Background: The 2009 Asian multicenter study for derivation of reference intervals (RIs) featured: 1) centralized measurements to exclude reagent-dependent variations; 2) inclusion of non-standardized analytes (hormones, tumor makers, etc.) in the target; and 3) cross-check of test results between the central and local laboratories. Transferability of centrally derived RIs for non-standardized analytes based on the cross-check was examined.
Methods: Forty non-standardized analytes were centrally measured in sera from 3541 reference individuals recruited by 63 laboratories. Forty-four laboratories collaborated in the cross-check study by locally measuring aliquots of sera from 9 to 73 volunteers (average 22.2). Linear relationships were obtained by the major-axis regression. Error in converting RIs using the regression line was expressed by the coefficient of variation of slope b [CV(b)]. CV(b) <10% was set as the cut-off value allowing the conversion. The significance of factors for partitioning RIs was determined similarly as in the first report.
Results: Significant sex-, age-, and region-related changes in test results were observed in 17, 15, and 11 of the 40 analytes, respectively. In the cross-comparison study, test results were not harmonized in the majority of immunologically measured analytes, but their average CV(b)s were <10% except for total protein, cystatin C, CA19-9, free thyroxine, and triiodothyronine. After conversion, 74% of centrally derived RIs were transferred to each local laboratory.
Conclusions: Our results point to the feasibility of: 1) harmonizing test results across different laboratories; and 2) sharing centrally derived RIs of non-standardized analytes by means of comparative measurement of a set of commutable specimens.