4. Secondary Reference Laboratories (SRLs)
SRLs work directly with manufacturers to assist them in standardizing their methods and in providing comparison data for certification of traceability as outlined in section C. Each SRL analyzes GHB by a precise method (documented total imprecision <3%) which can be calibrated to the Primary Reference Method. A proposed SRL must demonstrate traceability to the CPRL method.
a. Secondary Reference Method Calibration: Calibrator may be obtained from a manufacturer or prepared by the proposed SRL. Calibrator values for SRLs are either assigned by the CPRL (by direct analysis or based on fresh sample comparisons with the CPRL) or by the SRL method while calibrated with the initial or previous lot of calibrator.
b. Secondary Reference Method Quality Control: Each SRL prepares or purchases their own QC specimens (two levels; a non-diabetic level and a diabetic level of approximately 10% GHB) and sets their own limits and acceptance criteria using appropriate QC guidelines.
c. Certification of Traceability: A proposed SRL must perform precision testing following NCCLS EP5-A2 guidelines (1). The proposed SRL must also participate in a fresh blood sample comparison (n=100) for estimation of bias from the CPRL following modified NCCLS EP9-A2 guidelines (2). For bias estimation, fresh specimens will be collected by either the proposed SRL or the CPRL and shipped to the other party. All data for calculation of precision as well as estimation of bias is submitted to the NETCORE for analysis.
In order to be considered traceable to the CPRL: 1) total imprecision (CV) must not be statistically significantly >3%, 2) one within-method outlier is acceptable; one between-method outlier is acceptable, 3) the 95% confidence interval for the predicted bias should overlap the + 3% range of the CPRL at 2 GHB levels (6 & 9% HbA1c). All data analysis is performed by the NETCORE following NCCLS EP5-A2 and NCCLS EP9-A2. All outliers are investigated by the NETCORE to determine if the discrepancy could be due to characteristics of the specimen, e.g. presence of carbamylated adducts, hemoglobin variants, etc. If there is more than 1 between-method outlier, and the discrepancy is found to be due to the characteristics of the specimen, the discrepant samples can be replaced. In this case, the data is reanalyzed. If the criteria are still not met, the proposed SRL may make changes as necessary, re-analyze another set of samples and send the new data to the NETCORE for re-evaluation. Final certification of traceability is issued by the NETCORE after review of the data by the Steering Committee.
References: