Abstract
Diagnosis and management of disease is increasingly driven by international efforts at consensus statements based on evidence of best practice. For endocrinology, diagnostic criteria based on the results of biochemical testing have for many years formed the mainstay to objective decision making, a recognition that biomarkers correlate better with morbidity and mortality risk than do signs, symptoms, and anthropometric parameters. Such testing becomes of increasing significance in an era of closer clinical scrutiny and accountability, and dictates the need for accurate, reliable results. But how strong is the evidence base for the diagnostic cutoff levels deployed in, for example, the investigation of GH secretion disorders? Our conclusion in reviewing GH assays’ performance is of a fragile state of play. Thus, Seth et al. (1) reported a 2-fold difference in values generated by the eight methods used by 85 participants in the United Kingdom National External Quality Assessment Scheme for GH. Where the most negatively biased method reports a value of 19.7 mU/liter (the preferred reporting unit in the United Kingdom), the most positively biased reports 38.1 mU/liter. From 1994–1998, variability across United Kingdom laboratories worsened from 18–30%, and although this has now stabilized the intermethod, bias persists. These observations are not new (2, 3), and are seen whenever a diversity of calibrants and/or antibodies is deployed. Other endocrine assays such as TSH (4), LH, and PTH with, respectively, interlaboratory variability of 12.5, 15.3, and 26.9% during 2006 also provide examples (United Kingdom National External Quality Assessment Scheme, personal communication). Typically, however, little has been achieved in reducing the variation in results due to these issues.

This publication has 9 references indexed in Scilit: