Error in laboratory reference limits as shown in a collaborative quality-assurance program.

Laboratories participate in collaborative quality-assurance programs to maintain and improve the quality of their diagnostic assays, but little attention has been paid to diagnostic quality in these programs. We used a national quality-assurance program to assess the quality of laboratory reference intervals as exemplified by triiodothyronine, thyroxin, and thyrotropin immunoassays. The limits of the reference intervals used by laboratories bear virtually no relationship to laboratory bias, i.e., whether assays read "high" or "low." Further, correcting assay results from different laboratories for the reference limits used increases rather than decreases interlaboratory scatter. We conclude that the analytical quality of immunoassays now exceeds the quality of the reference limits supplied to clinicians to assist diagnosis, and that nationally or internationally defined reference limits would therefore provide more information at less cost than do individual laboratory reference limits.