Reference values, why and how.

A review describing the fallacies of the concept of normal values and the reasons for its substitution with reference values. The "reference value philosophy", its terminology and recommended practical procedures are presented and it is warned not to use incorrect terms such as "normal reference range". The concepts of health and disease are discussed from the point of view of the laboratory. Reference values are not always derived from "healthy" persons. In order to be comparable, reference values and observed (i.e. patient) values should be produced in the same way, using the same analytical procedure, quality control, etc. The influence of preanalytical factors such as food intake, posture, use of tourniquet and freezing and storing samples is great and necessitates standardisation of specimen collection. Strategies for the selection of reference individuals are presented and the reasons and methods for subdivision of the data are described. Descriptions are also given of the individual reference values and the statistical treatment of collected data including multivariate analysis. Finally, the reporting of observed values in relation to reference values is discussed. There is a trend to avoid the use of the reference interval because of the temptation to regard the values falling outside it as pathological.