A plea for intra-laboratory reference limits. Part 1. General considerations and concepts for determination


    loading  Checking for direct PDF access through Ovid

Abstract

Accurate results for quantitative procedures can be useless if the reference limits for the interpretation of laboratory results are unreliable. Recent concepts for quality management systems require that laboratories pay more attention to identification and verification of reference limits. Scientific recommendations often claim that each laboratory should determine intra-laboratory reference limits, which should be reviewed periodically. This recommendation is currently neglected by most laboratories; instead they use reference limits from external sources, despite various problems of transference. Prospective and retrospective methods either using or neglecting disease prevalences (polymodal or unimodal concepts, respectively) and applying different statistical approaches for determining reference limits have been described. The various procedures are reviewed with regard to their diagnostic sensitivity, specificity and (non-)efficiency. The present gold standard is the reference limit concept according to IFCC recommendations (a unimodal prospective approach). This concept, together with trueness-based standardization, is the most useful basis for harmonization of the decision-making process with laboratory results, despite complex problems of traceability and transference. This harmonization is at present only achieved for a limited number of analytes for which SI units and traceability can be technically realized. For the majority of measurands in laboratory medicine, much research is still required and results cannot be expected in the near future. For these measurands, a need remains for internal, efficient and simple identification of population-based reference limits. Therefore, newer retrospective concepts were developed that use large data sets from laboratory information systems to derive intra-laboratory reference limits. These approaches appear promising and should be further developed.

    loading  Loading Related Articles