Impact of the reference values on the clinically-relevant cut-offs. The example of cortisol testing in children

Abstract

AbstractAfter modification of our routine cortisol assay, we questioned the reference ranges for basal and stimulated cortisol plasma concentration in children.We retrospectively addressed the relevance of using the manufacturer’s normal reference range for basal cortisol and investigated its response to glucagon-betaxolol testing.Basal morning cortisol was 260 (98–604) nmol/L [manufacturer’s normal range (185–624) nmol/L: 26% subjects had “low” basal cortisol]. Upon testing cortisol increased to 502 (117–856) nmol/L. If a recently described 100% specificity threshold (403 nmol/L) is used it would amount to 31% adrenal insufficient children in apparently unaffected children. Basal and stimulated cortisol obtained with our prior radioimmunoassay (RIA) in a sub-group of subjects were lower: 411 (187–1061) and 770 (329–1542) nmol/L. Using the 403 nmol/L threshold with the radioimmunoassay would result in only 5% adrenal insufficient children.This shows again that laboratories have to advertise the need to establish reference values for given populations, both for basal or stimulated hormone levels. Failure to apply this rule will elicit false-positive and more critically, false-negative results.