Validating common reference intervals in routine laboratories

Abstract

Common reference intervals for numerical pathology tests have been proposed for many years as an improvement over the common situation where individual laboratories establish or select and validate their own intervals. However it is important that any intervals that are developed for common use are themselves validated for use in individual laboratories. There are three main aspects to consider, the appropriateness of the interval, methodological factors and population factors. Techniques for assessing method biases are reasonably straightforward with the use of shared samples and appropriate external quality assurance schemes. Validating the local population, which also encompasses the laboratory's method, can be done using a number of healthy subjects, the more the better, or by various "data mining" techniques using the results of tests performed on routine patients. In any of these methods there is the need to consider the selection of subjects, the statistical approach and the acceptance criteria. Only if a proposed common reference interval can be shown to be appropriate in routine laboratories can it become widely adopted and become truly "common".

Keywords: Assay quality standards; Common reference intervals; Population result variation.