and Laboratory medicine (EFLM) stated that impact on clinical outcomes was the gold standard for determining acceptable assay performance.¹ This meeting rationalised and developed the findings from the landmark Stockholm conference held 15 years previously,² and has shaped internal quality control (IQC) process design since. Our assays must be controlled to support favourable patient outcomes. To achieve this, a hierarchy of preferred performance criteria sources is followed (Table 1). This defines the assay total allowable error (TE_a). Use of the TE_a, in conjunction with our local performance, allows us to work out to what standard we are working.

Measures of assay performance