Abstract
The philosophy and achievements of the first two and a half years of a national quality control scheme for serum growth hormone assays are described. Three serum samples were distributed to participating laboratories every two weeks. A computer-produced summary of the quality control results, which contained scattergrams and a statistical analysis, was returned to participants four weeks after despatch. The performance of 11 experienced (SAS) laboratories was found to fulfil the necessary accuracy criteria, and their results provided the ‘reference group mean’, which served as the target value. Gross positive bias exhibited by several laboratories was due to the use of assays with inadequate sensitivity, and this practice was eliminated during the first six months. During the course of the Scheme the average bias fell from 30 to 14%. With assays of adequate sensitivity, bias was almost invariably due to the misuse of standards. Median within-laboratory, between-batch precision (CV) improved from 23 to 15%. The best performers achieved between-batch CV of 7% which was 1·5 × their within-batch CV: the worst had three- to four-fold differences between within- and between-batch CV. Rudimentary quality control of interpretation was found to result in improvements in interpretation per se and also served to reinforce the desire to improve numerical agreement.
A ‘recommended procedure’ based upon supplied first and second antibody, and a communal ‘reference range’ from 280 untreated acromegalics, both proved valuable.